The Environment and Emerging Development Issues:

THE ENVIRONMENT AND EMERGING DEVELOPMENT ISSUES VOLUME 2

UNU WORLD INSTITUTE FOR DEVELOPMENT ECONOMICS RESEARCH (UNU/WIDER) was established by the United Nations University as its first re search and training centre and started work in Helsinki, Finland in 1985. The purpose of the Institute is to undertake applied research and policy analysis on structural changes affecting the developing and transitional economies, to provide a forum for the advocacy of policies leading to robust, equitable, and environmentally sustainable growth, and to promote capacity strengthening and training in the field of economic and social policy-making. Its work is carried out by staff researchers and visiting scholars in Helsinki and through networks of collaborating scholars and institutions around the world.

THE ENVIRONMENT AND EMERGING DEVELOPMENT ISSUES VOLUME 2

Edited by PARTHA DASGUPTA and KARL-GÖRAN MÄLER

A Study Prepared for the World Institute for Development Economics Research of the United Nations University(UNU/WIDER)

This book has been printed digitally and produced in a standard specification in order to ensure its continuing availability

Great Clarendon Street, Oxford OX2 6DP Oxford University Press is a department of the University of Oxford. It furthers the University’s objective of excellence in research, scholarship, and education by publishing worldwide in Oxford New York Auckland Cape Town Dar es Salaam Hong Kong Karachi Kuala Lumpur Madrid Melbourne Mexico City Nairobi New Delhi Shanghai Taipei Toronto With offices in Argentina Austria Brazil Chile Czech Republic France Greece Guatemala Hungary Italy Japan South Korea Poland Portugal Singapore Switzerland Thailand Turkey Ukraine Vietnam Oxford is a registered trade mark of Oxford University Press in the UK and in certain other countries Published in the United States by Oxford University Press Inc., New York Chapters 1-3, 5-22 © The United Nations University, 1997 Chapter 4 © Swets & Zeitlinger, 1992 Not to be reprinted without permission The moral rights of the author have been asserted Database right Oxford University Press (maker) Reprinted 2006 All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, without the prior permission in writing of Oxford University Press, or as expressly permitted by law, or under terms agreed with the appropriate reprographics rights organization. Enquiries concerning reproduction outside the scope of the above should be sent to the Rights Department, Oxford University Press, at the address above You must not circulate this book in any other binding or cover And you must impose this same condition on any acquirer ISBN 978-0-19-924070-8 Printed and bound by CPI Antony Rowe, Eastbourne

CONTENTS Volume 2 Part IV Reciprocal Externalities: Local and Global 10 Endogenous Fertility and the Environment: A Parable of Firewood Marc Nerlove and Anke Meyer (with Viktoria Dalko ) 11 Is Co-operation Habit-Forming? Paul Seabright 12 Efficiency Issues and the Montreal Protocol on CFCs Peter Bohm 13 CO2 and the Greenhouse Effect: A Game-Theoretic Exploration Michael Hoel Part V Unidirectional Externalities 14 Analysis and Management of Watersheds John Dixon 15 The Management of Coastal Wetlands: Economic Analysis of Combined Ecologic-Economic Systems John Dixon and Padma N. Lal 16 Urban Air Pollution in Developing Countries: Problems and Policies Alan J. Krupnick

259 283 308 339

371 399 425

vi Part VI Macroeconomic Policies and Environmental Resource-Use 17 Macroeconomic Policies and Deforestation Robert Repetto 18 Microeconomic Responses to Macroeconomic Reforms: The Optimal Control of Soil Erosion scott Barrett Part VII Valuation and Management 19 Valuation of Tropical Forests Anthony C. Fisher and W Michael Hanemann 20 The Management of Drylands Ridley Nelson 21 Management of Wildlife and Habitat in Developing Countries Gardner Brown 22 Public Policy toward Social Overhead Capital: The Capitalization Externality David Starrett Index

463 482

505 529 555 574

CONTENTS Volume 1 List of Contributorsxvii 1 The Resource-Basis of Production and Consumption: An Economic Analysis Partha Dasgupta and Karl-Göran Mäler Part I Rights and the Legal Framework 2 On a Clear Day, You Can See the Coase Theorem Barry Nalebuff 3 Common-Property Resource-Management in Traditional Societies Raymond Noronha Part II Accounting for Environmental Degradation 4 A Water Perspective on Population, Environment, and Development Malin Falkenmark 5 Environmental Statistics and the National Accounts Martin Weale 6 The Environment and Net National Product Partha Dasgupta, Bengt Kriström and Karl-Göran Mäler

1

35 48

73 96 129

viii 7 Can Computable General-Equilibrium Models Shed Light on the Environmental Problems of Developing Countries? Shanta Devarajan 8 Development Strategies and the Environment Irma Adelman, Habib Fetini, and Elise Hardy Golan Part III Decision Under Uncertainty 9 Choice under Uncertainty: Problems Solved and Unsolved Mark Machina Index

140 161

201

PART IV Reciprocal Externalities: Local and Global

This page intentionally left blank

10 Endogenous Fertility and the Environment: A Parable of Firewood 10.1 INTRODUCTION The purpose of this chapter is to explore the question of how environment and population interact when fertility is endogenous in the sense that parents choose the number of children they have in order to maximize their own welfare, that is, in a calculating way. In this chapter we assume parents have children, not because they derive direct pleasure from them, but only because children are useful in the production of goods the parents desire. In fact, our analysis is an extension of previous work on the so-called ‘old-age security hypothesis’, in which children are essentially viewed as a capital good. As Schultz (1974) has suggested, ‘children are the “poor man's capital”‘. Neher (1971) and Willis (1980) develop the idea that parents in less developed countries are motivated, in part, to bear and rear children because they expect children to care for them in old age. In order to focus attention on the implications of endogenous fertility we introduce a series of greatly restrictive simplifying assumptions. First, parents are assumed not to be altruistic in the sense of caring about their children's future welfare. That such an assumption is patently false is evidenced by the fact that we, as members of the present generation, are here involved in a discussion of the kind of environment which future generations (our collective children) will face. One result of this simplification will be effectively to disconnect adults in consecutive generations.1 Second, we assume that there are no environmental (or consumption) effects on the population's, or on each individual's, ability to reproduce, that is, for example, no effects of environmental deterioration on adult death-rates

1

In previous work, Nerlove, Razin, and Sadka (1987a, b) show that, when children are valued in part for their own sake, better access to capital markets, which permits more efficient transfer of resources from present to future consumption, may well increase rather than decrease the number of children.

260 prior to reproductive ages (‘Malthusian’ mortality). In non-human populations, and perhaps historically in human populations, death-rates or out-migration have been important sources of homoeostasis in the dynamic relation between population and the environment. Similarly there are no biological limits to the ability of each individual to reproduce. The effect of these very strong assumptions is that our model cannot be used to explain historical experience, although it may be, we hope, quite useful in understanding certain key aspects of the population–environment nexus in developing economies. Third, we measure time in generations. Each life is divided into two periods. In the first the individual is a child or young adult who supports his or her parents. All costs of child-rearing are subsumed in the net support children provide their parents. At the end of this period the aged parents die, the children become adults, have children, and retire. In order to rule out complications of marriage, two-sex reproduction, etc., we assume that all reproduction is by parthenogenesis and that all births survive. Thus, we will analyse a two-period overlapping-generations model. Fourth, as generally implicit in current discussions of economic development, we assume that population pressure adversely affects the state of the environment, in particular, that larger population sizes, above some level, are associated with increasing rates of environmental deterioration, whereas, population sizes smaller than that level are associated with environmental improvements. For a contrary view, see McNicholl (1990) who suggests that in certain types of environments, given proper institutional and social organization, large population sizes can be associated with environmental improvement. Thus we assume there exists some population size consistent with unchanging environmental quality. With respect to timing, we assume environmental deterioration to be related to the size of the parents’ generation, i.e. to the number of children working in the previous period. Parents’ expectations of the marginal productivity of an additional family member in gathering firewood, and their perceptions of the level of the family production function are both influenced by the existing state of the environment. These expectations are formed at the time each parent decides how many children to have. We assume such decisions reflect the state of the environment observed by the parent in her productive period, and that they do not take into account the possible behaviour of other parents, or the effects of the number of children born on future states of the environment. Indeed, since parents are assumed not to care about their children's welfare, this behaviour is consistent, because the effects of the size of the children's cohort will only be realized after the parents are dead. The model results in a system of two non-linear difference equations, the solution to which exhibits a great variety of dynamic behaviour. The dynamics of the solution depend upon the function relating environmental deterioration to population size, but also, more crucially, on the parameters of the family decision function. What really matters in this problem is the way in which

261 the environment affects family output directly, on the one hand, and indirectly through its effect on the perceived marginal product of an additional child, on the other. The principal conclusion of our analysis is that in the absence of parental altruism and/or ‘Malthusian’ mortality, and in a closed world, the relation between population and environment is inherently unstable when fertility is endogenous. That is, when the interest of parents in having children lies solely in what those children can contribute to the parents’ consumption, and when parents are otherwise totally unconcerned about the lives their children will lead, the unpriced nature of environmental resources leads parents to fertility decisions which, while optimal from their own selfish point of view, ultimately lead to environmental disaster. Of course, in the real world nothing like this will actually happen if only because populations can migrate to more favourable environments, or because repugnant ‘Malthusian’ mechanisms may come into play long before the environment deteriorates to science-fictional levels. As members of the present generation, we are concerned about the welfare of future generations. In the absence of social intervention, parental altruism is not sufficient of itself to stabilize the relation between population and the environment. This is because, being unpriced, the true social cost of having an additional child is not borne by individual parents. In the absence of a system of Pigouvian taxes and subsidies designed to equate the private and social marginal costs and benefits of having children, a laissez-faire population policy does not lead to a Pareto optimum from the standpoint of the present generation, in which no parent can be made better off without making some other parent worse off. On the other hand, there is no guarantee that social policies which force parents to bear the true social costs of having children will stabilize the relation between population and environment. Love is not necessarily enough. The remainder of the chapter will follow this plan. In Section 10.2, we develop the model and show how the relation between the state of the environment and the birth-rate depends on parents’ perceptions of the benefit of having an additional child. In general, the family decision process leads to a reduced-form equation which relates the forecast state of the environment during the period in which the children will be economically active to the birth-rate. The sign of the derivative of this function is shown to depend on the ratio of two quantities: (i) the elasticity of the marginal product of a child with respect to environmental deterioration, and (ii) the elasticity of family output with respect to the state of the environment. Thus, with respect to the dynamics of the model, what is crucial is not the perceived marginal benefit of an additional child but how the environment affects that marginal benefit in relation to how it affects the average product per child. In Section 10.3, we attempt to characterize the qualitative dynamics of the model and to demonstrate the existence of one or more steady states, which,

262 however, we show are always unstable. A graphical method for viewing the solution to the model is given in Appendix 2. Section 10.4 concludes with some observations concerning the implications of the model for some central empirical questions in the study of environmental–demographic interactions, and asks whether parental altruism towards their children can alter the dynamic properties of the model.

10.2 A MODEL OF FAMILY DECISION-MAKING UNDER CONDITIONS OF ENVIRONMENTAL CHANGE Let Nt−1 be the number of parents at the end of period t − 1 who will decide on the number of children each will have at that time, bt−1. The total population alive in t is thus Nt−1 + bt−1Nt−1. Parents are assumed to die at the end of period t, at which time their children, now adults, themselves become parents. Thus the rate of population growth is equal to the per-family birth-rate:(10.1)

Population is increasing, stable, or decreasing according to whether(10.2)

Each child is assumed to live for two periods. In the first period, children support their non-working parents by ‘gathering firewood’. At the end of this period, each child becomes herself a parent and lives off the fruit of her children's labours. Families are assumed to share equally the total harvest of firewood by the children of the family among family members. This introduces the only element of cost to having children which we consider. Each mother decides how many children to have on the basis of her perception of the relationship between family income, the number of children she has, and the state of the environment. This relationship embodies her beliefs about the productivity of her children, during the period in which she will be supported by them, on the basis of her observation of the state of the environment during the period t − 1. We assume that the marginal product of an additional child is positive but diminishing. Increased environmental deterioration, the level of which we denote by Z, reduces the family's expected harvest, ceteris paribus, but its effect on the perceived marginal product of an additional child may be either positive or negative. Let xt be the family's expected harvest. Then(10.3)

Where f ≥ 0, f1 > 0, f11 < 0, and f2 < 0, for all b ≥ 0 and Z > 0.2

2

A logarithmically linear (Cobb–Douglas) relation between family income, the number of children, and the state of the environment, may be ruled out in this model because such an assumption produces a constant per family birth-rate independent of environmental quality. There is only one parameter value for which a steady state exists.

263 If the state of environmental degradation, Z, could be assumed to behave like the inverse of an ordinary factor of production, it would be plausible to argue that:

so that improvements in environmental quality would be expected to increase the marginal productivity of the other factor, namely children. Similarly, the marginal effect of further environmental deterioration might be expected to diminish at greater levels of poorer-quality environment, i.e. the better the environmental quality, the greater the marginal effect of deterioration. In short,

On the other hand, arguments can go the other way. For example, as forests recede up the mountain sides, parents may perceive a greater benefit of having an additional child to gather firewood. More realistically, perhaps, as the environment deteriorates there may be a shift from crop production to livestock, in which activity children have, arguably, a comparative advantage. Similarly, environmental deterioration may accelerate the perceived adverse effects on family income. In very poor-quality environments the effects of a given change in the quality of the environment may be larger than when the environment is in good shape. Parents are assumed to choose the number of their offspring so as to maximize their own selfishly determined utility in the retirement period. Utility is assumed to be a monotone-increasing function of the parent's own expected consumption of firewood, and nothing else. Thus, the parent maximizes

with respect to bt−1 taking Zt−1 as given. For simplicity, we drop temporal subscripts in the following derivation. The first-order condition for a maximum of family per capita income is(10.4)

or

or

Since b = 0 implies f(0, Z) = 0, b = 0 cannot be a solution to (10.4) as long as f1 is strictly positive, even at b = 0. For b > 0,

264 as a consequence of the diminishing marginal product of children and f(b, Z) > 0 for b > 0. Thus (10.4) determines a unique value of the birth-rate which is always positive, but which may be greater than, equal to, or less than 1 according to whether , at b = 1 (see Figure 10.A2). And thus (10.4) defines a function h which relates each state of the environment expected, Zt−1, to a unique positive value of the birth-rate at the end of period t − 1:(10.5)

Before we examine the further properties of h, note that the second-order condition for a maximum is always satisfied because f11 < 0. Now,

making use of the first-order condition.3 How are the properties of the function h in (10.5) related to those of f, which is the basis for parents’ decision-making with respect to fertility? First, let us determine h′. From (10.4)

so that(10.6)

Now 3

We also note that a Cobb–Douglas function is unacceptably restrictive for this problem. In this case, we have

so that

or

independently of the state of the environment. Thus, population increases, is constant, or decreases over time, by (10.1) at a constant rate according as order condition

for 0 < α < 1 and b > 0, is also satisfied. This example shows why Cobb–Douglas is not an interesting functional form in this context.

. The second-

265

and

Thus,(10.7) Since f11 < 0 and f > 0, b > 0, the sign of h′ is determined by(10.8)

and since f1 > 0, and f2 < 0, and f > 0, for b > 0, we can say unambiguously(10.9) that is, if the perceived marginal product of an additional child increases as the environment deteriorates, birth-rates increase as the environment worsens. Good environmental quality is not like an ordinary factor of production. However, a more general interaction of the environment and the marginal product of a child is possible since environmental quality may, at some levels, be like an ordinary factor of production. In any case, it is not under the control of the decision-making parent. From (10.7) we find after some manipulation, that(10.10)

where

is the elasticity of the expected marginal product of a child with respect to environmental deterioration, and

is the elasticity of family output with respect to environmental deterioration. We assume f1 > 0, f2 < 0, f > 0 and f11 < 0 for all b > 0 and, hence, for all Z. Thus(10.11)

Since f2 < 0, ξ < 0. As we saw, f12 > 0 implies h′ > 0 and also η > 0, but f12 < 0 implies η < 0. So only when f12 < 0, (that is, when environmental deterioration reduces the perceived marginal benefit of having an additional child) is the possibility open that such deterioration reduces the birth-rate. But, even in this case, the effect is not unambiguous since, by (10.11)(10.12)

266 We thus see that if the elasticity of family income is greater in absolute value than the elasticity of the marginal product of a child, then h′ > 0; whereas, if the opposite is true, then h′ < 0. Moreover, these elasticities may change as the environment deteriorates, which opens up the possibility that h′ changes sign. For example, suppose that f12 is positive at low levels of environmental deterioration, (that is, environmental deterioration enhances the marginal productivity of children, due to a shift, say, from crops to livestock), then h′ is positive at low values of Z. Thus η > 0 and since ξ < 0, (1 − η/ξ) is positive. But now as the environment deteriorates more and more one may argue that at some point f12 becomes negative, opening up the possibility that h′ falls and eventually becomes negative. Other factors may come into play as well. At very low levels of environmental quality, the costs of having a child which survives to the age at which she may be productive may rise steeply, thus also contributing to the decline and eventual negativity of h′. In the future, we propose to explore an alternative, survival model of family fertility decisions which also leads to a decline and eventual negativity of the function h′. Thus h′ may start out being positive but become negative as the environment deteriorates. Because the interaction between environmental quality and perceived family income and the marginal product of a child may be complex, the possibility of many changes in the sign of h′ exists. The key question is whether a solution to(10.13)

exists and if so, how many such solutions there are. If h is monotonic with either h′ > 0 or h′ < 0, then, provided the conditions on f stated in Appendix 1 are satisfied, equation (10.13) has exactly one solution at which b = 1. The value b = 1 implies population is stationary. Of course, the dynamics of the model still depend on the relationship between population and environmental deterioration, a relationship as yet to be specified. When h is non-monotonic with derivative which changes in sign with varying Z, (10.13) may have more than one solution, one solution, or no solutions at all. The possibilities are illustrated in Figure 10.1 for cases in which the sign of h′ changes exactly once in the range of 0 < Z < ∞. As we show in the next section, the dynamic properties of the model at the points , at which b = 1, depend on the sign of the derivative h′, or whether it is zero. Finally, to close the model we assume a relation between population and environmental quality. We suppose that the state of the environment in period t depends on the number of children during and at the end of t − 1, i.e. Nt−1, and in this way on the total population, and on its previous state. More specifically, let the rate of environmental deterioration depend on Nt−1:

267

Fig. 10.1

(10.14)

where g′ > 0. For g(N) > 1, the environment is deteriorating, whereas for g(N) < 1, the environment is improving, g″ may be either positive or negative. If g″ > 0, the rate at which the environment deteriorates accelerates in response to larger population; while, if g″ < 0, the rate of environmental deterioration decelerates for larger population sizes. We assume g is monotonic in population. As stated earlier, we assume that this relation is not taken into account by parents in their decisions.

268

10.3 THE DYNAMICS OF THE MODEL: ENVIRONMENTAL–DEMOGRAPHIC INTERACTIONS The dynamic system resulting from the model of family decision-making and demographic–environmental interactions derived in the preceding section consists of two simultaneous non-linear difference equations, (10.5) and (10.14):(10.15)

and

We assume that g is continuous, monotone, increasing, and that there is some value of N for which g(N) < 1. It follows that there exists a value such that

and g < 1 at and g > 1 at . The existence of a steady state in which neither the environment nor population are changing depends on whether there simultaneously exists a for which

As we showed above, the equation h(Z) = 1 may have many, one, or no solutions. Clearly, no steady state exists if there are no solutions to h(Z) = 1, but, if one or more solutions exists, any pair

for which is a steady-state solution to (10.15). For any such pair, the population will be unchanging and so will the state of the environment. It is not possible to say a priori whether this will be a state in which the population is small and environmental quality good, or whether it is one in which the environment is badly deteriorated and the population large. We can, however, say a good deal about whether such a steady state is attainable starting from a particular initial condition (N0, Z0) different from and whether, if once attained, such a state is stable or unstable, in the sense that small perturbations from it result in a return to it or a further departure from it. Systems of non-linear difference equations may be analysed qualitatively in the vicinity of a stationary point such as by standard methods. (On this point, see Iooss and Joseph (1990: ch. 4).) Essentially we proceed by linearizing the system in the neighbourhood of such a point. Such linearization works well except in the neighbourhood of a point at which h′ = 0. But this is exactly the case of one solution when h is non-monotonic (see Figure 10.1). This is clearly a most unlikely case; it is far more likely that there are no

269 solutions at all, i.e. h is not just tangent to the line Z = 1. If we rule out such cases, the following analysis applies to all equilibria which exist. But we will return to the case , at a value for which , in Appendix 1. For the linearized system corresponding to (10.15), the qualitative stability characteristics depend on the characteristic values of the matrix(10.16)

which are the two roots of(10.17)

or(10.18)

Note that when h′ = 0, both roots are one. The cases are as follows (see Iooss and Joseph, 1990:49): 1. The roots are real and both are less than 1 in absolute value. Then is a stable equilibrium. This case never occurs in our model. If h′ > 0, recall g′ > 0, then one root is greater than 1 and the other less; we are in case: 2. One root is greater than 1 in absolute value, the other less. Both are real. is a saddle-point. There exists a saddlepath in the (N, Z) plane, on which initial conditions lead to trajectories which go into . Any other initial conditions lead away from that point. Such a saddle is always unstable. This case occurs when h′ > 0. When h′ < 0, we are in case: 3. When the two roots are complex, the properties of the system depend on the modulus of these roots:

This case occurs only when h′ < 0. The modulus is always greater than 1 and the point All trajectories which begin at points near by are spirals away from .

is always an unstable focus.

The global characteristics of the system can be illustrated by a phase diagram in the (Z, N) plane. To draw the phase lines, first note that the phase line must satisfy(10.19)

and

From (10.15), the conditions of (10.19) are satisfied when

270 that is by

which are straight lines parallel to the axes in the (Z, N) plane. The intersection(s) are the stationary points of the system, which we label S1 and S2. As we have shown above, the dynamic behaviour of the system around points such as S1 and S2 depends crucially on the sign of h′ at those points. Moreover, as was shown in the previous section, if an equilibrium point exists for h′ ≠ 0 and if h′ changes sign just once, there will be two points S1 and S2, near each of which h′ will be of a different sign. To analyse the dynamic behaviour of (10.15), therefore, we begin by looking at just one curve line and one equilibrium point Si. We now proceed to study cases 2 and 3. Case (2): h′ > 0. In this case, everywhere to the right of the curve , population is increasing, whereas to the left, population is decreasing. Moreover, for , since g′ > 0, environment is deteriorating, i.e. ΔZ > 0, whereas for it is improving. Case (3): h′ < 0. In this case, everywhere to the right of the curve , population is decreasing, whereas to the left it is increasing. The behaviour of the environment over time is as in Case (2). These two cases are illustrated in the trajectories described in Figure 10.2. In Case (2), Si is a saddle-point: except along the saddle-path, perturbations off the equilibrium in any direction result in further departures from it; only on the saddle-path do perturbations result in a return to equilibrium. In Case (3), Si is an unstable focus, where perturbations generally result in an explosive spiral away from the point Si. always, or when f12 > 0, the case of abnormal behaviour Thus, in the case in which h′ is always positive, i.e. when of the perceived family production function for firewood, there exists a stationary equilibrium in which population and environment are in balance. But this equilibrium is an unstable saddle-point. Small perturbations from equilibrium in almost every direction result in ever greater departures from it leading either to explosive population growth and environmental deterioration, or to population extinction with, of course, environmental improvement which is enjoyed by nobody. When h′ is always negative, on the other hand, i.e. when always, then there also always exists a stationary equilibrium of population and environment, but such an equilibrium is a highly unstable focus. Small departures from equilibrium generally lead to ever wilder fluctuations in the environment and population until such point as the population becomes extinct. The whole process then stops. When h′ changes sign there may be more than one stationary equilibrium or none. (The case h′ = 0 at h = 1 is discussed in Appendix 1.) If there are

271

(a) Case 2: h′ > 0

(b) Case 3: h′ < 0

Fig. 10.2 equilibria and we start out at one, a small perturbation may push us into the vicinity of the other and a different form of dynamic behaviour may develop. When there is more than one equilibrium point, the global behaviour of the system can be extremely complex. If h, for example, is convex, h(Z) = 1 may have two distinct solutions, one, or none. In the case of two solutions, three types of global behaviour are possible (see Figure 10.3): (a) A homoclinic loop (Guckenheimer and Holmes, hereafter GH, 1986: 46), panel (a). Both population and environment remain within a limited region, enclosed by the loop, if the initial value was in that region, which is not a set of measure zero. If the initial value is outside of the region enclosed by the loop and not on the saddle-path, then population increases without bound and there is ecological disaster. (b) A saddle-source connection (GH: 367), panel (b). There is only one, unique path which results in a bounded solution. Population and the environment

272

(a) Homoclinic loop

(b) Saddle-source connection

(c) Limit cycle

Fig. 10.3 remain on this path only if the initial condition happens to be on it. The path is a set of measure zero. (c) A limit cycle (GH: 72), panel (c). When the initial point lies within the region enclosed by a loop, their values remain bounded, but may cycle. Thus, the cycles become ever more uniform with the passage of time as they approach the limit cycle inside the boundary of the region. If the initial value lies outside the region the situation is as described in (a): ecological disaster. When h is concave rather than convex the same qualitative behaviours are possible except, rather than an exploding population, mankind becomes extinct and the system comes to a halt.

273 The issues of bifurcation, hysteresis, and whether a stable limit cycle around a node such as S2 exists are discussed in Appendix 1. Appendix 2 presents a graphical method for determining the time-paths of environmental quality and population themselves, not just the qualitative properties of these paths, given only the general shapes of the functions h and g.

10.4 CONCLUSIONS: LOVE IS NOT ENOUGH We have shown that the dynamic interaction of environment and population when fertility is endogenous depends crucially on the way in which parents view the effects of environment on family income as against their view of its effects on the perceived benefit of an additional child. Depending on the expected relation of the elasticity of total family income to the elasticity of the marginal product of a child, the relation between the rate of population growth or decline, and the state of environmental degradation may be either positive or negative. Moreover, the derivative of the function relating the two may change sign. When the relationship between the birth-rate and environmental quality is monotonic, and the family production function satisfies certain restrictions, then a steady-state equilibrium exists, but is not stable. When a small perturbation is introduced at an equilibrium point, population either explodes, possibly cycling wildly, or mankind becomes extinct. When the function is not monotonic a steady-state equilibrium may fail to exist or multiple equilibria may exist. When such equilibria exist, however, they are not stable either. The analysis to this point suggests a focus for empirical research on this topic. The dynamics of environmental–demographic interaction when fertility is endogenous depend primarily on the way parents perceive the benefits of having children and not principally on the effect of population size on the environment, as long as the environment is adversely affected by larger population. In earlier work on endogenous fertility, Nerlove, Razin, and Sadka (1987a, b) show that parental altruism toward their children makes a big difference in a number of economic problems. They show, for example, that the ‘old-age security hypothesis’ need no longer imply that introduction of a more efficient method of transferring resources from present to future consumption will unambiguously reduce the birth-rate. What would be the implications of parental love for their children in this model? Love, in this sense, means that parents are concerned about the future welfare of their children beyond their ability to support their parents. This question is left for future consideration, but it seems clear that love is not enough to alter our main conclusion that the equilibrium between population and the environment is inherently unstable. Love has two effects. First, it provides another motive for having children and, as such, will only increase the perceived benefits of

274 having an additional child ceteris paribus, albeit presumably at a diminishing rate, from the standpoint of the individual parent. Second, however, concern for a child's future welfare by the parent introduces a link between generations which is not present in our model, in which parents are assumed not to care about what happens after their own deaths. When parents care about the future welfare of their children, however, it seems apparent that laissez-faire produces a Pareto-suboptimal solution from the standpoint of the present generation; when environmental resources go unpriced, a wedge is driven between the parents’ perceived net benefit of having an additional child and the true social benefit. Thus, while love is unlikely to be enough to lead to stability, it may provide a rationale for social intervention. Such intervention would take the form of a per capita tax on children, returned to parents in the form of a subsidy which would force parents to take account of the costs of environmental deterioration in terms of the welfare of future generations caused by having an additional child. Thus, the optimal system of taxes and subsidies would, in principle, depend on both the weight which parents place on the future welfare of their children (and children's children, and so on, in infinite regress), and on the way in which aggregate population affects environmental deterioration. Even with such a corrective Pigouvian system of taxes and subsidies, it is not clear that the dynamic relation between population and environment would be stabilized. That would depend on both the extent of parental altruism and the environmental effects of population growth.

Appendix 1: Existence of Steady-State and Periodic Solutions, Bifurcating Solutions, and the Effects of Hysteresis 1. EXISTENCE OF A STEADY-STATE SOLUTION The existence of a steady-state solution to (10.15) is shown in Section 10.3 to depend on the existence of solutions to the equation(A1.1) Clearly if(A1.2) or

for all Z no solution exists. h(Z) > 1 all Z can occur if(A1.3) for all Z. Similarly h(Z) < 1, for all Z, if for all Z(A1.4) To guarantee a solution to (A1.4) if h(Z) is continuous, it is sufficient that (A1.3) hold for some values of Z and (A1.4) for other values. However, this condition does not say anything about the derivative of h at a solution to (A1.4). In the text, we discussed only those cases for which , so that, given also that , where , the system could be linearized in the neighbourhood of . Because it is possible for h to be non-monotonic, it can happen that . At this point(A1.5)

and, either

or

for all . In this case, the equilibrium is called non-hyperbolic and both of the characteristic roots of J in (10.16) are 1. The qualitative properties of (10.15) cannot be studied locally by linearizing the system. As suggested in the text, we do not regard a non-hyperbolic equilibrium as a likely possibility in the absence of hysteresis or of policies designed to shift the function f and the marginal benefits associated with an additional child, f1. However, if such shifts are introduced, the equilibrium may also shift,

276 passing through such a solution at which h′ = 0, thus giving rise to bifurcating solutions. We take this possibility up below. Even in the case of a non-hyperbolic equilibrium, however, the phase diagram can still be constructed as in Figure 10.2. Let(A1.6) be a neighbourhood of the point at which

and

The phase diagram for h(Z) > 1, all , Z in , is given in panel (a) of Figure 10.A1 and the diagram for h(Z) < 1, all , Z in , is given in panel (b). Both kinds of non-hyperbolic equilibria are unstable. (a) h(Z) > 1

(b) h(Z) < 1

Fig. 10.A1

2. PERIODIC SOLUTIONS When for , and for time-path of points (Z, N) spirals around

, we show in the text that the stationary point is an unstable focus. The . In this case, there may be a cycle of, say, period k > 0. Then(A1.7)

277 For example, if k = 2, the conditions imposed on g and h are(A1.8)

where(A1.9) and(A1.10)

If there exist and , which satisfy (A1.8–10), they form a 2-cycle. Similarly, for general k, we have , which satisfy the 2(k + 1) equations(A1.11)

and

Where, if there exist to which satisfy (A1.11), they form a k-cycle. At the time of writing, it seems extraordinarily unlikely to us that such a complex set of conditions would ever be satisfied by the functions h and g likely to be encountered in practice. In any case, such a periodic solution, if it exists, may or may not be locally stable. The determination of the local stability properties of such a periodic solution are discussed in Grandmont (1989).

278

3. BIFURCATING SOLUTIONS Suppose h(Z) is a convex function of Z. In this case (A1.1) has either Two solutions: , and such that

and

One solution: such that

or No solutions.

Write(A1.12) There exists a unique(A1.13) for satisfying (A1.1), such that(A1.14) The reparametrization of h in (A1.12) permits us to discuss the nature of the solution to (A1.1) for a given g function in terms of the parameter c. For all , there are two steady states: (1) a saddle and (2) an unstable focus. For , there is one steady state which is neither a saddle nor a focus, but which is unstable in any case. If , there is no steady-state solution. If c changes for some reason, there is the possibility of a rather abrupt change in the nature of the steady states which exist and, in any case, in the time-path of (Z, N) generated by (A1.1). Changes in c can result from social policies or other factors (e.g. the development of parental altruism), which shift the functions f and f1. Figure 10.A2 shows a shift in the ratio f/f1 to for a given value of Z. The solution for the birthrate so determined changes from to (A1.15)

As drawn, c2 < c1 since

, or(A1.16)

A tax on children for example, returned to parents in the form of a lump-sum subsidy would lower f1 relative to f, reducing the birth-rate for any state of the environment. For different levels of the tax subsidy the f/f1 function is shifted with corresponding changes in h; thus, changes in c, as suggested, can result in rather peculiar changes in the dynamic behaviour of the system.

4. HYSTERESIS Variations in the function h may result from changes in social policy or in parents’ perceptions of the value of children. Such changes are likely to be reversible. Shifts in the function g, however, which relate the rate of environmental derioration to population

279

Fig. 10.A2

size may not, however, be so benign. Let the function h in (A1.1) be fixed, but consider a change of g to g*, at say some critical population level N* such that now(A1.17)

In the phase diagram drawn in Figure 10.2, there is now no line in which ΔZ = 0. In fact ΔZ > 0 everywhere. Even if the population now falls below N*, the environment will continue to deteriorate forever. More complicated forms of hysteresis may arise. All those connected with upward shifts in g result in steady states, which, if they exist, are characterized by lower levels of environmental quality.

Appendix 2: Graphical Determination of the Time-Paths of Population and Environmental Quality The purpose of this appendix is to demonstrate a graphical method for the determination of the time-paths of population and environmental quality and the stationary equilibria, if such exist. First, we divide the plane into four quadrants as follows: I II III IV

North-east: Z, N North-west: g(N), N South-west: Workspace South-east: h(Z), Z.

In quadrants II, III, and IV we construct 45°-lines. These lines will be used to project various quantities, represented by points, from one quadrant to another. In quadrants II and IV we also construct the unit lines parallel to the N and Z axes, respectively, along which g(Z) ≡ 1 and h(N) ≡ 1. Next, draw in the functions g(N) and h(Z) in quadrants II and IV, respectively. g(N) has been drawn monotonic increasing in Figure 10.A3. h(Z) has been drawn so that it is non-monotonic with first h′ > 0, then h′ < 0. (Remember the graph is ‘upside down’.) In this case there are two stationary equilibria S1 and S2, one a saddle-point, the other a spiral node, determined as follows. Find the intersections of h(Z) with the unit line to determine the points on the Z axis. Then find the intersection of g(N) with the unit line in quadrant II to determine the point . The two equilibria, plotted as S1 and S2 in quadrant I are the points . Choose a point (Z0, N0) as a starting point such that (Z0, N0) is not either S1 or S2. To find the next point in the sequence, proceed as follows: (1) Project N0 through the N axis to the curve g(N) in quadrant II to find g(N0) on the g axis. (2) In order to multiply Z0 by g(N0) to find

project Z0 around the 45°-line in II to the g axis. (3) To multiply two numbers x and y geometrically, both of which lie on the abscissa in the plane, construct a line with slope y as follows: Project y around the 45°-line to the ordinate then construct a line through y parallel to the abscissa cutting the unit line. This part determines a line through the origin with slope y. Draw a perpendicular through x to this line. The corresponding point on the ordinate is the product xy. (4) Construct a line with slope g(N0) with respect to the g axis in quadrant II as in (3). Using the procedure outlined there, find

which will lie on the h axis.

281

Fig. 10.A3

(5) Project this point around the 45°-lines in quadrants III and IV into quadrant I. (6) Now project Z0 through the Z axis to the curve h(Z) to find h(Z0) on the h axis. (7) In order to find

proceed as in steps (3), (4) and (5): Construct a line with slope h(Z0) with respect to the h axis in quadrant III; multiply by N0, as in step (3), to obtain N1 on the g axis; project around to quadrant IV and I. Thus, we have found the next point in the sequence, (Z1, N1), in quadrant I. Now you can repeat steps (1)–(7) to obtain (Z2, N2), and so on. But be careful! Since all the equilibria are unstable you will shortly project yourself right off the paper with probability 1.

282

REFERENCES GRANDMONT, J. M. (1985), ‘On Endogenous Competitive Business Cycles’, Econometrica, 53: 995–1045. GUCKENHEIMER, J., and P. HOLMES (1986), Nonlinear Oscillations, Dynamical Systems, and Bifurcations of Vector Fields (New York: Springer-Verlag). IOOSS, J., and D. D. JOSEPH (1990), Elementary Stability and Bifurcation Theory (New York: Springer-Verlag). MCNICHOLL, G. (1990), ‘Social Organization and Ecological Stability under Demographic Stress’, Working Paper, 11, (Washington, DC: Population Council) NEHER, P. A. (1971), ‘Peasants, Procreation, and Pensions’, American Economic Review, 61: 380–9. NERLOVE, M., A. RAZIN, and E. SADKA (1987a), Population Policy and Individual Choice: A Theoretical Investigation Research Report 60 (Washington, DC: International Food Policy Research Institute). NERLOVE, M., A. RAZIN, and E. SADKA (1987b), Household and Economy: Welfare Economics of Endogenous Fertility (NewYork: Academic Press). SCHULTZ, T.W (1974), Economics of the Family: Marriage, Children and Human Capital (Chicago: University of Chicago Press for the National Bureau of Economic Research). WILLIS, R. J. (1980), ‘The Old-Age Security Hypothesis and Population Growth’, in T. K. Burch (ed.), Demographic Behavior: Interdisciplinary Perspectives on Decision Making (Boulder, Colo.: Westview Press).

11 Is Co-operation Habit-Forming? 11.1 INTRODUCTION Do people who have co-operated in the past find it easier to co-operate in the future? Do people who have betrayed each other's trust in the past find it hard to establish the trust necessary for co-operation in the future? Common sense suggests the answer to both questions is ‘yes’. It suggests also that the answer may matter very much for the design of institutions that depend on co-operative behaviour. Environmental problems often raise precisely such questions of institutional design. And furthermore, the scope and range of environmental external effects sometimes means that their resolution depends on co-operation between people whose previous relations have been characterized by indifference or even hostility. In short, care of the environment requires not only co-operation, but often new forms of co-operation. Can new forms be easily found? Or is it easier to promote co-operation within the framework of old and familiar ways of life? It is a truism that action to alleviate the harmful consequences of environmental externalities requires collective action on the part of the affected parties. But sometimes all that is necessary is for them to negotiate their way to an agreed solution, which can then be enforced through systems of penalties and incentives that induce individual parties to act in a way consistent with an efficient outcome. More commonly, however, credibly enforceable agreements cannot be made. Efficiency then requires truly collective action—action in which individuals forgo immediate benefits for themselves, in the interest of the group of which they form a part. What might explain why individuals should behave in such a public-spirited manner? Could it also explain a tendency for past co-operation to make future co-operation easier? Anthropologists and sociologists have tended to focus on the way in which individual behaviour is governed by rules and codes of conduct, the genesis of which is often explained by their conduciveness to the interests of the group. There is little doubt that such rules play an important role in bringing about

284 co-operative behaviour; human beings respond to normative signals from each other, even when these override what they perceive to be their own interests. Likewise, there can be little doubt that in close-knit societies, individuals often care enough about the interests of others for them to seek co-operation even when their own interests are not thereby advanced. But within economics the focus of explanation has been different: there has been in recent years much effort to explain co-operative behaviour in terms of a more sophisticated understanding on the part of individuals about where their (individual) long-term interests really lie. In particular, individuals face problems of collective action not once but repeatedly: the knowledge that pursuit of their short-term interests can harm their long-term aims by affecting the reaction of others may be a powerful inducement to behaviour that displays apparent solidarity with the interests of the group. Recent work in game theory (and particularly the theory of repeated games) has done much to make the nature of this inducement precise.4 This chapter will do four things. First, it will discuss the light which the theory of repeated games can throw upon the tendency of co-operative behaviour to enhance the prospects for successful further co-operation, and will develop a simple formal model to give a precise illustration of this effect. In particular, it will show the existence of large sets of outcomes in which players’ beliefs about the trustworthiness of others are corroborated by the others’ behaviour—even though the only factor affecting whether agents are trustworthy is whether they believe others are trustworthy! Secondly, it will discuss the literature on co-operatives in developing countries, to see to what extent empirical studies corroborate the predictions of the game-theoretic literature. Thirdly, it will report the results of a study of the determinants of the success and failure of co-operative societies in South India; this study gives support to the view that past co-operation is in itself conducive to co-operation in the future. Finally, it will discuss the implications for policy.

11.2 CO-OPERATION IN REPEATED GAMES 11.2.1 General arguments A familiar argument in game theory (one by now familiar to many who are not economic theorists) is that efficient cooperative outcomes may be sustained as equilibria of repeated games even when the only equilibria of the associated one-shot games are inefficient.5 But not all repeated games have this property: an equally familiar backward induction argument shows that if a game with a

4

This does not mean it has undermined the validity of arguments that appeal to altruism or to social norms; these different explanations are complementary, in that altruism, social norms, and long-term self-interest may be mutually reinforcing motives in the appropriate circumstances, though their relative importance will need careful, if difficult, empirical investigation.

5

Sabourian (1990) provides a valuable survey of the theory of repeated games.

285 single inefficient equilibrium is repeated a known and finite number of times, the inefficient equilibrium will be played in all periods. In order for repetition to enable co-operation, one of a number of conditions must hold: (i) the one-shot game may have multiple Nash equilibria over which all players have a strict preference ordering;6 (ii) the number of repetitions of the game, though possibly finite, may be uncertain, with a sufficiently high probability at each period of the game's being repeated for the risk of punishment to be an inducement to present co-operation; (iii) the game may be infinitely repeated; (iv) there may be incomplete information; and (v) players may have bounded rationality. In all cases, the possibility of co-operation depends upon players not discounting future pay-offs too heavily; if they do, the gains from short-term self-interested behaviour may be too great for any future inducements to outweigh. In the limit, when the complete information game is repeated infinitely often and there is no discounting of the future, the Folk Theorem states that any individually rational pay-offs (i.e. those that make continued participation preferable to withdrawing from the game) can be supported as an equilibrium, by a suitable choice of strategies to punish players who deviate from the equilibrium behaviour. The Folk Theorem is couched in terms of Nash equilibrium strategies (and may therefore rely on threat strategies that are not credible out of equilibrium). But an extension by Aumann and Shapley (1976) and Rubinstein (1976) shows that any individually rational pay-offs can also be supported as a subgame perfect equilibrium. The idea is to construct strategies that punish players who fail to play their part in punishing those who deviate from equilibrium behaviour; the infinite horizon ensures that any player can always be punished for long enough to prevent any deviation from being worth while. Unfortunately this result is not necessarily robust in the presence of even very slight discounting of the future, although Fudenberg and Maskin (1986) show that it will be so under certain conditions (namely that the dimension of the space of individually rational payoffs is as great as the number of players).7 When there is significant discounting of the future it is not possible to say anything very general about the kinds of cooperation that can be sustained. Nevertheless, it is clear what we must look for: it must be possible for credible retaliatory action by other players to outweigh the immediate benefits to any individual of failing to co-operate. This requirement in turn has two

6

Benoit and Krishna (1985); Friedman (1985); Fraysse and Moreaux (1985).

7

Abreu et al. (1990) prove important and intuitive results for the case of repeated games with discounting and imperfect monitoring, including the proposition that the equilibrium average-value-set is monotonic in the discount factor (which means, roughly, that an increased degree of concern for the future is always by increased benefits of co-operation).

286 components: first, other players must have at their disposal retaliatory strategies that ‘hurt’ the deviator sufficiently in future periods, even when future pay-offs are discounted; secondly, these retaliatory strategies must be credible, which means that, once an individual has deviated, it must be in the others’ interest to put the retaliation into effect. This may be true naturally (retaliation may be what they would naturally do in the circumstances, as when it involves playing a Nash equilibrium of the stage game), or else it may be true because of an agreement between the affected parties to put the retaliation into effect. In the latter circumstance, retaliation is itself a form of collective action, which must therefore be credible if the original collective action is to be credible. It is in this respect that one can think of the setting up of police forces and similar institutions as a central form of collective action. What can theory say about whether co-operation is habit-forming? One consideration that immediately springs to mind is that all co-operative equilibrium strategies in repeated games must be to some extent history-dependent, if only in the simplest of ways: the possibility of retaliation depends on actions that are sensitive to what other players have done in previous periods. So, a breakdown of co-operation in one period would be expected to lead to a failure of cooperation in a future period, by way of retaliation. Unfortunately, this suggestion is not very useful as a way of explaining a tendency for co-operation to be habit-forming, for two reasons. The first is that, to be credible (at least in infinitely repeated games), punishment strategies typically must not continue indefinitely; after a period of punishing deviators, players return to the co-operative equilibrium. This means that, although non-co-operation by a deviating player leads to a period of further non-co-operation by others, this period is itself succeeded by further co-operation. Over the whole stretch of time, therefore, it might not be possible to conclude a general tendency for co-operation to be habit-forming. Secondly (and more importantly), the history-dependence of strategies does not imply the historydependence of outcomes. To sustain co-operation, strategies need to be history-dependent off the equilibrium path (they specify what to do in the event of deviations from equilibrium behaviour). But if the only outcomes actually observed are equilibrium outcomes, there need be no observed history-dependence in these at all. The simplest example is one in which two players first co-operate and then play ‘tit-for-tat’ (each doing what the other has done in the previous period). In such a game the players would always cooperate, and there would be no way to tell whether their cooperation was in any sense dependent on the fact that they had co-operated in the past, since one would not observe any instances of non-co-operation against which to make a comparison. This suggests that the question: ‘is co-operation habit-forming?’ needs to be made more precise. We need to know whether in two situations in which the costs and benefits of co-operation are in every other respect identical, one could expect to observe different outcomes dependent on the degree of

287 co-operation in the past, with high previous co-operation making present co-operation more likely. Here theory can be of considerable help. The very strength of the Folk Theorem and related results is a reminder that repeated games have potentially a great multiplicity of equilibria: any individually rational pay-offs can be supported under the Folk Theorem conditions. So even though the costs and benefits of co-operation are identical in two situations, the observed outcomes may be different. Now in itself this does not mean that in circumstances where there has been co-operation in the past, future co-operation is more likely. For instance, a sequence of alternating co-operative and non-cooperative outcomes can be supported under the Folk Theorem just as much as a sequence of continually co-operative or a sequence of continually non-co-operative outcomes. But if, when faced by a potentially large number of equilibrium outcomes, players use past behaviour as a guide to choosing (as a co-ordination mechanism, in other words), then it will be easy to explain why previous co-operation should make future co-operation more likely. Nevertheless, for practical purposes, it is hard to believe that the observed tendency for co-operation to be habitforming represents merely a means whereby individuals co-ordinate on one of several possible co-operative outcomes. First of all, in the kinds of co-operative institution that are typically established to overcome problems of collective action, there is no difficulty about communication. On the contrary, members may spend a long time communicating with each other (or squabbling, to put the matter less clinically), but may still fail to resolve their difficulties in implementing successful collective action. Secondly, if individuals are seeking to co-ordinate their actions, it is hard to understand why they should ever choose to co-ordinate on any but efficient outcomes. So if we observe failures of cooperation in the past followed by failures of co-operation in the future, it seems perverse to imagine that the reason for this is that players have chosen to co-ordinate on an equilibrium with little co-operation (when they might have chosen to co-ordinate on one with more).8 We still need to explain why failures of co-operation in the past might make nonco-operation preferable for all parties in the future. In practice we frequently observe failures of co-operation even when the opportunities for communication are good. In other words, we are looking at the outcomes of games in which co-operation is not easy but hard to sustain, so that players can sometimes manage to do so and sometimes not. This suggests we are dealing with situations in which either discount rates are high

8

An interesting recent paper by Crawford and Haller (1990) explores features of co-ordination games in which players use past behaviour to assist their co-ordination among multiple equilibria in the future. Unfortunately their arguments do not carry over (at least, not straightforwardly) to the present context. In their framework, there are multiple equilibria of each stage game, whereas here there is only a single dominant strategy equilibrium in each round; the multiplicity is in the equilibria of the repeated game as a whole. Since the game with multiple equilibria is not itself repeated, arguments about the value of equilibria attained once as a guide to future equilibrium play are not relevant.

288 (relative to the frequency with which opportunities occur for repeated co-operation), or the one-off benefits from defection are high relative to the per period costs to the defector of retaliation.9 Again, this accords with common sense. Suppose an institution is established to protect common grazing land in a village. It may take some time to discover that the rules of grazing are being flouted or that the officers have embezzled the funds set aside to put up fences. Even though the previously co-operative members may now withdraw their co-operation in retaliation, the dishonest officers or the uncooperative grazers may have benefited by enough in the meantime for this retaliation to leave them no worse off than they would have been by co-operating. In circumstances like these—namely where the sustainability of cooperation is a marginal matter—it is likely that the presence or absence of trust will affect the extent to which co-operation succeeds. The very fact that makes cooperation hard to sustain (the large benefits to one-period defection) also means that it makes a significant difference to individuals whether they co-operate anticipating similar behaviour on the part of others, or choose instead to defect without waiting for others to do so first. And here it will be very important what expectations individuals have about each other's behaviour, and therefore what degree of trust they have in one another. If an individual believes that others are unlikely to co-operate, there may be no point in waiting for these gloomy expectations to be confirmed; the individual may as well defect immediately. Contrariwise, if co-operation on the part of others is confidently anticipated, defection now might be foolhardy for it would threaten the viability of the whole co-operative enterprise.

11.2.2 A model of trust This intuition can be formalized with the help of a very simple model, cast in the form of a two-person game. Suppose that the benefits to co-operation are given by the pay-off matrix in Fig. 11.1. The one-shot game takes the form of a prisoner's dilemma. Here the only restrictions are that Y > X > 0 and that Z > 0. There is a discount factor g. Then we know that provided Y − X < gX/(1 − g) there exists a retaliation strategy which consists of playing Defect for a finite number of periods in the event that the other player has played Defect after an agreement to co-operate, and which ensures that the other player is no better off from the defection. Let T be the lowest integer such that Y − X = gX + g2X + … + gT. Then T is the smallest number of periods for which each player must threaten to retaliate in order for the threat credibly to sustain co-operation.

9

This point can be put very simply as follows: for co-operative behaviour to be sustainable, the one-period gains to any agent i of defecting must be less than the discounted costs to that agent of retaliation for some finite number T of periods. Let Di be the gains from defection and Ri be the costs per period of retaliation, with g the discount factor. Then co-operation is sustainable iff Di < g ·Ri + g2 ·Ri + … + gT ·Ri , which implies Di < gRi /(1 − g ). Then, given Ri , co-operation will be unsustainable if g is low or Di is high.

289 Fig. 11.1

Now consider how incentives for co-operation might be affected by the degree of trust between the players—specifically, by whether each player believes that the other will co-operate in the current period. Assume the players have communicated and agreed to play the equilibrium (Co-operate, Co-operate) in perpetuity thereafter, with the credible threat that if one defects the other will subsequently play Defect for t periods, where t > T. Nevertheless, the players are not certain whether to trust each other in the current period (for the time being we assume that this uncertainty affects the current period only; in subsequent periods the players will establish with certainty what outcome path they are following). Each player believes with subjective probability p that the other will play Co-operate in the current period (p may, of course, be 1 in the case of complete trust). Then, the expected pay-off to each player from playing Co-operate is(11.1)

which is equal to p times the pay-off from playing (Co-operate, Co-operate) for ever, plus (1 − p) times the pay-off from playing Co-operate while the other player defects, then retaliating for t periods and then returning to (Co-operate, Co-operate) for ever. The expected pay-off to each player from playing Defect will likewise be(11.2)

and by subtracting (11.2) from (11.1) we see that the net benefit from playing Co-operate is(11.3)

which is negative for sufficiently low p, and positive for p close to unity by the fact that t > 7. This means that sufficient pessimism about the prospects for co-operation by the other player in the current period will lead each player to defect, while sufficient optimism will lead that player to co-operate. The critical value of p at which P(C) − P(D) = 0 is given by(11.4)

290 which is increasing in Y and Z and decreasing in X. This means that a slight lack of trust (p slightly below unity) is more likely to harm the prospects for co-operation, the lower are the gains from co-operation (provided these are positive), the higher are the benefits from defecting while the other player co-operates, and the higher is the cost of co-operating while the other player defects. So far, of course, p has been introduced simply as an exogenous parameter and the possibility of multiple equilibria established, but nothing has been said about how players might form their beliefs. A more interesting question, therefore, is whether there exist equilibria in which these beliefs are self-confirming, in the sense that if players hold certain subjective probabilities of co-operation they will in fact co-operate with exactly these probabilities. It might seem that in the example given above this is unlikely except for corner solutions; given p, players either co-operate or do not, so that only values of p of zero or unity could be self-confirming. But if co-operation depends on other stochastic factors (if pay-offs, for instance, are uncertain) there may be a large set of interior self-confirming equilibrium beliefs. This much is unsurprising. More interestingly, perhaps, there may be a large set of interior selfconfirming equilibrium beliefs even without any extrinsic uncertainty, in a manner reminiscent of models with sunspots. These two cases are explored below.

11.2.3 Self-conrming beliefs with extrinsic uncertainty The natural way to introduce extrinsic uncertainty is to derive the subjective probabilities of co-operation from each player's uncertainty about the other player's pay-offs. Given probability distributions over the values of X, Y and Z, the value of p would then be derived as the probability that P(C) − P(D) is positive. An equilibrium would consist of a value of p such that, given the distributions of X, Y, and Z, the probability that P(C) − P(D) > 0 was equal to p. Since the expression P(C) − P(D) is itself a function of p, there may well again be multiple solutions. It is difficult to say anything general about the nature of solutions for the value of p.10 But one special case is of interest. Suppose that the values of two of the pay-offs are known (more specifically, that they are common knowledge between the players). Without loss of generality let these be X and Z. The only uncertainty concerns the value of Y; furthermore, it is common knowledge that this can take only two values, Y1 and Y2, where Y2 > Y1. Because p*(Z, X, Y) is increasing in Y, it follows that p*(X, Y2, Z) > p*(X, Y1, Z); we can also assume that p*(X, Y2, Z) < 1. Now suppose that Y takes the value Y1 with probability q and Y2 with probability (1 − q); each player knows the value of Y in his own case, while knowing only the distribution in the case of his

10

It is not even possible to be sure that interior solutions exist, because of potential discontinuities in the value of pr (P (C ) − P (D ) > 0).

291 opponent. If p > p*(X, Y2, Z) then both types of player co-operate with probability one; if p < p*(X, Y1, Z) then both types defect with probability one. However, if p*(X, Y1, Z) < p < p*(X, Y2, Z) then players for whom Y = Y1 will cooperate while players for whom Y = Y2 will defect. So the probability of co-operation will be the probability that Y = Y1, namely q. This implies that for any probability q such that p*(X, Y1, Z) < q < p*(X, Y2, Z), then provided this is the probability that Y = Y1, there is an equilibrium in which each player's subjective probability that the other will co-operate is equal to the true probability that the other co-operates, namely q. There are also two corner equilibria: one in which each player believes with certainty that the other will co-operate (and both players indeed co-operate with certainty); and a third equilibrium in which each player believes with certainty that the other will defect, and in consequence both players indeed defect with certainty.

11.2.4 Self-conrming beliefs without extrinsic uncertainty Interior values of p can be self-confirming in equilibrium without extrinsic uncertainty if a player's subjective probability in any one period (which determines whether or not he will co-operate) is itself not deterministic, but fluctuates randomly. Suppose that each player fluctuates between a state of optimism and a state of pessimism about the prospects for co-operation. In particular, let the players fluctuate between the states shown in Table 11.1. Table 11.1 Beliefs and states Player

State

1 1 2 2

Optimism Pessimism Optimism Pessimism

Subjective probability that other will co-operate P1 P2 Q1 Q2

Probability of being in state Q 1−Q P 1−P

We assume 1 > P1 > P*(X, Y, Z) > P2 > 0 and 1 > Q1 > P*(X, Y, Z) > Q2 > 0 (call this assumption A). This implies that each player always cooperates when optimistic, defects when pessimistic. This implication is consistent with selfconfirming probabilities strictly greater than zero or strictly less than unity provided there is less than perfect correlation between the states of the two players. For instance, player 1's belief when optimistic that player 2 will cooperate with probability P1 is confirmed if the probability that player 2 is optimistic, conditional on player 1's being optimistic, is P1. This, in turn, is consistent with player 2's having an unconditional probability P of being optimistic if

292 (11.5)

and the analogous necessary condition for player 2's belief to be confirmed (i.e. the unconditional probability that player 1 is optimistic) is(11.6)

which through substitution yield(11.7)

and(11.8)

It follows from assumption A that P and Q themselves must lie between zero and unity, and it is evident that provided 7 and 8 hold any sets of probabilities within the relevant range can be self confirming. In fact we can state: Proposition 1: Any four real numbers p1, p2, q1, q2 can constitute a self-confirming equilibrium of beliefs in which P1 = p1, P2 = p2, Q1 = q1, Q2 = q2 provided (i) P1, P2, Q1, and Q2 thereby satisfy assumption A; and (ii) each number is the true probability, conditional on the player's being in the relevant state, of that player's opponent being in the optimistic state. Proof: Given assumption A each player co-operates with probability 1 when optimistic and with probability zero when pessimistic. Therefore if p1 is the true probability, conditional on player 1's being optimistic, of player 2's also being optimistic, it is also the true conditional probability of player 2's cooperating. Therefore P1 = p1 is a self-confirming belief. The same argument holds for p2, q1, and q2. One can easily imagine how such beliefs might evolve. For instance, suppose one player is always optimistic when it is sunny and pessimistic when it is rainy. Another is optimistic when it is calm and pessimistic when it is windy. Since there is a positive but imperfect correlation between rain and wind, the first player's belief that the second player is more co-operative when it is sunny is confirmed (and vice versa). Furthermore, provided each player started out with subjective probabilities leading him always to co-operate when optimistic and defect when pessimistic, updating these probabilities in the light of observed frequencies is not likely to affect the actual frequency with which either player cooperates (and will therefore be consistent with convergence of the updating process). Of course, if either player were aware of the existence of a signal perfectly correlated with the state of the other, he could improve the precision of his forecasts. But in the meantime he may well feel that his predictions have been adequately confirmed. This intuition in turn provides the basis for a more formal result concerning stability: Proposition 2: Suppose Proposition 1 is true. Agents have prior probability distribution functions (p.d.fs with means and variances

, and

,

293 , and

. They play the game repeatedly, updating their prior beliefs using Bayes’ Rule. If: (i) and (ii) , then the probability that the means of the posterior p.d.f ’s converge to the equilibrium beliefs p1, p2, q1, and q2 tends to 1 as , and tend to 0. Proof: If (i) holds at any repetition of the game, player 1 will co-operate with probability q1 when player 2 is optimistic and with probability q2 when player 2 is pessimistic (this follows from the definition of q1 and q2 in Proposition 1 as equilibrium beliefs). If (i) holds for all repetitions of the game, therefore, and will converge to q1 and q2 by the Central Limit Theorem. Likewise, if (ii) holds for all repetitions of the game, and will converge to p1 and p2. It remains to be shown that, if (i) and (ii) hold for any repetition of the game, the probability that they hold for all repetitions of the game tends to unity as , and tend to 0. To show this, note that for any finite n and for any e > 0 the probability that the absolute difference between the mean of the prior and the mean of the posterior distribution is greater than e tends to zero as the variance of the prior distribution tends to zero. Next, for any finite n, there exists an e > 0 such that the probability that the absolute difference between the mean of the prior distribution and P*(X, Y, Z) is less than e tends to zero. Therefore, since the prior means satisfy conditions (i) and (ii), the probability that the posterior means fail to satisfy (i) or (ii) as appropriate itself tends to zero as , and tend to 0. The intuition behind the proof is straightforward: if prior beliefs are very diffuse, then the posterior p.d.fs of the players will be heavily affected by the observation of the first few plays of the game, so that even if their prior probabilities are initially in the region in which co-operation occurs if and only if there is optimism (which is the stable region of the updating process), a few rounds of bad luck may move the posterior probabilities out of this region. The probability that this will occur becomes vanishingly small as the prior beliefs become less and less diffuse. So, to sum up, we have shown that even in the absence of extrinsic uncertainty,11 the state of trust between the players may affect the probability of future co-operation. There are three important features of such a finding: first, high levels of trust are associated with high probabilities of co-operation; secondly, the degree of trust may be confirmed by the players’ behaviour even if the only stochastic factor affecting their behaviour is the degree of trust itself; and thirdly, players may differ in their degree of trust for each other (in equilibrium) even if all the objective factors affecting their pay-offs are symmetrical between the players. So a symmetric game can have asymmetric interior equilibria.

11

To say that the uncertainty is intrinsic does not imply that it is endogenous (the same point is true of intrinsic uncertainty in sunspot models). But the uncertainty does not directly affect any of the pay-offs.

294

11.2.5 Trust as a strategic variable It is natural to suggest that players might derive their subjective probabilities of co-operation by observing each other's behaviour; this would indeed be the obvious way in which previous co-operative history could influence the prospects for present co-operation. In order to model this we must abandon the assumption that it is only in the present period that there is uncertainty about whether or not there will be co-operation. This complicates the analysis considerably; in particular, it implies that the benefits of sustaining co-operation in the present period are not as great as if present cooperation could ensure co-operation with certainty in future periods. So to keep matters simple I shall assume the following: (i) If either player fails to co-operate in the present period both players move with certainty to a retaliation phase whose present value to each of them is R*. (ii) Each player has a probability q of co-operating and ascribes to the other a probability p of co-operating (in equilibrium p will equal q). These probabilities derive solely from extrinsic uncertainty about pay-offs, which is symmetrical between players, so that players’ subjective probabilities are likewise identical. We can derive the expected present value to each player of participating by the principle of dynamic programming. Let V(p, q) be the present value of the benefits of participating, given that the opponent co-operates with probability p and the player with probability q. Then V(p, q) is equal to benefits of co-operation times the probability of co-operating, plus those of defecting times the probability of defection. The benefits of co-operation are the one-period benefits X plus the discouted benefits of participating in the next period W(p, q) times the probability that the other player also cooperates, less the (probability-weighted) costs of co-operating if the other player does not do so. Formally:(11.9)

In a steady state W(p, q) will equal V(p, q) so we can solve for V(p, q):(11.10)

Likewise in equilibrium p = q so in steady-state co-operative equilibrium(11.12)

and q will be given by the probability that P(C) − P(D) > 0, i.e. that(11.13)

which again may have a number of solutions depending on the probability distributions of X, Y, Z, and R*.

295 However, once present behaviour influences players’ perception of the probability that the other will co-operate in the future, there may be incentives for strategic co-operation, designed to influence the value of p in a manner favourable to the player himself. That there are incentives to do so may be seen by differentiating (11.9) with respect to p; for small changes in p we may assume (by the envelope theorem) that Wp(p, q) = Vp(p, q), so differentiating (11.9) is equivalent to differentiating (11.10). Doing so yields:(11.14)

This expression is evidently positive in X, Y, Z, and negative in R*. This indicates that players have more reason to cooperate in order to increase trust, the higher are the benefits to co-operation, but also the higher are the benefits of cheating and the costs of being cheated upon (and the lower the value of being in the retaliation phase).

11.2.6 Trust and renegotiation The literature on repeated games12 has drawn attention to the possibility that some strategies that support co-operative behaviour may rely on threats of retaliation that are not credible—not because it would be in any one player's interest to deviate from the retaliatory strategy, but because if the retaliatory strategies result in inefficient outcomes, players would have an incentive to renegotiate their way to Pareto-superior outcomes. The importance of trust suggests why this may not be an important problem for co-operative behaviour in circumstances like those captured in the present model. In the context of a simple repeated prisoner's dilemma, individuals might be tempted to cheat because they reason that, having cheated, they can persuade other players to let bygones be bygones, and start co-operating again immediately rather than go through an inefficient period of non-co-operation. However, when trust matters for the possibility of co-operation, bygones may not be bygones, for the simple reason that previous cheating may persuade other players that a new agreement to co-operate is not credible even if they had previously thought the original one was indeed so. So, by cheating, a player may reduce the probability value p in future periods sufficiently to prevent other players from co-operating even if an agreement to do so is reached. This will ensure that the retaliatory strategy is credible, and the original co-operative agreement may be credible after all.

11.2.7 Co-operation and the assurance game Some writers on practical problems of collective action in developing countries have suggested that these may often be modelled better as an assurance

12

Farrell and Maskin (1989).

296 game than a prisoner's dilemma (see Runge, 1986, and Stevenson, 1991: esp. 73–6). The former differs from the latter in that non-co-operation is not a dominant strategy in the one-shot game; instead, it is in players’ interests to cooperate provided they can be assured that others (or enough others, where multi-person games are in question) will do the same. Now it is obviously an empirical matter whether particular situations are indeed better modelled as one type of game rather than another. An analysis in terms of the assurance game will not provide much help in those situations which are indeed prisoners’ dilemmas, and the externalities involved in problems of common-property resources suggest that the prisoners’ dilemma will often be the better model for the one-shot game. However, one way of viewing the argument of this chapter (and, more generally, the literature on repeated games) is as analysing the circumstances under which the threat of retaliation transforms a prisoner's dilemma in the one-shot game into a supergame whose overall pay-off structure is in fact an assurance game. Repetition does two things: first, by allowing for the possibility of retaliation it transforms pay-offs in such a way as to make co-operation rational if enough other players are believed to be likely to do the same; and secondly, by making this common knowledge between players it increases the likelihood that other players will co-operate and therefore the likelihood that co-operation will in fact be individually rational.13 Some of the applied literature has stressed that co-operation can be sustained with only a subset of the parties involved. A generalization of the model of this chapter to the n-person game would explain this naturally in terms of a higher probability of co-operation by any individual leading to a higher incidence of co-operation in any multi-person group. It would emphasize that the pay-offs to any one person will depend upon how many of the others can be expected to co-operate in any one period.

11.3 THE EXPERIENCE OF CO-OPERATION IN DEVELOPING COUNTRIES The performance of co-operative enterprises as a response to problems of underdevelopment has not been an encouraging one.14 Throughout the world, the obstacles to collective action have been systematically underestimated by governments and other agencies promoting co-operative enterprise as an alternative to exclusive reliance on privatesector entrepreneurial capitalism. Sometimes this has been evidently because co-operatives were

13

I am grateful to Bhaskar Vira for this point.

14

Lele (1981) surveys the literature up to the end of the 1970s. More recent contributions (not all pessimistic) include Attwood (1989), Attwood and Baviskar (1988), Berkes (1989), Feeny et al. (1990), Gill (1983), Levi and Litwin (1986), Nolan (1988), Onwuchekwa (1985), Ostrom (1990), Putterman (1985), Sengupta (1991), Stevenson (1991), Wade (1987), among many others.

297 advanced as an appropriate form of organization for too wide a range of economic activities—a belief often accompanied, as in China or other socialist-inclined societies such as Tanzania, with a hostility to anything resembling individual material incentives. Sometimes co-operatives have been granted a more specific role (in agricultural marketing, for instance) and have been run on less doctrinaire lines, but with results that have been at best mixed. Even where co-operative credit societies, for instance, continue to function, they usually do so at the price of heavy subvention from the State. Their record as an efficient form of intermediation is poor (often loan-recovery rates are particularly low), and there is even doubt whether they contribute to a more equitable access to economic resources, since local élite groups are often observed to dominate the activities of co-operatives to a considerable degree. Nevertheless, while co-operatives as a blanket solution to development problems have been effectively discredited, a belief in their appropriateness in certain specific areas of economic life is enjoying something of a resurgence. Nowhere is this more true than in the field of environmental economics. It is here that market failures due to private competition have shown themselves most starkly. When problems of overgrazing, deforestation, or water pollution rise, the alternative to collective action is not usually private action; it is complete inaction.15 So the pessimistic conclusions of studies that have found cooperative enterprises inferior to ordinary private enterprises at undertaking tasks that in principle either could do equally well,16 do not carry over simply to these contexts where private action has already demonstrated its inadequacy. Furthermore, participatory management of collective assets such as environmental resources has increasingly appeared essential for mobilizing the broad support necessary for their conservation.17 So the question under what conditions participatory management can be effective has assumed an increased urgency with the growing awareness of the importance of environmental issues in development. Fortunately, the literature on co-operative enterprises is not uniformly bleak.18 In particular, comparative studies have often highlighted factors systematically affecting the results of efforts at collective action, whether in the case of officially sponsored and financed development schemes (Dore and Mars, 1981) or local initiatives coming from within a community (Wade, 1987).

15

This undoubtedly has something to do with the large numbers of agents that are typically involved, which makes Coasian bargaining solutions infeasible. Runge (1986) stresses the point that private solutions to the problem of the commons have often failed to stop over-exploitation.

16

Stevenson (1991) is a recent example which demonstrates econometrically the lower productivity of common-property than private-property pasturing in Switzerland, while nevertheless accepting that transactions costs may make privatization infeasible in some circumstances.

17

See Chopra et al. (1989), for instance. Participatory management as a general theme is far from new, of course, but after the initial enthusiasm of the early 1970s it, too, was more soberly appraised in the 1980s.

18

Bratton (1986) and Thomas (1987) are just two of a number of recent studies with success to report.

298 On balance these studies provide considerable support for the view that prior experience of successful collective action strengthens the basis for collective action in the future. For instance, Dore (1981) discusses a number of cases that provide support for the view that ‘community solidarity helps to promote community-wide responses to development programmes’, though his interpretation of this phenomenon is as revealing the existence of entrenched norms of behaviour rather than as affecting individuals’ calculation of their own long-term advantage. Likewise a number of studies have blamed excessive heterogeneity of membership for a breakdown of co-operative action (Gaikwad, 1981: esp. 286–8; Onwuchekwa, 1985: 226; Litwin, 1986: 39–40), though they have usually stressed that if the gains from cooperation are sufficiently great the disadvantages of heterogeneous membership can be overcome.19 Some studies have specifically mentioned the fact that success appears to breed success (and vice versa) in co-operative development (Sargent, 1982: esp. 49–52). Even where case-studies have reported successful co-operative action even in the presence of strong factional divisions (Attwood, 1989, for instance), this appears to be due to competition among faction leaders in circumstances where members have sufficient democratic control to choose among them freely, and in no way invalidates the claim about the importance of trust. The argument that that prior experience of successful collective action strengthens the basis for collective action in the future should not be confused with the related claim that co-operatives work best when grafted in some way on to ‘indigenous’ forms of co-operation.20 Prior experience of indigenous co-operation is likely to strengthen the prospects for future co-operation provided it strengthens trust—but it could have exactly the opposite effect if the prior experience has been adverse. For example, it may be that encouraging local religious organizations to be involved will help to promote collective economic action—but not if on previous forays into economic activity these organizations have acquired a reputation for corruption or incompetence. The literature on co-operative enterprises in development does appear, then, to offer support for the view that cooperation is habit-forming. But the overwhelming dependence of this literature upon case-study material (and the corresponding difficulty in drawing comparable conclusions from different cases) suggests it may be fruitful to explore other kinds of evidence based on somewhat larger samples. This is the purpose of the next section.

19

Young et al. (1981) in a study of Ugandan co-operatives comment in some surprise that members in their survey appeared to believe co-operatives were working effectively even though only 30 per cent of them thought they always behaved honestly. The key seems to lie in the evidence they also report that only 3.4 per cent believed private Asian traders always behaved honestly and only 5.3 per cent thought so of African traders: the relative degree of trust in co-operatives was therefore quite substantial.

20

See Young et al. (1981, ch. 2), for a discussion of this argument, as well as two references cited there: Bennett (1975), and Seibel and Massing (1974).

299

11.4 A STUDY OF CO-OPERATIVE SOCIETIES IN SOUTH INDIA This section reports illustrative results of a study of milk producers’ co-operative societies in Tamil Nadu State in South India. These results suggest that the presence or absence of trust may play an important role in explaining the success or failure of collective action, independently of the objective gains to participation in such collective action. Milk producers’ co-operative societies have been a central component in the implementation of the Government of India's Integrated Rural Development Programme: through them are channelled subsidized loans for the purchase of livestock, and complementary facilities such as feed and veterinary care. A society's costs are recouped through collection of milk produced by members’ animals, from the value of which a charge is deducted before members are reimbursed. In spite of substantial subsidies, many co-operative societies have started up only to fold again relatively quickly. Aggregate statistics for the extent of co-operative society collapse are hard to come by: the Tamil Nadu Cooperative Milk Producers’ Federation reports a total of 6,615 functioning co-operative societies in the State in 1989, with a further 1,516 dormant and 730 liquidated altogether. But the figures for dormancy are almost certainly a massive understatement of the true position. A sample of seventy-two societies based in and around a market town in the centre of Trichy district in April 1990 contained nineteen that were no longer functioning, none of which was on the list of dormant societies and only one of which had been inactive long enough to be among the liquidated societies recorded in the district.21 A further eleven societies were in serious difficulty, with active membership reduced to a handful of people. The causes of these difficulties are various.22 Some societies suffered from financial irregularities, with officers embezzling the funds received for the sale of milk rather than returning them to members. A more common problem was that members would avail themselves of society membership in order to receive subsidized loans, but would then sell milk to private milk-traders rather than to the society. Both of these kinds of ‘cheating’ were likely to lead to the eventual breakdown of the society, an outcome that in the opinion of many villagers left them collectively worse off. Sometimes members of the society would decide that it was not worth continuing to farm milch cattle, and would sell their animals so that active membership dwindled; although hardly describable as ‘cheating’, such actions typically made it harder for the co-operative society as whole to continue functioning. Many of the bank and government officials responsible for implementing the IRDP scheme in general, and the system of loans to livestock co-operative societies in particular, tend to describe the problem of co-operative breakdown

21

Anecdotal evidence suggests that district officials substantially understate the incidence of dormancy in making reports to the state-level authorities.

22

These are analysed in greater detail in Seabright (1990).

300 as though it is simply a matter of the good faith and integrity of the individuals concerned. The policy problem on this view consists simply in finding individuals of sufficient calibre and appealing sufficiently to their public spirit. While there can be little doubt that such factors are of significance, it is rapidly evident from a comparative study that there are other economic and social factors systematically associated with the success and failure of co-operative action. One feature that is striking at first glance is the proportionately greater incidence of breakdown in the areas of irrigated and relatively commercial agriculture, compared to those characterized by dryland farming. Table 11.2 illustrates. Of the seventy-two societies in the sample, three have been excluded as being of too recent formation (less than one and a half years) for any conclusions to be drawn about their success or failure. The remaining sixty-nine were classified into five categories: (1) (2) (3) (4) (5)

Defunct In serious difficulty Performing adequately but with some difficulties Healthy Outstanding.

Table 11.2 Performance of co-operative societies by agroclimatic zone State of society Defunct In difficulty Adequate Healthy Outstanding Total

Dryland

Mixed

Wetland

Total

5 5 7 14 5 36

2 2 2 1 1 8

12 4 3 6 0 25

19 11 12 21 6 69

Table 11.2 reveals how societies in the dryland areas suffered from considerably fewer problems than those in the wetland regions. The most likely explanation on the face of it is that the returns to co-operative activity are greater in the dryland regions—not because livestock farming undertaken by co-operatives is absolutely more profitable but because it is more profitable relative to the meagre alternative income-earning opportunities faced by farmers in the dryland regions. In other words, the differential between the gains to private activity and the gains to co-operative activity is greater in the dryland areas. The economics of this contrast between agroclimatic zones is explored in greater detail in Seabright (1990), but depends upon the fact that the initial scale economies in the provision of production, transport, and marketing infrastructure have been to a greater degree exhausted in the irrigated zones, leaving less for specifically co-operative action to achieve.

301 Nevertheless there are clearly a number of other factors at work in explaining the success and failure of co-operative societies. In order to gain some idea of the relative significance of these different factors, I estimate an ordered probit model of co-operative society performance. An ordered probit model estimates an equation:(11.15)

subject to the constraint that the values of Y are not directly observed. What is observed is the value of a discrete variable Z which is related to Y by a series of conditions:

(11.16)

and so on. Here the values of Z correspond to the five possible states of a society, and the procedure estimates both the coefficients on the various independent variables, and the threshold values Y1, Y2, etc. of the dependent variable. Essentially the ordered probit procedure is doing no more than provide a multivariate analogue of the kind of suggestive information contained in a univariate format in Table 11.2. It should be stressed that the parameter estimates I report are merely illustrative of the kinds of causal factor that may be responsible for co-operative society performance, and that in a sample of only sixty-nine it is not possible to test causal hypotheses rigorously. In particular, the t-statistics in an ordered probit model are valid only asymptotically. The small sample properties of ordered probit estimators are not well understood, but most practitioners accept that much larger samples are required than for OLS estimation before any degree of confidence can be placed in test statistics.23 Nevertheless the results are interesting in so far as they suggest an independent role for economic factors and for factors representing trust. The independent variables considered are the following: 1. 2.

23

ZONE: a dummy variable representing the agroclimatic zone: 0 = dry, 1 = mixed, 2 DISPUTES: an index of the extent to which the village(s) in which the society functions

= wet. is (are) characterized by a previous record of breakdown of trust or of collective action. A village scores 1 if the co-operative society secretary describes it as ‘seriously divided politically’ or if there is a known history

Nevertheless, I do want to claim that the procedure followed here is preferable to reliance entirely upon case-studies, which has been the predominant methodology of the literature dealing with collective action. Limited dependent-variable analysis using large samples has been principally confined to data at the individual or household level and not (for obvious reasons) to data involving whole societies or communities.

302

3. 4. 5. 6. 7. 8.

of political feuding;24 otherwise it scores zero. It scores an additional 1 if serious disputes between villagers are typically resolved by calling the police rather than by appeal to a committee of village notables. It scores a further 1 if different caste groups within the village hold religious festivals separately rather than co-operating to hold joint festivals for the village as a whole. SCHEDULED CASTE: a dummy variable taking the value 1 if a majority of the society's members belong to the scheduled castes, and zero otherwise. OFFICER’S EDUCATION: number of years of schooling of the principal officer in charge of running the society. LANDLESS: a dummy variable taking the value 1 if a majority of the society's registered members are landless labourers (provided these number seventy-five or more), and zero otherwise. PRICE DISADVANTAGE: takes the value 1 if there are private traders in the village offering higher milk prices than the society (otherwise zero). QUALITY CONTROL: a dummy variable taking the value 1 if the society used a testing system for the fat content of milk, and paid vendors accordingly (otherwise zero). PRIVATE TRADERS: the number of private milk-traders visiting the village each day.

Table 11.3 illustrates the results of five alternative equation specifications. The most striking feature of the table is that the coefficient on the DISPUTES variable is consistently negative (with a t-value well over two) in all equations.25 This means that, controlling as far as possible for objective economic factors, the villages in which there is a prior history of disputes or breakdowns of collective action are more likely to observe breakdowns in collective action in the future, exactly as the model of trust in co-operation predicts. The economic variables in the table have the expected sign. The coefficient on ZONE is negative (which implies that success probabilities are lower in the wetland regions, where the gains to co-operation are lower). So is that on PRICE DISADVANTAGE (implying that high private gains from defection make failure more likely). The sign of the coefficient on PRIVATE TRADERS has the same implication. The positive coefficient on LANDLESS probably reflects the fact that landless labourers, with fewer alternative economic opportunities, have more to gain from participating in co-operative societies. The positive coefficient on OFFICER’S EDUCATION might appear to support the view cited above about the importance of the characteristics of individuals involved in the society. But when QUALITY CONTROL is added in equation (3)

24

This inevitably involves a degree of subjective judgement (as indeed does the classification of societies by degree of success). However, it was encouraging that respondents in general had no difficulty in understanding the questions put to them nor in telling us whether the village was ‘seriously’ divided. Conversation with them soon made the distinction between mild and serious political divisions pretty unmistakable.

25

A few other equation specifications were also tried and rejected because of problems of multicollinearity or endogeneity bias; the coefficient on DISPUTES remained high and negative in all of them.

303 Table 11.3 Determinants of success or failure of co-operative societies Independent vari- Eq. 1 ablea Zone − 0.29 (− 1.92) Disputes − 0.42 (− 2.34) Scheduled − 0.48 caste (− 1.76) Officer's 0.11 education (6.00) Landless — Price disadvantage Quality control Private traders Disputes (wet zone) Disputes (dry zone) Threshold 1 Threshold 2 Threshold 3

Eq. 2

Eq. 3

Eq. 4

Eq. 5

− 0.32 (− 1.82) − 0.47 (− 2.40) − 0.45 (− 1.40) 0.15 (4.89) 0.72 (2.00) − 0.67 (− 1.68) —

—

—

− 0.26 (− 1.40) − 0.49 (− 2.85) − 0.57 (− 1.62) 0.004 (0.11) 0.53 (1.45) − 1.00 (− 2.45) 0.94 (3.14) —

− 0.36 (− 1.22) [—]

—

− 0.35 (− 2.11) − 0.58 (− 3.21) − 0.47 (− 1.48) 0.12 (6.23) 0.84 (2.45) − 0.73 (− 1.81) —

—

—

—

—

− 0.08 (− 1.44) —

—

—

—

—

0.54 (3.80) 1.07 (5.66) 2.32 (8.28)

0.62 (3.74) 1.23 (5.83) 2.52 (9.14)

0.79 (3.56) 1.45 (5.39) 2.74 (8.73)

0.65 (3.76) 1.28 (5.85) 2.58 (9.36)

—

− 0.40 (− 1.14) 0.12 (5.84) 0.72 (2.40) − 0.78 (− 1.79) —

− 0.52 (− 0.80) − 0.42 (− 2.30) 0.60 (3.52) 1.19 (5.42) 2.46 (8.85)

t-ratios in parentheses Note: number of observations: 69. a

the coefficient on OFFICER’S EDUCATION drops to barely above zero. This suggests that the main reason why individuals of high education levels have been able to contribute to successful collective action is that they perceived the importance of devising individual incentives for co-operation. So this turns out to be an economic variable after all. The interpretation of SCHEDULED CASTE is ambiguous. It may be either an economic variable or a ‘trust’ variable, or may represent elements of both. Seabright (1991) reports evidence from the same region that members of the scheduled castes tend to be discriminated against in livestock markets, so it may be that these groups have less to gain from belonging to co-operative societies.26 Alternatively, it could be that since members of the scheduled castes 26

The issue is somewhat more complex than this brief summary makes it appear. The study cited reported that scheduled castes faced discrimination in livestock markets generally (whether or not they were IRDP beneficiaries), but not necessarily more than recipients of IRDP loans. Since co-operative societies consist partly but not entirely of IRDP recipients, this would imply that scheduled castes might be expected to own lower-quality livestock on average. This would not of itself explain a greater tendency to ‘cheat’ by selling milk to private traders (since the low quality of the animal is independent of whether the owner sells milk privately or to the society). But it might explain a greater tendency to give up livestock farming altogether after once taking it up.

304 have tended in the past to be systematically excluded from access to the benefits of political and social participation., they are themselves less confident that the system of co-operative societies will yield them any benefits this time. Their trust may be therefore lower, and their willingness to co-operate correspondingly reduced. Finally, it is possible that the social marginalization of the scheduled castes has given them less experience at forms of collective action among themselves as well as between them and other groups. The degree of intra-caste as well as inter-caste trust may be low.27 Finally, the DISPUTES term is interacted with ZONE; it was predicted above that, provided the benefits from cooperation were positive, a lack of trust was more likely to harm co-operation the lower these benefits were. This would imply that the coefficient on DISPUTES in the wet zone would be more negative than that in the dry zone. While the coefficients reported have the right relative sign, the difference between them is too small for any significant conclusions to be drawn.28 At all events, the main point is that the experience of co-operative societies in this sample suggests that, even controlling for differences in objective economic conditions, successful collective action in the past increases the chances of successful co-operation in the future. While the sample is too small for this suggestion to remain more than tentative, this at least suggests a direction for further corroborative research.

11.5 CONCLUSIONS AND POLICY IMPLICATIONS This chapter has argued both on theoretical and empirical grounds that the success or failure of co-operative action in the past can be expected to have a significant effect on the prospects for co-operative action in the future, by affecting the degree of confidence or trust that individuals have in the likelihood that others will play their part in a co-operative agreement. Theoretically, it

27

Indeed, when the model is estimated with separate coefficients for societies with a majority of scheduled caste members and societies with only scheduled caste members, there is little difference between the coefficients (with the latter being marginally more negative).

28

In addition to the equations reported, a number of alternatives to equation (2) were tried to test for robustness. A dummy variable taking the value 1 when caste membership in the society was heterogeneous and 0 when it was homogeneous was only weakly positive, though it slightly lowered the coefficient on SCHEDULED CASTE . Replacing PRICE DISADVANTAGE by the difference in price offered between private milk-traders and the society raised its t -ratio, as well as the coefficient on LANDLESS . Replacing LANDLESS by the number of landless labourers in the village leads to the latter having a t -ratio of only 1 : 3, and halves the coefficient on SCHEDULED CASTE and reduces that on ZONE by a third, suggesting the presence of multicollinearity problems. Finally, excluding from the sample all known cases of embezzlement leads to a slight reduction in all coefficients except OFFICER'S EDUCATION . But since this is only slight it suggests that there is no reason to think that the determinants of co-operative breakdown through embezzlement are systematically different from the determinants of other failures of co-operation.

305 has argued that individuals may come to have beliefs about the trustworthiness of others that are apparently confirmed by events, even though the only cause of non-co-operation is fear about others’ trustworthiness. Empirically, it has cited case-studies and has presented primary data from a small sample survey to suggest that co-operation is indeed habit-forming to a degree. Several policy conclusions might tentatively be drawn from this. First, where the promotion of collective action is being considered, it makes sense to tailor its form to that of institutions that have successfully promoted collective action in the past. This does not necessarily mean ‘indigenization’, since indigenous institutions might have acted as an obstacle to co-operation of the requisite kind. Nor does it necessarily mean that pre-existing institutions themselves need to be involved in organizing co-operative action: it may be enough for the co-operation to be organized between people who have previously co-operated; overlapping membership may be all that is necessary. But where indigenous institutions have been successful at promoting co-operation among people whose future co-operation is required, those institutions should be incorporated as far as possible into the new arrangements. Secondly, some communities have a comparative advantage in co-operation. This means that it may be more efficient to concentrate on promoting co-operative institutions in a relatively small number of locations than to spread resources in an indiscriminate attempt to promote co-operation everywhere. Likewise, for watershed management or deforestation programmes it may be wise to concentrate on relatively small and manageable plans of action among groups that can be expected to trust each other rather than require cooperation across a wide range of communities.29 Thirdly, the importance of education and training in the promotion of collective action can hardly be stressed enough. Many developing countries’ primary-education syllabuses contain almost no exercises designed to give schoolchildren experience in undertaking tasks collectively. Whether or not the Battle of Waterloo was won on the playing fields of Eton, the battle of the watershed will almost certainly be won or lost in the schools of the developing world.

REFERENCES ABREU, D., D. PEARCE, and E. STACCHETTI (1990), ‘Toward a Theory of Discounted Repeated Games with Imperfect Monitoring’, Econometrica, 58. ATTWOOD, D. (1989), ‘Does Competition Help Co-operation?’, Journal of Development Studies, 26.

29

For instance, rather than require many villages to co-operate in the preservation of trees in a forest area, it may be wise to divide the area up into plots to which different villages have title. There is still a collective-action problem within each village, but a less serious one.

306 ATTWOOD, D. and B. BAVISKAR (1988), Who Shares? Cooperatives and Rural Development (New Delhi: Oxford University Press). AUMANN, R., and L. SHAPLEY (1976), ‘Long-Term Competition and Game-Theoretic Analysis’, mimeo. BENNETT, J. (1979), ‘Agricultural Cooperatives in the Development Process: Perspectives from Social Science’, California Agricultural Policy Seminar Monograph, No. 4, Davis, University of California (cited in Young, Sherman, and Rose, 1981). BENOIT, J.-P., and V. KRISHNA (1985), ‘Finitely Repeated Games’, Econometrica, 53. BERKES, F. (1989), Common Property Resources (London: Belhaven Press). BRATTON, M. (1986), ‘Farmer Organizations and Food Productions in Zimbabwe’, World Development, 14. CHOPRA, K., G. KADEKODI, and M. MURTY (1989), ‘People's Participation and Common Property Resources’, Economic and Political Weekly, 24. CRAWFORD, V., and H. HALLER (1990), ‘Learning How to Cooperate: Optimal Play in Repeated Coordination Games’, Econometrica, 58. DORE, R. (1981), ‘Introduction’, in Dore and Mars (1981). DORE, R. and Z. MARS (1981), Community Development (Paris: UNESCO). FARRELL, J., and E. MASKIN (1989), ‘Renegotiation in Repeated Games’, Games and Economic Behavior, 4. FEENY, D., F. BERKES, B. MCCAY, and J. ACHESON (1990), ‘The Tragedy of the Commons: Twenty-Two Years Later’, Human Ecology, 18. FRAYSSE, J., and M. MOREAUX (1985), ‘Collusive Equilibria in Oligopolies with Finite Lives’, European Economic Review, 27. FRIEDMAN, J. (1985), ‘Equilibria in Finite Horizon Non-Cooperative Supergames’, Journal of Economic Theory, 35. FUDENBERG, D., and E. MASKIN (1986), ‘The FolkTheorem in Repeated Games with Discounting’, Econometrica, 54. GAIKWAD, V. (1981), ‘Community Development in India’, in Dore and Mars (1981). GILL, M. (1983), Agriculture Cooperatives (New Delhi: Vikas). LELE, U. (1981), ‘Cooperatives and Development: A Review’, World Development, 9. LEVI, Y., and H. LITWIN (1986), Community and Cooperatives in Participatory Development (Aldershot: Gower). LITWIN, H. (1986), ‘Correlates of Community Collaboration’, in Levi and Litwin (1986). NOLAN, P. (1988), The Political Economy of Collective Farms (Cambridge: Cambridge University Press). ONWUCHEKWA, C. (1985), Agricultural Cooperatives and Problems of Transition, University of Stockholm, Dept. of Business Administration. OSTROM, E. (1990), Governing the Commons (Cambridge: Cambridge University Press). PUTTERMAN, L. (1985), ‘Extrinsic versus Intrinsic Problems of Agricultural Cooperation: Anti-Incentivism in Tanzania and China’, Journal of Development Studies, 21. RUBINSTEIN, A. (1976), ‘Equilibrium in Super-Games’, Discussion Paper, Centre for Mathematical Economics and game Theory, Hebrew University of Jerusalem. RUNGE, C. (1986), ‘Common Property and Collective Action in Economic Development’, World Development, 14. SABOURIAN, H. (1990), ‘Repeated Games: A Survey’, in F. Hahn (ed.): The Economics of Missing Markets, Information and Games (Oxford: Clarendon Press).

307 SARGENT, M. (1982), Agricultural Cooperation (Aldershot: Gower). SEABRIGHT, P. (1990), ‘Co-operative Societies and Private Competition: A Theoretical and Empirical Study’, University of Cambridge, mimeo. SEABRIGHT, P. (1991), ‘Quality of Livestock Assets under Selective Credit Schemes: Evidence from South Indian Data’, Journal of Development Economics, 37. SEIBEL, H., and A. MASSING (1974), Traditional Organizations and Economic Development: Studies of Indigenous Cooperatives in Liberia (New York: Praeger), cited in Young et al 1981. SENGUPTA, N. (1991), Managing Common Property: Irrigation in India and the Philippines (New Delhi: Sage Publications). STEVENSON, G. (1991), Common Property Economics: A General Theory and Land Use Applications (Cambridge: Cambridge University Press). THOMAS, B. (1987), ‘Development through Harambee: Who Wins and Who Loses? Rural Self-help Projects in Kenya’, World Development, 15. WADE, R. (1987), Village Republics: Economic Conditions for Collective Action in South India (Cambridge: Cambridge University Press). YOUNG, C, N. SHERMAN, and T. ROSE (1981), Cooperatives and Development: Agricultural Politics in Ghana and Uganda (Madison: University of Wisconsin Press).

12 Efciency Issues and the Montreal Protocol on CFCs ABSTRACT The Montreal Protocol, designed to regulate CFC emissions, is particularly interesting now as a checklist or blueprint for future treaties on global pollution other than CFCs. This chapter takes a look at the properties of the Protocol from the point of view of global efficiency. Although it cannot be proven, of course, that the Protocol deviates from the feasible ‘best’ observing all political constraints as well as all real or perceived transactions costs, it can be shown that the Protocol has a couple of inefficiency properties that could have been avoided through package design: the Protocol is inefficient even if signed by all nations, e.g. because allocation of pollution among signatories is inefficient, especially through the way of compensating developing countries for their adjustments in CFC use. Also, the control variables of the Protocol are imperfectly linked with the relevant target variables, CFC emissions. The Protocol implicitly punishes signatories who like to speed up the process of reducing CFC use. Moreover, the existence of nonsignatories creates loopholes in the Protocol. The chapter also comments on two attempts to model more rigorously some efficiency-related issues relevant to the Montreal Protocol.

12.1 INTRODUCTION In principle, global public bads (GPBs), such as depletion of the ozone layer or global climate change, affect everybody and can be influenced by anybody on the globe. Hence, a constitution, aimed at an efficient level of GPB reduction, would need to involve everybody in order both to heed their valuations and to control their actions. However, even if we had such a constitution or ‘world government’ and designed it according to some definition of efficiency, it could still fail due to deviations from a fully co-operative behaviour on all sides. There are two kinds of uncooperative behaviour: the individual (or the national Government) may cheat or he may openly break the rules of the constitution.

309 In the absence of a ‘world government’, we can have a treaty among national Governments. As one extreme, the implications of a treaty could match that of a perfect world constitution. As another, the treaty could run into three kinds of failure in terms of uncooperative behaviour: in addition to individual signatories (i) cheating or (ii) openly breaking the rules, we have that parties can (iii) choose not to be signatories. The third item may imply that cheating options grow with the possibility of signatories forming coalitions with non-signatories. Furthermore, it may complicate the process for setting rules and, as a consequence, increase the cheating options. Since a treaty among national Governments is the best institution available for controlling GPBs, it would be interesting to know whether there are treaty options with systematic differences in their ability to lead the world economy to a state of efficiency. Relevant information could be acquired from an analysis of existing treaties related to GPBs. More specifically, checking the efficiency problems of existing treaties would be important for two reasons. It would allow an evaluation of the treaty, revealing if possible how well it performs in its role of controlling the GPB; such an evaluation may contribute to a discussion of ways in which the treaty may be improved. In addition, investigating the properties of an existing treaty could contribute to formalizing a checklist that would be useful in drafting similar treaties for other GPBs. In this chapter, we investigate from the economist's perspective—essentially disregarding ‘political constraints’ as well as various enforcement aspects—whether there are any incentives to uncooperative or inefficient behaviour to be found in the 1987 ‘Montreal Protocol on Substances that Deplete the Ozone Layer’ (MP). This treaty was drafted in an attempt to reduce the releases of CFCs (chlorofluorocarbons or ‘freons’) and halons into the atmosphere, hence reducing the risks of depleting the ozone layer. Initially, when CFC technology seemed hard to replace in a number of important applications, especially in foam-blowing and as a solvent or refrigerant, there existed a strong reason for analysing how effective and efficient this treaty version would be in providing incentives for reducing global CFC use. The costs for CFC substitution have now gone down so much (see e.g. Markandya, 1990) that there may no longer be any strong incentives for uncooperative or distorting behaviour here.30 Therefore, and as long as new applications of CFCs do not arise, for which costs of substitution would be significant, the primary reason for analysing the MP and its efficiency aspects, would be to improve our understanding of the problems facing the design of a future treaty combating other GPBs. Specifically, it could provide an input to a treaty on global climate change, an issue which in many ways is similar to the ozone-depletion

30

This does not rule out of course that a change in the global political climate could give rise to uncooperative behaviour among countries which are now signatories to the MP, and that such a behaviour could spread.

310 problem but likely to entail compliance costs of a much higher order of magnitude. The design of a treaty like the MP can be seen as the result of a negotiation game to which game theory could be employed. That is, all clauses of the treaty could be regarded as being parts of some equilibrium outcome of negotiations among a group of players with given objectives. This kind of analysis can be found in the literature (see Barrett, 1989; and Hoel, 1990) and we shall also make reference to parts of that literature in what follows. But as just indicated, the topic of this chapter is simply one which tries to identify ‘counter-productive’ incentives and other causes of inefficiency in the MP, regardless of why this treaty was given the design it now actually has. Had there been only one conceivable hypothesis for the actual treaty design, this choice of topic could be questioned, since it would not lead us to observe that some ‘imperfections’ actually were the result of intentional compromises during the negotiations. It seems pretty safe to argue, however, that there are far more than one realistic and sufficiently well-specified version of a negotiation game leading up to the MP. One special reason for this uncertainty is the fact that external conditions—in particular, the perception of the damage to the ozone layer and of the size of the substitution costs—changed during the negotiation period to an extent that also the relevant game-model may have changed over the negotiation period as a whole. Moreover, at least some parties seem to have felt that time for action was running short and that being able to use as much CFCs as possible in one's own country, given the global CFC-use, was no longer a top priority. In addition, the role of ‘domestic political considerations’—emphasized by Hahn and Richards (1989)—in shaping the positions of the negotiators is hard to incorporate in a negotiation-model. Hence, the treaty as a whole may not be entirely the result of negotiations for which a well-specified game-theoretic framework could be profitably used. The main conclusions of the chapter will probably sound familiar to those well acquainted with efficiency aspects of national environmental policy. The conclusions may be summarized by stating that, although the MP, at the time it was drafted and all things considered, may have been the ‘best’ feasible treaty for controlling CFC emissions, there seems to be some incentives missing, to which the Protocol could have been adjusted to the (potential) benefit of all parties. More specifically, the main points are as follows: 1. The loss of CFC control due to a large number of countries not signing the MP is probably small compared to the inefficiency of CFC control allowed by the Protocol itself. The main reasons are that most non-signatories are developing countries (DCs) which the MP anyway allows a substantial increase in CFC-use in the foreseeable future, and that non-CFC technologies developed in the signatory industrialized countries (ICs) to replace CFCbased technologies are likely to become highly competitive in the long run.

311 2. The MP does not take into account the differences in net costs of CFC-use reductions among signatories; especially when ICs and DCs are compared, these differences are likely to be significant. 3. The Protocol tries to control CFC-use, not CFC emissions; some emissions occur with certainty at the time when the CFCs are used, while others occur with uncertainty some twenty years or so later. Control of crude estimates of emissions, based on the form of CFC-use (thus avoiding prohibitively high costs of administration), would not only be more relevant, but also help to control emissions from CFCs which remain in products and which, sooner or later, may prove to be cost-efficient to have safely destroyed. 4. The MP allows trade in CFC-based products among signatories in a way that discourages them from doing better than required by the MP. 5. Although CFC emissions from countries which are not signatories to the MP are relatively small, the existence of non-signatories creates a possibility for signatories to counteract the spirit of the MP by importing certain CFC-based products from the non-signatories. Sections 12.2 and 12.3 present the general benchmark case of a ‘perfect’ or efficient treaty controlling a global public bad and compare this case with that of having no treaty at all. Section 12.4 summarizes the main properties of the MP. Section 12.5 reviews previous work that attempts to explain why all potential signatories have not signed the MP.31 The question of whether the treaty would have been efficient had all countries actually been signatories is addressed in Section 12.6. How the existence of non-signatories can distort signatory behaviour is discussed in Sections 12.7 and 12.8, while Section 12.9 shows how unconstrained trade among signatories can stop CFC-use from being reduced below the level required by the Protocol. Section 12.10 addresses one aspect of the question of whether the MP may actually prevent a better global treaty from being reached. The MP solution to providing compensation to DC signatories is analysed in Section 12.11. (Table 12.1 gives an overview of the main efficiency issues to be discussed in the chapter.) Conclusions are presented in Section 12.12. Table 12.1 Efficiency issues of partial co-operation (treaty not signed by all): The case of the Montreal Protocol 1. 2. 3. 4. 5.

31

Inefficient even if all had signed? Specifically: Is compensation (redistribution to DCs) inefficiently designed? Are signatories deterred from doing more than required? Non-signatories make loopholes effective? Does a Protocol not signed by all now make future global CFC treaties less effective?

Sect. 12.6 Sect. 12.11 Sect. 12.8 Sects. 12.7, 12.8 Sect. 12.10

We disregard here that the MP is in a dynamic process of being redrafted and—especially for this reason—of having new countries joining the treaty.

312

12.2 THE PERFECT TREATY A treaty which is signed by all nations and which is ‘efficient’ and ‘fair’ will here be called a perfect treaty. For a treaty to promote a globally efficient reduction of a global public bad, it would need to observe abatement costs as well as benefits for all potential signatories in the world. An efficient treaty may then be defined as one which maximizes a weighted sum of global benefits net of global costs, using, for example, some definition of exchange rates as international weights and assuming the benefits of future generations to be incorporated in the benefits as estimated by the present national Governments. Alternatively, if it is not possible to even approximate the benefits of reducing the global public bad, an efficient treaty would be one that minimizes the global abatement costs for any given reduction. Efficiency objectives such as the two now indicated do not necessarily mean that all countries should take abatement actions. Abatement costs in a given country may be so high that the globally optimal abatement policy would require no action to be taken there. Another case is that national abatement costs as part of a globally efficient abatement package do not need to be paid for by the country where the abatement occurs. Strictly speaking, the efficient distribution of abatement activity and the distribution of the financing of this activity are two separate matters. Thus, a country that has low-cost options for abatement may certainly have to use them as part of the efficient abatement activity of a perfect treaty but need not—as a consequence of the chosen definition of fairness or optimal international income distribution—have to pay for (all of) the abatement costs. Conversely, countries that do not have enough lowcost abatement options to meet the abatement requirement for which they have a financial responsibility in a perfect treaty may still have to bear the costs of abatement and, presumably, would only welcome being able to utilize less costly abatement options in another country. Although it may be required that a perfect treaty takes account of equity or fairness issues such as those now indicated, we shall reserve the term ‘perfect’ for global treaties that are efficient, thus assuming them to be independent of the distribution of the financial burden among countries.

12.3 A SIMPLE MODEL Contributions to a global public bad are, in all circumstances, likely to differ among countries. For the specific cases of greenhouse gases and substances that deplete the ozone layer, the relevant country characteristics seem to be primarily population size and stage of development (or GNP per capita). Taking these characteristics into account when modelling optimal or equilibrium behaviour in the global economy places a heavy burden on making such a model transparent and tractable.

313 A model of the global economy that may provide a simple starting-point for analysing perfect treaties as well as imperfect ones is that where all countries are assumed to be identical. Here, we shall make use of a slightly revised version of a model of this kind presented in Barrett (1989, 1990). Following Barrett, we formulate the global efficiency problem as one where total abatement benefits minus total abatement costs are maximized for all countries. (For details see Appendix 1.) Marginal abatement costs, MCi, increase with the level of abatement activity as shown in Figure 12.1. Furthermore, Barrett assumes that marginal abatement benefits, when all countries increase their abatement level concurrently, fall as indicated by MBci for country i, and by MBni, when country i acts alone. The optimal national abatement level at full international co-operation (qco) is that where marginal costs equal aggregate marginal benefits (which in this model of identical countries equal the individual country's marginal benefits of the co-ordinated abatement activity of all countries), MBci. Optimal individual behaviour in the case of no cooperation (qn) is where marginal costs equal marginal benefits of the country acting alone, MBni. As shown in Appendix 1 and as illustrated in Figure 12.1 for two different marginal cost curves, the effect of co-operation on abatement activity, qco − qn, as well as the gains from co-operation, shown by the shaded areas in Figure 12.1, eventually decrease the lower the marginal costs, given the marginal benefits associated with abatement. The Montreal Protocol (UNEP 1987) is so far the only treaty that exclusively deals with a global public bad. Let us now see to what extent this treaty approaches the optimum using this simple model. Fig. 12.1 Gains from co-operation

314

12.4 THE MONTREAL PROTOCOL The aim of the MP is to control the emissions of CFCs and halons, hereafter CFCs for short. The treaty deals with the CFC substances, not with the products made from these substances. In essence, the treaty states that (a) CFC consumption (i.e. production plus imports minus exports of bulk chemical)32 by a signatory should not exceed certain levels after certain dates; (b) ditto for CFC production; (c) a developing country is entitled to delay its compliance by ten years and to reach a consumption level of 0.3 kg. per capita during the first ten years33 to ‘meet its basic domestic needs’; and (d) parties may transfer CFC production reduction requirements among them;34 (e) imports of controlled substances from non-signatories are banned, and products containing CFCs should be listed and no later than January 1993 be banned from being imported from non-signatories; and (f) no developing country may export controlled substances to a non-signatory (beginning January 1993). There are a number of additional rules in the Protocol, most of them rather vague. Some rules are likely to have been amended by the time this is being read. However, the original rules indicated above are the ones which our discussion will focus on. Referring to the model presented above, we may note that, in the Protocol, countries are distinguished as being developing (DCs) and industrialized (ICs). Hence, countries are not treated as being identical. Moreover, countries are distinguished also in that some may choose to be signatories, others not. The fact that not all countries have signed (and ratified) the MP is of course a major reason why this treaty deviates from a perfect one. When the MP was executed in January 1989, almost all ICs, among them the most important user and producer countries, and a minor part of the DCs had signed the MP, covering more than 80 per cent of global CFC-use. Barrett (1989) uses his model to explain this relative success of the MP. Before addressing the questions (a) whether or not the MP would have been a perfect treaty had all countries signed it and (b) what the existence of non-signatories means for the performance of the MP signatories, let us analyse the validity of this explanation.

32

After 1 January 1993 exports to non-signatories are no longer to be subtracted. This, however, means that up to four years after the Protocol took effect, January 1989 being the earliest possible date, signatories are free to export any amount to non-signatories.

33

World average CFC consumption at the time the treaty was negotiated was some 0.25 kg. per capita.

34

The limits to these transfers that originally existed were removed in 1990.

315

12.5 THE MONTREAL PROTOCOL AS AN OPTIMUM TREATY FOR A WORLD OF IDENTICAL COUNTRIES Barrett shows how it may be in the interest of a stable subset of the identical countries to co-operate, subject to the non-co-operative behaviour of the non-signatories. He points out that even if net benefits from (limited) co-operation are small, many countries are likely to co-operate and sign the treaty. In the case of the MP, part of the reason for this behaviour could be that recent evaluations have indicated that marginal abatement costs are small. The fact that so many countries have signed the MP may well be explained by Barrett's argument that the non-co-operative abatement effort is not too far from the globally optimal level. In Figure 12.1, this is illustrated by case 2, where qn and qco are quite close (assuming the number of signatories to approach the maximum). The relevance of the model used here may however be questioned. One prerequisite for the very small effect of cooperation shown in this model is that marginal abatement benefits are significantly reduced by increased abatement and eventually reach zero (see Figure 12.1). However, marginal benefits are unlikely to approach zero even at a 100 per cent abatement level, i.e. with no new CFC emissions. This is because there are (a) significant amounts of earlier emissions that will continue to deplete the ozone layer for a long period to come and (b) other substances affecting the ozone layer that will continuously be released since they are outside the control of the (original) MP or any similar treaty. If the marginal benefits of reduced CFC emissions are nowhere near zero, the result of a reduction of marginal abatement costs may well be that the gains from co-operation will increase. In Appendix 2, this is shown to be consistently so for marginal benefits being constant over the relevant range of abatement activity. The case of an equilibrium number of actual signatories being smaller than the number of potential signatories can arise in the revised model as well, as illustrated in Appendix 3. However, hardly any model, in which countries are assumed to be identical, is likely to be suitable for explaining the actual number of MP signatories. While simple models assuming countries to be identical could give instructive results for specific issues, there are other issues for which actual differences among countries now and in the past play a decisive role. More specifically, observing various differences among countries would not add much insight into the two extreme cases of full co-operation and no co-operation; in characterizing these two cases and the possible existence of stable equilibria for cases of partial co-operation, models with identical countries will do (see Appendices 2 and 3). By contrast, the impact of (large) differences among countries is likely to be significant when analysing specific outcomes between these two extremes, i.e. for imperfect treaties such as the MP. At least four reasons may be highlighted: First, although the opportunity costs of using CFCs are small for most uses,

316 relative opportunity costs may still differ a great deal among uses.35 Thus, differences with respect to volume and sectoral distribution of CFC-uses may give rise to substantial differences in abatement costs among countries. In some cases, efficient abatement of CFC-use can cause considerable social or political costs in terms of labour lay-offs, e.g. when a depressed region is affected and no suitable replacement for CFC-based production can be found. Hence, real costs—especially relative to the size of the economy—may vary considerably among countries. Second, countries may differ as well, and even more, in their benefit valuations of changes in the ozone layer. In particular, the health effects of such changes differ depending on the pigment of the skin and average longevity in the country. (Such differences may be particularly significant between industrialized and developing countries.) In addition, certain parts of the world, e.g. countries close to Antarctica, seem to be more at risk of ozone depletion than others. Governments may also differ in their evaluation of the probabilities of health effects or effects on productivity in agriculture. Hence, differences in benefit estimates may be substantial. Third, an aspect related to the preceding one is that benefits from own reductions of emissions may be significant for some (heavy-user) countries, primarily the USA, and completely negligible for others. In the early phases of negotiations concerning the ozone-depletion problem, the USA accounted for more than 50 per cent of global CFC emissions. Differences in this respect are likely to influence the outcome of the negotiations and the willingness to sign a treaty like the MP. Finally, there are likely to be relevant differences among countries in areas that are entirely outside the simple model and relate to things other than the costs and benefits discussed so far. For historical or political reasons, some countries may choose to act as followers of actions taken by certain other countries and these countries only. As a case in point, DCs have been reluctant to follow the reductions proposed by the ICs since the CFCs which have now accumulated in the stratosphere are the results of actions almost exclusively taken by the ICs for a number of years. Therefore, an invitation to introduce a bound to CFC-use in all countries may be unacceptable to those which have used almost no CFCs in the past. For the reasons mentioned, the willingness to co-operate may vary a great deal among countries. Modelling negotiation behaviour concerning global pollution and pollution abatement when countries differ in relative abatement costs, benefit valuations, and even in their past use of CFC is obviously not an easy task. In addition, the relevance of any such attempt would be difficult to

35

At the time of the original drafting of the MP, appropriate examples would be CFC for aerosol use, where opportunity costs were around zero (with the exception for certain special uses), and automobile air-conditioning, for which CFCs were clearly superior. Note here that some signatories already had abolished the first type of CFC-use by 1986, the base year for CFC reductions according to the Protocol, while others had not.

317 evaluate. Hence, instead of trying to model the behaviour toward ozone-layer protection and remain uncertain about the relevance of such a model for determining the efficiency of the actual design of the MP, we now turn to investigating the efficiency properties of the MP by identifying what incentives to non-co-operative or globally inefficient behaviour the MP actually contains. Does the MP make signatories willing to keep uncontrolled CFC-based imports from non-signatories at a minimum? Does it make signatories willing to use as little as possible of their CFC quotas or does it stop them from unilaterally increasing their abatement activity? And, more generally, does it make signatories attain the resulting reduction in global CFC emissions in the most efficient way? A treaty which does all these things would obviously be a better one than one that does not.

12.6 WOULD THE MP BE EFFICIENT IF SIGNED BY ALL? There can be widely diverging views on what are the national, and hence, the global, benefits associated with global CFC reduction. A number of Governments may argue that the MP is imperfect in the sense that its cutbacks of CFC production and consumption are based on an underestimate of these benefits. Conversely, there are likely to be Governments that do not consider the benefits large enough to outweigh any noticeable sacrifice. These Governments would of course find it better to abstain from signing the MP. Still, the question can be raised: could the MP have been a perfect treaty, if it had been signed by all nations?36 At least three reasons can be given why the answer is in the negative: 1. Noting the clause which allows countries to trade reductions of CFC production among one another (see (d) above), we can see that the MP allows a given (reduced) level of CFC output for the world as a whole and with all nations as signatories to be reached at minimum cost. However, the MP does not allow a similar arrangement for trade in quotas of CFC consumption.37 As is perhaps the most common lesson taught by environmental economics, regulating the size of pollution from different sources without heeding the abatement costs involved, will (except by pure chance) lead to a situation where marginal abatement costs differ among polluters. Thus, it follows that there is no guarantee that global abatement activity would be efficient even if all nations had joined and abided by the Protocol. In fact, as we shall return to in Section 12.11, the MP is supporting CFC use in DCs where abatement costs are relatively small, hence making it unlikely that efficiency will be attained, regardless of the number of signatories.

36

Here, we disregard the fact that the MP does not cover all substances, the emissions of which contribute to the depletion of the ozone layer, and that the use of substances not covered may increase as a result of certain CFC-uses now being controlled.

37

Trade in CFC-consumption quotas is allowed among member countries of the European Community (EC), the reason being that the EC in this respect is regarded as one party to the MP.

318 2. The MP does not provide incentives for all kinds of CFC-release control. CFC releases can be controlled in four ways: by limiting the use of CFCs in the production of new products, by recovering and reusing CFCs from products under repair or servicing, by recovering and reusing CFCs in products when scrapped, and by safe destruction of used CFCs. The major part of CFC releases occurs during manufacturing or when repairing/servicing CFC-based products, and this may continue to be the case for some time. Such releases are controlled by the MP through its limits on CFC-use. Recovery for reuse of CFCs is encouraged by the mere existence of binding constraints imposed on CFC-use; this creates—if not a formal price in the form of a charge or a permit price—a shadow price on CFCs as long as the signatory has not chosen to use only such regulatory instruments (design standards) which leave no room for incentive effects on CFC-reuse. Recovery for destruction would be encouraged in an efficient manner if countries chose to subject CFC destruction to a subsidy equal to the CFC charge (or shadow price). The composition of the four sources to CFC releases would be efficient with such a pricing policy. When charges or permits are levied on the use of new CFCs, this policy can be seen to amount to a deposit-refund system with a deposit (charge/permit price) on new CFCs and a refund on all CFCs recovered for reuse (automatically via the absence of a charge/permit price for reuse) and destruction (the subsidy).38 Destruction of CFCs in scrapped cooling equipment and closed-cell foams does not seem to have been an attractive proposition so far, although pilot-plant testing of various destruction methods have met with some success (UNEP, 1989). This may change over time when CFC-use is further reduced, increasing the marginal costs of CFC-use reductions, hence increasing the optimal subsidy rate for destroyed CFCs. More important, perhaps, is that new and profitable methods for safe destruction (e.g. of halon products, or CFCs in cooling units in areas with a high density of such equipment) are not likely to come forth, rapidly or at all, without a subsidy on this kind of CFC-release control. The MP, however, provides no incentive for signatories, who use but do not produce CFCs, to submit destruction technologies for used CFCs for approval and to use such technologies to attain an efficient level of CFC-release control.39 By contrast, deducting CFCs destroyed—using technologies approved by the signatories—is accepted for determining net CFC production (see Montreal Protocol, Art. 1.5). If signatories were given the right to deduct from their consumption of new CFCs those CFCs which had been safely recovered from old equipment by methods and monitoring as accepted by the MP signatories—thus defining a value of net CFC consumption

38

Cf. Bohm (1981, 1988).

39

This can be contrasted with the incentive effect that the MP has had on R & D of non-CFC technology. This effect, which has been in force at least since the time the MP was drafted, has most likely contributed a great deal to the reduction in costs of such technology.

319 close to the real target variable, CFC emissions—this could have led to a significant change in the composition of those emissions in two ways: First, incentives would then have been provided for CFCs used in products made now, and those remaining in the products until scrapped, to be recovered for safe destruction; this could also influence the design of such products, facilitating future CFC recovery and destruction. Second, the significant amount of CFCs in products existing at the time the MP took effect would also have been subject to an incentive for safe CFC destruction when scrapped.40 In sum, CFC emissions from old sources are only partly covered by the MP, leading to an inefficient composition of measures to reduce CFC emissions in the long run. Providing incentives for efficient reductions of CFC emissions from old sources would allow a higher use of new CFCs without exceeding the limits of total CFC emissions implied by the MP or allow total CFC emissions to be reduced more than has been possible so far. 3. For a treaty to be perfect it must take account of the fact that different activities involving controlled substances may have different potentials for polluting the environment. The levels of production and consumption allowed for in the MP are calculated based on the varying ozone-depletion potentials (ODPs) of the different controlled substances. However, the MP does not observe that the dates at which CFCs are released into the atmosphere differ among CFC-uses. In some cases, the CFCs are released at the time they are used as inputs in commodity production. In other cases (e.g. in foams and as refrigerants), a substantial amount of CFCs remains enclosed in the products for many years, say, up to twenty years. Obviously, the environmental costs of releases today differ from those twenty years later. Aside from ordinary time preferences, one reason may be that we may become more efficient in recovering CFCs from existing products twenty years from now; another is that over time efficient ways may be found to reduce the detrimental effects of increased UV radiation. This could very well warrant significant differences in value among CFC-uses, even larger than some of the ODPs which are observed in the Protocol. Since the MP refers to CFC production dates and use dates, we have that the Protocol is neutral in dealing with the expected release dates for CFCs used for manufacturing and consequently it does not discourage CFC uses that involve early CFC releases. In addition, as discussed in Section 12.9, the MP does in fact discourage signatories from reducing CFC-use below the level required in the Protocol. Thus, we may conclude, that even if the Protocol had been signed by all nations, it would not operate in a way that could be called efficient. We now turn to the question whether or not the existence of non-signatories increases the malfunctions of the Protocol, over and above those now observed.

40

We may add here that, when the MP is amended so that CFCs are to be phased out at some date, CFC releases from old equipment would continue after that date at least as long as there are no incentives for reducing these releases through reuse or destruction.

320 More specifically, does the Protocol in the presence of non-signatories make signatories behave less favourably than if all nations had signed?

12.7 WOULD SIGNATORIES RESPOND TO INCENTIVES TO CHEAT ON THE PROTOCOL? Signatory CFC-related behaviour is of course likely to differ a great deal from that of non-signatories. In a formal sense, a non-signatory country can do whatever its Government wants to do, although in actual practice its behaviour may be influenced by considerations to keep the option open to join later or to avoid running the risk of being economically hurt by ‘retaliatory’ trade policies by signatory countries. Thus, non-signatory behaviour may to some extent be—from an ozone-protection point of view—favourably influenced by signatory behaviour. Signatory behaviour, as we shall see in the subsequent section, can be unfavourably influenced by the existence of nonsignatories, given the loopholes that the MP has created for signatories to ‘cheat’ on the Protocol. Except for those countries that keep their CFC-use strictly below the accepted maximum level (likely to be true at least for some DCs), signatories are subjected to free-rider incentives to cheat. Why would countries, choosing to join the MP, be willing to respond to such incentives? At least three reasons can be given: 1. A signatory who would willingly abide by a stringent MP, only if all countries had co-operated and signed, may no longer be willing to do so when this proviso is not fulfilled and especially when countries in a similar position (say, other NICs) have not signed the Protocol. 2. Signatory behaviour may differ according to the reason why the country has chosen to be a signatory. In Barrett (1989) and Hoel (1990), a country is assumed to join the treaty if abatement costs fall short of estimated environmental benefits. But in addition, a reason for joining may be that countries simply want to be part of a co-operative movement in order to benefit from its side-effects (e.g. avoiding risks of losing partners for other forms of international co-operation). The side-effects of joining the MP could conceivably be greater than the net abatement costs, i.e. the excess of abatement costs over environmental benefits. 3. Another case is that where a country has chosen to join the treaty, but all along has been endorsing its own version of an optimal treaty—one which allows an abatement level well below that of the MP. Such a signatory may be tempted to cheat, owing to pressing domestic problems and if the ways to cheat are subtle and palatable. We shall not deal here with cheating in the form of false reporting of CFC-use rather focus on the ‘loopholes’ for circumventing the spirit of the MP.

321 This can happen through CFC-related trade between non-signatories and signatories which the MP has failed to consider.

12.8 THE ROLE OF THE EXISTENCE OF NON-SIGNATORIES FOR SIGNATORY ‘CHEATING’ BEHAVIOUR? Being able to trade with non-signatories opens up some possibilities for signatories to circumvent the spirit of the MP. To illustrate, let us take a version of case 3 just mentioned. This is the case of a country that has accepted having its CFC-based production activity constrained by the MP and therefore has joined the treaty, but finds the MP requirements ‘too demanding’. Say the Government of this country would have liked the MP to be less stringent towards countries of the type this signatory belongs to, e.g. the DCs, and thus would like to do less of a reduction of net CFC-use if it somehow could ‘get away with it’. According to the MP, a signatory can, at least for the time being, import CFC-based products from non-signatories. Here, we focus on such products which are produced with, but do not contain, any CFCs.41 The MP states (Art. 4.4) that signatories shall, no later than 1993, determine ‘the feasibility of banning or restricting the imports’ of such products from non-signatories. However, the Protocol does not state when actions would be taken, should banning or restricting imports be found feasible. Thus, for a significant period of time, the MP allows signatories to meet the MP requirements by reducing CFC-use in this type of production and instead import the products from a non-signatory.42 Over a longer period of time, this trade activity could also encourage producers to leave the signatory home country and establish production in a non-signatory country, then from there export the products back to the signatory home country or other signatories. The kind of trade activities now mentioned can take place without the Government of the signatory country knowing it or doing anything about it. Or, the Government could be actively supporting it. One reason for doing so may be that influential domestic producers and/or employment in the country are dependent on the supply of the CFC-based products and that the Government is hard pressed to accommodate such interests. Could this loophole have been eliminated? There is no doubt that policy actions of the type referred to in the MP, Art. 4.4, are feasible, although all parties may not agree to implement them now. One example is to have potentially CFCbased products be treated as actually CFC-based products as long

41

The monitoring problem is much smaller and the period for unregulated imports is much shorter for products containing CFCs (see MP, Art. 4.3). Therefore, we abstract from this case here.

42

Production activities possibly relevant for this kind of trade are cleaning or sterilization of electronic products and the manufacturing of certain foam products.

322 as signatory importers cannot present proof of the contrary as obtained from their trading partners. (Given that the MP bases compliance checks in general on self-reporting, enforcement techniques and enforcement costs would not be a major problem here.) Such a rule should probably have been introduced in the ‘package’ of the original MP, since later on it may not be feasible to get sufficient agreement on a separate point like this one. (This, of course, does not necessarily mean that the original number of signatories would remain the same.) Another way, similar to that now discussed, to circumvent the MP is for signatories to replace CFC-use in the manufacturing of certain products—which are to be exported to non-signatories—by having the products, or relevant parts in them, manufactured in the importing countries.43 Also this loophole could have been reduced by trade restrictions, e.g. banning signatories from exporting products designed for CFC-use to non-signatories.44 To sum up the main argument here, we have seen that the existence of non-signatories presents some trade opportunities involving CFC-use that signatory countries may want to take advantage of, and by doing so obstruct the spirit of the MP. We will now show that the opportunities for trade between signatories allowed by the MP also present obstacles for an efficient international abatement policy.

12.9 INTERSIGNATORY TRADE DISCOURAGING UNILATERAL REDUCTION OF CFC-USE An efficient treaty (disregarding enforcement costs) would see to it that signatories are not prohibited from undertaking CFC abatement, the full costs of which they are willing to bear. As a case in point, Sweden has decided to do more than the MP requires it to do by speeding up its reduction of CFC-use and to phase out CFCs by 1995; to this end, it is instituting bans and similar regulatory instruments to stop Swedish firms from using CFC-based technology. Firms using CFCs, e.g. as a blowing agent for making foams, are

43

US congress (1988) gives an interesting example, where air-conditioned cars exported to other countries could have the air-conditioners charged in the importing country. Another example is when ships from signatory countries have servicing operations, involving CFC (and halon) use, moved to harbours in non-signatory countries.

44

It should be noted that this problem need not vanish completely, if all countries were signatories. With signatories having excess CFC-use capacity—especially DCs since they were given extra capacity by the MP—it would not. Excess CFC-use capacity would provide an incentive for a country to start CFC-based production for exports. (In fact, this is still a possibility even after the explicit proposal by some signatories to let DCs expand their CFC-use also for exports was rejected.) Although the MP formulation, saying that DCs may delay compliance by ten years ‘to meet its basic domestic needs’ (Art. 5.1), now has been clarified to mean ‘not for exports’, this can hardly guarantee that additional CFC-use is not directly or indirectly linked to increased exports. What has been said now also forms a basis for arguing against providing compensation in the form of extra room for future use of CFCs to countries that so far have not used much CFCs (see Section 12.11).

323 required to shift to non-CFC-based techniques, of which only more expensive or less attractive substitutes are available. Since the CFC-based products now could be imported from other signatory countries, albeit at a higher price than that previously charged by the Swedish producers, but still below the competitive price of the new substitutes, firms as well as environmental policy administrators complained. Not allowing a signatory in this situation to control imports from other signatories thus threatens to eliminate the new substitute products that the signatory is trying to encourage in the spirit of the MP. Not only does this outcome deter others from following the example set for substituting certain CFC-based products, which must be the main purpose of the measure taken by the signatory, it may also discourage signatories from taking similar steps for other CFC-based products. It should be noted here that this is a problem which obviously would not disappear even if all countries were signatories. Thus, a treaty being complete in this sense would not have helped. But could the MP have been designed so as to avoid the counteracting effect on initiatives for promoting the replacement of CFC-based technology by substitutes? If it could, the treaty design would not have stopped countries from exceeding the ambitions of the MP and possibly from playing a lead role in increasing the speed by which CFC-use is cut back, assuming that the aggregate reductions advocated by the MP fall short of a globally efficient level. Allowing the use of an import ban would promote efficiency in this case and remain in line with the GATT rules which recognize a country's right to use trade barriers to protect its environment as long as this does not favour similar goods produced domestically (cf. OECD, 1989). The MP could therefore have explicitly stated the rights of signatories to introduce trade barriers to protect their attempts to reduce CFC, as long as such trade restrictions are compatible with existing treaties. In addition, given the ‘vague and cursory wording’ (OECD, 1989) of the relevant GATT Article XX, the interpretation of this rule may have turned out more favourable for the objective of the MP, had the Protocol provided this backing of trade control.

12.10 DOES THE MP PRECLUDE REACHING A FUTURE OPTIMAL-FEASIBLE CFC TREATY? A potential problem with the MP is that the early commitments of the signatories may have put them in a worse position when bargaining with the non-signatories to achieve a truly global CFC treaty. As shown by Hoel (1990) for a two-country model, where the ‘unselfish’ country 1 has taken unilateral actions to reduce its emissions, it may no longer be possible to reach, in a second step, the co-operative solution with the ‘selfish’ country 2 that would have been attainable in the absence of the unilateral action and that would have offered a lower level of total emissions.

324 As a simple illustration of this outcome, let group S represent the group of early MP signatory countries (essentially ICs) and group NS represent the non-signatory countries (mainly DCs).45 Let us say that, in 1986, before the MP was drafted, group S accounted for 90 emission units and group NS for 10 units. Furthermore, let us assume that the predicted levels for 1995, are 50 units for each of the two groups, whereas the emission levels in the absence of the MP would have been 100 units and 50 units, respectively (see Table 12.2). Table 12.2 Emission levels (in ‘units’) for two groups of countries, the original MP signatories (S) and the nonsignatories (NS). Group

(I) With MP S 90 50 5

1986 Predicted for 1995 Global treaty for 2000 and on Global emission level 30

(II) Without MP NS 10 50 25 15

Now, assume that, for year 2000 and on, group NS can offer at most a 50 per cent cut in its emissions (i.e. from 50 units to 25) in exchange for a reduction by group S from 50 to 5 units (i.e. by 90 per cent from the 1995 level). If no unilateral action had been taken, i.e. in the absence of the 1987 MP, group NS would have accepted to undertake, we assume, an 80 per cent reduction (from 50 to 10 units) in exchange for a 95 per cent reduction by group S (from 100 units to 5). Thus, with the 1987 MP, the eventual global emission level would be 30 units in contrast to only 15 units for the case without the MP. While there is a risk that the outcome of the MP is negative as illustrated here, the relevant question is what role, if any, should be given to this risk when drafting treaty proposals like the MP. For example, should the requirement of a minimum number of signatories for the MP to take effect have been increased, say, to cover all ‘potentially important’ countries? Hoel's conclusion is that a country should give ‘careful consideration’ to this risk before it ‘departs from following its own self-interest’. Putting this advice into practice is none the less hampered by the presence of a number of other possible implications of ‘unilateral’ actions, most of which are cited by Hoel but, for obvious reasons, left out from his bargaining model: 1. The action would set an example, influencing the behaviour of the rest of the world in the direction of increasing its willingness to take similar actions. This may be particularly relevant for our application of the Hoel model to the

45

Hoel acknowledges that his analysis also covers the case where a ‘country’ is ‘a group of countries which have co-ordinated their policies’.

325

2.

3. 4. 5.

MP case, given the significance of the ‘founding fathers’ of the MP in terms of CFC production and consumption. (Behaviour is, of course, likely to be influenced also by considerations working in the opposite direction, implying that if country 1 reduces its emissions there is less need for country 2 to do so, assuming marginal benefits of reduced emissions are decreasing; this effect, however, is already incorporated in Hoel's model.) The fact that a country or a group of countries documents a willingness to make sacrifices is likely to improve the climate for international negotiations and the chances for reaching a co-operative solution later on. This would be particularly important where all those countries that have been significantly contributing to the ozonedepletion problem over the years are signatories to the MP and those which have not are non-signatories (see also Bohm, 1982). Taking action makes it clear to others that the actor estimates the costs of reductions in CFC use to be in some sense surmountable. Furthermore, the practical implications in this respect would become more conspicuous as time passes. (Points (1) to (3) are observed in Hoel, 1990.) Country 2, or the non-signatories of the MP, may eventually become aware of the retaliatory power built into the Protocol (concerning CFC-related trade and certain forms of technology transfer to non-signatories, see MP, Art. 4) and other unfavourable trade measures that signatory countries may take. Most important perhaps is that, since CFC emissions have accumulative environmental effects, lasting far into the future, an early start in reducing CFC emissions is valuable in itself. Even if there are assurances of reaching a treaty better than the MP, the value of this time gain from unilateral action, may still outweigh its drawbacks.

The aspects now mentioned are not only of potential importance in the case of the MP, but rather likely to have been among the driving forces behind the drafting of the MP. Even more relevant for our discussion is that these aspects must be considered when evaluating the efficiency of the Protocol. In that perspective, it hardly seems justified to criticize the MP for making an eventual, truly global, treaty less efficient.

12.11 DOES THE MP PROVIDE EFFICIENT COMPENSATION TO DCS? Forming a treaty to control CFC emissions raises, as mentioned earlier, a distribution problem regarding the size of the sacrifices to impose upon the different signatory countries. The MP deals with this problem in three ways: 1. through the choice of basic control measures, in particular the pace at which CFC-use and production are to be reduced, implying a specific

326 distribution of sacrifices among the countries to which these basic measures apply (MP, Art. 2); 2. by singling out a set of countries, namely DCs as defined by the Protocol, for a less demanding set of the control measures, essentially a ten-year delay of compliance, during which period annual CFC-use may rise up to 0.3 kg. per capita (average DC use being at most some 0.05 kg. per capita in the mid-1980s, see Markandya, 1990); and 3. by calling upon the parties to the Protocol to facilitate bilaterally or multilaterally the provision of aid to signatory DCs for a transfer to technologies alternative to those based on CFCs (MP, Art. 5). Here, we shall discuss the form of compensation mentioned under (2), given the choice of basic measures under (1) and given the group of countries to which the less demanding set of rules should apply. But first, it should be noted that the extent of the concessions mentioned under (2) is quite significant in terms of the additional CFC-use permitted. Thus, even if all DCs were to sign the MP, their CFC-use could increase substantially such that global CFCuse by 1998 could be much larger than that of 1986—the year on which the MP bases its reductions of CFC-use for ICs (to 50 per cent by 1998).46 For example, if a set of DCs with a population of 3 billion people had signed the MP and attained their maximum permitted CFC-use level by 1998, their use of controlled substances would equal approximately that of the whole world for 1986. Second, it should be noted that the reductions for DC signatories after 1998 are to be based on their average annual CFC-use for the 1995–7 period. In this way, the MP provides incentives to DC signatories, who like to keep their CFC-use options on a high level after 1998, to exceed their otherwise optimal CFC-use level for 1995–7, wherever this level is not effectively bound by the 0.3 kg. per capita limit. An alternative to the MP would have been to let post-1998 CFC-use be based on (but possibly being many times larger than) some historical CFC-use figure, say, the CFC-use level in 1986 (as is the case for IC signatories). However, in the discussion below, we assume for simplicity that, without the special treatment described in (2), DC signatories would have had to follow the same rules as those for IC signatories. With the MP giving DCs a compensation in the form of the special treatment stated in (2) above, the question to ask is whether this is in fact an efficient form of compensation. As an alternative, we shall consider offering signatory DCs a money transfer equal to the estimated value of the delayed compliance to the Protocol. (We abstract here from effects on CFC-related trade with non-signatories.) Assume, to begin with, that an accurate prediction could be made of the CFC-use in individual DCs as well as of the cost of replacing CFC by

46

As noted earlier, the signatories are in a continuous process of adjusting the MP reduction targets. In 1990, these targets were significantly augmented.

327 non-CFC technology.47 Given that the objective is to improve the welfare of the signatory DCs from what they would have attained if they were not given the ‘favour’ of delayed compliance, we can note that a money transfer equal to the cost difference would represent an overcompensation. The reason is that the recipients of the transfer are likely to prefer using part of the transfer for other things than the end-products (primarily refrigeration) for which the CFCs would have been used. Given that this overcompensation amount is easier to compute than the exact compensation and that this amount is likely to be the only feasible version of a money transfer of the type now suggested, the question is, would the (signatory) ICs prefer to pay that amount in exchange for an increase in their CFC-use equal to the reduced CFC-use of the signatory DCs under this scenario?48 The answer is ‘yes’ if the IC willingness to pay for the extra amount of CFC-use exceeds the DC's extra costs for giving up that amount of CFC-use (i.e. if IC costs of giving up that amount of CFC-use is higher than that of the DCs). The efficiency problem now presented is illustrated in Figure 12.2(a) and (b). Here, we have the total global signatory CFC-use, say for the period 1989 to 1998, shown by qNMP in the absence of the MP (Figure 12.2(a)) and by qMP when the MP is in force (Figure 12.2(b)). In the latter case, CFC-use for the ICs are limited to qIC*, while assuming that the room for DC use of CFCs provided by the Protocol is still large enough for them to use CFCs up to the point, qMP − qIC* where the marginal benefits of such use is zero. From this it follows that there is at least a volume qopt − qIC* that should be transferred from the DCs to the ICs to obtain an efficient distribution of the given global CFC-use volume, qMP. At least part of this efficiency gain would have been achieved by identifying the uniform CFC-use reduction path of the alternative version of the MP that would have approached the efficient CFC-use distribution, qopt and qMP − qopt. Hence, the change of compensation system considered here would be called for. It should be noted that in addition to the reason already pointed out, there is at least one other that contributes to making marginal-use (net) benefits relatively much larger for ICs than for DCs over the relevant CFC-use range. Before the implementation of the MP, ICs already had in place a production capacity adjusted to CFC technology, part of which would become unused as a consequence of joining the MP. DCs, in contrast, will have to increase their capacity of production using CFCs to obtain their optimal CFC-use level under the MP. On the other hand, costs of shifting to the new non-CFC technology would probably not differ very much between ICs and DCs, at least

47

For a recent elaborate calculation of this kind, covering the period up to 2008, see Markandya (1990).

48

We assume that the set of DC signatories will be at least the same as that of the present MP after a change in the form of compensation.

328

Fig. 12.2 Efficient distribution of CFC-use: (a) No MP; (b) MP at work

not to the extent that would outweigh the short-term capital cost difference now discussed. Returning to one of the crucial assumptions made earlier, it is of course not possible to estimate the exact CFC-use in DCs for long periods like ten to fifteen years or more. It would be even more difficult to estimate cost differences for CFC and non-CFC technologies over such long periods (see, however, Markandya, 1990). Thus, had the parties that drafted the MP in the mid-1980s tried to do this, the result would have been much less accurate than now assumed. However, this uncertainty also affects the MP as it came to be designed in the sense that the implications of Article 5 (see (2) and (3) above) for DC CFC-use, and hence for global CFC-use, could not be known. Since the uncertain outcome of the MP in this respect proved to be acceptable to the parties, the uncertain adequacy of a money-transfer mechanism of the type discussed here might also have been accepted. This is especially so, since the alternative arrangement would have the important effect of making the treaty effect on global CFC-use from signatories known with certainty (disregarding the problem of inaccurate international statistics in this field, see US congress, 1988). The compensation approach discussed here involves estimating the costs saved for ICs being able to raise its CFC-use to an extent equal to the extra

329 DC CFC-use arising from (2). These cost savings should have been relatively easy to estimate with reasonable precision, granted that the proposed MP was already available and was confronted with the alternative discussed here. Given then an approximate estimate of the willingness to pay from individual ICs as their minimum contributions (aside from those envisaged under (3)) and given the advantage of knowing the ‘exact’ CFC-use to be expected from DC signatories, amending the MP in the way discussed here can be expected to have been politically acceptable. Two further aspects need to be observed. One is that the relevance of the estimates of what individual ICs should be ready to provide as a minimum to an international fund for transfers to DC signatories, would be jeopardized if the ICs withdraw part or all of that money from ordinary aid to DCs. While this is possible, similar incentives for withdrawal of aid could be expected to arise from the relatively tougher standards that the present MP imposed on its IC signatories and the preferential treatment accorded to DCs. The other is a question of how the fund should be allocated among the DCs. This distribution problem, however, will also be encountered under the present version of the MP. As noted in (3) above, the MP parties are to decide how much aid should be provided to signatory DCs in order to assist them in shifting to non-CFC technologies. Thus, while the distribution problem may be less significant and arise at a later stage, it is an issue that also the MP must address. The difficulties now suggested would be significantly reduced, if not eliminated, by letting the CFC-use quotas allowed by the MP be tradable. The DC quotas should be based on some estimate of their expected CFC-use under the present version of the MP and not on the maximum of 0.3 kg. per capita. To illustrate, say that there only two countries, IC and DC, which the MP allows qIC* and qMP − qIC*, respectively. IC may now buy qopt − qIC* from DC at a total price, at least equal to the minimum compensation required by DC. In the real world, with a large number of traders/ signatories (and, if necessary, some protection against imperfect competition), a market price equal to p in Figure 12.2(b) is established, determining the actual compensation to DCs as equal to area B in the figure. (Area A is the gain of the ICs.) Here, we accomplish both the certainty of the volume of total CFC-use for signatories and an efficient allocation of this volume. Two other possible implications of this approach should be mentioned. First, it is possible that DCs which have remained non-signatories may have been attracted by this version of the treaty, if they perceive that it offers them overcompensation (as defined) in convertible currency. Second, for signatories who in these new circumstances are willing to make the same sacrifices as those they accepted when they joined the MP, further reductions of total CFCuse (below qMP) could be accomplished. To sum up, we have shown that ICs, as far as economic principles go, would be willing to pay an amount sufficient to compensate DCs for lowering their

330 CFC-use from the level permitted by the MP; this is in addition to the aid to facilitate the replacement of CFC-based technologies required by the MP. As a ‘by-product’, better information would be available concerning CFC emissions from signatory DCs. Thus, from the point of view of efficiency, it is hardly possible to find the MP solution to the distribution problem to be efficient.

12.12 SUMMARY AND CONCLUDING REMARKS While the Montreal Protocol is admired by many (including the present author) as a remarkable achievement of international negotiation for controlling global pollution, it has come under some criticism. Thus, Barrett (1989) argued that the MP was made possible mainly because its effects on the environment and the sacrifices required to implement its rules were so small. Hoel (1990) has developed a model showing the risks that unilateral actions by a country or a group of countries, which may be interpreted as the MP signatories, can worsen the prospects for a true global treaty. The relevance of these two results has been questioned here, however. The first result is questioned because the benefits are unlikely to have been represented in an appropriate fashion in Barrett's model and also because the uniform-country model does not permit the proper evaluation of the MP, considering that there are major differences among countries as far as CFC-use is concerned. The relevance of the second result is questioned because the negative effects implied by Hoel's model are likely to be outweighed by the positive effects of a ‘unilateral’ action that can come about through the MP. Here, the efficiency of the MP has been questioned on a number of points, most of them summarized in Table 12.3. In particular, we have seen that the MP discourages signatories from reducing their CFC-use below the maximum level permitted by the MP and from reducing their net CFC-use by bringing about efficient methods for destruction of CFCs in scrapped products. We have seen that there are both short-run and long-run incentives to avoid doing better and even to do worse (via trade) than that intended by the MP. In the short run, signatories are likely to refrain from being more ambitious than strictly required by the MP due to competition from CFC-based production in other signatory countries. Moreover, they are induced to buy CFC-based products from non-signatories and from signatory DCs with CFC-use quotas in excess of their own requirements. In addition, incentives to develop methods, and invest in plants, for CFC destruction have been kept at a suboptimal level, while at the same time incentives have been present for CFC-based firms in signatory ICs to move to non-signatories and signatory DCs. Moreover, we have pointed out that by basing permitted DC use of CFCs after 1998 on the level of use during a future period (1995–7), instead of a

331

Table 12.3 The Montreal Protocol: summary of main conclusions 1.

2.

3.

4.

5.

Inefficient even if all had signed? Yes, allocation of CFC-use among (IC) signatories is inefficient and control variables (use and production) are not firmly linked to target variables (emissions) Is compensation (redistribution to DCs) inefficiently designed? Yes, future CFC-use quotas are based on data the DC signatory can control and DC compensation could have been made larger and/or IC sacrifices smaller by allowing quota transfers or quota trade Are signatories deterred from doing more than required? Yes, signatory trade counteracting unilateral attempts to further reduce CFC-use is not allowed to be regulated Non-signatories make loopholes effective? Yes, non-signatories can increase the output of products, made with but not containing CFCs, for trade with signatories, hence allowing signatories to use their CFC quotas for other products Does a Protocol not signed by all now make future global CFC treaties less effective? No, hardly to an extent that would make the Montreal Protocol not worth while.

historical period, the MP could induce DC signatories to increase their CFC-use. The preferential treatment given to DCs which is allowed for in the MP in tackling the distribution problem, has been found to be inefficient. It is inefficient because the resulting global CFC emissions could have been obtained at smaller costs to the ICs or with larger compensations to the DCs or both, by transferring CFC emissions from DCs to ICs so that the marginal willingness to pay for CFC-use would tend to be equalized between the two groups. Alternatively, for the same level of national sacrifices that the MP imposes on the parties, a lower global CFC-emission volume could have been attained. Compensating the DCs by money transfers equal to the value of CFCs for redistribution to the ICs would also permit that the maximum future CFC-use by signatories be known with some certainty; this is not possible under the concessions currently used in the MP. This last observation suggests that the MP does not attempt to minimize the costs of global CFC-use reductions in a way that would have been formally possible, e.g. with a system of globally tradable CFC-use permits. There may have been strong political reasons for not doing so, but it is far from clear that the implicit costs of these reasons have been fully taken into account. It may be argued, of course, that since the MP is still in the process of being redrafted, obvious inefficiencies can be corrected later. However, it must be borne in mind that, given the initial design of the treaty package, some piecemeal adjustments for improving the package may no longer be feasible. Fixing the starting-point for future negotiations by choosing a particular imperfect treaty design creates a set of vested interests that may preclude such adjustments from being feasible. Therefore, it remains important to have the

332 efficiency properties of alternative initial treaty designs carefully investigated before one is selected.

12.13 EPILOGUE: THE 1990 REVISIONS OF THE MONTREAL PROTOCOL The Montreal Protocol of 1987 was explicitly designed as a flexible and dynamic treaty to be amended over time. This is likely to be an attractive feature from several points of view, not least with respect to efficiency. The first revisions were made in London in June 1990. A main objective of the London meeting was to attract additional signatories, in particular China and India, whose CFC-use so far has been small, but is expected to grow rapidly. By the time of the meeting, ‘Fifty-eight governments plus the European Community, representing 99 per cent of world production and 90 per cent of consumption, had ratified or acceded to the Protocol’ (Benedict, 1991). Of these, twenty-eight were developing countries. •

• •

A fund, to be administered by the World Bank, was set up to help finance projects and programmes to meet the incremental costs of developing countries complying with the Protocol. An amount of $US160 million was agreed upon for an initial three-year period, plus an additional $US 80 million if China and India soon became signatories. As a result, these two countries have now signed the Protocol. Governments of IC signatories accepted a responsibility to transfer new technology to DC signatories. Additional substances that deplete the ozone layer were added to the Protocol: ten new CFCs, carbon tetrachloride, and methyl chloroform. For the original CFCs and halons, reductions and the eventual phase-out of production and consumption should be stepped up.

The efficiency problems of the original protocol as discussed here (cf. points 1–4 in Table 12.3) remain. However, the fact that additional countries have now joined the Protocol means that the problem connected with the existence of non-signatories (point 4 in Table 12.3) has been reduced. To the remaining efficiency-related issues should be added the following two, highlighted at the London meeting: • •

No agreement was reached on the control of HCFCs, another ozone-depleting substance, to which signatories are now shifting to some extent. The data on controlled substances produced, exported, and imported, as reported by many signatories, were found to be inadequate.

Appendix 1: The Identical-Country Model Barrett (1989) formulates the global-efficiency problem as one where total abatement benefits minus total abatement costs,(A1.1)

are maximized for the given number of countries, N. Here, benefits of the individual country, Bi’ are a function of total abatement , whereas abatement costs, Ci, are a function of the country's abatement level, qi. Using for simplicity quadratic cost and benefit functions,

and

respectively, we maximize (A1.1) with respect to Q for (qi = Q/N), producing the familiar first-order condition of public goods (here abatement of global pollution) saying that the marginal global benefits should equal marginal national abatement costs:(A1.2) Thus, under full co-operation (co), each country should be required to set its level of abatement at

This blueprint for a perfect treaty can be contrasted to the treaty-free Cournot equilibrium, where every country maximizes its abatement level given that of all other countries.(A1.3) which yields the non-co-operative abatement level, qn, where each country's marginal benefits of its own abatement activity equal its marginal abatement costs, i.e.

That is,

which is clearly less than qco for N > 1. The difference in abatement level between the perfect-treaty and no-treaty cases is

We may note right away that this difference is small for c/b close to zero or to infinity, i.e. for relatively insignificant marginal abatement costs or relatively insignificant abatement benefits.49 In Figure 12.A1, where MBni are the marginal benefits of 49

Note that changes in c and b imply proportional changes in marginal costs and benefits.

334

Fig. 12.A1 Unilateral and co-operative abatement adjustments

unilateral abatement adjustments (i.e. when all qj are given, equal to qn) and where MBci are the marginal benefits when all countries adjust under full co-operation (i.e. qi = qj), we show qco and qn for three levels of c, given b (see cases (1), (2), and (3)). (Each non-co-operative equilibrium is shown by the intersection of MBni and MCi; the co-operative equilibrium is shown by the intersection of MBci and MCi.) Here, we can see that country net benefits from cooperation are small when c is small or large relative to b. (Compare the shaded areas with the striped one, illustrating the case of middle range c/b values, given a and N.)

Appendix 2: Constant Marginal Abatement Benets Here, we treat total abatement benefits for the individual country as equal to bQ. Hence, in a world of identical countries, global benefits equal NbQ. With costs as in Appendix 1, we maximize

with respect to Q to get the optimal abatement activity under full co-operation:

For the case of no co-operation, we maximize

with respect to qi, with optimal abatement activity without co-operation given by

From this we have that the abatement effect of co-operation,

grows consistently larger the smaller the marginal costs or the cost parameter c. (See Figure 12.A2 for N = 3 and c(2) > c(1).) This is contrary to the case with decreasing marginal benefits and perhaps intuitively more appealing. Fig. 12.A2 Abatement effects of co-operation

(We can easily see here that no additional information of relevance for the issues now discussed would be obtained by assuming that the countries were different, say, with N = 3 and b1 > b2 > b3 and c1 < c2 < c3.)

Appendix 3: An Equilibrium Number of Signatories below Full Co-operation A simple illustration of the case where we have an equilibrium number of signatories NS, where 1 < NS < N, is shown here. Assume N = 3. Benefits (B) for each country is as in Appendix 2

Marginal costs are

, see Figure 12.A3 for an example.

Fig. 12.A3 Equilibrium number of signatories

(1) Full co-operation implies, as before, marginal global benefits equal to marginal costs and hence

(2) No co-operation is, as before, given by marginal costs equal to marginal benefits of the country moving alone, i.e.

(3) If two countries departing from qi (1) = qn, i = 1, 2, would start to co-operate with no. 3 not co-operating, thus q3 remaining at qn, qi would shift to

See Figure 12.A3 for an illustration. For qi(2) as in this illustration, this shift is worth while for nos. 1 and 2 (and even better for no. 3, the non-co-operating country). (4) If no. 3 wants to join the two others, it would first have to take the step up to the abatement level nos. 1 and 2 have agreed on, qi(2). This implies a net loss (I) for no. 3, of course, since the new level deviates from the optimum level, qn; see the shaded area

337 (I) in Figure 12.A3. As the next step all three move to qco, the optimum abatement level when all countries co-operate, which implies a gain (II) for each country; see the shaded area (II) in Figure 12.A3. Hence, it pays country No. 3 to join if the net effect of the two steps is positive. If the kink in the marginal cost curve is outside the qco − qn interval the net effect is zero. If it is inside this interval, it follows from simple geometry that the loss in step I exceeds the gain in step II. We can use this model to show how to analyse the issue, when it pays a country to defect from a treaty. Assume to begin with full co-operation among the three countries. Defection (a move from qco to qn) pays, say, for country No. 3 (a) if the remaining signatories adjust to their now optimal abatement level (qi(2)), exactly for the reasons stated under (4) above, and (b) even more, if the remaining signatories would stay put at their now suboptimal abatement level qco. However, defection does not pay if co-operation breaks down altogether as a result of the defection (a variation of the prisoners’ dilemma); each country now loses an amount corresponding to areas III and IV in Figure 12.A3. The model can be used also to show how it could pay signatories to bribe a non-signatory party to join the treaty or a particular signatory not to defect. Each signatory, Nos. 1 and 2, gains areas II and III from No. 3 joining (or not defecting in case (a)) which is far more than needed for compensating No. 3. The results presented in this appendix hold even for the case of more countries than three and even if there are nonsignatories aside from the countries under discussion.

REFERENCES BARRETT, S. (1989), ‘On the Nature and Significance of International Environmental Agreements’, Working Paper (London: London Business School). BARRETT, S. (1990), ‘The Problem of Global Environmental Pollution’, Oxford Review of Economic Policy, 6. BENEDICT, R. E. (1991), Ozone Diplomacy: New Directions in Safeguarding the Planet (Cambridge, Mass. : Harvard University Press). BOHM, P. (1981), Deposit-Fund Systems: Theory and Applications to Consumer Policy (Baltimore: Johns Hopkins University Press). BOHM, P. (1982), ‘CFC Emissions Control in an International Perspective’, in J. Cumberland et al. (eds.) The Economics of Managing Chlorofluorocarbons (Washington, DC: Resources for the Future). BOHM, P. (1988), Economic Instruments for Reducing CFC Emissions (Copenhagen: Nordic Council of Ministers). GRUBB, M. (1989), The Greenhouse Effect: Negotiating Targets (London: Royal Institute of International Affairs). HAHN, R. W, and G. MCGARTLAND (1989), ‘The Political Economy of Instrument Choice: An Examination of the US Role in Implementing the Montreal Protocol’, Northwestern University Law Review, 83. HAHN, R. W, and K. R. RICHARDS (1989), ‘The Internationalization of Environmental Regulation’, Harvard International Law Journal, 30.

338 HOEL, M. (1989), ‘Global Environmental Problems: The Effects of Unilateral Actions taken by One Country’, Dept. of Economics, University of Oslo, mimeo. HOEL, M. (1990), ‘Efficient International Agreements for Reducing Emissions of CO2’, Dept. of Economics, University of Oslo, mimeo. MARKANDYA, A. (1990), ‘The Cost to Developing Countries of Entering the Montreal Protocol’, Dept. of Political Economy, University College, London, mimeo. MORRISETTE, P. M. (1989), ‘The Evolution of Policy Responses to Stratospheric Ozone Depletion’, Natural Resources Journal, 29. UNEP (1987), Montreal Protocol on Substances that Deplete the Ozone Layer: Final Act 1987 (Nairobi: UN Environment Programme). UNEP (1989), Report of the Technology Review Panel (Nairobi: UN Environment Programme). UNEP (1990), ‘Transfer of Technology and the Financing of Global Environmental Problems: The Role of Users' Fees’, Discussion Paper (Nairobi: UN Environment Programme). US Congress (1988), ‘An Analysis of the Montreal Protocol on Substances that Deplete the Ozone Layer’, Staff Paper (Washington, DC: Office of Technology Assessment).

13 CO and the Greenhouse Effect: A Game-Theoretic Explorathion 2

13.1 INTRODUCTION Some of the most serious current environmental problems have a global character. Perhaps the most serious environmental problem in the next century will be climatic changes caused by the greenhouse effect. Increased atmospheric concentration of greenhouse gases, of which CO2 is the most important, may increase the average temperature by 1.5–5 degrees within the middle of the next century. The greenhouse problem is a typical global environmental problem. The climatic changes throughout the world depend only on world-wide aggregate emissions of climate gases, and not on how these emissions are distributed between countries. The consequences of climatic changes may of course differ strongly between different countries, but the climatic changes themselves depend only on world-wide emissions. Global environmental problems are difficult to solve. Each country's own contribution to world-wide emissions is small, there is therefore little each country can do by itself. Although each country benefits from reduced emissions from all other countries, it is in no country's self-interest to make significant sacrifices through large reductions in its own emissions. The reason is that given the emissions of other countries, any single country's own emissions contributes only negligibly to aggregate emissions. We thus get a ‘prisoners' dilemma’ type of situation. This situation is analysed in more detail in Section 13.2, where CO2 emissions are modelled as the outcome of a non-co-operative game. In this game, each country wants to maximize its own income minus its own environmental costs. The non-cooperative equilibrium is defined as the set of emission levels which are such that each country is content with its own emission level given the emission level of the other countries. For this set of emission levels, the marginal cost of reducing emissions for

340 each country is equal to the country's own marginal environmental cost of its emissions. Small countries will have practically no influence on total emissions, and therefore on their own environmental costs. For such countries, equilibrium emissions thus give marginal costs of reducing emissions close to zero. This means that emission levels for small countries are practically equal to what they would have been in the absence of any environmental considerations. Larger countries will have some influence on total emissions, and will thus choose emission levels somewhat lower than what they would have chosen in the absence of environmental concerns. Nevertheless, without any co-operation between countries, total emissions from all countries will not be much lower than they would have been without any concern for the environment. This chapter is based on research at the Centre for Research in Economics and Business Administration (SNF), Oslo, and the Centre for International Climate and Energy Research, Oslo (CICERO). Parts are based on material from Hoel (1991a,b,c). I am grateful to Rolf Golombek and several participants of the WIDER workshop for useful comments on an earlier version. Section 13.2 also derives the first-best social optimum. This optimum is defined as the allocation of emissions which maximizes the total income of all countries minus the sum of all the countries’ environmental costs. In this optimum, each country's marginal cost of reducing emissions is equal to the sum of all countries’ marginal environmental costs. This implies a particular allocation of emissions between countries, and thus a particular distribution of income in the absence of side payments. For this concept of a social optimum to be meaningful, one must therefore assume that some form of side payments is possible. In order to reach a first-best social optimum, one needs some kind of co-operation between countries. However, not all kinds of international climate agreements will lead to a social optimum. In particular, international co-operation in the form of an agreement among the co-operating countries to cut back emissions by some uniform percentage rate compared with a specified base-year will generally not give the first-best social optimum. One reason for this is that such an agreement will imply an allocation of emission which makes marginal costs of reducing emissions differ between countries. Moreover, it is argued in Section 13.3 that not all countries will find it in their interest to participate in this kind of agreement. The limited participation aggravates the efficiency loss associated with this type of agreement. The cost difference between a cost-efficient agreement and a ‘uniform percentage reduction’-type of agreement may be quite large. The chapter therefore explores the properties of two other types of international agreements, namely (i) an international CO2 tax and (ii) tradable CO2 emission permits. Section 13.4 introduces an international CO2 tax into the static non-co-operative game presented in Section 13.2. More specifically, it is assumed that the Government of each country pays a tax, proportional to its CO2 emissions, to an international agency. The tax revenue (minus administrative costs) are reimbursed to the Governments of the participating countries according to some reimbursement rules. In Section 13.4, these rules are specified as a set of fixed shares of the total tax revenue. These reimbursement

341 shares as well as the tax rate on CO2 emissions are determined through negotiations between the participating countries. In practice, it may be very difficult to reach an agreement on the size of these variables. It would probably be even more difficult to reach an agreement if one opened up the possibility of different countries facing different tax rates. The present analysis is therefore restricted to a uniform CO2 tax, although this may give an allocation of emissions which differs slightly from the first-best optimum. The reason for this deviation from the first-best optimum is that some or all countries may be so large that their own CO2 emissions contribute non-negligibly to the greenhouse effect, and thus their own welfare. In practice, however, a uniform tax at an appropriate level will give an allocation of emissions which is very close to the allocation in the first-best optimum. Under the tax scheme, the distribution of total costs (i.e. costs of reducing emissions plus net taxes) between the participating countries is determined by the reimbursement rules. Since the allocation of emissions between countries is (almost) efficient no matter how taxes are reimbursed, the reimbursement rules can be determined purely from considerations of fairness. In particular, it is in principle possible to choose reimbursement rules which make each country better off under the agreement than without any international co-operation. As argued in Section 13.3, this is a plausible minimum requirement that an agreement must satisfy in order to achieve broad voluntary participation in an agreement. In Section 13.5 it is shown that a system of tradable emission permits is very similar to a system of an international CO2-emission tax. If all countries had been ‘small’ (in a sense defined in Sections 13.4 and 13.5), the two systems would in fact be identical. Although some countries in fact are ‘large’, the difference between these two types of international agreements are likely to be of minor importance. Sections 13.2–13.5 consider the CO2 problem as a static issue. However, in reality CO2 emissions affect the climate through cumulative emissions. This is modelled in Section 13.6, where CO2 emissions are given as an outcome of a nonco-operative two-period game. Different solution concepts for the game (open loop and perfect) are discussed, where time-paths of CO2 emissions in the absence of a CO2 tax are derived. It is shown under which conditions a uniform CO2 tax gives the first-best social optimum. In particular, it is shown that the tax giving the optimal solution is the same for the open loop and the perfect equilibrium, in spite of the fact that these two solutions differ in the absence of a CO2 tax. The interpretation of this surprising result is given in the end of Section 13.6. Future climatic changes depend on emissions of other climate gases (CFC, CH4, O3, N2O) as well as on CO2. A fully efficient agreement should therefore be related to emissions of all climate gases, weighted together with their impact on the climate. The main reason for only treating CO2 in this chapter is expositionary purposes. All of the arguments remain valid if instead

342 of CO2 conventions one considered more general conventions on all climate gases. Whatever type of international agreement one reaches during the next decade, it will probably cover only CO2, and not other climate gases. The reason for this limitation is that CO2 emissions are relatively easy to monitor, at least indirectly through monitoring each country's use of fossil fuels. Emissions of other climate gases are much more difficult to monitor. Although efficiency considerations suggest agreements encompassing all climate gases, practical considerations may thus force one to limit an agreement to CO2, at least initially. (See e. g. Hoel (1992) and Michaelis (1991) for a discussion of some of the complications which arise when greenhouse gases other than CO2 are included in the analysis.)

13.2 NON-CO-OPERATIVE AND CO-OPERATIVE OUTCOMES Consider N countries, each with an income function Rj(vj), where vj is the emission of CO2 from country j. Other inputs are held constant, and the income function is assumed to be increasing (up to some level ) and strictly concave in vj. We shall call , which is the emission level chosen in the absence of environmental concerns, the maximal emission level of country j. Rj(vj) measures income in excess of the income level with no emissions, i.e. Rj(0) = 0. Total CO2 emissions (S) are given by(13.1)

and each country is assumed to have an increasing and strictly convex environmental damage function Dj(S). The net benefit of country j is(13.2)

Consider first the non-co-operative Nash equilibrium of the one-shot game in which all countries choose their CO2 emissions simultaneously. The non-co-operative Nash equilibrium implies that Bj in (13.2) is maximized with respect to vj, taking the other emission levels vi (i ≠ j) as given and with (13.1) inserted into (13.2). This gives(13.3)

i.e. each country chooses its emission level so that marginal income of emissions is equal to the country's own marginal environmental cost. Denote the emission levels implied by (13.3) by v1*, … vN*. The smaller a country is, the less can it influence total emissions, and thereby the environment. Intuitively, we therefore expect non-co-operative emissions vj* to be closer to maximal emission level the smaller country j is. In Appendix 1 it is shown that this intuition is correct, i.e. that other things equal, the ratio is closer to 1 the smaller a country is.

343 The sum of net benefits for all countries is(13.4)

A first-best social optimum follows from maximizing (13.4) with respect to CO2 emissions, which gives(13.5)

where we have defined(13.6)

In other words, marginal revenues are equal for all countries, and equal to the sum over all countries of the marginal damage of CO2 emissions. Notice that (13.5) gives a particular distribution of CO2 emissions, and thus a particular distribution of net benefits between countries in the absence of side payments. With side payments, however, the conditions (13.5) do not restrict the possible distributions of net benefits between countries. It is therefore only when side payments of some form are permitted that (13.5) is the obvious candidate for a co-operative equilibrium. Comparing (13.3) with (13.5) it is easily verified that the sum of emissions from all countries is lower in the first-best optimum than in the non-co-operative equilibrium. To prove this formally, use superscripts * and ° for the non-cooperative and co-operative outcome respectively. Assume S° ≥ S*. Since this gives D′(S°) > Dj′(S°) ≥ Dj′(S*), we must have vj° < vj* for all j, which implies S° < S*. This contradiction proves that we must have S° < S*. Intuitively, we might expect that emissions from all countries are lower under the first-best optimum than they are in the non-cooperative case. For the general case, however, this need not be true. A simple example demonstrating the possibility of a country having higher emissions in the first-best optimum than in the non-co-operative equilibrium is given in Appendix 2. To illustrate some of the results of this section, consider the following numerical example. All countries are assumed to have equal environmental cost functions. Moreover, it is assumed that the marginal environmental costs are constant, i.e.(13.7)

where M is some positive constant. The countries may be identified by an index r ∈ [0,1] over the countries' relative numbers, defined by(13.8)

Assume that all countries have quadratic income functions

344 (13.9)

i.e.(13.10)

In the absence of environmental considerations, all countries would choose v = 1. For any v < 1, marginal incomes will differ between countries with different c(r)-values. We shall assume that the c(r)-function is given by(13.11)

In other words, a country's marginal income of emissions for a given emission level (< 1) is higher the higher r-value the country has, i.e. the higher its index number (cf. (13.8)). In Appendix 3, it is shown that (13.11) implies that if country r could choose a uniform emission level for itself and all other countries, it would choose v = r. In other words, the most preferred emission level for different countries is uniformly distributed over the range [0, 1]. The non-co-operative equilibrium follows from (13.3), which in the present example takes the form

or, using (13.11)(13.12)

When N is large, all v*(r) are thus practically equal to 1,50 i.e. the emission level chosen in the absence of environmental considerations. The first-best solution follows from (13.5), which in the present case takes the form c(r)(1 − v) = M for all r, i.e. using (13.11)(13.13)

In this example the first-best optimum implies that each country set its emission level equal to its most preferred uniform emission level. Optimal emission levels are thus distributed uniformly over the interval [0, 1].

13.3 EQUAL PERCENTAGE REDUCTIONS OF EMISSIONS International co-operation often takes the form of an agreement among the co-operating countries to cut back emissions by some uniform percentage rate compared with a specified base-year. This type of agreement has two disadvantages. In the first place, it is well known from environmental economics

50

If e.g. N = 100, the range for v * is [0.99, 1].

345 that equal percentage reductions of emissions from different sources usually gives an inefficient outcome, in the sense that the same environmental goals could be achieved at lower costs through a different distribution of emission reductions. This source of inefficiency might in fact be quite large, although the estimates in the literature vary. On the one hand, Kverndokk (1991) studies a reduction of world-wide CO2 emissions to 80 per cent of the 1990 level. A uniform percentage reduction in all countries gives only 13 per cent higher costs in 2020 than a cost-efficient allocation of emission reductions. On the other hand, Barrett (1991) considers the cost for the EC of stabilizing its emissions at the 1988 level by 2010. He finds that if each EC country were required to stabilize its emissions, the total cost would be almost fifty times that of the cost-efficient solution! A second problem with agreements of equal percentage reductions is that not all countries will find it in their interest to participate in such agreements. A likely minimum requirement for a country to participate in an agreement is that the country is better off under the agreement than it is without any international agreement. The reason for such a requirement is the following. As long as there is no international law to force countries to participate in an agreement, each country can choose to be a free-rider outside the agreement instead of participating in the agreement. If the country stands outside the agreement, it can enjoy (almost) the same benefits of reduced emissions as if it participates in the agreement, while it doesn't bear any of the costs of reducing emissions. An important motive for a country to participate in an agreement instead of being a free-rider is that by being a free-rider it increases the risk of the whole agreement breaking down. This motive for participating in the agreement is stronger the more the country has to lose from the agreement breaking down. Obviously, a country which doesn't lose anything from the agreement breaking down has no incentive to participate in the agreement, and it will therefore instead choose to be a free-rider. To study the participation issue in more detail, consider the following two-stage process. In the first stage, each country decides whether or not to participate in an agreement. In the second stage, emission levels of the non-participating countries and the uniform percentage reduction of the participating countries are determined. We start by considering the second stage. Denote the set of participating countries by P. If these countries choose to cut back their emission levels by (1 − α) 100 per cent compared with their emission levels in the non-co-operative equilibrium, each of these countries gets an emission level equal to Non-participating countries act as they do in the non-co-operative equilibrium, i.e. country k ∉ P chooses its emission level vk(α,P) so that(13.14)

is satisfied, where total emissions S(α, P) are given by

346 (13.15)

Assume that the participating countries choose α so that their total income minus their total environmental costs is maximized.51 This gives(13.16)

When countries decide whether or not to participate in an agreement, they take account of (13.16). As explained above, we assume that a country chooses to participate if and only if the agreement gives it a higher pay-off than it obtains in the non-co-operative equilibrium. Formally, defining country j's pay-off as(13.17)

the set P of participating countries is denned by(13.18)

where Pk is the set P plus country k. Condition (13.18) means that all participating countries are better off under the agreement they subsequently choose than they would have been without any co-operation. Moreover, each of the non-participating countries would have got a pay-off below their non-co-operative pay-off had they chosen to participate in the co-operating group. It is not obvious that any P satisfying (13.18) (with at least two members) exists. Nor can the possibility of several P-sets (each with at least two members) be ruled out for general functions Rj and Dj. In Appendix 4 it is shown that if one or more sets of P (with at least two members) exists, each of these sets may be supported by a perfect equilibrium of a two-stage game in which the decisions of whether or not to participate is made in the first stage. In Appendix 3 it is shown that for the numerical example given in the end of Section 13.2, the set of participating countries will be the 65 per cent of the countries which have the lowest costs of reducing emissions. These countries will cut down their emissions by 40 per cent, implying a total reduction of emissions equal to 26 per cent compared to the first-best reduction of 50 per cent.

13.4 INTERNATIONAL EMISSION TAXES An alternative to an international agreement on uniform reductions of emissions is some kind of agreement on international emission taxes. It is useful

51

An alternative assumption about α, considered in Hoel (1991c), is that α is given by the median value of the most preferred α-values of the participating countries.

347 first to consider two ways in which an international CO2 tax should not be designed. Perhaps the closest analogue to a domestic CO2 tax is a CO2 tax which is administered and enforced by an international agency, and applies to all CO2 emissions in all countries, just like a CO2 tax administered and enforced domestically. One reason why such a tax is unacceptable is that it would require a large and costly international bureaucracy. Moreover, most countries would not accept an international agency to have the necessary juridical power within their own country. An alternative form of taxing CO2 emissions is an agreement to harmonize domestic CO2 taxes across countries. However, it is not clear what such an agreement implies, in particular if some of the participating countries are nonmarket economies. Even in market economies, it is not clear whether such a harmonized CO2 tax should be in addition to various other domestic taxes on fossil fuels the countries might have, or instead of all existing taxes on fossil fuels. If existing taxes reflect domestic externalities in transportation etc., an internationally agreed upon CO2 tax ought to be an addition to existing taxes, while one could argue that the CO2 tax should take the place of other taxes on fossil fuels if these are pure revenue-raising taxes. A problem with an international agreement requiring equal domestic CO2 taxes is the free-rider problem. As mentioned earlier, it is in each country's interest to have little or no restrictions on their own CO2 emissions, given the CO2 emissions from other countries. If a country is required to have a CO2 tax through an international agreement, it is therefore in the interest of the country to try to make this tax be as ineffective as possible. One way to do this is to reduce other domestic taxes on fossil fuels, e.g. taxes on gasoline which several countries have for domestic purposes. Even if a country doesn't directly reduce such domestic taxes, it might raise them less than it would have, had it not been for the imposed CO2 tax. Another way to reduce the effect of the imposed CO2 tax is to manipulate prices of other domestic goods. Roughly speaking, a country should tax close substitutes to fossil fuels and subsidize complements. Obvious examples are taxes on other types of energy (e. g. hydroelectric power) and subsidies on automobiles and air-conditioning. This type of price policy will reduce the effect of an imposed CO2 tax on a country's consumption and production pattern, and thereby reduce the cost for the country, even though the country in a formal sense is sticking to the international agreement. Instead of harmonizing domestic CO2 taxes, an international agreement could specify the following type of international CO2 tax: The Government of each country pays a tax, proportional to its CO2 emissions, to an international agency. The tax revenue (minus administrative costs) is reimbursed to the Governments of the participating countries according to some specified rules. To simplify, we ignore administrative costs, and assume that the tax revenue is reimbursed to the countries in fixed shares β1, β2, …, βN, where βj > 0 and

348 Σiβi = 1. The determination of the βj's is beyond the scope of this chapter, although it is natural to think of the β vector being determined through international negotiations. A natural requirement for the β vector is that each country's βj is so large that the country is better off with the tax scheme than it is in the non-co-operative equilibrium without any tax on CO2 emissions, cf. the discussion in Section 13.3. Denoting the tax rate by t, the net benefit of country j is(13.19)

The term in square brackets represents taxes minus reimbursements for country j, and may be positive or negative. Consider the non-co-operative Nash equilibrium of the one-shot game in which all countries choose their CO2 emissions simultaneously. Each country takes the tax rate t and the reimbursement vector β as given. The non-cooperative Nash equilibrium implies that Bj in (13.19) is maximized with respect to vj, taking the other choices vi(i ≠ j) as given and with (13.1) inserted into (13.19). This gives(13.20)

We want to see if there exists a tax rate which gives the first-best optimum. If such a tax rate exists, it follows from (13.5) and (13.20) that(13.21)

i.e.(13.22)

From (13.22) it is clear that only if(13.23)

(at the optimal S) is it possible to find a tax rate which makes the non-co-operative equilibrium coincide with the social optimum. Whether or not the condition (13.23) holds depends on the β vector. In principle, the β vector could have been chosen so that (13.23) holds in equilibrium. With arbitrary Dj-functions, however, this could violate the requirement that all countries are at least as well off in this first-best optimum as they are in the non-co-operative case without any CO2 tax. In particular, there could be a country which would not be negatively affected by the climatic changes following from increased atmospheric concentration of CO2, i.e. Dj′ ≤ 0 for this country. Clearly, such a country would need to have βj > 0 to be willing to participate in an arrangement with a positive tax on CO2 emissions, i.e. β > Dj′(S)/D′(S). An interesting special case is the case in which all Dj-functions are identical except for a factor of proportionality. More precisely, Dj(S) = αjD(S) for all j

349 in this special case, where D(S) is given by (13.6), and where αj ≥ 0 and Σiαi = 1. One would expect the αj-coefficients to be closely related to the size of the countries, i.e. for αj to be higher the larger country j is. Since the β vector also will be related to the size distribution of the countries, it is not unconceivable that that the β vector is chosen so that it is identical to the α vector. If this is the case, condition (13.23) always holds. If the condition (13.23) holds, if follows from (13.22) that the tax rate(13.24)

leads to the first-best optimum. In other words, the tax rate should be equal to the sum of marginal environmental costs. Whether or not (13.23) holds (13.24) inserted into (13.20) gives(13.25)

or(13.25′)

It follows from (13.25′) that countries with high marginal damage functions relative to their size (as measured by βj), will have less CO2 emissions than what follows from the first-best optimum and conversely for countries with relatively low marginal damage functions. Notice also that for all countries which are small, the terms Dj′(S)/D′(S) and βj will be close to zero. For these countries it is therefore clear from (13.25′) that the tax given by (13.24) leads to CO2 emissions which are approximately equal to the CO2 emissions in the first-best optimum. For most countries, these small-country assumptions are valid. However, there are some countries and groups of countries for which the small-country assumptions are not valid. Obvious examples are USA, USSR, and EC, which have 23 per cent, 18 per cent and 13 per cent of global CO2 emissions, respectively. For large countries, Dj′(S)/D′(S) and βj cannot be approximated by zero. Nevertheless, Dj′(S)/D′(S) − βj may be close to zero also for these large countries. In the numerical example considered in the end of Section 13.2, all countries had the same marginal environmental costs. From the analysis above we know that a first-best optimum for this case is achieved by a tax scheme in which all countries receive the same reimbursement share (i.e. β1 = … = βN). In Appendix 3, it is shown that for this example equal-reimbursement shares also make all countries better off with the agreement than without.

13.5 TRADABLE EMISSION PERMITS For domestic environmental problems, it is well known that a system of tradable emission permits has the same desirable cost-efficiency properties as a system of emission taxes. The same is true for global environmental problems.

350 An alternative to an international CO2 tax is a system of CO2-emission permits which each country is free to use itself or to sell to other countries who can use them instead. When there are many countries participating in the CO2 agreement, and each country is relatively small, a competitive market for CO2-emission permits is likely to develop. In this case each country will regard the price of CO2-emission permits as independent of its own CO2 emissions. Denoting this price by q, the net benefit of country j is thus(13.26)

In this expression, the term qvj represents the total costs to country j CO2 emissions. Whatever the initial distribution of CO2-emission permits is, the unit cost of emissions is equal to the market price of CO2-emission permits. If the country is buying CO2-emission permits, it is obvious that the unit cost of CO2 emissions is equal to the price of CO2emission permits. The same is true if the country is selling CO2-emission permits: The unit cost of CO2 emissions is the income forgone from the additional sale of a CO2 quota, i.e. equal to the market price of a CO2 quota. The term βjqS is the value of the initial allocation of emission permits to country j, assuming that it receives a share βj of the total emission permits S. The expression (13.26) is valid for all countries. In a non-co-operative equilibrium, each country chooses vj so that Bj is maximized. In this maximization, S is taken as given, since it is determined by the total number of emission permits which are allocated to the countries. A small country will also regard the market price of emission permits (q) as exogenous, so that maximization of (13.26) gives(13.27)

which gives vj as a declining function of q, i.e.(13.28)

The equilibrium price q is determined by(13.29)

Notice that whatever value of S is chosen, (13.27) implies that all countries have equal marginal incomes of CO2 emissions. In other words, total income is maximized subject to the constraint on total emissions. Moreover, if S is chosen equal to the socially optimal level S°, it is clear that the present non-co-operative equilibrium coincides with the first-best optimum. Notice the similarity between a system of tradable emission permits and an emission tax. Provided the term in square brackets in (13.25′) is close to one, (13.24) and (13.25′) are equivalent to (13.27). As long as all countries are small in the sense that the term in square brackets in (13.25′) is close to one and q is regarded as exogenous in the system of tradable emission permits, the

351 two systems are in fact isomorphic. The gross CO2 tax paid by a country corresponds to the market price of CO2emission permits multiplied by the country's CO2 emissions, while the country's reimbursement corresponds to the market price of CO2-emission permits multiplied by the initial emission permits the country gets. A tax scheme with an emission tax t ′ giving total CO2 emissions equal to S ′ and tax reimbursements to the N participating countries proportional to the vector (β1,…, βN) with Σiβi = 1 is thus equivalent to a system of tradable emission permits where the initial emission permits to the N countries are (β1S′, …, βNS′). In this system of tradable emission permits the market price of the emission permits becomes identical to the emission tax t ′ in the corresponding tax scheme, and the initial distribution of tradable CO2-emission permits corresponds completely to the reimbursement shares in the CO2tax scheme. The exact equivalence between a CO2 tax and a system of tradable emission permits no longer holds when some of the countries involved are large. The reason for this is partly that we cannot be sure that the term in square brackets in (13.25′) is close to 1 for a large country, and partly that a large country usually will not consider the market price of emission permits as exogenous. To see the implications of the latter issue, consider first the case in which a large country behaves like a traditional monopolist or monopsonist, depending on whether it is a seller or a buyer of CO2emission permits. A large country will take the effect of its own emission level on the market price q into consideration, so that maximization of (13.26) now gives(13.27′)

dq/dvj is positive for a large country, since the price of permits is higher the more permits the country buys (or the less it sells). The marginal incomes are therefore equalized across all countries only if vj = βjS for all large countries. It is therefore in principle possible to find an initial distribution of permits making the system of tradable emission permits correspond with the first-best social optimum. However, this β vector could very well violate the requirement that all countries should be at least as well off under the system of tradable emission permits as they are in the non-cooperative case without any kind of international environment convention. Notice also that the β vector making the system of tradable emission permits equivalent to the first-best optimum is in general different from the β vector giving equivalence between the equilibrium with a CO2 tax and the first-best social optimum. A large country will not necessarily behave like a traditional monopolist or monopsonist. Assume that the large country under consideration is a seller of CO2-emission permits. A similar discussion applies also for a large country buying CO2-emission permits. A traditional monopolist faces a large number of indistinguishable, anonymous buyers. This forces the monopolist to offer all buyers the same price schedule. If transactions costs between buyers are

352 small, this price schedule must be linear, i.e. the price a buyer pays must be independent of how much it buys. A large country selling CO2-emission permits is in a slightly different position. Although the number of buyers is relatively large, they may all be identified. This opens up for the possibility of the large country giving different offers to different countries, and for the offers to be more sophisticated than simply announcing a price and letting the buyer determine the quantity. The outcome of such a process depends on several factors, one of which is how much information the large country has about cost functions of other countries. In addition to the considerations above, one may have large countries on both sides of the market for emission permits. In such a bilateral monopoly or bilateral oligopoly market it is not obvious what the outcome is for emission levels. Although the outcome of such complex market systems cannot be predicted without a more precise description of the market, the possibility of reaching a cost-efficient outcome cannot be ruled out. Disregarding large countries, a CO2-tax scheme and a system of tradable CO2-emission permits are equivalent. With large countries the strict equivalence breaks down. However, also for this case both systems may give allocations of CO2 emissions which are close to the cost-minimizing allocation of CO2 emissions. Since both systems have good allocative properties and equal distribution properties, other considerations must determine which of the two systems one should choose. An obvious objection against a CO2 tax is that it is difficult to know exactly which tax rate corresponds to an agreedupon level of total CO2 emissions. However, for the case of CO2 one is not particularly concerned about the exact emissions in any particular year, but rather about the development over several years. Once one has agreed upon a desirable development of CO2 emissions, one can decide upon a corresponding initial CO2 tax rate and a tentative future development of the CO2 tax. If it turns out that the initial CO2 tax gives a different level of CO2 emissions than expected, the CO2 tax at later dates can be adjusted up or down compared with the original plan for the CO2 tax development. Through this type of procedure, one should relatively quickly be able to reach a path of CO2 emissions which is close to the desired path. With CO2-emission permits, the desired path of CO2 emissions can be obtained accurately and immediately. However, in this case the problems are pushed over to each individual country: it is difficult for each country to achieve CO2 emissions which are exactly equal to its emission permits. Since unused CO2-emission permits could have been sold at an earlier date, most countries will not design their CO2 policies with a large security margin relative to their CO2emission permits. One must therefore expect frequent cases of CO2 emissions in excess of the emission permits. Unlike the CO2-tax scheme, the system of CO2-emission permits has no built-in mechanism for treating CO2 emissions in excess of what a country had planned.

353 A possible way to treat emissions above or below the level required by the CO2-emission permits, is for the international agency in charge of the CO2-emission permits at the end of each year to buy unused emission permits or sell necessary additional emission permits at the market price for the emission permits. With this modification, however, the system of CO2-emission permits no longer has the advantage of making it possible to achieve a desired level of CO2 emissions with 100 per cent accuracy. The discussion above suggests that a CO2-tax scheme in some respects might be slightly simpler than a system of tradable CO2-emission permits. On the other hand, a system of tradable emission permits is a relatively small formal extension of more traditional types of international conventions, such as, for example, uniform emissions reductions for all countries. The difference between an international tax and tradable emission permits is however quite small. Both systems probably give large efficiency gains compared to rigid international agreements of the ‘uniform percentage reduction’-type. The important decision is therefore not which of the two cost-efficient systems (an international tax or tradable emission permits) one chooses, but that one chooses one of them and not more traditional, inefficient methods of reducing total CO2 emissions.

13.6 A DYNAMIC GAME OF CO EMISSIONS 2

So far, CO2 emissions have been analysed as a static problem. However, the climate is in reality not affected only by the current emissions of CO2. The climate is affected by the atmospheric concentration of CO2, which in turn depends on cumulative emissions. In order to study this issue, we therefore consider a simple two-period model of CO2 emissions in this section. Using small and capital letters for period 1 and 2, respectively, we have(13.30)

(13.31)

We thus ignore cumulative emissions prior to period 1 (since they are historically given) as well as natural degradation from period 1 to period 2. The benefit function of country j at the beginning of period 1 is(13.32)

where δ∈ (0,1) is a discount factor which is common to all countries. The first-best optimum follows from maximizing ΣiBi subject to (13.30)–(13.32), and gives(13.33)

(13.34)

354 for all j. Notice in particular that this optimum is dynamically consistent. Reoptimizing after period 1 gives the same optimal values V1, …, VN as in the initial optimization problem, provided actual emissions in period 1 were equal to their optimal values. Using (13.34), (13.33) maybe rewritten as(13.33′)

The interpretation of (13.33′) is as follows: the marginal income (in present value) of moving one unit of emissions from period 2 to 1 for any country should be equal to the total marginal environmental cost of adding this emission to the cumulative stock in period 1 instead of period 2. The interpretation of (13.34) is similar to that of (13.5) for the static case: the marginal income of emissions from any country in period 2 should be equal to the total marginal environmental costs of emissions in period 2. We next turn to the non-co-operative game. In a dynamic game of the present type, some choice of equilibrium concept must be made. A natural candidate is the dynamic analogue of the Nash equilibrium of Section 13.2. Each country chooses the time-path of CO2 emissions which is an optimal reply to the other countries’ strategies, which are specified as similar time-paths. In the literature of dynamic games, this type of equilibrium is often called an open-loop equilibrium. The problem with this equilibrium is that it is not a subgame-perfect equilibrium (in the sense of Selten (1975)) unless the countries can commit themselves to specific time-paths of CO2 at the beginning of the planning period (Reinganum and Stokey, 1985). An alternative equilibrium concept is to model each country as choosing a decision-rule strategy, making its CO2 emissions at any time depend on the history of the game up to this point of time. In the present context this means that emissions in period 2 depend on emissions in period 1. With this type of strategy, often called a closed-loop strategy, we can derive subgame-perfect equilibria. An open-loop strategy in the present model implies that each country chooses its emission levels for both periods, taking emission levels in both periods from all other countries as given. Consider first the open-loop equilibrium. In this equilibrium, each country's emission path maximizes its net benefits (i.e. Bj from (13.32)) given the time-paths of each other country's emissions. This gives(13.35)

(13.36)

By comparing (13.35)–(13.36) with (13.33)–(13.34), it is easily verified that total cumulative emissions (S) are higher in the open-loop equilibrium than in the first-best optimum. This equilibrium is dynamically consistent, in the sense that as long as s is equal to the equilibrium value implied from (13.30)–(13.31) and (13.35)–(13.36), the equilibrium values Vj from these equations constitute a Nash

355 equilibrium of the game starting in the beginning of period 2. However, for an equilibrium to be perfect it is not enough that this consistency property holds along the equilibrium path: subgame perfection requires that the strategies for emissions in period 2 constitute a Nash equilibrium when viewed from any total emission level s in period 1, i.e. not only from the s-value which follows from (13.30)–(13.31) and (13.35)–(13.36). (See e.g. Fershtman (1989) for a further discussion of the relationship between time consistency and subgame perfection.) The perfect equilibrium is found by first considering period 2. In the beginning of period 2, s is given. Emissions in period 2 are determined as a standard static non-co-operative equilibrium, i.e.(13.37)

for all j. This Nash equilibrium of the subgame for period 2 defines emission levels as functions of total emissions in period 1 (s), i.e.(13.38)

In period 1, each country chooses its emission level taking as given each other country's emission level in period 1 and each other country's emission function (13.38) for period 2. In other words, country j maximizes(13.39)

which gives(13.40)

Equations (13.30), (13.31), and (13.40) determine the equilibrium emission levels in period 1. Let us compare the open-loop equilibrium, given by (13.33) and (13.34), with the perfect equilibrium, given by (13.37) and (13.40). It is clear that these equilibria could only be equal if for all j. In Appendix 4 it is proved that this cannot be the case. Moreover, it is argued in this appendix that s and S usually are higher in the perfect equilibrium than in the open-loop equilibrium. Consider next how emission taxes work in the present dynamic game of CO2 emissions. Assume that the emission taxes are t and T in the two periods, and that the tax reimbursement vector in both periods is (β1, …, βN). The benefit function of country j at the beginning of period 1 is now(13.41)

where Bj is given by (13.32). Consider first the open-loop equilibrium. In this case country j considers t, T, βj, and as given. Maximization of (13.41) therefore gives(13.42)

(13.43)

356 Comparing (13.42)–(13.43) with (13.33)–(13.34), it is easily verified that the non-co-operative equilibrium with emission taxes coincides with the first-best optimum (denoted as before by the superscript °) if and only if(13.44)

and(13.45)

From (13.44)–(13.45) it is clear that only if(13.46)

is it possible to find a uniform tax rate which makes the non-co-operative equilibrium coincide with the social optimum. It is clear from (13.46) that there may be no constant reimbursement vector (β1, …, βN) satisfying (13.46), since Dj′(s°)/D′(s°) may differ from Dj′(S°)/D′(S°). If, however, we consider the special case of Dj(x) = αjD(x) for all x (cf. Section 13.4), (13.46) holds provided the reimbursement vector is chosen so that βj = αj, for all j. If (13.46) holds, the optimal taxes are given by(13.47)

Notice that for δ sufficiently close to 1 and D′(s°) > 0, the optimal emission tax declines over time. However, this feature follows from the assumption of a finite horizon: in Hoel (1991b) it is shown that in an infinite-horizon model of the present type, the optimal tax rises over time as long as emissions are positive. (Cf. also Ulph et al. (1991) for a further discussion of the dynamics of an optimal carbon tax.) For an arbitrary vector (β1, …, βN), inserting the emission taxes (13.47) into (13.42)–(13.43) gives(13.48)

(13.49)

357 For small countries, the terms βj, Dj′(S)/D′(S) and (Dj′(s) + δDj′(S))/(D′(s) + δD′(S) are close to zero. Even if βj, Dj′(s)/ D′(s), and Dj′(S)/D′(S) are relatively large for large countries, the square brackets in (13.48)–(13.49) may be close to one also for these countries. Emission taxes given by (13.47) may therefore give an equilibrium which is quite close to the first-best social optimum even in the presence of large countries. So far, we have only considered the open-loop equilibrium. The benefit function of country j is given by (13.41) no matter what equilibrium concept we use. However, while country j regards Vi(i ≠ j) as given in the open-loop solution, it takes (13.38) into consideration in the perfect equilibrium, i.e.(13.50)

Although country j takes vh (h ≠ j) as given, it is clear from (13.50) that vj affects all Vi. The first-order condition (13.43) remains valid in the perfect equilibrium. Since Vi now depends on vj, (13.42) is modified to(13.42′)

If (13.46) holds so that the emission taxes (13.47) make the open-loop equilibrium equivalent to the social optimum, the term in square brackets in (13.42′) is zero. But this means that the emission taxes given by (13.47) make also the perfect equilibrium coincide with the social optimum. The result above is quite surprising. Without taxes the perfect equilibrium usually gives higher CO2 emissions than the open-loop equilibrium. Intuitively, we would therefore expect that a higher Pigouvian tax is necessary to achieve the full optimum in the perfect equilibrium than in the open-loop case. The result above shows that this intuition is wrong. To understand the result above, it is useful to consider why the perfect equilibrium has higher CO2 emissions than the open-loop equilibrium in the absence of taxes. In both equilibria each country knows that its CO2 emissions contribute to total cumulative emissions. In the open-loop equilibrium additional cumulative emissions by definition have no effect on the time-path of CO2 emissions from other countries. In the perfect equilibrium, on the other hand, additions to the CO2 stock resulting from a particular country's emissions in period 1 tend to reduce the emissions in period 2 from other countries (for ).52 Since the country under consideration prefers emissions from other countries to be low, it has an incentive to choose higher CO2 emissions than it would have chosen had the time-paths of CO2 emissions from other countries been fixed. When there is a CO2 tax, it is no longer obvious that reduced future CO2 emissions from other countries benefit a particular country. The reason is that lower future CO2 emissions from other countries also means lower future tax

52

The interpretation is valid also for

, except that all effects have opposite signs.

358 revenue from other countries, part of which is redistributed to the country under consideration. With the optimal tax, the negative effect of lower future tax reimbursements (the term in (6.14′)) is exactly equal to the positive environmental effect of other countries reducing their future emissions (the term in (13.42′)). Each country therefore chooses its CO2 emissions without taking the effect on future emissions from other countries into consideration, just as in the open-loop equilibrium.

Appendix 1: Country Size and Emission Levels Consider the case in which the only difference between the different countries is their size. In this case we may use each country's maximal emission level as a measure of its size. More precisely, we assume that(A1.1)

implying that(A1.2)

The function ρ(x) is assumed to satisfy ρ(0) = 0, ρ″(x) < 0, ρ′(x) > 0 for x < 1 and ρ′(1) = 0. In the absence of any environmental concern, the revenue of all countries (relative to the case of no emissions) is thus proportional to their maximal emission levels. The marginal revenues of emissions are also equal for two countries if their ratios of actual emissions (vj) to maximal emissions are equal. Assume also that(A1.3)

i.e. that the environmental damage cost in a country is proportional to its size, as measured by . For this case it follows from (13.5)–(13.6) that the first-best optimum is characterized by(A1.4)

In other words, optimal emissions as a proportion to maximal emissions are equal for all countries in this case. The non-co-operative equilibrium follows from (13.3), i.e.(A1.5)

From (A1.5) and ρ″ < 0 it immediately follows that Combining (A1.4) and (A1.5) yields(A1.6)

is closer to 1 the smaller the country is (i.e. the smaller is).

For a small country, the right-hand side (A1.6) is approximately equal to zero, since follows from the properties of the function ρ that , , for such a country.

for such a country. It

Appendix 2: Emission Levels in the Non-cooperative Equilibrium and in the Social Optimum To see that a country may have higher emissions in the first-best optimum than in the non-co-operative equilibrium, consider the two-country case in which both countries A and B have equal quadratic revenue functions v − v2/2, i.e. R′(v) = 1 − v. The environmental damage functions are(A2.1)

and(A2.2)

where a∈ (0,1) and b∈ (0,1/(1−a)). The non-co-operative equilibrium follows from (A2.2)(A2.3)

giving(A2.4)

The restrictions on the parameters a and b imply that both countries have positive emission levels. From (A2.4) it follows that(A2.5)

The first-best optimum follows from (A2.5):(A2.6) i.e.(A2.7)

and(A2.8)

which confirms the general result proved in Section 13.2. Comparing (A2.7) with (A2.4), we immediately see that vA0 < vA*. However, for b close enough to 1/(1 − a), we get vB0 > vB*, i.e. country B has higher emissions in the first-best optimum than in the non-co-operative equilibrium.

Appendix 3: Derivation for the Numerical Example Given by (13.7)–(13.11) To see the implication of the assumption (13.11), consider the most preferred emission level of country r, if it could choose a uniform emission level for itself and all other countries. This emission level is found by maximizing

which gives

or, using (13.11), v = r. In other words, the most preferred emission level for different countries is uniformly distributed over the range [0, 1]. Total welfare, i.e. total income minus total environmental costs, is given by(A3.1)

where V is total emissions, and F(r) is the cumulative distribution function for the countries, i.e. F(r) says how many countries have an index number lower or equal to r. We therefore also have(A3.2)

In our example F(r) = rN. Average welfare is thus(A3.3)

or, inserting (13.9) and (13.11)(A3.4)

The absolute level of average welfare is of no particular interest. Instead, we normalize so that average welfare with maximal emissions, i.e. v(r) = 1 for all r, is zero. As mentioned in Section 13.2, v(r) = 1 for all countries is also a good approximation of the non-co-operative equilibrium for our example. Denoting this normalized welfare level by w, it follows from (A3.4) that(A3.5′)

or(A3.5)

362 From Section 13.2, we know that v(r) = r in the social optimum. Inserting this into (A3.5) gives(A3.6)

With no co-operation i.e. v(r) = 1 for all r, the environmental cost per country is M. The average welfare increase per country of achieving the first-best optimum instead of the non-co-operative outcome is thus 25 per cent of this environmental cost. Consider next the case of uniform percentage reductions. Since v*(r) = 1 for all r in this example, uniform reductions of emissions also implies uniform emissions. Assume that a proportion p of the countries participate in an agreement. Since countries are indexed by increasing costs of reducing emissions, it will be the countries with lowest index numbers which participate in an agreement. Total welfare for the pN participating countries is given by an equation which is similar to (A3.1):(A3.7)

where V now instead of (A3.2) is given by(A3.8)

since the (1 − p)N non-participating countries have emission levels equal to 1. Using the same procedure which led to (A3.4), we now find average welfare for the pN participating countries to be(A3.9)

Since we by assumption have uniform emissions for the pN co-operating countries, i.e. v(r) = v for r ≤ p, (A3.9) may be rewritten as(A3.10) The emission level which maximizes the participating countries’ average pay-off follows from maximization of (A3.10), and gives

or(A3.11) This function v(p) in (A3.11) corresponds to the function α(P) in (13.16) for the general case. It is easily verified that v(p) is declining in p for p < 0.70 and increasing in p for p > 0.70. The minimum value of v(p) is 0.59, and v(Q) = v(1) = 1. We next turn to the determination of the participation rate p. From the definition of participation given by (13.18) for the general case, the country indexed p in the current example must be equally well off under the agreement as under the non-cooperative equilibrium.

363 Under the agreement, the welfare of country p is(A3.12)

With no agreement, country p's welfare is(A3.13)

Setting wA(p) = wN(p) gives

or(A3.14) To determine the equilibrium value of p, we insert v(p) from (A3.11)(A3.15)

which gives(A3.16)

Compared with the non-co-operative case, average welfare for all countries follows from the general formula (A3.5), which in the present case may be written as(A3.17)

Inserting (p, v(p)) from (A3.16) gives(A3.18) The benefits of a ‘uniform reduction’-type of agreement are thus about 29 per cent lower than the benefits of a firstbest social optimum (which was 0.25 M, cf. (A3.6)). Consider next a CO2 tax of the type analysed in Section 13.4. The tax given by (13.24) is in the current example equal to M. With this tax rate, we get the first-best optimum provided βj = 1/N, cf. (13.23). As pointed out previously, a tax scheme has little hope of success unless all countries feel they are better off with the tax scheme than they are in the non-co-operative equilibrium. In the current example, the difference in welfare for country r between the tax equilibrium and the non-co-operative equilibrium is (since v0(r) = r in the tax equilibrium and v*(r) ≈ 1 in the noncooperative equilibrium):(A3.19)

364 In (A3.19), the first term (in square brackets), represents the revenue loss due to the reduced emissions. The second term is the environmental gain (average emissions are 1/2 in the tax optimum as opposed to 1 in the non-co-operative equilibrium), and the third term is the net taxes (i.e. taxes minus reimbursements) paid by country r. Inserting for c(r) from (13.34) and t = M, we get(A3.20)

In other words, all countries are better off under the tax equilibrium than they are under the non-co-operative equilibrium.

Appendix 4: The Decision of Whether or not to Participate in an Agreement Modelled as a Two-Stage Game Consider a two-stage game in which all countries decide whether or not to participate in an agreement in the first stage, and the participating countries decide upon a common reduction of emissions (relative to vj*) in the second stage. The second stage is trivial, various types of bargaining games can give outcomes such as α(P) from (3.3). In the first stage of the game, countries correctly foresee how emission levels will be determined in the second stage. In deciding whether or not to participate, countries thus take account of the function α(P). If the countries simultaneously have to decide once and for all whether or not they will participate in an agreement, we get a situation similar to the one analysed by Barrett (1989). In this case a country will find it better to participate in the agreement than to be a free-rider only if the agreement is so much better when it participates that this environmental benefit exceeds the costs of the country participating. As shown by Barrett (1989), Carrano and Siniscalco (1992), and Hoel (1991c), the equilibrium of this game may have very few countries participating in an agreement. The assumption that countries must simultaneously decide once and for all whether or not they wish to co-operate is rather restrictive. An alternative could be that countries decide whether or not to participate in an agreement sequentially, and that each country can change its decision after seeing the choices of the other countries. The first stage of the game ends when no countries want to change their participation decision. Assume that there exists an equilibrium set P with at least two members satisfying (13.18). Then this set P is an equilibrium to the game described above, i.e. with sequential decision in the first stage of the game. To see this, consider the following strategies of the countries in the first stage of the game: k ∉ P: decide not to participate no matter what other countries decide. j ∈ P: decide to participate provided no other countries i ∈ P have decided not to participate; otherwise decide not to participate.

It is obvious that any country k ∉ P cannot improve the outcome for itself by choosing any other strategy than the one described above. This is also true for any country j ∈ P. The strategies above imply that there will not be any agreement if any country i ∈ P decides not to participate, country j can therefore not do better than to ‘not participate’ in this case. Since country j ∈ P by construction is better off with an agreement than without, it is in this country's interest to behave so that there will be an agreement, i.e. to decide to participate as long as all other countries i ∈ P have decided to participate.

Appendix 5: Properties of the Subgame Perfect Equilibrium Define(A5.1)

and(A5.2)

We want to prove that F−j′(s) ≠ 0 for some j. We start by proving that F(s) < 0. The function fi(s) is given by (13.37), which may be written as(A5.3)

where gj′ < 0 since Rj″ < 0 and Dj″ > 0, cf. (13.37). But this means that(A5.4)

where G′ < 0. (A5.4) implicitly defines ∑iVi as a function of s, it is this function we have called F(s). We thus have, using F=d(∑iVi)/ds

i.e. F′(s) < 0, since G′ < 0. Moreover, since ∑iVi=S−s, it follows from (A5.4) that

i.e.(A5.5) Assume that F−j′(S) ≥ 0 for all j. Then it follows from (A5.2) that F′(s) ≥ fj′(s) for all j. Summing over all j gives NF′(s) ≥ F′(s), which contradicts F′(s) < 0. We therefore must have F−j′(s) < 0 for some j. If all countries are equal, F−j(s) will be the same for all countries. But since F−j′(s) < 0 for some countries, this inequality must hold for all countries if countries are equal. When countries differ, we may have F−j′(s) > 0 for some countries. Nevertheless, it is of interest to see the implication of F−j′(s) ≤ 0 for all j, which obviously holds also for many cases in which countries differ. Eq. (A5.5) follows from (13.37), which holds both under the open-loop and the perfect equilibrium. Using superscripts O and P for the open-loop and the perfect equilibrium, respectively, it therefore follows that SP − SO and sP − sO have the same sign. If F−j(s) ≤ 0 for all j, then SP > SO and sP > sO. To see this, assume the opposite. Comparing (13.35) with (13.40), sP ≤ sO, SP ≤ SO and (with a strict inequality for at least one j) must imply that for all j, with a strict inequality for at least one j. But this implies sP > sO, which contradicts our assumption. We thus have sP > sO and SP > SO if F−j(s) ≤ 0 for all j.

367

REFERENCES BARRETT, S. (1989), ‘On the Nature and Significance of International Environmental Agreements’, (London: London Business School), mimeo. BARRETT, S. (1991), ‘Reaching a CO2 Emission Limitation Agreement for the Community: Implications for Equity and Cost-Effectiveness’ (London: London Business School), mimeo. CARRARO, C., and D. SINISCALCO (1992), ‘The International Dimension of Environmental Policy’. European Economic Review, forthcoming. FERSHTMAN, C. (1989), ‘Fixed Rules and Decision Rules: Time Consistency and Subgame Perfection’. Economics Letters 30. HOEL, M. (1991a), ‘Efficient International Agreements for Reducing Emissions of CO2’. The Energy Journal, 12/2. HOEL, M. (1991b), ‘Emission Taxes in a Dynamic Game of CO2 Emissions’, in R. Pethig (ed.), Conflict and Cooperation in Managing Environmental Resources, Springer, forthcoming. HOEL, M. (1991c), ‘International Environment Conventions: The Case of Uniform Reductions of Emissions’. Environmental and Resource Economics, forthcoming. HOEL, M. (1992), ‘How Should International Greenhouse Gas Agreements be Designed?’. In P. Dasgupta, K. -G. Mäler, and A. Vercelli (eds.), Economics of Transnational Commons (Oxford: Oxford University Press). KVERNDOKK, S. (1991), ‘Global CO2 Agreements: A Cost-Efficient Approach’, Centre for Research in Economics and Business Administration, Bergen, mimeo. MICHAELIS, P. (1991), ‘Global Warming: Efficient Policies in the Case of Multiple Pollutants’, The Kiel Institute of World Economics, mimeo. REINGANUM, J. F., and N. L. STOKEY (1985), ‘Oligopoly Extraction of a Common Property Natural Resource: The Importance of the Period of Commitment in Dynamic Games’. International Economic Review, 26/1: SELTEN, R. (1975), ‘Re-examination of the Perfectness Concept for Equilibrium Points in Extensive Games’. International Journal of Game Theory, 4. ULPH, D, A. ULPH, and J. PEZZEY (1991), ‘Should the Carbon Tax Rise or Fall over Time’, University of Bristol and University of Southampton, mimeo.

This page intentionally left blank

PART V Unidirectional Externalities

This page intentionally left blank

14 Analysis and Management of Watersheds Water resources have a long tradition of both physical and economic analysis. Some of the earliest applied welfare economics focused on development of water resources in large watersheds: benefit-cost analysis was developed in the USA in response to a 1936 legal requirement for economic analysis of the benefits and costs of proposed waterresource projects. The famous Green Book, published in 1950 (US, 1950), codified the general principles of economic analysis that resulted (Hufschmidt et al., 1983). Analysis of water-resources development attracted increasing attention from professional economists in the following years and the issues raised became a major take-off point in natural-resource and environmental economics. Among those writing on these issues were Eckstein (1958), Krutilla and Eckstein (1958), Hirshleifer (1960, 1965), McKean (1958), Maas et al. (1962), and Dorfman (1965, 1972). The focus was on practical issues such as multi-purpose river development, the optimum size of a project, the sequencing and timing of projects, the apportioning of benefits and the determination of cost-allocation rules, and the conjunctive use of surface water and groundwater resources. More recently innovative thinkers have applied game-theory approaches and negotiating approaches to the allocation of water rights within and between countries (Young et al., 1982; Dufournaud and Harrington, 1990; Rogers, 1991). Although watershed development (the development of water resources in a watershed) and watershed management are not the same, there is a great deal of overlap between the two, both physically and conceptually. The writings mentioned above largely dealt with capital investment decisions, and the fact that these activities occurred in a watershed was almost incidental. At the same time, other researchers were concerned with socio-economic issues found in a watershed and how inhabitants used land and water resources to produce income. The concept of ‘integrated rural development’ that was popular several decades ago was a recognition of the links between people, resources, and economic growth. The modern concept of watershed

372 management has grown out of these two approaches—managing water resources and managing socio-economic systems. What is watershed management? Watershed management, therefore, is more than just the proper economic analysis of water-resource projects. Although watersheds are physically defined units that have an inherent economic logic, and the analysis and protection of water resources is an important aspect of watershed management, an understanding of the interactions between land, water, and people for the production of goods and services is an equally important part of this topic. Traditionally the field of watershed management had a strong hydrologic focus, particularly on the use of structural (engineering) as well as nonstructural (vegetative) practices to control the quantity, quality, and timing of water flows (Brooks et al., 1991; Pereira, 1989). Watershed-management practices were defined in terms of changes in land use, vegetative cover, and other actions that affected the hydrology of the watershed. Although economic factors entered, they were usually given secondary importance. Another view places greater emphasis on the economic/social systems found in a watershed and how individuals and societies interact with physical resources. This socio–economic perspective considers water-resource management and its economic analysis as a key component, but not sufficient in and of itself (Easter, Dixon, and Hufschmidt, 1986). In this view watersheds are approached as complex biophysical and socio-economic units that require management that combines both economic and hydrologic factors. Watersheds have been studied from various perspectives. Water-resource economists study the issues related to economic efficiency of water use and investments. Physical scientists examine the causes and magnitudes of phenomena such as stream flow, flood run-off, soil erosion, and sedimentation. Watersheds are also studied by social scientists as units containing various social and political groups. The challenge of watershed management. The history of watershed management has been very mixed. The major success story that is usually cited is that of the Tennessee Valley Authority—the TVA; this large system was developed to provide power, flood control, and economic development benefits to a largely rural, economically depressed region of the USA. It has been successful on both accounts—as a hydrological system and in terms of generating economic growth.53 Other successful examples of integrated watershed management, especially in developing countries, are more difficult to find. More often one hears of problems related to watersheds and their management. Much of this

53

There are a number of other examples of macro-level watershed planning including the Danube River Commission in Europe, the Damodar Valley Authority in India, and the Mekong Commission in South-east Asia. Although the Danube River Commission has had some success, the Damodar experience has been more mixed and the Mekong Commission has been severely hampered by the Vietnam War and its aftermath.

373 concern is due to high and growing levels of soil erosion and resulting downstream impacts. Whereas the TVA was able to reduce land-use pressures and soil erosion by promoting economic development in a depressed region (largely by the provision of cheap electric power), many watersheds today are experiencing worsening soil-erosion problems even as major investments are being made to improve their management. In a survey of Asian watersheds (many of which are densely populated and intensively used) Magrath and Doolette (1990) identified six major watershed problems: loss of agricultural productivity due to erosion; deforestation; population and poverty; downstream sedimentation; flooding; and reduced dry-season stream flows. These problems are also common in watersheds in other parts of the world. Many of these problems are especially acute in poorer developing countries where growing populations are exerting intense pressure on increasingly scarce land and water resources. The causes of soil erosion and accelerated run-off of water and soil are several: more people live and work on the land, and, with growing populations, existing cultivated land is used more intensively and more marginal lands are being brought into cultivation. This expansion of the agricultural frontier results in the cultivation of steeper lands in the upper watershed and the shortening of fallow periods in other areas. Combined with intense tropical rains, the predictable result is increased soil erosion. The same population and economic pressures result in the overgrazing of other areas, with similar devastating impacts on soil structure and soil-erosion rates. Deforestation is a contributing factor in many areas. Not only does it result in the removal of valuable natural habitats and the frequent waste of large amounts of the biomass at the site, logging is also very erosive, particularly the construction of logging access roads. Logging roads are commonly built in areas with steep topography, and are often built very hastily and are poorly aligned. They are designed to be used for a very short period and then abandoned. As a result, they generate tremendous amounts of soil erosion. In a logging area in Palawan in the Philippines, logging roads accounted for less than 3 per cent of the drainage basin area but created 84 per cent of total surface erosion (Hodgson and Dixon, 1988). Logging roads also provide access to previously inaccessible sites, resulting in the immigration of people and consequent agricultural development, leading frequently to further land clearing and soil erosion. Watershed management poses many challenges: land- and water-management decisions create environmental effects that are felt both as production changes in upland fields and through the impacts of soil and water downstream. There are also social impacts, due to both cultural differences between upland and lowland residents and the differential. This chapter examines watersheds as both physical management units and as economic units. Although there is a welldeveloped literature on each

374 approach, analysis of the intimate relationship between the two, and the social-welfare implications of management decisions, yields valuable insights for the design of improved management systems.

14.1 WATERSHEDS AS PHYSICAL SYSTEMS Watersheds vary tremendously in size, ranging from as small as 100 km2 to ones that include large parts of entire countries. Although technically all land belongs to one or another watershed, the focus is usually on smaller units, especially those in upper watersheds where the interactions of physical and social forces can be most dramatic. Some authors divide watersheds into categories determined by slope; Magrath and Doolette (1989), for example, define the ‘upper watershed’ as land having a slope exceeding 30 per cent, and the ‘lower watershed’ as land with a slope of 8–30 per cent. Flatter areas with slopes of less than 8 per cent are excluded. Table 14.1 defines several watershed-related terms. These definitions differentiate between the physical units—the watershed and the larger river basin—and the process of planning Table 14.1 Definitions of watershed-related terms A watershed is a topographically delineated area that is drained by a stream system. The watershed is a hydrologic unit that has been described and used both as a physical-biological unit and as a socio-economic and sociopolitical unit for planning and implementing resource-management activities. A river basin is similarly defined but is of a larger scale (e. g. the Mekong River Basin, the Amazon River Basin, and the Mississippi River Basin). Integrated watershed management is the process of formulating and implementing a course of action involving natural and human resources in a watershed, taking into account the social, political, economic, and institutional factors operating within the watershed and the surrounding river basin and other relevant regions to achieve specific social objectives. Typically, this process would include: (1) establishing watershed-management objectives; (2) formulating and evaluating alternative resource management actions involving various implementation tools and institutional arrangements; (3) choosing and implementing a preferred course of action; and (4) thorough monitoring of activities and outcomes, evaluating performance in terms of degrees of achievement of the specified objectives. Watershed-management practices are those changes in land use, vegetative cover, and other non-structural and structural actions that are taken in a watershed to achieve watershed-management objectives. The watershed approach is the application of integrated watershed management in the planning and implementation of resource management and rural development projects or as part of planning for specific resource sectors such as agricultural, forestry, or mining. Imbedded in this approach is the linkage between uplands and lowlands in both biophysical and socio-economic contexts. Source: Easter and Hufschmidt (1985); Brooks et al., (1991).

375 for management of the watershed or parts of the watershed. The definition of the ‘watershed approach’, the combination of biophysical and socioeconomic variables, is very similar to the approach used here. In a geographic sense, watersheds are topographical units defined by the flow of water: the hydrologic cycle is a key factor in any watershed. As seen in Figure 14.1 this cycle is defined by rainfall, evapotranspiration, and stream-flow. All water enters or leaves the system through one of these three means. Waterflow can be both overland, along the surface of the earth, as well as subsurface, with eventual recharge to the water-table. Both surface and subsurface water (groundwater) form part of the return flow to streams, lakes, or the ocean. Land-use patterns and changes have direct effects on the movement of water and on soil erosion. The hydrologic and soil aspects of watersheds are discussed in Hamilton with King (1983), Brooks et al. (1991) and, in the case of tropical watersheds, Pereira (1989). Although even undisturbed soils and natural vegetative systems have some degree of natural soil erosion, a major management issue is the effect of human actions on the rate of soil erosion in a watershed. According to Hamilton and Pearce (1986) the biophysical effects of different land uses, both natural and disturbed, in a watershed can be summarized under six headings: Fig. 14.1 The hydrologic cycleSource: Hamilton and Pearce in Easter, Dixon, and Hufschmidt (1986).

376 • • • • • •

soil erosion at the land-use site harmful sediment off-site pollution of water by chemicals changes in total water yield in streams in the watershed changes in the distribution and timing of water delivered in the watershed—low flows as well as floods changes in the water-table.

Of the six factors only the third—pollution of water by chemicals—is entirely the result of human actions. The other processes have both natural and human origins. Soil erosion and resulting sedimentation have received the most attention as watershed-management problems. This is due to both the productivity effects of soil erosion (the on-site impacts) and the costs that are associated with sedimentation downstream (off-site impacts). These are now considered in greater detail.

14.1.1 Soil erosion Soil erosion is the physical process of removing part of the soil in one place and transporting it to another location. Soil erosion is a flow concept, and is measured as a rate: x tons/ha. /yr. Although most of the world's richest agricultural areas are the result of soil erosion and sediment deposition by major rivers (e.g. the Nile, the Ganges, and the Yellow River), prehistoric delta and floodplain development occurred in a setting where there was little if any man-made infrastructure that could be damaged by floods or sedimentation. Today there are major investments in infrastructure—e.g. dams, irrigation systems, and ports—cities and other facilities in most downstream areas. Losses of property from flooding, and damage to human health are both potential costs of changes in water quality and quantity. The costs of soil erosion are thus large and growing as increased populations settle in these downstream areas. As explained by Hamilton and Pearce (1986), erosion is a serious problem because it has impacts on both upstream and downstream areas. Upstream, soil erosion leads to loss of productivity and the loss of water-storage capacity on eroded sites. Downstream, the eroded soil is deposited as sediment although in some cases this may prove to be beneficial through the creation of fertile deltas or river-bottom land. More often, sedimentation results in damage to downstream fields, river channels, and capital infrastructure such as dams, water systems, and irrigation channels, thereby imposing costs on others. Soil erosion includes a range of phenomena including sheet erosion (the removal of thin layers of topsoil from an eroded site), gully erosion (the formation of incised gullies in a hillside), and mass wasting (the structural failure of part of a hillside as in the case of a landslide). Although all of these types of erosion occur in nature and are natural processes, man's use of watershed lands, particularly steeper upland areas, can greatly increase the extent and rate of

377 erosion. A wide range of activities can lead to increased soil erosion: agriculture, logging, and construction activities, for example, roads, dams, or homes. Natural vegetative cover is one of the best forms of protection against accelerated erosion. Forests, grasslands, and mixed vegetation provide the canopy to intercept rain and absorb part of its energy; vegetative litter on the soil surface also helps to reduce the erosive force of falling or flowing water. When water is slowed down by vegetation, roots, and litter as it flows across the surface of the soil, it has a greater chance to be absorbed into the soil. The root structures of trees and plants and their ability to hold the soil together and to allow water to penetrate to the water-table is of even greater importance in steep areas. The impact of vegetative cover is illustrated by the Universal Soil Loss Equation(USLE).54 The USLE is usually defined as GE = R × K × LS × C × P where gross erosion (GE) is determined by multiplying factors for the energy content of rainfall (R), soil erodibility of specific soil types (K), a topographic factor based on slope and length of the field (LS), a dimensionless cropping management factor (C), and erosion control practices (P) (Hufschmidt et al., 1983; Brooks et al., 1991). The cropping practices factor, C, is an annual value based on a numerical ratio that relates annual soil loss from a vegetated area to that from the same area if it were bare. These values are country- or region-specific and depend on soil types and other variables. Their range is very large: in West Africa, in comparison with a base value of 1.0 for bare soil, Roose (1977) estimated annual average C factors that ranged from 0.2–0.8 for first-year cassava or yams, to 0.1–0.2 for rice, to as low as 0.01 for savannah and 0.001 for forest, dense shrub, or high mulch crops. These numbers are very instructive—bare soil has 1,000 times the erosion potential of forest or dense shrub; when crops are grown the difference decreases: bare soil is potentially five to ten times as susceptible to erosion as rice and not a great deal more erodible than first-year cassava. Leaving soils with their natural cover is obviously good for preventing soil erosion but often less efficient in terms of producing the many goods and services demanded by watershed inhabitants than cultivated fields. The trade-off is thus between more intensive land use and the potential for increased erosion. As the West Africa figures show, however, there is a wide range of potential land uses, some of which are much more erosive than others. Actual erosion rates vary from less than 1 ton/ha./yr. on forested lands to more that 100 tons/ha./yr. on sloping, open fields. Values from 5 to 30 tons/ha./yr. in agricultural areas are not uncommon. In the hilly Phewa Tal catchment of Nepal, for example, estimated erosion rates ranged from 8 tons/ha./yr. for

54

Although the USLE was formulated based on observations in the north-east USA, it captures the general process of erosion and has proved useful in many parts of the world. Researchers have made estimates for the different factors to reflect soil and rainfall characteristics in many other countries.

378 forest lands, to 10 tons/ha./yr. for terraced fields, 15 tons/ha./yr. for scrub forest land, and 35 tons/ha./yr. for open grazing land (Flemming, 1983). (Of course soil erosion is affected by much more than merely the C factor. The other variables in the USLE have their impact and some are amenable to change, such as cropping practices and field size and slope, while others, such as the rainfall and soil quality variables, are given. For an excellent discussion of the impact of different patterns of forest use and land conversion on erosion rates and water flows in a tropical watershed, see Hamilton with King, 1983.) Costs of soil erosion. Estimates of the costs of soil erosion in developing countries have traditionally focused on the onsite, productivity costs: the decrease in agricultural production as soils become thinner and lose valuable nutrients. More recently, analysts have also begun to include the off-site, downstream impacts, which, though important, are frequently smaller than on-site impacts. In an analysis of soil erosion on Java, Magrath and Arens (1989) found that the major portion of the annual cost of soil erosion was agricultural productivity losses on eroded fields (over 90 per cent of the total annual loss of about $340 million) and that off-site, downstream costs (largely due to sedimentation) to irrigation systems, harbours, and reservoirs were relatively small, less than 10 per cent of the total. (This study is presented in greater detail later.) In a similar analysis of soil-conservation activities in the Loess Plateau region of China, Magrath (1992) found that the on-site agricultural productivity benefits of soil conservation, just as in Java, were much larger than the downstream benefits. The internal rate of return of the project was about 14–16 per cent if only on-site, upper watershed benefits were included; if the downstream benefits from reduced sedimentation are also included, the IRR marginally increases to 16–18 per cent. In a study that focused on only the on-site costs of soil erosion in Mali, Bishop and Allen (1989) used land-resource information and the USLE to estimate the annual loss of net farm income from soil erosion. Their estimate of annual losses ranged from $4.6 to $18.7 million. Since erosion in one year has cumulative productivity impacts (since soil is not regenerating fast enough to replace lost soil), the estimate of the present value of current and future net farm income lost (using a ten-year time-horizon and a 10 per cent discount rate) ranges from $31 to $123 million, 4–16 per cent of Mali's agricultural GDP. These estimates of losses were then compared to the costs of soil-erosion prevention measures. The authors found that selected investments could be justified on economic grounds, but noted that the low adoption rates indicated that either farmers were using a much higher implicit discount rate or there were other factors that had not been taken into account in the analysis.

14.1.2 Sedimentation If soil erosion is a watershed-management problem largely caused by human use of land and water in the watershed, sedimentation is the downstream,

379 off-site, physical result of this process. Sedimentation can be thought of as the stock analogy of soil erosion; it is defined as the deposition of soil particles by water. Soil may also be transported by wind but this is usually less of a concern in most tropical watersheds. Sedimentation is not an easily defined process. One can measure erosion rates in the upper part of a watershed and sedimentation rates in a reservoir or irrigation channel downstream. In between, however, there is a complicated set of interactions between the soil suspended in the water, the stream channel, and the stream bed. In all watersheds, large as well as small, sediment transport is a process whereby soil is deposited and released from numerous points or storage sites within the system. In larger systems the transport process can take months, weeks, or years from initial soil removal upstream to final deposition in the lowlands or at the river mouth on the coast. Observed sedimentation today is the result of actions taken several periods earlier. Major storm events greatly effect sedimentation. In some cases 50 per cent or more of yearly sediment delivery to a site occurs in one major storm event. Although we tend to think of soil erosion and resulting sedimentation as linear and deterministic, these processes are more often non-linear and stochastic, with a great variability in mean levels. This makes measurement of cause and effect more difficult and increases the uncertainty about the efficacy of proposed remedial measures. Since tremendous amounts of soil are already in movement in larger watersheds, preventing all future upstream erosion, even if it were possible, may have no appreciable effect on sedimentation rates downstream. Sediment imposes various costs on society. Increased sediment in streams, lakes, and reservoirs can harm or kill valuable aquatic life; impair water quality for industrial and domestic uses; reduce reservoir-storage capacity for hydropower, flood control or agricultural purposes; damage turbines and other equipment; and build up stream beds resulting in increased risk of flooding. The process of reservoir sedimentation is of particular interest because of the very large capital investments in dams and reservoirs, investments that are lost if reservoirs prematurely silt up. Mahmood (1987) presents an excellent overview of the impact, extent, and mitigation of reservoir sedimentation. Controlling sedimentation. Sedimentation can be minimized by preventing soil erosion in the first place. Note, however, that the long time-periods involved in sediment transport and delivery may well mean that any soil-erosion control measures will not have an impact on sediment delivery rates for a considerable period of time. This problem is compounded when economic analysis is performed; any positive discount rate will quickly reduce the value of benefits that accrue a long time in the future. Nevertheless, some watershed experts state that in many cases soil-erosion measures should not be undertaken to prevent sedimentation of existing dams or other structures, but rather, to protect future investments. While it may be an extreme position to

380 state that so much sediment is already in transport that soil-erosion control measures are unlikely to help protect existing investments, the idea does contain a great deal of truth. The optimal time to begin soil-erosion control measures is before the dam or irrigation system is built, not once excessive sediment starts to build up behind a dam or in an irrigation canal. (There are some measures that can be usefully implemented once a dam is built but before serious problems arise. These include vegetative barriers along streams and the edges of the reservoir, and limited structural works.) Physically trapping or removing sediment once it is in transport is an expensive ‘after-the-fact’-type approach. In addition to the standard soil- and water-conservation practices promoted for upland areas, the replanting of vegetation in the degraded areas and the use of buffer strips of vegetation along streams can greatly reduce the inflow of sediment into streams, and hence into the system (Doolette and Smyle, 1990; Banerjee, 1990). Buffer strips work by both slowing the overland flow of water and reducing its erosive energy as well as by physically trapping soil. Vetiver grass is being touted as an effective, inexpensive natural means of controlling both soil erosion and sediment transport. It is being used in many parts of the world including Fiji, India, China, and Indonesia (World Bank, 1990; Magrath and Doolette, 1990). The use of other grass and vegetative barriers to obtain similar benefits is being actively pursued. Many approaches have been tried to reduce sediment inflows or remove sediment once it has been deposited in a reservoir. Mahmood (1987) discusses the use of watershed management in the upper watershed and the use of ‘debris dams’ (small dams on major sediment-contributing tributaries). These dams are seldom cost-effective given their small size and short useful life. Once sediment reaches a reservoir it can be removed by flushing, sluicing, and dredging. Flushing is the process of hydraulically clearing accumulated sediment by opening dam gates to allow water to leave quickly, picking up deposited sediment with it. The physics of this process are such that it is only rarely used successfully.55 Sluicing also has many limitations (see Mahmood, 1987). Dredging is very expensive; Mahmood estimated the cost of dredging alone (without disposal costs) as $2–$3 per m3, about twenty times the cost of constructing replacement storage capacity. In some selected cases dredging may be viable, but generally it is not economically attractive.56 An appropriate

55

In China, authorities have had considerable success in managing major dam/reservoir systems on heavily sediment-laden rivers. The process involves using variations in annual flows and selective opening of gates to maintain a dynamic situation that keeps sediment entrained in the river's water. Referred to as ‘storing sediment in the dry season, flushing sediment in the wet season’, this approach is a practical response to the sediment problem and operates the dams on, what is in effect, a ‘run-of-the-river’ basis.

56

Since dams and reservoirs are classic examples of sunk costs, the value of the last units of useful storage capacity may be very high and justify sediment-removal costs. In general, however, there is an order of magnitude difference in the costs of either preventing sedimentation (or increasing dead storage capacity in the design phase) and removing it once it is deposited; the latter is almost invariably more expensive.

381 control strategy combines soil-conservation measures in the upper watershed with adequate provision of dead storage capacity in new reservoirs.

14.2 WATERSHEDS AS SOCIO-ECONOMIC SYSTEMS Even if the hydrology and physical processes of water and soil movement are very important factors, watersheds can also be described as socio-economic systems that are, in many cases, coterminous with the physical system. This overlay of the physical dimension by the socio-economic element is seen in Figure 14.2. The natural system on the left of the figure has its counterpart in the social system on the right-hand side. Both dimensions need to be explicitly considered; the physical factors determine the rates of soil and water movement while the socio-economic factors determine the size and Fig. 14.2 A natural and social system schematic of a watershedSource: Hamilton and King (1984).

382 distribution of benefits and costs. The interplay of physical processes and human use of the resources within the watersheds creates both the problems of watershed management as well as the potential for the production of goods and services that lead to increases in human welfare. The socio-economic side is often ignored because of the difficulties involved in dealing with social issues (Blaikie, 1985). The socio-economic dimension can usefully be broken down into social and economic components: the social dimension defines the institutions and organizations involved in implementing policies and programmes while the economic dimension identifies the size and distribution of outcomes. The concept of environmental externalities has a one-to-one correlation with that of economic and social externalities; all are commonly encountered in the watershed. Physical flows of water and soil are usually unidirectional, from the upper watershed to areas downstream. Therefore, most of the externalities are imposed by upstream areas on to those areas further downstream. Monetary flows can go in either direction. The general pattern, however, is for political and economic power to be concentrated in the downstream areas and reverse monetary flows—from downstream to upstream—are usually quite small.

14.2.1 The sociology of watersheds Changes in land use and the development of new infrastructure have been identified as two of the major causes of soil erosion and sedimentation in many watersheds. Economic analysis helps identify the benefits and costs of alternative management patterns. Implementing suggested improvements, however, is frequently complicated by social and political factors. A divergence between the incidence of benefits and costs, combined with political mistrust, often results in poorly implemented programmes. Although not true for all watersheds, in many countries there is a marked difference between those groups inhabiting the upper watershed and those who live further downstream. This may be seen in the location of political (and financial) power and the relative flows of benefits. Dani (1986) examined these interactions in the case of the HinduKush, Himalayan region, and identified the patterns of decision-making, resource inputs, and flows of benefits. As seen in Figure 14.3, these patterns define the political and social interactions of the lowland and upland inhabitants in a watershed, often to the benefit of lowland communities. Upper watersheds, in addition to being physically remote, are often politically remote as well (Magrath and Doolette, 1990). Upland groups, not infrequently members of minority groups, are viewed as ‘different’ and outside of the national mainstream. They are either ignored or absorbed into the dominant national groups, often leading to the alienation of the upland group. Alienation can be both cultural and material. This process, as seen in Figure 14.3, results in an unequal division of the benefits from improved

383

Fig. 14.3 Highland-lowland relationships in watershed managementSource: Dani in Easter, Dixon, and Hufschmidt (1986).

watershed management. A disproportionate share goes to the lowland group, even though the resources required efficiently to implement the policies are disproportionately paid for by the upland group. This separation of benefits and costs is not surprising given that the main opportunities to conserve soil are in the upper watershed while a larger share of population and infrastructure are located in the lower part of the watershed. One result of this process is the underdevelopment of the upland group. This is frequently accompanied by political and social tensions, and an ‘usthem’ attitude. Although the foregoing is a stylized version of a reality, there is enough truth in this view to help explain the rather dismal record of many watershed-management projects. Those who are being called upon to implement new cultivation or land-management practices often do not see themselves as benefiting from the changes. Those downstream who are negatively affected by existing upland cultivation or land-management patterns see upland residents

384 as a difficult, somewhat foreign group that needs to be ‘managed’. Obviously these tensions do not contribute to effective communication or effective resource management. Increased public participation of individuals and groups in the design and implementation of projects that will affect them is a common-sense, but often ignored requirement for successful development. The 1992 World Development Report of the World Bank, Development and the Environment, stresses the importance of participation in creating effective programmes and viable resource-management institutions. This theme is explored at length in an FAO Conservation Guide (Bochet, 1983) that focuses on the role of mountain communities in the design and implementation of watershed-management programmes in upland areas.

14.2.2 The economics of watershed management In spite of the complex interactions of physical and social forces within a watershed, economics has much to contribute to improved analysis and management, even if in many cases one is in the world of second-best solutions rather than in the optimal world of unconstrained maximization. The economics of watershed management is not inherently complicated; in fact, most of the approaches used are based on standard project analysis techniques. Valuation of certain goods and services may be an issue, but considerable progress has been made in this area. Frequently, the biggest analytical problem is understanding the complex biophysical interactions within a watershed, e.g. what are the impacts of soil erosion on crop production and downstream areas? Which management measures and institutional frameworks really work? Nevertheless, the power of economic analysis, especially social-welfare-based analysis, is that it presents a coherent, rigorous way to incorporate the various activities within a watershed. By using the watershed as the unit of account (the ‘project’ under consideration), few of the impacts of alternative management actions are external to the system. Almost all impacts are now within the boundaries of the unit of account and, since the social-welfare-based analysis is comprehensive, all benefits and costs are included in the analysis. Economic externalities largely disappear. The measurement and monetary evaluation of the various activities of interest within the watershed are discussed in a number of references (see Tolley and Riggs, 1961; Easter, Dixon, and Hufschmidt, 1986; and Gregersen et al., 1987 for examples of the approach and selected case-studies). The rapidly developing fields of applied resource and environmental economics are providing useful examples of valuation of environmental and resource impacts associated with watershed-management activities (e.g. the studies of Magrath and Arens (1989), and Bishop and Allen (1989) cited earlier). Three cases are presented here: the environmental impacts of dams and reservoirs

385 and their economic analysis; the economic valuation of soil erosion on Java; and economic analysis of management alternatives for three ecologically linked, resource-dependent industries in a Palawan watershed. Dams and the environment. Dams are a classic case of a large, capital-intensive, water-resource investment in a watershed. Dams, particularly large ones, are built by Governments to supply various social benefits—water supplies, power, irrigation, and flood control among others. Dams are also one of the more disruptive infrastructural investments found in a watershed. Dams frequently require that access be opened up to remote or previously inaccessible sites, often in the upper watershed. The process of construction and associated infrastructure development may result in a large immigration of people into the area. Roads themselves can be very destructive and lead to severe erosion problems; in addition, roads open up other areas, public or not, to settlement and development. This can also lead to soil erosion and sedimentation downstream. Dams also lead to the displacement of people from the reservoir site. This is particularly true of dams located in the developing nations of Asia, Africa, or Latin America. The process of resettlement imposes both costs and social strains on the Government and those who are to be resettled. The population in the receiving area may also resent the new settlers. Resettlement problems, both financial and institutional, are among the most difficult ones to address adequately for new dam projects. The ongoing controversy over the Narmada River dams in India largely focuses on the resettlement issue.57 In terms of environmental effects, dams are affected by what takes place above them in their watershed. Increased erosion in the upper watershed and resulting sedimentation in the reservoir may reduce the useful life of the dam. Dams themselves also have major impacts on the environment and on those living on the dam site and in the areas below the dam. A recent World Bank Technical Paper, Dams and the Environment (Dixon, Talbot, and LeMoigne, 1989), examined the environmental impact of dams within the physical, social, and economic content of the watershed. The argument is made that dams should be evaluated within the larger watershed context to account more fully for their economic benefits and costs, wherever they occur. The primary benefits and costs are those associated with the construction and operation of the dam and the direct production of project outputs, such as power, water, and irrigation. A complete economic analysis, however, must take into account the benefits and costs of the externalities created by the dam. (The Hells Canyon

57

The Sardar Sarovar Dam on the Narmada has provoked a great deal of controversy, primarily centred on the resettlement of those affected by the reservoir. While the dam authorities have a responsibility adequately to resettle those affected, very little attention in the public debate has focused on the millions of potential beneficiaries of the project. One estimate is that the reservoir and water distribution system will affect about 98,000 ha. and about 240,000 people while the project will irrigate from 1.2 to 1.8 million ha., benefiting about 40 million people (Seckler, 1992). Proper analysis of such a project should consider the economic welfare of the area with and without the project, not only the cost to those who must be resettled.

386 study of Krutilla and Fisher (1975, 1985) is a classic study of a proposed dam, power-generation alternatives, and environmental benefits and costs.) In some cases, environmental impacts act on the dam itself, e.g. land use in the watershed above the dam site, while more often the environmental impacts of the dam fall on those living around the dam site or downstream. Each impact can be identified and, in many cases, measured and valued. Table 14.2 presents a list of selected environmental effects and their valuation. For each effect the economic impact is noted and whether or not it is a benefit or cost. Representative valuation techniques are also noted.58 Many of the environmental effects can be both beneficial and harmful. For example, the impact of sediment trapped in the reservoir can result in a reduction of the silt load downstream of the reservoir. This can be both a benefit and a cost to different downstream groups. Reduced silt in the water means fewer nutrients carried in the water (a loss) while less sediment may also mean reduced sedimentation of downstream irrigation canals (a benefit). Similarly, involuntary resettlement is usually only considered as a cost whereas in some cases, if the quality of life and social services provided after resettlement are superior to the situation before, there may be net benefits from resettlement. Not all of the environmental effects of large dams can be valued in monetary terms. Still, many can be accommodated within an expanded economic analysis, especially when externalities are explicitly taken into account. Soil erosion in the uplands of Java. In an attempt to answer the question ‘what are the magnitudes of the economic costs of soil erosion in the uplands of Java?’, Magrath and Arens (1989) estimated both the on-site, productivity costs of soil erosion and the off-site, sedimentation-related costs of soil erosion on Java. In order to understand the physical processes involved, the impacts of those processes on the production of desired goods and services, and the responses of those affected, three linked models were developed: (1) a geographic information-based system (GIS) model was developed to integrate data on soil type, topography, rainfall, and land use to estimate the levels and distribution of erosion; (2) a soil erosion-productivity model was developed for rainfed agricultural land to estimate on-site productivity impacts; and (3) an economic model of farmer response to falling productivity was used to value the effects of erosion. Additional analysis was carried out on the economic costs of downstream sedimentation and waterquality impacts on reservoirs, irrigations systems, and coastal areas. Separate estimates were made for the four provinces on Java; both on-site, agricultural productivity impacts were measured as well as off-site sediment-linked costs due to siltation of irrigation systems, harbour dredging to remove

58

The representative valuation techniques listed are not all-inclusive. In the case of health and resettlement impacts, for example, difficult valuation issues like the value of a human life or the value of psychic costs associated with resettlement are not really covered. This is not to say that these issues are unimportant; but merely to indicate that they are not well handled by available economic valuation techniques.

387 Table 14.2 Selected environmental effects of large dams and their economic impacts Environmental effect

Economic impact

Environment on dams 1. Soil erosion—upstream, Reduced reservoir capacity; sedimentation in reservoir change in capacity; change in water quality; decrease in power Dams on the environment 1. Chemical water quali- Increased/reduced treatty—changes in reservoir ment cost-reduced fish catch, loss of production and downstream 2. Reduction in silt load, Loss of fertilizer, reduced downstream siltation of canals, better water-control 3. Water temperature Reduction of crop yields changes (drop) (esp. rice) 4. Health—water-related Sickness, hospital care, diseases (humans and ani- death; decrease meat and milk production mals) 5. Fishery—impacts on fish Both loss and increase in irrigation, spawning fish production 6. Recreation—in the res- Value of recreation opporervoir or river tunities gained or lost, tourism 7. Wildlife and biodiversity Creation or loss of species, habitat and genetic resources

Benefit (B) Cost (C)

Representative valuation technique

B, C

Change in production, preventive expenditures, replacement costs

B, C

Preventive expenditures, changes in production

B, C

Replacement costs, preventive expenditures avoided Changes in production

C B, C B, C B, C B, C

8. Involuntary resettlement Cost of new infrastructure, B, C social costs 9. Discharge variations, Disturbs flora and fauna, C excessive diurnal variation human use, drownings, recession agriculture 10. Flood attenuation Reduces after-flood culti- B, C vation; reduces flood damage. Source: Dixon, Talbot, and LeMoigne (1989).

Loss of earnings, healthcare costs, changes in production survey techniques Changes in production, preventive expenditures Travel-cost approach, property-value approach Opportunity-cost approach, tourism values lost, replacement costs, contingent valuation Replacement-cost approach, ‘social costs’, relocation costs, contingent valuation Relocation costs, changes in production Changes in production, flood damages avoided.

This page intentionally left blank

389 silt deposits, and costs of reservoir sedimentation in terms of lost power output and reduced availability of irrigation water. For example, the estimate of the costs of sediment deposition in irrigation systems is based on information on the share of total annual operation and maintenance (O and M) expenditures for irrigation systems that go to silt removal, adjusted for increased removal to maintain efficient operations. (The details can be found in Magrath and Arens, 1989.) Although some of the off-site costs have been estimated others have not, e.g. flooding, damage to coastal fisheries, disruption of urban water supplies, and water pollution by agricultural chemicals costs were not included in the analysis. Table 14.3 presents total estimated annual costs of soil erosion on Java. Costs are estimated at $340–406 million per year, or about 0.5 per cent of total GDP. Over 95 per cent of the costs are the on-site costs of declining soil productivity, affecting millions of farmers. Even though there is considerable uncertainty about the reliability of much of the data, the results indicate that soil erosion is imposing large and growing costs on the users of Java's soils and represents a significant mining of the resource base. This information, in turn, can be used to analyse watershedmanagement alternatives. Successfully Table 14.3 Total estimated annual costs of soil erosion on Java ($m.) West Java On-site 141.5 Off-site Irrigation 1.7–5.7 system siltation Harbour dredging 0.4–0.9 (1984/85) Reservoir sedi- 9.0–41.3 mentation TOTAL 152.6–189.4

Central Java 29.1 0.8–2.7

Jogyakarta 5.7 0.1–0.5

East Java 138.6 1.2–4.0

Java 315.0 7.9–12.9

0.1–0.3

[—]

0.9–2.2

1.4–3.4

3.5–16.3

[—]

3.8–17.3

16.3–74.9

33.5–48.4

5.8–6.2

144.5–162.1

340.6–406.2

Source: Magrath and Arens, 1989; Magrath and Doolette, 1990.

implementing programmes that affect millions of small farmers, however, is a major institutional challenge. Competing resource-users within a watershed in Palawan. The El Nido watershed on the island of Palawan in the Philippines presents the case of a small watershed that contains several competing resource-users. In this case the watershed drains into Bacuit Bay (see Figure 14.4), the home of a successful artisanal fishery and a growing resort industry based on scuba-diving. The water flowing into the bay can affect the health of the coral reefs and, in turn, the productivity of the fishery and the clarity of the water. Divers are attracted by clear water, abundant fish life and healthy reefs.

390 Fig. 14.4 Bacuit Bay and the El Nido watershed in Palawan, the PhilippinesSource: Hodgson and Dixon (1988).

A Philippine logging company has a large concession on the northern part of Palawan. This concession includes part of the El Nido watershed and has resulted in considerable soil erosion from the construction of logging roads and logging activities. The eroded soil is in turn deposited in Bacuit Bay, damaging the coral reef reducing the number of fish in the bay; both impacts reduce the bay's attractiveness to scuba-divers.

391 This case is particularly instructive since each resource-user—the logging company, the fishermen, and the resort industry—has a legal right to operate in the bay or its watershed. And yet, the generation of unilateral economic externalities from soil erosion and sedimentation imposes large social costs on those downstream of the logging operation. Each resource-user is maximizing based on his or her perception of reality. The end-result, however, is the removal of forest cover, accelerated soil erosion, and considerable long-term damage to the fishery and tourism industries. The costs of the environmental externalities created by logging are paid for by the fishermen and the resort operators, and are not included by the loggers in their operating costs. Since many of the logs are exported, these environmental costs represent an export subsidy as the prices of the exported logs do not include all costs associated with their harvesting and processing. (This point is discussed more fully in Dasgupta and Mäler, 1991). A broader economic analysis that incorporates the various benefits and costs of alternative management practices was carried out by Hodgson and Dixon (1988). In the analysis, two extreme scenarios were compared: continued logging with consequent damage to the other two industries (fisheries and tourism) and a logging ban that would help protect the fishery and the tourism industry but at the cost of lost timber production and loss of export earnings. A series of assumptions were made in order to estimate the generation of gross revenues under each scenario. For example, tourism revenues were based on existing and planned facilities, known occupancy rates and average daily expenditures for food, lodging, and services. Logging revenues were based on the quality and quantity of the timber in the concession area, and observed FOB prices. Similarly, fishery revenue is based on historical catch data and observed market prices. (For more details see Hodgson and Dixon, 1988 or 1992.) It was not possible to estimate net benefits (the best measure for making a comparison) because reliable cost data were not available. A ten-year time-horizon was chosen and a 10 per cent discount rate was used. (Sensitivity analysis was carried out with a 15 per cent discount rate but did not change the results.) The results are presented in Table 14.4; these estimates are rough but give an indication of the orders of magnitude involved. Option 1, the logging ban, was clearly superior to Option 2, continued logging. Not only did Option 1 generate larger gross revenues at all discount rates, but the benefit stream under Option 1 could continue indefinitely into the future whereas the logging revenues under Option 2 would disappear after five or six years. The major cost of Option 1 is the lost logging income. The major costs of Option 2 are the decrease in fishery production due to coralreef death, the loss of fish biomass, and the major contraction of the resort business. Not only would the existing highclass resorts go out of business but planned expansion would not take place. This explains the almost $40 million difference in tourism revenues between the two options.

392 Table 14.4 Gross revenues from the tourism, fishery, and logging industry under Option 1 (logging ban) and Option 2 (continued logging). Calculated over a ten-year period with 10–15 per cent discount rates (×$1,000) Gross revenue Tourism Fisheries Logging TOTAL Present value: 10% Tourism Fisheries Logging TOTAL Present value: 15% Tourism Fisheries Logging TOTAL

Option 1

Option 2

Option 1–2

47,415 28,070 0 75,485

8,178 12,844 12,885 33,907

39,237 15,226 (12,885) 41,478

25,481 17,248 0 42,729

6,280 9,108 9,769 25,157

19,201 8,140 (9,769) 17,572

19,511 14,088 0 33,599

5,591 7,895 8,639 22,125

13,920 6,193 (8,639) 11,474

Source: Hodgson and Dixon (1988).

Tourism revenue is very important in this case. As seen in Table 14.4, the generation of gross revenues for fisheries and logging combined is almost the same for Options 1 and 2, with Option 2, continued logging, marginally higher. The potential for tourism revenue really swings the analysis to favour a logging ban. (One could also argue for a logging ban on grounds of equity and sustainable production.) This case points out how a broader watershed-wide analysis includes effects that are normally considered as externalities and leads to a different management decision than would be the case of a micro, component by component approach. The case also raises some interesting questions. If the benefits from a logging ban are so large why don't the fishermen and the resort operators pay the logging company to stop activities in the watershed and prevent further logging? This Coasian solution would be optimal in a world with well-established institutions and adequate information. Unfortunately, in the Philippines institutions are weak and information is far from perfect. The fishermen are poor and have very limited resources and the resort owners are not from the area and lack detailed knowledge of how to operate in the local political situation. As a result, no action was taken and logging continued with heavy sedimentation of the reef. The predicted negative impacts on the fishing and tourism industry are now happening. It is too late for action, the watershed has been logged, the eroded soil is already in transport, and it is too expensive to take preventive measures. Perhaps this lesson can help prevent similar occurrences in other locations.

393

14.3 IMPLICATIONS AND CONCLUSIONS The problem of analysis and management of watersheds is fundamentally one of mixed impacts, mixed institutional jurisdictions, and mixed signals as to what are benefits and what are costs. Even though economic analysis can incorporate the various physical and economic impacts within the unit of account, the reality of the world is that implementation of programmes and activities on an integrative scale is very difficult. In addition, just as there is suspicion between lowland and upland inhabitants, there is also rivalry between various government departments. Governmental institutional structures do not usually follow the physical watershed boundary; they more often slice across the watershed. There are few incentives for cross-departmental cooperation (see Dixon, 1989, for an elaboration on these and the following themes). Management responses. Traditional management measures of regulation and direct control are usually only effective for large individual projects or operations that can be easily identified and monitored (if there is political will). Although we prefer an institutional minimalist approach, there are certain activities and interventions that can be useful. Land-use planning and zoning, for example, can be important in directing developments away from sensitive areas. In other cases, an active government presence, such as in public forest lands, helps prevent encroachment and reinforces government ownership and management. However, these cases are the exceptions rather than the rule. In many watersheds the main source of environmental degradation is the action of thousands of individual decision-makers—farmers, fishermen, and gatherers of forest products. The existence of technical inefficiency creates the opportunity for pure Pareto-superior changes that do not involve trade-offs. In order effectively to influence the actions of large numbers of individuals one must use other measures, usually some form of market-based system, to send signals throughout the economy. Policy reform is frequently needed. The economist's prescription of ‘getting your prices right’ as well as ‘getting your policies right’ is a direct result of the realization of the futility of trying to control directly or micro-manage large numbers of individuals. Policy and price reform include price changes for inputs or outputs, incentives given to promote or discourage the production of certain crops, and the harmonization of government programmes that may be working at crosspurposes. One example of the latter is the not uncommon occurrence of one government department encouraging terracing to prevent soil erosion and increase crop yields, and another government agency distributing heavily subsidized fertilizer, thereby reducing the incentive to undertake expensive soil-conservation measures to increase crop yields. If fertilizer is available and priced cheaply enough, it can be used in increasing quantities to mask the negative effects of soil erosion on crop yields. (The potential for using

394 agricultural price policy to control environmental degradation is explored in Barbier and Burgess, 1992.) Multi-level management. Management of watershed resources will therefore benefit from an intermediate approach that builds on both micro-level analysis, that identifies how and why individuals use resources in a certain way, and macrolevel analysis in search of appropriate policy measures. The negative effects on the environment created by both large projects and individual resource-users must be taken into account. Unilateral externalities should not become unpriced subsidies. The dichotomy between micro-management of large projects and macro-management of individuals creates the need for multi-level analysis and management. At the analytical level, both large projects and the actions of individual farmers must be analysed to understand what is happening. For this purpose the watershed is frequently an appropriate unit of analysis. At the policy or management level, however, a different approach is necessary. Whereas the large project can be directly regulated or controlled, the individual resource-user can be reached effectively only by the use of macro-policies and incentive systems such as price policies, subsidies or taxes, or land reform. In the case of true subsistence farmers who are largely outside the market system, it may be necessary to use community-based group projects. The following lessons can be drawn from an examination of the recent experience with watershed management: First, an integrated, multidisciplinary approach is required. Given the complicated interactions between resources and people, any programme designed from a single perspective (geological, economic, hydrologic, or political), is unlikely to be successfully implemented. A wide variety of expertise is required to identify how resources are being used, what the likely effects of that use will be, and why these actions are being taken. Second, economic analysis can play a key role in such an integrated approach. Since many resource-users are acting on their own perception of benefits and costs, it is essential to examine these different users to understand what macro-level policies or programmes may be effective in improving their use of land and water resources. Economic analysis from the individual or private perspective is of key importance. It is also necessary to examine the broader macroeconomic framework and its impact on decision-making within the watershed (Sfeir-Younis, 1986). For example, trade and pricing policies send signals through the market that affect resource-use within a watershed. Even if such policies do not affect subsistence farmers in the uplands, they may help prevent the displacement of labour from lowland agriculture to the uplands. Third, environmental externalities should be included, to the extent possible, in the analysis of social welfare. Recent advances in valuation methodology have made it increasingly possible to include monetary estimates of environmental effects. The monetary value of environmental benefits or costs, both at present and in the future, should be included in the economic analysis.

395 Fourth, multiple interests should be reflected in policy-making. Just as several disciplines need to be involved in the analytical process, many different and sometimes conflicting interests need to be accommodated in policy formulation. (In watersheds there are often as many political externalities as there are environmental externalities!) Economic criteria can be used to design the best solution, but there will frequently be valid reasons that make it unacceptable. Concerns for income distribution, minority rights, or political balance may dictate another approach. Economic analysis, carefully done, can indicate the trade-offs and economic costs of second-best solutions. Fifth, integrated analysis does not mean integrated implementation. Integrated resource-management programmes can rarely be implemented by using existing institutional or organizational frameworks. The existing structures are usually organized on sectoral lines and may divide the watershed among different organizations. Although this obviously complicates programme implementation, it is inadvisable to set up a new, integrated organization. Attempts to integrate over wide areas or across sectors have usually been unsuccessful. Organizations such as the Tennessee Valley Authority in the USA are rare exceptions. In some cases the use of integrated oversight committees to co-ordinate individual departmental work has been successful. The lack of an integrated implementing authority is not a major problem if an integrated analysis is done to identify problems and desired solutions. Although programme implementation is along existing organizational lines, integrated analysis can result in actions being taken by a certain ministry (for example, terracing in the uplands via the publicworks department) even when the costs clearly exceed the benefits to that ministry. In this example, the irrigation and hydropower authorities would reap large downstream benefits even though they do not pay for or implement the terracing programme. The Government may need to provide incentives to promote interministerial cooperation in order to facilitate project implementation. Sixth, the role of macro-policies should not be overlooked. Although direct regulation works for large projects, it is not effective in dealing with large numbers of individuals. Macro-level policies that rely on the market to send signals do not face the same institutional and organizational constraints. In many cases, the appropriate role for Government is to provide needed incentives and assistance for individuals or groups to undertake the desired actions themselves (e.g. to grow certain crops or build terraces) These policies are the Government's way of providing the link between cause and effect, and of connecting costs and benefits over space and time. Some combination of taxes, subsidies, direct investment, or price changes may be needed. For example, in the case of promoting upland terracing to protect lowland reservoirs, any of a number of policies could be used: regulations requiring farmer-built terraces, subsidies for farmer-built terraces, communal construction of terraces (with or without compensation), government construction of terraces, increased taxes on unterraced lands, reform of land tenure,

396 dredging of lowland reservoirs, increased taxes on lowland water- and power-users to pay for the other options, changed input or output prices to encourage farmer-built terracing, and incentives for increased interministerial cooperation. Although direct intervention in terracing produces immediate results, it cannot provide the extensive coverage usually required. Some degree of reliance on market incentives, usually within the rural economy, will be more effective. The challenge for the analyst is to identify the policy alternatives that are acceptable given the social, cultural, and political contexts of each case, and to determine the policy that can be implemented effectively. The economist examines the economic implications of alternatives, calculates the cost and benefits of each, and determines who ‘wins’ and who ‘loses’. Ultimately, it is the political decision-maker who decides, but the integrated approach will help in this process and in the policy-maker's search for effective, efficient, and equitable solutions to land- and water-resources management.

REFERENCES BANERJEE, A. K. (1990), ‘Revegetation Technologies’, in Doolette and Magrath (1990). BARBIER, E. B., and J. C. BURGESS (1992), ‘Agricultural Pricing and Environmental Degradation’, Background Paper for the World Development Report 1992 (Washington, DC: World Bank). BISHOP, J., and J. ALLEN (1989), The On-Site Costs of Soil Erosion in Mali, Environment Department Working Paper No. 21 (Washington, DC: World Bank). BLAIKIE, P. (1985), The Political Economy of Soil Erosion in Developing Countries (New York: Longman Group). BOCHET, J.-J. (1983), Management of Upland Watersheds: Participation of the Mountain Communities, FAO Conservation Guide No. 8 (Rome: FAO). BROOKS, K. N., P. F. FFOLLIOTT, H. M. GREGERSEN, and J. L. THOMAS (1991), Hydrology and the Management of Watersheds (Ames, Ia: Iowa State University Press). DANI, A. A. (1986), ‘Annexation, Alienation and Underdevelopment of the Watershed Community in the Hindu Kush-Himalaya Region’, in Easter, Dixon, and Hufschmidt (eds.) (1986). DASGUPTA, P., and K.-G. Mäler (1991), ‘The Environment and Emerging Development Issues’, in Proceedings of the World Bank Annual Conference on Development Economics 1990 (Washington, DC: World Bank). DIXON, J. (1989), ‘Multilevel Resource Analysis and Management: The Case of Watersheds’, in G. Schramm and J. Warford (eds.), Environmental Management and Economic Development (Baltimore: Johns Hopkins University Press). DIXON, J., L. TALBOT, and G. LE MOIGNE (1989), Dams and the Environment: Considerations in World Bank Projects, Technical Paper (Washington, DC: World Bank).

397 DOOLETTE, J. B., and W. B. MAGRATH (eds.) (1990), Watershed Development in Asia, Technical Paper, 127 (Washington, DC: World Bank). DOOLETTE, J. B., and J. W. SMYLE (1990), ‘Soil and Moisture Conservation Technologies: Review of Literature’, in Doolette and Magrath (eds.), 1990. DORFMAN, R. (ed.) (1965), Measuring Benefits of Government Investments (Washington, DC: Brookings Institution). DORFMAN, R. (1972), ‘Conceptual Model of a Region's Water Quality Authority’ in R. Dorfman, H. D. Jacoby, and H. A. Thomas, Jr. (eds.), Models for Managing Regional Water Quality (Cambridge, Mass.: Harvard University Press). DUFOURNAUD, C. M., and J. J. HARRINGTON (1990), ‘Temporal and Spatial Distribution of Benefits and Costs in River-Basin Schemes: A Cooperative Game Approach’, Environment and Planning A, 22: 615–28. EASTER, K. W., and M. M. HUFSCHMIDT (1985), ‘Integrated Watershed Management Research for Developing Countries’, East-West Center Workshop Report (Honolulu: East-West Center). EASTER, K. W., J. A. DIXON, and M. M. HUFSCHMIDT (eds.) (1986), Watershed Resources Management: An Integrated Framework with Studies from Asia and the Pacific (Boulder, Colo.: Westview Press). ECKSTEIN, O. (1958), Water Resource Development: The Economics of Project Evaluation (Cambridge, Mass.: Harvard University Press). FLEMMING, W M. (1983), ‘Phewa Tal Catchment Program: Benefits and Costs of Forestry and Soil Conservation in Nepal’, in L. S. Hamilton (ed.), Forest and Watershed Development and Conservation in Asia and the Pacific (Boulder, Colo.: Westview Press). GREGERSEN, H., K. BROOKS, J. A. DIXON, and L. H. HAMILTON (1987), Guidelines for Economic Appraisal of Watershed Management Projects, FAO Conservation Guide No. 16 (Rome: FAO). HAMILTON, L. S., with P. N. KING (1983), Tropical Forested Watersheds: Hydrologic and Soils Response to Major Use or Conversions (Boulder, Colo.: Westview Press). HAMILTON, L. S., with P. N. KING (1984), ‘Watersheds and Rural Development Planning’, in J. Hanks (ed.) Traditional Life-Styles, Conservation and Rural Development, IUCN Commission on Ecology Papers 7 (Gland, Switzerland: International Union for the Conservation of Nature and Natural Resources). HAMILTON, L. S., and A. J. PEARCE (1986), ‘Physical Aspects of Watershed Management’, in Easter, Dixon, and Hufschmidt (1986). HIRSHLEIFER, J. (1965), ‘Investment Decision under Uncertainty: Choice-Theoretic Approaches’, Quarterly Journal of Economics, 79/4. HIRSHLEIFER, J., J. C. DEHAVEN, and J. W. MILLIMAN (1960), Water Supply: Economics, Technology and Policy (Chicago: University of Chicago Press). HODGSON, G., and J. A. DIXON (1988), Logging versus Fisheries and Tourism in Palawan: An Environmental and Economic Analysis, EAPI Occasional Paper No. 7 (Honolulu: East-West Center). HODGSON, G., and J. A. DIXON (1992), ‘Sedimentation Damage to Marine Resources: Environmental and Economic Analysis’, in J. B. Marsh (ed.), Resources and Environment in Asia's Marine Sector (Washington, DC: Taylor & Francis). KRUTILLA, J. V., and O. ECKSTEIN (1958), Multiple-Purpose River Development (Baltimore: Johns Hopkins University Press). MAASS, A. A., M. M. HUFSCHMIDT, R. DORFMAN, H. A. THOMAS JR., S. A. MARGLIN, and G. M. FAIR (1962), Design Water Resource Systems (Cambridge, Mass.: Harvard University Press).

398 MAGRATH, W. B. (1992), ‘China: Loess Plateau Soil Conservation Project, Sedimentation Reduction Benefit Analysis’, Draft Paper, Asia Technical Department (Washington, DC:World Bank). MAGRATH, W. B. and P. ARENS (1989), The Costs of Soil Erosion on Java: A Natural Resource Accounting Approach, Environment Department Working Paper No. 18 (Washington, DC: World Bank). MAGRATH, W. B. and J. B. DOOLETTE (1990), Strategic Issues for Watershed Development in Asia, Environment Department Working Paper No. 30 (Washington, DC: World Bank). MAHMOOD, K. (1987), Reservoir Sedimentation: Impact, Extent, and Mitigation, Technical Paper, 71 (Washington, DC: World Bank). MCKEAN, R. N. (1958), Efficiency in Government Through Systems Analysis, with Emphasis on Water Resource Development (New York: John Wiley & Sons). PEREIRA, H. C. (1989), Policy and Practice in the Management of Tropical Watersheds (Boulder, Colo.: Westview Press). ROGERS, P. (1991), ‘International River Basins: Pervasive Unidirectional Externalities’, presented at a conference on the Economics of Transnational Commons, Siena, Italy, 25–7 April. ROOSE, E. J. (1977), ‘Application of the Universal Soil Loss Equation of Wischmeier and Smith in West Africa’, in D. J. Greenland and R. Lal (eds.), Soil Conservation and Management in the Humid Tropics, 177–87 (Chichester: John Wiley & Sons). SECKLER, D. (1992), ‘The Sardar Sarovar Project in India: A Commentary on the Report of the Independent Review’, Center for Economic Policy Studies, Discussion Paper No. 8 (Rosslyn, Va.: Winrock International). SFEIR-YOUNIS, A. (1986), ‘Economic Policies and Watershed Management’, in Easter, Dixon, and Hufschmidt (eds.), 1986. TOLLEY, G. S., and F. E. RIGGS (1961), Economics of Watershed Planning (Ames, Ia: Iowa State University Press). US Federal Interagency River Basin Committee (1950), Subcommittee on Benefits and Costs, Proposed Practices for Economic Analysis of River Basin Projects (Washington, DC: World Bank). World Bank (1990), Vetiver Grass: The Hedge Against Erosion, 3rd edn. (Washington, DC: World Bank). YOUNG, H. P., N. OKIDA, and T. HASHIMOTO (1982), ‘Cost Allocation in Water Resources’, Water Resources Research, 18/3:463–75.

15 The Management of Coastal Wetlands: Economic Analysis of Combined EcologicEconomic Systems Coastal wetlands are complex ecosystems. They represent an intermediate phase between terrestrial and aquatic environments and are very sensitive to changes in hydraulic conditions (Bennet and Goulter, 1989). Whether a freshwater swamp or a saline mangrove forest, the combination of land and water, and the fact that environmental effects are transmitted through water, mean that economic analysis of such systems is usually more complicated than for resources that are purely terrestrial (e. g. agricultural field-crops), or aquatic (e. g. a fishery). The Ramsar Convention59 identifies seven landscape units in which wetlands are an important component: estuaries, open coasts, floodplains, freshwater marshes, lakes, peatlands, and swamp forests (Dugan, 1990). Wetlands are valuable for their natural functions, the products that they produce, and their attributes, especially as protectors of wildlife and biodiversity. Table 15.1 lists the main functions, products, and attributes associated with these seven wetland types and indicates the relative importance of each. Mangroves are an especially extensive and important wetland category that depend on flows of fresh and salt water and are commonly found in tropical and subtropical sheltered coastlines and estuaries. They are affected by activities upstream that influence the quantity and quality of inflowing fresh water and by actions within the wetland. In turn, coastal wetlands like mangroves also affect the near shore and off-shore areas in various ways. An economic analysis of these areas must take these links into explicit account. The analysis is complicated by the existence of complex ecosystems interactions, the production of multiple products and services (only some of which are physically found in the wetland itself), and problems with valuation of goods and services.

59

The Ramsar Convention, the Convention on Wetlands of International Importance Especially as Waterfowl Habitat, is an international convention of countries and parties concerned with wetland conservation. The Convention has received increasing support; it had twenty-three contracting parties in 1980 and now has more than fifty-five.

400 Table 15.1 Functions, products, and attributes of wetlands

Functions 1. Groundwater recharge 2. Groundwater discharge 3. Flood control 4. Shoreline stabilization, erosion control 5. Sediment/toxicant retention 6. Nutrient retention 7. Biomass export 8. Storm protection/windbreak 9. Micro-climate stabilization 10. Water transport 11. Recreation, tourism Products 1. Forest resources 2. Wildlife resources 3. Fisheries 4. Forage resources 5. Agricultural resources 6. Water supply Attributes 1. Biological diversity 2. Uniqueness to culture/heritage

E s - (wit- Opet u a - hout n ries man- coasg r o - ts ves) Mangroves

Floodpla ins

Fres- La k - Peat- Swalands m p hwa- es fort e r ests marshes

○ • • • • • • • ○ • •

○ • ▪ ▪ ▪ ▪ ▪ ▪ • • •

○ • ○ • • • • • ○ ○ ▪

▪ • ▪ • ▪ ▪ ▪ ○ • • •

▪ ▪ ▪ ▪ ▪ ▪ • ○ • ○ •

○ ▪ ▪ • ○ ○

▪ • ▪ • ○ ○

○ • • ○ ○ ○

• ▪ ▪ ▪ ▪ •

▪ •

• •

• •

▪ •

▪ ▪ ○ ▪ • • ○ • • •

• • • ○ ▪ ▪ ○ ○ ○ ○ •

• ▪ ▪ ○ ▪ ▪ • • • ○ •

○ ▪ ▪ ▪ • •

○ • ▪ ○ • ▪

○ • ○ ○ • •

▪ • • ○ ○ •

• •

▪ •

• •

• •

Key: ○ = absent or exceptional; • = present; ▪ = common and important value of that wetland type. Source: Dugan (1990).

This chapter focuses on the use of economic analysis to guide policymakers in making choices between development and conservation of tropical coastal wetlands, primarily mangrove forests. Examples of economic valuation of mangrove ecosystems, largely from the Asia-Pacific area, are presented. Some of the analyses value selected wetland goods and services while others attempt to value the complete ecosystem; studies on Fiji (Lal, 1990) and Irian Jay a, Indonesia (Ruitenbeek, 1991) represent fairly comprehensive valuation efforts. The coastal wetlands of Kosrae, an island state of the Federated States of Micronesia in the western Pacific, are used to illustrate the application of an alternative, macro-environmental, safe minimum standard (SMS) approach (Lal, 1991; Dixon, 1991).

401

15.1 COASTAL WETLANDS: PRODUCERS OF GOODS AND SERVICES Although coastal wetlands are found in both tropical and temperate countries, mangrove ecosystems are an especially important sub-group found in the tropical and subtropical belt throughout the world; about eighty species of plants are recognized as mangroves. In the mid-1980s mangroves covered about 170, 000 km2 world-wide, with 40 per cent in Asia, 4 per cent in Oceania, 25 per cent on the east coast of the Americas, 11 per cent on the west coast of the Americas, and 20 per cent in Africa (Hamilton, Dixon, and Miller, 1989). (Coastal saltmarshes, the other main type of coastal wetlands, are mainly confined to temperate regions with some limited occurrence in the tropics, especially in arid or monsoonal regions. In the tropics, saltmarsh wetlands are generally found on the landward side of mangrove forests.) Being located in the land–water–air interface, wetlands, including both mangroves and saltmarshes, consist of terrestrial and aquatic sub-systems with the movement of water providing the linkage between the two. The examples in this chapter are largely drawn from mangrove systems located in the tropics. Even when we refer to ‘mangroves’, however, the lessons are usually applicable to all wetlands, tropical as well as temperate. (For example Turner (1988, 1991) discusses the economics and ethics of wetland conservation, largely in the context of temperate countries, but his analytical approach is very similar to the one presented here.) 15.1.1 Wetland goods and services Societies value coastal wetlands for their ecological services as well as for the extractable products they provide, both within the wetland and those found outside of the wetland area. For example, mangrove wetlands support diverse communities of micro-and macro-flora and fauna, both terrestrial and aquatic (Saenger et al., 1977, 1983; UNDP/ UNESCO, 1987). Many of the plant and animal species are of direct and indirect economic and social value. Economic benefits of mangroves include the harvest of forestry and fisheries products and the hunting of game animals for subsistence. The Matang mangrove forest in Malaysia has been managed on a sustainable basis since 1908. Table 15.2 lists the direct and indirect products produced by mangrove forests. The direct forest products are derived from the mangrove trees themselves while the indirect products (and services) are from the fish, crustaceans, mammals, and reptiles found in the mangrove (Hamilton and Snedaker, 1984; Hamilton, Dixon, and Miller, 1989). Coastal wetland ecosystems also provide a variety of non-consumptive services or functions ranging from recreational use and aesthetic benefits, both of which are derived from their unusual flora and fauna, to coastal protection, flood mitigation, and nutrient filtering. All of these services are dependent on

402 Table 15.2 Products of mangrove forests Direct forest products • Firewood • Charcoal • Construction materials (timber, poles, pilings) • Fodder • Green manure • Cellulose for paper • Food, drugs, and beverages (sugar, alcohol, tea substitutes, medicines) • Tannins and dyes Indirect forest ecosystem products • Finfish (many species) • Crustaceans (prawns, shrimp, mangrove crabs) • Molluscs (oysters, mussels, cockles) • Bees • Birds • Mammals • Reptiles • Other fauna (e.g. amphibians, insects) Source: adapted from Hamilton and Snedaker, 1984, as reported in Hamilton, Dixon, and Miller, 1989.

the continued sustainability of the ecosystem. Table 15.1 listed various functions provided by wetlands, highlighting the importance of mangroves to shoreline protection and stabilization, protection of inland areas from storm surges, and as ‘processors’ of incoming sediment and pollution and ‘exporters’ of nutrients to coastal waters.

15.1.2 Development pressures In spite of the wide variety of valuable goods and services associated with wetlands, they are often prime candidates for conversion to other uses. This paradox is due to several factors: wetlands are commonly publicly owned; they are considered as low-valued, difficult-to-manage resources (largely because many of their benefits are external to the wetland itself); and their location along the coast makes them attractive sites for new infrastructure and development. Mangroves and other wetlands are frequently destroyed and filled for residential, infrastructure, or industrial uses; used for the construction of sea walls and ports; and drained or converted to agriculture or aquaculture development. Furthermore, mangroves are also indirectly damaged by upland

403 development. Soil erosion in upstream areas, resulting in increased sediment loads carried into the mangrove by freshwater flows, can smother the air-breathing roots of mangroves and kill the trees (Hamilton and Snedaker, 1984). Development within the mangrove forests, especially the construction of roads and other structures that block water flows, has also resulted in sedimentation and mangrove death (Lal, 1989). Conservation of wetlands for conservation's sake is not desirable. Wetlands are economic resources that must be managed like any other resource. The analytical question, therefore, is under what conditions should mangroves or other wetlands be left intact and protected, and when should they be converted to other uses. In addition, if mangroves (or other wetlands) are to be converted, what forms of conversion are most sustainable? Are there conversion patterns that are more desirable than others? What portion of an existing wetland can be converted without causing loss of the off-site goods and services whose existence depends on the wetland? There are both ecological and economic dimensions to these questions. The ecologic analysis helps determine what the impacts of conversion are, both to the remaining wetland area as well as to the overall production of goods and services dependent on the wetland system. The economic analysis evaluates the generation of net benefits under different management alternatives. These questions can be addressed by use of a social-welfare-based form of project analysis that is sometimes referred to as ‘extended benefit-cost analysis’.

15.2 THE USE OF EXTENDED BENEFIT-COST ANALYSIS Economic analysis of wetlands has generally focused on the choice between wetland conservation on the one hand and conversion or development on the other, and depends on techniques for valuing the respective benefits and costs of each. Some variation of benefit-cost analysis (B-CA), which provides a systematic method of identifying, measuring, and comparing economic benefits and costs over time, has generally been used. A social-welfare-based B-CA includes the external and environmental benefits and costs of a project (Krutilla and Fisher, 1985). This is in contrast to more traditional project analyses that largely confine themselves to the measurement of direct project benefits and costs.

15.2.1 Benet-cost analysis in theory An implicit assumption in the extended B-CA framework is that a society will be economically efficient in its use of resources when all costs and benefits, both direct and indirect, tangible and intangible, are included in the analysis. The objective function is the maximization of social (or societal) net benefits.

404 The model assumes that individuals regard environmental attributes and services as commodities which can be traded. Because a market price for a complete mangrove ecosystem does not exist, it is assumed that the demand for the ecosystem is the derived demand for the goods and services that the system supports or provides. Thus, for example, the value of the mangrove ecosystem is the sum of net benefits derived from forestry and fisheries’ harvests, filtering of nutrients, aesthetic benefits, and benefits derived from other services of the ecosystem as a whole. Societies are composed of individuals and the relevant measure of benefits depends on estimation of changes in individual welfare (Hufschmidt et al., 1983). Individual benefits also include non-marketed items and certain intangible benefits. (The identification, valuation, and inclusion of these items are discussed later.) The sum of individual benefits constitutes the collective economic welfare of a society (Mishan, 1976). If goods and services are lost, as is the case with mangrove or wetland conversion, a loss is imposed on society. Whether or not this change is justified depends on the net benefits generated by the alternate use of the mangrove or wetland and the distribution of those benefits. A decision-maker with perfect information will compare the opportunity cost of protecting a certain wetland resource today (e.g. the net development benefits that are lost by conserving the wetland) with the net benefits of the goods and services that will be lost if development takes place. The latter is the sum of the functions, products, and attributes listed in Tables 15.1 and 15.2.

15.2.2 Benet-cost analysis in practice In practice, the application of benefit-cost analysis to wetlands has frequently resulted in partial project evaluations that tend to favour development. The direct costs of conservation, in terms of project or development benefits forgone, have been highlighted while the indirect benefits from conservation, many of which appear as economic externalities, have tended to be ignored. This results in a bias towards conversion and development. The reasons for this are not hard to find. First, even if the mangrove is analysed as a self-contained unit, analytical mistakes occur. For example, in some cases poor decisions are due to incomplete information or understanding of natural conditions, leading to overestimates of the benefits from development. In Fiji, for example, extensive mangrove areas were reclaimed without taking into account the inherent constraints of the acid sulphate soils. As a result, the land was still lying idle more than fifteen years after initial reclamation (Lal, 1989). In other cases, land-based commercial benefits and the direct costs of the development project may be included, while other benefits may be ignored. These include traditional subsistence uses of the ecosystem, benefits from products that are spatially removed from the wetland, or the ecological

405 benefits for which there are no measured market values. In the case of mangroves, for example, frequently only the value of commercial forestry products is included in the estimate of the on-site benefits that will be lost in the event of conversion (Burbridge et al., 1981). The value of subsistence uses of various mangrove products, such as mangrove fuelwood, which may be of equal or greater importance than the commercial forestry products, are often not included (Lal, 1989). In Fiji, it was found that the net benefits derived from subsistence harvest of fuelwood was almost twice that of the commercial harvest whereas the value of fisheries’ products was almost the same for commercial and subsistence uses. The inclusion of subsistence values is particularly important in many developing countries where coastal wetlands are an essential resource for local communities. Second, even if the mangrove is analysed as part of a linked ecosystem, the values of off-site effects may not be included. For example, the role of mangroves for the sustainability of coastal fisheries is widely understood and acknowledged; these interactions are best analysed within an ecological framework. However, the economic value of the off-site fisheries benefits is often not included in project evaluation (for the case of Fiji, see Ernest, 1983; ADAB, 1985). These off-site fisheries benefits are often an order of magnitude greater than the on-site forestry benefits. For a proper B-CA, the analyst must take into account both the location of goods and services and their valuation. This dichotomy between location and valuation is illustrated in Figure 15.1. This 2 × 2 matrix defines the location of goods and services on one axis (on-site and off-site) and the valuation problem on the other axis (market values available or not available). Traditional analyses of mangroves tend to focus on those resources included in Quadrant 1—those goods and services that are found on-site and are marketed, e.g. mangrove poles or crabs. Some of the Quadrant 2 resources—those found offsite (outside of the mangrove in adjacent waters) but with market prices—are also included, especially in more recent evaluations. The explicit valuation and inclusion in the analysis of fish or shellfish that depend on the mangrove for part of their life cycle but are caught in adjacent waters is an excellent example. In Fiji, the NPV of forestry production over a fifty-year period at a 5 per cent social discount rate was estimated to be about $200/ha. while the mangrove's contribution to the fisheries NPV (under the assumption that half of the combined commercial and subsistence catch is dependent on the mangrove) was estimated to be $2,700/ha. The off-site fishery benefits of the mangrove are much greater than the on-site forestry benefits (Lal, 1990). The resources in the last two quadrants are largely ignored. Quadrant 3 includes the important goods and services found in the mangrove, some of which may be collected and used by local communities, that do not enter into the market (e.g. medicines, collection of other minor forest products, fish nursery values). Recreational uses of mangrove may also fall into this quadrant. The fourth quadrant, off-site and non-marketed goods and

406

Fig. 15.1 Relation between the location and type of mangrove goods and services and their economic valuationSource: Developed by Dixon and Burbridge and reported in Hamilton and Snedaker (1984).

services, includes such difficult to measure and value impacts as nutrient flows to estuaries and the storm surge benefits that mangroves provide to coastal areas. The problems inherent in the identification and monetization of such effects pose a major valuation challenge for those concerned with more correctly identifying the total contribution of mangrove ecosystems to social welfare. The matrix in Figure 15.1 serves as a useful checklist for the analyst who is investigating the economic value of a mangrove or other wetland ecosystem. It is clear that the economist must work with natural and social scientists to identify, quantify, and value the varied goods and services produced by the natural ecosystem.

15.2.3 Valuation methodology The complex nature of wetlands, the uncertainties in the interactions between ecological and economic processes, and the absence of markets for many of the goods and services produced by wetland ecosystems, require the use of a number of different valuation and evaluation techniques. The underlying assumption is that the value of the ecosystem is based upon the demand for the goods and services produced. Various techniques can be used to value the different goods and services associated with a wetland. The main issues and approaches are nicely laid out by Turner in a recent report (Turner and Jones, 1991). Figure 15.2, adapted from Barbier (1989), uses the concept of total economic value (TEV) for

407

Fig. 15.2 Wetland benefits valuationSource: Adapted from Barbier (1989).

wetland benefit valuation; TEV is composed of use-values and non-use-values. For each subcategory of benefits in Figure 15.2 representative valuation approaches are listed. There is a substantial and growing literature on valuation methodology that can be used to place monetary values on many of the goods and services produced by wetland ecosystems (see Sinden and Worrell 1979; Hufschmidt et al., 1983; Dixon and Hufschmidt, 1986; Turner, 1988 among others). Valuation based on the production of goods and services. One of the most commonly used techniques for valuing the gross annual value of wetlands has been the incomes approach or products and services approach, whereby the physical production of goods and services is valued using actual or surrogate market prices of the resource (Sinden and Worell, 1979; Hufschmidt et al., 1983). This approach concentrates on the values associated with those goods and services found in Quadrants 1 and 3 of Figure 15.1 and is very similar to what are referred to as direct use or structural values in the TEV formulation presented in Figure 15.2. This approach has been used in the case of marshlands (e.g. Pope and Gosselink, 1973; Gosselink et al., 1974; Raphael and Jaworski, 1979; Costanza, Farber, and Maxwell, 1989), and to a limited extent in the case of mangroves (e.g. Christensen, 1982; Baines, 1979; Watling, 1985; Lal, 1990; Ruitenbeek, 1991).

408 In most of the studies mentioned above, it was assumed that the entire value of fishery and forestry products is attributable to the wetlands. This implies that the marginal value of labour and capital in the respective industries is zero. Moreover, these evaluations of fisheries benefits frequently fail to recognize the principle of ‘with and without analysis’ (Hufschmidt et al., 1983). Specifically, since the fish associated with mangroves are not unique to that ecosystem, the assumption that without the mangroves there will be no fisheries is not valid. Some fisheries will still be viable provided there are other ecosystems, such as those based on sea-grass beds or coral reefs. Furthermore, it is difficult to determine the value of the wetland ecosystem when it is important only at some period in the life cycle of the species. For example, the wetlands may act as a nursery ground or as a breeding site during certain periods. More information is needed on the exact link between fish production and mangroves. It is known that if an entire mangrove is destroyed there will be a negative effect on fish and crustacean production. What is not known, however, is the marginal impact of destruction or conversion of part of the wetland. In the presence of these uncertainties and in the absence of specific information about the marginal change in fish production directly attributable to mangrove conversion, a with and without analysis using different assumptions is useful. In Fiji, the assumption of a 50 per cent decline in the harvest rate of those fish and crustacean species known to be dependent on mangroves during some time in their life cycle and caught from the coastal waters was used to determine the marginal effect of mangrove reclamation. The NPV of commercial and subsistence fish and crustacean species dependent upon mangroves was estimated to be $2,700/ha. of mangroves over a fifty-year planning period with a 5 per cent discount rate or $150/ha./yr. in 1986 dollars. This is about ten times greater than the net benefits derived from the on-site subsistence and commercial harvests of mangrove fuelwood (Lal, 1990). Even under a 20 per cent decline in the fisheries harvest, the off-site benefit is over five times greater than the on-site value from direct harvest of mangrove timber products. Revealed preferences. Not all goods and services can be valued using change-in-production approaches and marketanalysis approaches, as is done for production of such products as poles, timber, crabs, and fish. A number of other valuation approaches can then be employed. For example, recreational and

409 tourism demand can be evaluated using the travel cost methods, where preferences are revealed through observation of actual use of wetland resources. This technique involves estimating consumer surplus associated with the recreational use of a site (e.g. Clawson and Knetsch, 1966; Farber and Costanza, 1987; Brink, 1973). Survey-based techniques. The sum of the calculated values of goods and services is still only a minimal estimate of the onsite value of coastal wetland ecosystems. The productivity approach depends on those values that can be measured in market prices or by means of appropriately estimated shadow prices. Not included are the option values associated with using the resource in the future, or the bequest or existence values of the ecosystem. Some of the environmental benefits have no natural units of measurement even though their physical measures may be available. For example, biodiversity can be measured in terms of diversity indices, but its economic value is tied in with option values, existence values, and bequest values. The contingent valuation method (CVM) is commonly used to value option, existence, and bequest values (see Figure 15.2). CVM involves creating a hypothetical market in which participants are asked questions designed to reveal their willingness to accept compensation (WTAC) for services provided by a system which will be lost in the event of development. Alternatively, consumers are asked about the price they are willing to pay (WTP) for the preservation of an ecosystem (Sinden and Worrell, 1979; Hufschmidt et al., 1983; Farber and Costanza, 1987). Sometimes these values can be fairly large. A study by Greenley, Walsh, and Young (1981) found that the sum of option, existence, and bequest values was greater than direct user-benefits. However, in their analysis it was noted that there was evidence of instrument bias (Nash and Bowers, 1988). The results were questioned on the basis that Greenley et al. did not distinguish between consumer surplus and option value. In another study, respondents were asked to separate their WTP for preservation or extension of wilderness areas in Colorado into various component parts: consumer surplus attributable to direct use, option value, existence value, and bequest value (Walsh, Loomis, and Gillman, 1984). Again, it was found that the sum of the latter three were of the same order of magnitude as the userbenefit. Although CVM is finding increasing acceptance for use in valuing environmental services, there are few examples of its application in the case of wetlands. In an analysis of the annual wetlands-based recreation value of Louisiana wetlands, Bergstrom et al. (1990) used a TEV approach to estimate the expenditures and consumer's surplus associated with on-site, current recreational uses of a 1.32 million ha. area of Louisiana wetlands. Based on both interview data and CVM surveys they estimated that aggregate current expenditures were about $118 million and aggregate consumer's surplus was about $27 million in 1986–7. The aggregate gross economic value was thus about $145 million or $11/ha. /yr.

410

15.2.4 Property rights, benet valuation, and assessing development benets In theory, well-defined property rights that can be freely traded capture most of the benefits associated with a resource. Actual sale prices should represent the social net benefits of the resource in question (Hufschmidt et al., 1983). This value should be greater than the sum of the net benefits derived from the goods and services supplied by the system since intangibles are also valued and included. This is the basis of the hedonic approach to estimating the total economic value of a resource. (An example of the application of this approach is found in the buying and selling of coastal fishing rights for filling and near-shore development in Tokyo Bay (Hanayama and Sano, 1986).) In the evaluation of the benefits of alternative uses of wetlands, the market selling price of developed land has often been used, especially in the USA, where wetlands are generally in private ownership. Similarly, mangroves are in demand because of their location along the coast: the expected benefits from alternative uses such as ports, industrial developments, or aquaculture are real. In such situations, locational values of developed and converted wetlands can also be estimated using hedonic methods (e.g. Abdalla and Libby, 1981; Shabman and Bertelson, 1979). In practice, given the absence of a formal market for reclaimed land, the present value of the expected net benefit stream after conversion is used as a proxy for the value of the resource. However, both the market land price and the capitalized benefit stream estimates do not usually reflect the value of the mangrove-dependent fisheries, although the value of the forestry products harvested for commercial sales may be included (Lal, 1989). It is important to make realistic assumptions about the sustainability of wetlands after they are converted to agriculture or aquaculture. There may be important physical factors that limit the long-term potential of mangrove soils to support other forms of agriculture or aquaculture development. This problem is widespread, e. g. in India (Bandhopadhyay, 1985; Umali et al., 1987); Thailand (de Glopper and Poel, 1972; Aksornkoae 1985); and the Philippines (Ponnamperuma, 1985). In Fiji, as described earlier, the long-term sustainability of reclaimed or converted mangrove soils was not considered in the benefit-cost analysis undertaken by the Government. Reclaimed wetland soils produce a maximum expected crop after a significant maturation period. Even then, the maximum yield can usually be sustained only on limited soil types and under ideal conditions (Lal, 1989). Furthermore, when projects are planned and developed in different stages, often the initial capital costs are treated as sunk costs, as was the case with a rice-irrigation project in Fiji (ADAB, 1985). The costs of maintaining sea walls and banks (particularly as a result of the activities of burrowing animals), the effects of drought conditions, and the costs of special management practices required on coastal wetland soils (in

411 particular the problem of acid sulphate soils) are also commonly ignored (Lal, 1989). Ignoring these ecological constraints can mean that a valuable, dynamic, natural ecosystem, the mangrove, is lost without obtaining in return any sustainable, productive use of the reclaimed lands. In Fiji, most of the mangrove lands which were reclaimed for rice or sugar cane are lying idle; even where some crops are grown, the returns are small or negative. In the largest and the longest-running mangrove reclamation project involving sugar cane and shrimp farming, the net present value of reclaimed land is negative if the effects of ecological constraints are taken into account (Lal, 1990). The returns from agriculture development were found to be negative even if the loss of the benefits of fisheries and forestry products were excluded. In some cases reclaimed lands are abandoned after the soils fail to produce expected yields (e.g. in Chittagong Plains in India (Amirul Islam, 1964)), or because of frequent crop failures after drought causes the capillary rise of salts from the subsoils, as happened in West Africa (Moormann and Pons, 1975). Aquaculture developments are also susceptible to the acid sulphate conditions common in mangrove areas resulting in the abandonment of ponds after a few years. This results in a ‘shifting’ aquaculture that progressively converts more and more mangroves to develop new ponds leaving abandoned unproductive soils in its wake.

15.2.5 Institutional factors Even if perfect information is available, the choice between conservation and development often depends on institutional factors. Just as the project boundary is determined by the physical location of the development project and not by the ecological characteristics of the ecosystem, government agencies undertaking an economic analysis are also guided by their own narrow, sectoral jurisdictions and interests. Often, the government agency which is mandated to develop a particular sector, such as agriculture, has a vested interest in undertaking partial evaluations of projects, often favouring development.60 Political expediency and short-term interests of government agencies are a universal problem (Buchanan and Tullock, 1964). For example, in Fiji the Department of Drainage and Irrigation within the Ministry of Agriculture and Fisheries is the implementing agency for projects to drain and irrigate mangrove lands for crops such as sugar cane and rice. This same agency also undertakes the analysis of these projects. Only engineering costs and direct

60

The fact that mangroves are often publicly owned, and hence ‘free’ to government planners, often encourages wetland conversion, even when other suitable land exists near by but is privately owned. A full economic analysis comparing these alternatives should compare the costs of privately held land with the cost of lost goods and services if wetlands are converted. It should be noted, however, that privately held wetlands will also be undervalued since the landowner will not normally be able to benefit from the off-site production of goods and services. In the case of wetlands, privatization itself does not normally imply improved resource management.

412 project benefits are included, ignoring the value of existing mangrove benefits that are lost. Since the survival of the drainage and irrigation division is dependent upon projects being approved and financed by the Government, it is not surprising that there is a strong bias towards project approval. Similar behaviour on the part of government agencies dealing with wetland drainage in England has also been noted (Bowers, 1988).

15.3 EXAMPLES OF MANGROVE VALUATION Two questions need to be asked when assessing the ‘development’ alternatives being proposed for mangrove areas. First, are the expected benefits from conversion large and sustainable, and, second, what are the true costs (direct and indirect) of losing the mangrove ecosystem? There is not an extensive literature reporting valuation estimates. Table 15.3 presents some examples for mangroves gleaned from the literature and reported in Hamilton and Snedaker (1984). The reported values (in $US/ha. /yr.) range from estimates for the complete ecosystem (Quadrants 1 to 4 in Figure 15.1) to estimates of forestry products (Quadrant 1 goods: on-site and marketed) to fishery products (Quadrant 2: off-site and marketed). The range of annual values reported is great, from as low as $25/ha. /yr. for forestry products in Malaysia to over $1,000/ha. /yr. in Thailand when fishery and forestry Table 15.3 Examples of values placed on mangrove systems and mangrove ecosystem products Type of resource or product and location Complete mangrove ecosystem Trinidad Fiji Puerto Rico Forestry products Trinidad Indonesia Malaysia Thailand Fishery products Trinidad Indonesia Fiji Queensland Thailand Recreation, tourism Trinidad

Date

Value placed on resource ($US/ha. / yr.)

1974 1976 1973

500 950–1, 250 1, 550

1974 1978 1980 1982

70 10–20 (charcoal and wood chips) 25 30–400

1974 1978 1976 1976 1982

125 50 640 1, 975 30–100 (fish); 200–2, 000 (shrimp)

1974

200

Source: Hamilton and Snedaker (1984), in Hamilton, Dixon, and Miller (1989).

benefits are included. Most of these values are for gross benefits; net benefits, after capital and labour costs are subtracted, would be less.

15.3.1 Thailand A carefully done case study in Chanthaburi Province in south-east Thailand (Christensen, 1982) examined a traditional

413 mangrove-based economy and presented annual values for a variety of products including forest products, nipa thatch, fisheries, oyster culture, shrimp farming, and agriculture. (Note that a mangrove is a naturally sustainable ecosystem while the alternative uses after conversion require continuous maintenance and management.) Values were estimated for each of the mangrove-dependent products, both for present productivity levels and for potential levels with improved management. These values are presented in Table 15.4. The shrimp-farming values are large, ranging from $200 to $2, 000/ha. /yr. These values are for commercial shrimp farms on converted land that are partially dependent on the remaining mangrove as a source of shrimp fry. Still, the total value per hectare from forest and fishery products from an intact mangrove ecosystem is substantial, from $160/ha. at present to a potential of over $500/ha. When compared to one alternative use, rice farming, one sees why a broader economic analysis is needed. The expected return from agriculture ($165/ha. /yr.) is large compared to the annual value per hectare from charcoal production ($30). Establishing agricultural fields requires the destruction of the mangrove. If this happens not only will the value of charcoal production be lost, but also other off-site benefits. When the present value of fishery products caught within and outside the estuary are added, the two uses are equivalent in value. When future increases in potential income are considered, and the contribution of mangroves to shrimp farming through the natural production of shrimp larvae is also included, the value of intact mangroves becomes Table 15.4 Tentative economic comparison of various forms of land use in the mangrove areas of Chanthaburi, Thailand Gross income Present $/ha. /yr. ‘Forestry’: Official charcoal production 30 Fishery inside estuary 30 Mangrove-dependent fishery outside 100 Oyster culture — Total forestry 160 Shrimp farming 206 Rice farming 165 Source: Christensen (1982), in Hamilton, Dixon, and Miller (1989).

Potential $/ha. /yr. 400 30 100 60 590 2,106 [—]

414 very substantial. In addition, forestry and fishery production are fairly labour-intensive and have important employment-generation potential. In the Thai case, a potential production system based on intensified management of the natural mangrove was about equal to shrimp cultivation in creating jobs and more employment-intensive than rice farming. Note that even these estimates for the value of mangroves in Thailand only include a limited number of the goods and services included in Tables 15.1 and 15.2. As such, they are minimum estimates of the yearly value of mangrove products. And yet, the decision as to whether or not mangroves should be converted (destroyed) is made by comparing this minimum, partial estimate with the total expected benefit from conversion. No wonder mangroves are being lost at such a rapid pace!

15.3.2 Malaysia Mangroves can be managed on a sustained yield basis as forest systems. In Malaysia, for example, the Matang Mangrove Forest Reserve has been managed for the production of fuelwood and poles since the turn of the century. Located in the state of Perak, the 40,000-ha. reserve has been managed on a thirty-year rotation since 1950. The Reserve is carefully harvested both to maintain important ecosystem functions and ensure regeneration while also producing income. Its value lies in both its forest products as well as the fish, prawns, and other sea foods that use the mangrove as a breeding ground and benefit from the food sources produced by the mangrove. In 1976, the direct government revenue collected from forest products was over $425,000 or about $12.70/ha. of productive forest. Management of the forest provided direct employment for 1,400 people and another 1,000 indirectly. In addition, the nearby fishing industry employed another 2,600 people directly and 7,800 indirectly. The gross value offish landings, mostly prawns, was about $12 million in 1977 (Dugan, 1990).

15.3.3 Indonesia Similar issues to those found in Thailand were raised in a report on coastal resources management in Indonesia (Burbridge and Maragos, 1985). Indonesia has the largest mangrove area of any country in the world, and conversion, both to dryland and to tambak, or fishpond culture, is widespread. Indonesian mangroves are highly productive and diverse and under severe pressure from overharvesting of forest products and conversion. In spite of an official policy against further tambak extensification, Burbridge and Maragos observed this occurring throughout the country. They noted that the potential for sustainable production from mangrove areas was largely ignored and that ‘a prevailing but erroneous view among many advocates of mangrove conversion is that the swamps are wastelands with little value in comparison to “higher” uses such as tambak and rice culture’ (ibid. 24).

415

15.3.4 Ecuador Large areas of mangrove forest have been converted to shrimp grow-out ponds in Ecuador. This development has been particularly rapid in the southern Gulf of Guayaquil—16 per cent of the mangrove forest was lost between 1966 and 1982. In addition to the physical loss of mangrove forest, researchers believe that this has also resulted in the measured decline in abundance of shrimp larvae in Ecuadorean estuaries (Lahmann, Snedaker, Brown, 1987). This is a major concern of shrimp-farm owners who rely on these wild larvae as a stock source for their grow-out ponds. As a result, productivity falls and shrimp farmers may have to use more expensive hatchery operations to produce shrimp larvae.

15.3.5 Fiji The Fiji study referred to elsewhere in this chapter contains comparisons of conservation and development options. The analysis compared the net benefits of converting mangrove lands to rice and sugar cultivation by estimating the various benefits of the mangrove that would be lost after conversion. In the analysis the main items included were the fairly small on-site benefits from commercial exploitation of mangrove timber resources and the much more important off-site benefits from fishing income that is directly dependent on the mangrove for its sustained productivity. Based on the assumption that half of the local fishing catch was dependent on the mangrove for its sustained production, a per hectare value for fishery losses ($F150) was added to the value of lost forestry production ($F9), for a total economic ‘loss’ of $F159/ha. /yr. for each hectare of mangrove lost. Conversely, if the land had not been reclaimed, annual production of fishery and forestry products worth about $F160/ha. would be expected. In the case of agricultural development on reclaimed mangrove land in Fiji, varying the discount rate from zero to 10 per cent did not change the decision that no development was the preferred option. Taking into account the ecological constraints of the reclaimed tidal wetland soils and the loss of fisheries and forestry products, the NPV of the agricultural project alternatives were negative for discount rates from 1 to 10 per cent (Lal, 1990). In the case of sugar cane projects, where the ecological constraints of the acid sulphate soils were extremely critical, decreasing the discount rate to zero still resulted in a negative NPV for the project. Even excluding the loss of off-site benefits of the mangrove, the net benefits of agricultural development were negative. In one site the per hectare annual ‘development benefit’ after conversion for agriculture and aquaculture uses was a negative $F516. Converting mangroves in Fiji to agricultural uses therefore represented a net economic loss of $F675/ha. /yr., or much more than the value of lost forestry and fishery products alone.

416 In this case, conversion clearly made no sense since the development alternatives were not sustainable and did not produce positive net benefits by themselves. Nevertheless, the on-site forestry benefits were very small, clearly leading to the interest in converting mangroves to other uses. As in the Thai case, the coastal fishery benefits are considerably larger than forestry benefits. Sales prices for mangroves also reflect the division between on-site and off-site benefits and the difficulty of capturing off-site benefits for the landowner. It was found that in most cases indigenous Fijians, with recognized coastal rights, received as payment for their mangrove lands only a small portion of the actual economic loss to society from decreased forestry and fishery production (Lal, 1990). If the future fishery and forestry benefits that will be lost due to conversion are valued over fifty years at a 5 per cent discount rate, they average $F2, 734/ha. for fishery products and from $F164 to $F217/ha. for forestry products, or a total of over $F2,900/ha. In contrast, landowners were paid only about $F520/ha. for non-industrial uses, or less than 20 per cent of forgone fishery and forestry benefits (but roughly three times the estimate of forgone forestry benefits alone). Land sold for industrial uses received a higher price (largely due to its location on the coast and not to any recognition of the goods and services associated with the mangrove). By considering only the on-site, terrestrial component (the Quadrant 1 benefits of Figure 15.1), it is very easy to undervalue the economic importance of mangroves. Fijian mangrove lands are also converted because of problems with land-ownership. In Fiji, 83 per cent of the land is communally owned by the indigenous Fijians and as such cannot be sold or bought. As a result, the governmentowned mangrove areas are particularly attractive for development. It was found that mangrove reclamation for residential housing development was uneconomical if alternative land (that is, in the situation ‘without’ the institutional constraint) were assumed to be available, albeit at a cost (Lal, 1990). Only in the case of industrial uses for reclaimed land, with a very high premium attached to the site, did the with and without analysis not make a difference.

15.3.6 Irian Jaya, Indonesia The development of a woodchip export industry poses a potential threat to a 300,000-ha. mangrove ecosystem in Bintuni Bay, located in north-west Irian Jaya in eastern Indonesia. At present the bay supports an important shrimpexport industry and the coastal areas are home to some 3,000 households in a mixed economy that depends on farming, wages, and extraction of traditional mangrove forest products. In a recent study Ruitenbeek (1991) examined the management options for this important resource. Based on a survey of 100 households in the area, Ruitenbeek concluded that (1) non-market traditional uses of the mangrove are significant and accounted for about 70 per cent of the total value of income derived from

417 mangroves, (2) traditional mangrove uses contribute proportionately more to low income households, and (3) expansion of the wage economy does not necessarily imply an equal reduction in dependence on the mangrove. Extrapolating the survey results to the larger Bintuni Bay population, Ruitenbeek estimated that traditional uses of mangroves for hunting, fishing, and gathering have a value of about $10 million per year and fishery products are valued at about $35 million per year. In comparison, clear-cutting of up to 80 per cent of the mangrove under a selective cutting scheme has a maximum value of $20 million per year. To evaluate the benefits and costs of these alternative uses and the extent that mangrove cutting should be permitted, six forestry options were evaluated in a benefit-cost analysis framework. The management options ranged from 100 per cent clear-cutting to a complete cutting ban. The clear-cut option is optimal only if linkages between ecosystem components are ignored. A cutting ban is optimal only if linear and immediate linkages exist between ecosystem components (i.e. cutting any mangrove would have direct and linear effects on both the fishery and the traditional coastal economy that depends on the mangrove). Using a more reasonable assumption of linear but delayed linkages (based on a five-year delay), selective cutting for woodchips of 25 per cent of the mangrove has a net present value that is $35 million greater than the clear-cutting option (that would result in the loss of important mangrove-dependent goods and services), and $1.5 million greater than the cutting ban option. Because of the ecosystem linkages, more extensive cuttings would yield no additional net benefits. The Bintuni Bay case illustrates some of the challenges that resource managers face. The traditional uses of the mangrove are important, especially to poorer residents and through the export-orientated fishery industry. The attraction of greater mangrove harvesting for woodchip exports is also strong: foreign investment and foreignexchange earnings from a remote, poor region. The analysis indicates that there are strong economic arguments for selective mangrove clearing. The dynamics of ecosystem linkages, especially between mangrove areas and the shrimp fishery, will be important determinants of the location and pace of selective mangrove harvesting activities.

15.4 AN ALTERNATIVE APPROACH TO EVALUATION OF DEVELOPMENT OPTIONS The extended benefit-cost analysis framework, despite its usefulness in systematically identifying and estimating the benefits and costs of a development project, is limited because of the need for monetary valuation of the benefits of an ecosystem, some of which may not be known or are difficult to measure. In situations where there is a great degree of uncertainty about species diversity or the rarity of a species or habitat, a ‘minimax approach’ adapted from

418 game theory has been suggested (Ciriacy-Wantrup, 1952; Bishop, 1978). Commonly referred to as a ‘safe minimum standard’, this approach seeks to minimize the maximum loss to present and future generations. This strategy also provides an economic justification for ‘preservation’, such that the endangered habitat or species is maintained unless the social cost of doing so (that is, its opportunity cost) is unacceptably high, as defined by society. Macro-environmental standards that relate ecological factors with social objectives can also be used to define land-use alternatives (Turner, 1988). In this context, the concept of safe minimum standards (SMS) can be applied to ecosystem zoning or ‘districting’. A wetland is ‘zoned’ according to the ecological capabilities of the ecosystem, which allow for both use and conservation and/or preservation, to meet the overall development objectives and aspirations of a community. The criteria use a rank ordering of sub-areas within various general use-categories. These categories include the following: sanctuary, recreational districts, traditional use, forestry, fisheries, and reclamation for alternative uses such as agriculture or urban developments. One of the limitations of this approach is that the rank order gives only ordinal measures and does not provide a cardinal measurement scale which would allow a finely tuned comparison of different uses. Consequently, for evaluating alternative projects within the constraints of the macro-environmental standards defined for each of the districts, an expanded benefit-cost framework would be a valuable tool. The use of macro-environmental standards to define districts (or ecosystem zones) was recently attempted in the case of the small island state of Kosrae in the Federated States of Micronesia (Lal, 1991; Dixon, 1991). Kosrae is a ‘high’ island that has a mountainous core with a narrow coastal plain and extensive offshore reefs. There are extensive mangroves along much of the coast that provide wood products, mangrove crabs, and help support the locally important coastal fishery. As Kosrae began to promote economic development the mangroves were immediately considered as a prime site for development. The planning question, therefore, was to identify the importance of the mangrove and determine how this land could be allocated among different uses (including conservation), without needlessly losing important benefits from intact mangroves. A macro-environmental approach was adopted and the Kosraen wetlands were divided into five districts defining different types of allowable uses. A ranking system was developed based upon the distribution of coastal wetlands, the major species alliances of the mangroves in each area, major soil types, and use-capabilities as defined by ecological characteristics. Present and projected distribution of population, resource-utilization patterns, and national development goals were taken into account. Five different use-districts were defined: sanctuary, recreation, forestry, traditional use, and urban development districts. Each of the districts has its own environmental and economic standards that will be used to evaluate

419 proposed projects. This information was then combined with other information on resources and economic opportunities to define resource-planning districts. These are shown in Figure 15.3 and include special areas for recreation and tourism (a potentially important industry in the future), inshore fisheries, conversion for aquaculture, agriculture or construction sites, and logging. The intention is to allow development to continue, but to ensure that it is balanced and that mangroves will not be converted unless there is a full evaluation of both the expected benefits from the conversion and also the costs associated with loss of mangrove. It is still too early to tell the effectiveness of this approach. A comprehensive Kosrae Coastal Management Programme is just now being formulated and implemented by the Government.

15.5 CONCLUSIONS Mangroves and other wetlands are under increasing pressure for conversion to other, supposedly higher-valued uses. In many cases conversion may be fully justified, but in others it results in short-term private gains at the expense of larger, long-term social losses. Although much more detailed analysis of the economics of wetlands in general and mangrove ecosystems in particular is needed, we have learned enough to draw the following general conclusions: • • • • •

Decisions on whether or not to convert mangroves to other uses are frequently based on the value of marketable forestry products produced by the natural mangrove. These values may be quite low. Fishery and marine products, both within the mangrove and in nearby waters, are frequently much more valuable than forest products. The linked land–ocean system of a mangrove creates complicated and far-reaching ecosystem linkages affecting the production of a wide range of socially valuable goods and services. Whereas a natural mangrove is a self-sustaining, productive ecosystem, many alternative uses have proved to be expensive to construct and maintain, or have produced disappointing economic results due to low and declining productivity. Normal market forces will almost always favour conversion of mangroves (and other wetlands) to other uses. This is a result of both locational and valuation factors. Because of the dispersed nature of the products found in the mangrove and those outside that are dependent on the mangrove, most analyses are narrow and exclude important components. The problem of assigning monetary values to some goods and services of the mangrove also exists; in addition to identifying the products and services found outside the mangrove, some ecosystem or biodiversity benefits may be very hard to value in monetary terms. Because of this ‘market failure’, government involvement is essential if mangroves are to be used in a socially optimal manner.

420 Fig. 15.3 Resource planning districts for KosraeSource: Recommendations for a Kosrae Island Resource-Management Program, Preliminary Report, 1990, University of Hawaii Seagrant Extension Service.

421 • •

New uses for mangroves through tourism development and as wildlife habitats may become increasingly valuable in the future. Subsistence production of various non-marketed goods and services may be very important in some areas but is rarely reflected in any economic analysis of a mangrove ecosystem.

In sum, a fuller economic analysis of wetlands such as mangrove forests, and the alternative uses being proposed, may well demonstrate that many mangroves yield greater social net benefits as natural ecosystems. In cases where conversion is clearly necessary or justified, sound physical-socio-economic analysis can help to plan conversions that reduce to a minimum the loss of mangrove benefits. A combined economic-ecologic analysis is needed to analyse properly management and development options for coastal wetlands.

REFERENCES ABDALLA, C, and L. W. LIBBY (1981), ‘Development Values of Michigan Wetlands’, in Brandt Richardson (ed.), Selected Proceedings of the Midwest Conference on Wetland Values and Management, 453–65 (St Paul, Minn.: Minnesota Water Planning Board). ADAB, (1985), Fiji Rice Development Study. A Report prepared by the Australian Agricultural Consulting and Management Co. Pty. Ltd for the Australian Development Assistance Bureau (ADAB) and the Ministry of Primary Industries, Suva, Fiji. AKSORNKOAE, S. (1985), ‘Success and Failures in the Use of Mangrove Reclaimed Land inThailand’, in UNDP/ UNESCO Report 1985, Report of the Workshop on the Conversion of Mangrove Areas to Paddy Conversion, 79–86, Los Banos, Laguna, Philippines, 1–3 April 1985. AMIRUL ISLAM, M. (1964), ‘Soils of East Pakistan’, in Proceedings of the UNESCO Symposium on Scientific Problems of the Humid Tropical Zone Deltas, Dacca, 83–7. BAINES, G. B. K. (1979), Mangroves for National Development: A Report on the Mangrove Resources of Fiji, A Report to the Government of Fiji, Suva, Fiji. BANDOPADHYAY, A. K. (1985), ‘The Sunderban Mangrove Experience on Reclamation’, in the UNDP/UNESCO Report 1985, Report of the Workshop on the Conversion of Mangrove Areas to Paddy Cultivation, 65–70, Los Banos, Laguna, Philippines, 1–3 April 1985, UNDP/UNESCO Regional Project RAS/79/002. BARBIER, E. B. (1989), Economic Evaluation of Tropical Wetland Resources: Applications in Central America, LEEC Working Paper (London: IIED). BARLOWE, R. (1973), Land Resource Economics: The Economics of Real Property (Englewood Cliffs, NJ: Prentice-Hall). BATIE, S., and L. A. SHABMAN (1982), ‘Estimating the Economic Value of Wetlands: Principles, Methods, and Limitations’, Coastal Zone Management Journal, 10/3: 255–78.

422 BENNETT, J., and I. G. GOULTER (1989), ‘The Use of Multiobjective Analysis for Comparing and Evaluating Environmental and Economic Goals in Wetlands Managements’, Geojournal, 18/2: 213–20. BERGSTROM, J. C., J. R. STOLL, J. P. TITRE, and V. L. WRIGHT (1990), ‘Economic Value of Wetlands-Based Recreation’, Ecological Economics, 2: 129–47. BISHOP, R. C. (1978), ‘Endangered Species and Uncertainty: The Economics of a Safe Minimum Standard’, American Journal of Agricultural Economics, 60: 10–18. BOWERS, J. (1988), ‘Cost Benefit Analysis in Theory and Practice: Agricultural Land Drainage Projects’, in R. E. Turner (ed.), Sustainable Environmental Management: Principles and Practice (Boulder, Colo.: Westview Press). BRINK, C. H. (1973), ‘A Comparison of Consumers’ Surplus and Monopoly Revenue Estimates of Recreational Value for Two Utah Waterfowl Marshes’, PhD Dissertation, Utah State University, Logan. BUCHANAN, J., and G. TULLOCK (1964), The Calculus of Consent (Ann Arbor: University of Michigan Press). BURBRIDGE, P. R., J. A. DIXON, and B. SOEWARDI (1981), ‘Forestry and Agriculture: Options for Resource Allocation in Choosing Lands for Transmigration’, Development Applied Geography, 1: 237–58. BURBRIDGE, P. R., and J. E. MARAGOS (1985), ‘Coastal Resources Management and Environmental Assessment Needs for Aquatic Resources Development in Indonesia’ (Washington, DC:IIED). CHRISTENSEN, Bo (1982), ‘Management and Utilization of Mangroves in Asia and the Pacific’, FAO Environment Paper, 3 (Rome: FAO). CIRIACY-WANTRUP, S. V. (1952), Resource Conservation (Berkeley and Los Angeles: University of California Press). CLAWSON, M., and J. L. KNTESCH (1966), Economics of Outdoor Recreation (Baltimore: Johns Hopkins University Press). COASE, R. H. (1960), ‘On the Problem of Social Cost’; Journal of Law and Economics, 3: 1–44. COSTANZA, R., S. C. FARBER, and J. MAXWELL (1989), ‘Valuation and Management of Wetland Ecosystems’, Ecological Economics, 1: 335–61. DIXON, J. A. (1991), ‘Coastal Resources in Kosrae: An Undeveloped Economic Resource’, in Kosrae Island Resource Management Plan, Pacific Island Network, University of Hawaii (Honolulu: Pacific Island Network and Seagrant). DIXON, J. A. and M. M. HUFSCHMIDT, (eds.) (1986), Economic Valuation Techniques for the Environment: A Case Study Workbook (Baltimore: Johns Hopkins University Press). DUGAN, P. J. (ed.) (1990), Wetland Conservation: A Review of Current Issues and Required Action (Gland, Switzerland: World Conservation Union). ERNEST, M. (1983), ‘Mangrove Reclamation?—Realities and Realism!’, in P. N. Lal (ed.), Mangrove Resource Management, Proceedings of an Interdepartmental Workshop, 24 April 1983: Technical Report 5 (Suva, Fiji: Ministry of Agriculture and Fisheries). FARBER, S., and R. COSTANZA (1987), ‘The Economic Value of Wetland Systems’, Journal of Environmental Management, 24/1: 45–51. GOSSELINK, J. G., E. P. ODUM, and R. M. POPE (1974), The Value of the Tidal Marsh, Publ. LSU-SG-74-03, Center for Wetland Resources, Louisiana State University, Baton Rouge, La.

423 GREENLEY, D. A., R. G. WALSH, and R. A. YOUNG (1981), ‘Option Value: Empirical Evidence from a Case Study of Recreation and Water Quality’, Quarterly Journal of Economics, 4. HAMILTON, L. S., J. A. DIXON, and G. O. MILLER (1989), ‘Mangrove Forests: An Undervalued Resource of the Land and of the Sea’, in E. M. Borgese, N. Ginsburg, and J. R. Morgan (eds.), Ocean Yearbook 8 (Chicago: University of Chicago Press). HAMILTON, L. S., and S. C. SNEDAKER (ed.) (1984), Handbook for Mangrove Area Management (Honolulu: East-West Centre). HANAYAMA, Y., and I. SANO (1981), Valuation of Losses of Marine Product Resources Caused by Coastal Development of Tokyo Bay (Tokyo: Institute of Technology). HODGSON, G., and J. A. DIXON (1988), Logging Versus Fisheries and Tourism in Palawan. Environment and Policy Institute, Occasional Paper, 7 (Honolulu: East-West Center). HUFSCHMIDT, M. M., D. E. JAMES, A. D. MEISTER, B. T. BOWER, and J. A. DIXON (1983), Environment, Natural Systems, and Development: An Economic Valuation Guide. (Baltimore: John Hopkins University Press). KRUTILLA, J. V., and A. C. FISHER (1985), The Economics of Natural Environments: Studies in the Valuation of Commodity and Amenity Resources (Baltimore: John Hopkins University Press). LAHMANN, E. J., S. C. SNEDAKER, and M. S. BROWN (1987), ‘Structural Comparisons of Mangrove Forests near Shrimp Ponds in Southern Ecuador’, Interciencia, 12/5. LAL, P. N. (1983), ‘Institutional Aspects of the Management of Mangroves in Fiji’, in P. N. Lal (ed.), Mangrove Resource Management, Proceedings of an Interdepartmental Workshop, 24 April 1983, Technical Report 5 (Suva, Fiji: Ministry of Agriculture and Fisheries). LAL, P. N. (1989), Conservation or Reclamation: Economic and Ecological Interaction within Mangrove Ecosystems in Fiji, Unpublished PhD Dissertation, Dept. of Agricultural and Resource Economics, University of Hawaii at Manoa, Honolulu. LAL, P. N. (1990), Conservation or Conversion of Mangroves in Fiji. Environment and Policy Institute, Occasional Paper, 11 (Honolulu: East-West Center). LAL, P. N. (1991), ‘Utilization and Management of Coastal Wetland Resources in Kosrae’, in Kosrae Island Resource Management Plan, Pacific Island Network, University of Hawaii (Honolulu: Pacific Island Network and Seagrant). MISHAN, E. J. (1976), Cost-Benefit Analysis (New York: Praeger). MOORMANN, F. R., and L. J. PONS (1975), ‘Characteristics of Mangrove Soils in Relation to their Agricultural Land Use Potential’, in G. E. Walsh, S. C. Snedaker, and M. J. Teas (eds.), Proceedings of International Symposium of Biology and Management of Mangroves, University of Florida. NASH, C, and J. BOWERS (1988), ‘Alternative Approaches to the Valuation of Environmental Resources’, in R. E. Turner, (ed.), Sustainable Environmental Management: Principles and Practice (Boulder, Colo.: Westview Press). PONNAMPERUMA, F. N. (1985), ‘Use of Mangle Land in the Philippines for Wetland Rice Production’, in UNDP/ UNESCO Report 1985, Report of the Workshop on the Conversion of Mangrove Areas to Paddy Conversion. Workshop Organized by the UNDP/UNESCO Research and Training Pilot Programme on Mangrove Ecosystems in Asia and the Pacific (RAS/79/002), in co-operation with the National Mangrove Committee of the Philippines. Los Banos, Laguna, Philippines, 1–3 April 1985.

424 POPE, R. M., and J. G. GOSSELINK (1973), ‘A Tool for Use in Making Land Management Decisions Involving Tidal Marshlands’, Coastal Zone Management Journal, 1/1:65–74. RAPHAEL, C. N., and E. JAWORSKI (1979), ‘Economic Value of Fish, Wildlife, and Recreation in Michigan's Coastal Wetlands’, Coastal Zone Management Journal, 5/3: 181–94. RUITENBEEK, H. J. (1991), Mangrove Management: An Economic Analysis of Management Options with a Focus on Bintuni Bay, Irian Jaya (Halifax, Nova Scotia: Dalhousie University). SAENGER, R, E. J. HEGERL, and J. D. S. DAVIE (1983), ‘Global Status of Mangrove Ecosystems’, Environmentalist, 3. (Suppl. No. 3): 1–88. SAENGER, R, M. M. SPECHT, R. L. SPECHT, and V. J. CHAPMAN (1977), ‘Mangal and Coastal Salt-Marsh Communities in Australasia’, in V. J. Chapman (ed.), Ecosystems of the World: I. Wet Coastal Ecosystems (Amsterdam: Elsevier). SHABMAN, L. A., and S. S. BATIE (1978), ‘Economic Value of Natural Coastal Wetlands: A Critique’, Coastal Zone Management Journal, 4/3: 231–47. Also see ‘Rebuttals’ by E. P. Odum and H. T. Odum and the reply to the rebuttals by Shabman and Batie in the Coastal Zone Management Journal, 5/3: 231–44. SHABMAN, L. A., and M. K. BERTELSON (1979), ‘The Use of Development Value Estimates for Coastal Wetland Permit Decisions’, Land Economics, 55/2: 213–22. SINDEN, J. A., and A. C. WORRELL (1979), Unpriced Values: Decisions Without Market Prices (New York: Wiley Interscience). STAPLES, D. J., D. J. VANCE, and D. S. HEALES (1985), ‘Habitat Requirements of Juvenile Penaed Prawns and Their Relationship to Offshore Fisheries’, pp 47–54. In P. C. Rothlisbergy, B. J. Hill, and D. J. Staples (eds.), Second Australian National Prawn Seminar (Cleveland, Australia: National Parks Service). TURNER, R. K. (ed.) (1988), Sustainable Environmental Management: Principles and Practice (Boulder, Colo.:Westview Press). TURNER, R. K. and T. JONES (eds.) (1991), Wetlands: Market and Intervention Failures (London: Earthscan Publications). UMALI, R. M., M. A. EUSEBIO, F. O. TESORO, H. T. CHAN, S. IBRAHIM, and J. E. ONG (1987), ‘Management Techniques and Methodologies’, in UNDP/UNESCO Report, Mangroves of Asia and the Pacific: Status and Management. Technical Report of the UNDP/UNESCO Research and Training Pilot Project on Mangrove Ecosystems in Asia and Pacific (RAS/79/002) (Quezon City, Philippines: Natural Resources Management Centre and National Mangrove Committee, Ministry of Natural Resources). UNDP/UNESCO, (1987), Mangroves of Asia and the Pacific: Status and Management. Technical Report of the UNDP/ UNESCO Research and Training Pilot Project on Mangrove Ecosystems in Asia and Pacific (RAS/79/002) (Quezon City, Philippines: Natural Resources Management Centre and National Mangrove Committee, Ministry of Natural Resources). WALSH, R. G., J. B. LOOMIS, and R. A. GILLMAN (1984), ‘Valuing Option, Existence and Bequest Demands for Wilderness’, Land Economics, 60/1. WATLING, D. (1985), A Mangrove Management Plan for Fiji: Phase I (Suva, Fiji: Fiji Government Press).

16 Urban Air Pollution in Developing Countries: Problems and Policies 16.1 INTRODUCTION Air-pollution control has traditionally been a low priority environmental issue in developing countries. Electricity generation and vehicle use (particularly cars), both major sources of air pollution in developed countries, are generally less significant on a per capita basis in developing countries. And, although industrial emissions tend to be less controlled in the developing world, lack of emissions-monitoring data has made it difficult to focus attention on this type of pollution. In contrast, the problems of deforestation, desertification, and water pollution-related diseases have dominated the policy agenda of both developing countries and their donors. Nevertheless, for a variety of reasons, air pollution in urban areas of developing countries is of growing concern. Population and output have been increasing rapidly in the primary cities, such as Mexico City, Calcutta, Lima, etc., bringing with them increased levels of air pollution. In addition, the number of large cities is growing. According to the World Health Organization (WHO/UNEP, 1988) there were thirty-five cities with populations over 4 million in 1980, but by the year 2000, there will be sixty-six, and by the year 2025, 135. In many of the large cities, concentrations of the conventional air pollutants far exceed US ambient standards and WHO guidelines, and respiratory disease rates are on the rise amidst a fall in the incidence of the traditional waterborne diseases. Particularly in China and India, air pollution from burning coal has attained levels high enough to bring international attention from organizations such as the World Bank and AID. In addition, greater concerns about both global warming and acid rain have focused worldwide attention on fossil-fuel burning by both developing and developed countries. Finally, increasing attention is turning to the poor quality of indoor air, particularly in those rural areas where poor-quality fuels and stoves are used for cooking and heating (Smith, 1987, 1988). Yet, in the midst of a worsening air-pollution situation, the policy response to these problems has been slow and, when steps are taken, they tend to be in the direction of command and control policies (such as setting

426 technology-based standards) rather than economic-incentive approaches (such as effluent fees or tradable pollution permits), even though the latter promises lower costs of environmental improvements. This preference for command and control policies is unfortunate in a developed country, but doubly unfortunate in a developing country because it is less able to afford the loss of resources implied by such inefficient policies.61 This chapter reviews the evidence for air-pollution problems in urban areas of developing countries and considers alternative policies for addressing these problems in light of the economic, institutional, and environmental differences between developing and developed countries.

16.2 WHAT ARE THE MOST SERIOUS AIR-POLLUTION PROBLEMS? One may search for evidence of air-pollution problems in developing countries anywhere in the chain of causation leading from emissions, to ambient concentrations (and their comparison with ambient standards), to exposure, to health and other physical effects, and to the monetized value of these effects. Data inadequacies and other problems common to most developing countries confound efforts to explore some of these links quantitatively, particularly emissions, exposures, and monetized values.62 Still, the contribution of various types of sources to air pollution is reasonably well understood. In addition, reasonably comparable and reliable, if not up-to-date, data on urban airpollution concentrations is available from WHO/UNEP and elsewhere and several very recent studies of disease in developing countries permit at least a partial picture to be drawn of how air pollution is affecting health.

16.2.1 Pollutant categorization The pollutants of potential concern may be classified into conventional air pollutants (sulphur dioxide (SO2), particulates, ozone (O3), nitrogen dioxide (NO2), and carbon monoxide (CO)), air toxics (including lead, benzene, 1, 3 butadiene, and others), and the regional and global pollutants (acid rain, carbon dioxide (CO2) and associated atmospherically reactive gases, and chlorofluorocarbons and similar compounds). The regional and global pollutants are not considered in this chapter.

61

See Hahn (forthcoming) for political economy-based explanations of the preference for command and control approaches in developed countries. Recent legislative and regulatory steps taken both in the USA (The Clean Air Act Amendments of 1990 permit electric utilities to trade sulphur dioxide allowances among themselves) and Europe (Sweden's replacement of some value added taxes with environmental taxes) suggest a swing towards economic-incentive approaches to pollution control.

62

China has reasonably complete emissions inventories for some air pollutants (see China, 1988). In addition, some work has been done (e.g. Smith, 1987, 1988) on characterizing the determinants of exposure in urban areas of developing countries.

427 The conventional pollutants have a wide range of possible effects which may be classified as effects on: health (acute morbidity, risks of developing chronic disease, and risks of premature mortality); a variety of productive activities (such as crop production); economic assets (such as building materials); and environmental assets (such as endangered species and parkland). Because of our concern about the effect of air pollution in cities and the clear preference of most environmental laws for protecting health, the focus here is on the health effects of urban pollutants.

16.2.2 Sources of urban air pollution There are three major categories of sources of air pollution in urban areas: point sources (industrial sources and power plants), mobile sources (cars, buses, motorcycles, trucks, etc.), and domestic sources (home heating and cooking). In developed countries, such as the USA (see Figure 16.1 and USEPA, 1989), coal-burning by power plants and industry are the major sources of SO2. Industry is also a major source of particulates and air toxics.63 The automobile is the primary source of CO, and an important source of Fig. 16.1 US emissions, by source, 1989 (teragrams per year)Source: USEPA (1991b).

63

The US chemical industry, according to the EPA's Toxic Release Inventory, June 1989, is responsible for four times more emissions of air toxics (946 million pounds annually) than its nearest competitor, the primary metals industry.

428 VOCs and NOx (along with industry and power plants), as well as some air toxics. Indeed, in the USA, Graham (1990) estimates that US cancer risks from exposure to air toxics are predominantly related to motor-vehicle emissions. Domestic heating and cooking is done primarily with natural gas, oil, and electricity in developed countries—all relatively clean sources of energy. In developing countries, the importance of various sources of pollutants differs from that of developed countries. Coal-burning for industrial processes and heating and for domestic heating (particularly in northern regions of China and India) and cooking are chief sources of SO2 and particulates.64 Localized sources of these pollutants, such as smelters and steelworks, also contribute large shares. Tall stacks from major industrial and electric-power sources mitigate local emissions of SO2, NO2, and particulate pollution at the expense of creating regional acid-rain problems. Of the mobile sources, diesel trucks and buses are an important source of particulates (Faiz et al., 1990), because there are proportionally more of them relative to cars and diesel engines generate ten times more respirable particulates than petrol engines (WHO/UNEP, 1988) per kilometre travelled, although diesel buses are less polluting than petrol passenger vehicles per person-trip (Rallis, 1988). In addition, in some countries, such as Mexico, diesel fuel is high in sulphur, creating SO2 problems. Two-stroke motorcycles and scooters also create large quantities of the conventional pollutants in some cities and on a per-kilometre-travelled basis.65 The contribution of biofuels as sources of particulates and other pollutants in urban areas varies by fuel mix and pollution characteristics of the fuels. Dunkerley et al. (1981: 49) examines cooking fuels by region in 1976. Urban fuel consumption is weighted more heavily towards biofuels in Sub-Saharan Africa than in east Asia and Latin America. More recently, the important role of biofuels in Sub-Saharan Africa was echoed by Barnes (1990), who notes the pattern of deforestation radiating out from urban areas in Malawi and in Burkina Faso and the highly important role played by wood in urban households in Sub-Saharan Africa generally. He finds that ‘in many cities, 90 per cent of urban households use wood fuels [including charcoal] for cooking’ (ibid. 8), this in spite of the wider range of fuels available in urban areas. On the other hand, Smith (1987), who provides a comprehensive discussion of the use and environmental consequences of fuelwood and other biofuels in rural areas, is evidently referring to other regions when he notes that indoor air in urban areas is not a major problem because the urban population uses relatively clean fossil fuels for cooking, e.g. propane (LPG) and coal briquettes.

64

However, in China many people in cities cook with coal briquettes, some of which are low in ash and contain a sulphur inhibitor.

65

Scooters generate twenty-two times the amount of VOCs per km as cars. They are particularly a problem in cities such as Singapore, which has one motorcycle for every 3.5 cars (Faiz et al. , 1990).

429 Smith also considers the pollution characteristics associated with alternative cooking fuels. On an exposure basis, respirable particulates (RSP), CO, NOx, and formaldehyde (HCHO) are all considerably higher when cooking with a wood-fired stove than with kerosene, and the latter results in significantly larger exposures than gas. Industry is responsible for much of the emissions of so-called air toxics; however, concern about these emissions should be kept well in check. The recently completed US EPA inventory of toxic air pollutants revealed emissions of such chemicals high above what had been expected. Nevertheless, a recent analysis of these data (Graham, 1990) showed that reducing these emissions by 90 per cent from their uncontrolled levels would result in a reduction of only 500 statistical cases of premature mortality. Since US toxic emissions are, like those of developing countries, currently uncontrolled (although US industry presumably emits less pollution per unit output because of use of more efficient production processes), these findings suggest that toxic emissions from industry may not be of primary concern in developing countries. The discussion of cancer rates below will add support to assigning a lower priority to air toxics, at least for the near term. In at least the largest urban areas of developing countries petrol-using cars probably contribute most of the CO pollution, and a large share of the VOCs and NOx, pollutants of concern in their own right and as precursors to ambient ozone. Because, in most urban areas of developing countries, car ownership and use are low,66 CO and ozone concentrations are currently of lesser concern.67 In developing countries without unleaded fuel, cars and diesel buses and trucks would also produce much of the urban lead emissions, while diesels would emit polycyclic aromatic hydrocarbons (PAH) (USEPA, 1990), and petrol engines would be more responsible for emissions of benzene and 1,3 butadiene. Lead emissions are primarily caused by burning fuel with a lead additive to enhance octane. Few, if any, developing countries have phased out lead in gasoline, although nearly all developed countries have taken or are taking actions to do this (UN, 1989b). In addition, poor controls on particulate emissions at lead smelters and of battery-manufacturing activities in developing countries and their continued use of lead-based paint point to lead as a major problem.

66

According to UN (1989a), cars per 1,000 people in cities are nearly always less than 100 and generally less than fifty in developing countries, while ranging from 200 to 500 per 1,000 in developed countries.

67

The share of mobile-source pollution is likely to rise in developing countries, increasing the importance of this source of air pollutants. From UN (1989a), of six developing countries reporting vehicle-kilometres-travelled (VKT) in 1977 and 1987, all experienced increases in VKT, ranging from 23 to 250 per-cent. In Indonesia, vehicle ownership tripled from 1970 to 1981, in Brazil and Lagos it more than doubled, in Nigeria it increased five times. Bus and truck traffic can also be expected to increase. Some 600 million trips per day were made by buses in cities in developing countries in 1980; this is expected to double by 2000 (UN, 1989b ).

430

16.2.3 Evidence on concentrations At issue are concentrations of pollutants in urban air, in blood (lead), and indoors. Ambient concentrations. The concentrations of several of the conventional pollutants in selected urban areas of developing countries are available from the WHO/UNEP Global Environmental Monitoring System (GEMS) (WHO/UNEP, 1988). One advantage of this database is its reliability and comprehensiveness. Identical equipment and measurement protocols are used for each city. Furthermore, usually three monitors are located in each city, one in an industrial area, another in a commercial area, and the third in a residential area. The disadvantages are that only concentrations of SO2 and suspended particulate matter (SPM) (rather than the finer, respirable particulates) are consistently monitored and reported, the published data are six years old, and the reported concentrations are averages over all monitors in a city, not just those from the same type of area (e.g. residential). Thus, the severity of ambient exposures may easily be exaggerated. In addition, the choice of cities to monitor in a given country is made primarily on the basis of geography and city-size, not on the basis of pollution problems. Thus, it is likely that there are major cities with worse pollution that are not in the GEMS system.68 Because of these drawbacks, the WHO/UNEP data are supplemented below by more recent and detailed data from China. Two measures are most important for assessing health risks—the annual average concentration is needed to gauge chronic health risks, and the number of days exceeding the daily ambient standard is needed to gauge risks of acute health response. For the first measure, WHO (1988) presents for each city the minimum and maximum annual average of monitored ambient concentrations during the 1980–4 period and the grand average computed over all monitoryears during this period. There are fifty-four cities monitoring SO2 and forty-one monitoring SPM. In addition, WHO has conducted special studies of other pollutants, such as lead (Pb), CO, and NO2. No data are systematically collected for ambient ozone. Assuming that the residential monitors always register the lowest concentrations,69 and that residential monitors best characterize population exposures, the annual average concentrations associated with the minimum monitor-years give the best indication of chronic health risks. The situation for SO2 is better than for SPM (Figures 16.2 and 16.3). All cities but Rio de Janeiro, Seoul, and Milan show at least one monitor-year that does not exceed the WHO annual average guidelines for SO2 (upper limit of 60 μg/m3). However, for SPM, about half of the cities report a minimum

68

For instance, Tianjin, China, appears to have worse particulate pollution than the Chinese cities in GEMS. See below.

69

An assumption that may not be reasonable for all cities, see WHO/UNEP (1988: 16) for Japanese cities

431 monitor-year annual average SPM concentration that exceeds the WHO guidelines (upper limit of 90 μg/m3). Cities in developing countries are disproportionately represented in this group. It should be noted, however, that dust levels in some of these cities are high and inflate the monitored values. In fact a recent study in Beijing found that in the summer wind-blown dust accounts for 60 per cent of SPM, while in the winter, it accounts for only around 20 per cent (World Bank, forthcoming). Considering the number of days exceeding the WHO daily guidelines, only in three cities does the residential concentration exceed the daily SO2 guideline (150 μg/m3), and of these, only one is in a developing country (Shenyang, China) (Table 16.1). For particulates, on the other hand, nearly all the cities in developing countries show minimum monitor-year daily concentrations exceeding the WHO daily SPM guideline (230 μg/m3), most exceeding the guideline for a substantial number of days. For instance, all five cities in China and the three Indian cities in GEMS report minimum monitor-year daily SPM concentrations exceeding the guideline (Table 16.2). Of these, the worst are Xian, China, which exceeds the daily guideline 189 days per year and Delhi, India, which exceeds the guideline 212 days per year. Recent data obtained from the Chinese National Academy of Sciences provide annual average concentrations for five cities for 1988 (Table 16.3). These data are particularly interesting because they include sulphate and nitrate concentrations, in addition to SPM. The first two compounds are most frequently used as a proxy for acid aerosol concentrations—the pollutants now thought to be more closely associated with mortality than SPM (see below). Table 16.3 indicates that SPM levels in these cities are high—545 μg/m3 in Beijing and 720 μg/m3 in Tianjin. However, if dust levels contribute about 40 per cent to particulate levels, on average, anthropogenic concentrations would be ‘only’ 327 and 432 μg/m3 in these cities, respectively. In addition, this dust serves a useful purpose. Its alkalinity is thought to neutralize the acid aerosols that would be created by emission in and around these two cities, yielding rain that is close to neutral pH. In this case, the high SO4 and NO3 concentrations in these cities (15 and 19.5 μg/m3, respectively compared to the Los Angeles average of between 4 and 6 μg/m3) are not good indicators of acid sulphates and nitrate concentrations. Of more concern are the high sulphate levels in 1988 in Liouzhou (in southern China), which is located near an area of very high acid-rain concentrations (pH = 4.1). The three-day summer SO4 concentration in air was 28 μg/m3, while with a longer averaging time (ten days) in spring, the concentration was 18.8 μg/m3. The latter concentrations compare unfavourably with daily peak SO4 concentrations for Los Angeles, which range from 4 to 22 μg/m3. Concentrations of lead in blood. Surveys of lead levels in blood in the USA since the lead phase-out programme began in 1975 show a startlingly clear

432

Fig. 16.2 Summary of the annual SO2 averages in GEMS/air cities, 1980–4Source: WHO, Global Environment Monitoring System, Global Pollution and Health, 1987

This page intentionally left blank

434 Fig. 16.3 Summary of the annual SPM averages in GEMS/air cities, 1980–4Source: WHO, Global Environment Monitoring System, Global Pollution and Health, 1987.

This page intentionally left blank

436 Table 16.1 Number of days per year with SO2 levels above 150 μg/m3 Country

City

No. of site-years

No. of days over 150 μg/m3SO2 Min.

Avg.

Max.

Australia

Melbourne Sydney

13 12

0 0

0 2

0 11

Belgium Brazil Canada

Brussels Sao Paulo Hamilton Montreal

13 11 8 10

0 0 0 0

12 12 3 10

32 32 7 32

Toronto

9

0

1

3

Vancouver

5

0

0

0

Santiago Beijing Guangzhou

9 8 12

0 0 0

19 68 30

55 157 74

Shanghai

10

0

16

32

Shenyano

7

43

146

236

Xian

7

4

71

114

Colombia

Cali Medellin

1 3

0 0

0 0

0 0

Denmark Finland France Germany, Fed. Rep.

Copenhagen Helsinki Gourdon Frankfurt Munich

3 8 4 6 3

0 0 27 8 0

0 2 46 20 0

0 7 64 38 1

Greece Hong Kong India

Athens Hong Kong Bombay Calcutta

3 10 13 8

1 0 0 0

9 15 3 25

15 74 32 85

Delhi

12

0

6

49

Iran Ireland Israel Italy Japan

Tehran Dublin Tel Aviv Milan Osaka Tokyo

15 6 9 8 20 15

6 0 0 66 0 0

104 1 3 29 0 0

163 3 24 167 0 0

Malaysia Netherlands New Zealand

Kuala Lumpur Amsterdam Auckland Christchurch

1 10 12 12

0 0 0 0

0 1 0 0

0 5 0 2

Philippines Poland

Manila Warsaw Wroclaw

4 13 15

3 3 1

24 10 8

60 19 22

South Korea Spain Thailand UK

Seoul Madrid Bangkok Glasgow London

6 7 3 5 6

5 0 0 4 0

87 35 0 14 7

121 95 0 21 17

USA

Illinois St Louis

4 3

0 1

1 3

2 8

New York City

12

1

8

22

Houston

3

0

0

0

Caracas Zagreb

8 15

0 3

0 30

0 80

Chile China

Venezuela Yugoslavia

Source: World Health Organization, Global Environment Monitoring System, Global Pollution and Health, 1987.

437

Table 16.2 Number of days per year with gravimetrically determined SPM levels above 230 μg/m3 Country Australia Brazil Canada

China

Colombia Denmark Finland Germany, Fed. Rep. India

Indonesia Iran Japan Malaysia Philippines Portugal Thailand USA

Yugoslavia

City Melbourne Sydney Rio de Janeiro Hamilton Montreal Toronto Vancouver Beijing Guangzhou Shanghai Shenyang Xian Medellin Copenhagen Helsinki Frankfurt

No. of site-years No. of days over 230 μg/m3SPM Min. Avg. 4 0 0 10 0 3 6 0 11 10 0 8 15 0 0 14 0 1 12 0 0 8 145 272 10 7 123 10 19 133 13 117 219 10 189 273 3 0 0 6 0 0 11 0 19 3 0 0

Max. 0 19 35 14 6 7 7 338 283 277 347 327 0 1 75 0

Bombay Calcutta Delhi Jakarta Tehran Osaka Tokyo Kuala Lumpur Manila Lisbon Bangkok Birmingham Fairfield Chicago New York City Chatanooga Houston Zagreb

12 8 12 7 15 20 15 5 7 7 12 9 5 7 12 16 7 15

207 330 338 268 347 2 4 59 225 28 209 28 0 14 0 17 0 57

23 189 212 4 8 0 0 10 0 4 5 0 0 0 0 0 0 13

100 268 294 173 174 0 2 37 14 12 97 7 0 6 0 1 0 34

Source: World Health Organization, Global Environment Monitoring System, Global Pollution and Health, 1987.

438 Table 16.3 Concentrations of pollutants in Chinese cities, 1988 City Beijing Tiajin Langfang Tanggu Jixian Guangzhou

Liouzhou

Pollutants (μg/m3) TSP 545 720 533 248 529 223

SO4 15.00 19.50 9.10 11.90 12.70 21.80

NO3 5.78 6.29 5.02 6.60 4.31 9.60

127

8.92

3.08

189

18.82

2.14

260

28.01

3.23

Note Annual daily average Annual daily average Annual daily average Annual daily average Annual daily average 10-day average, spring 3-day average, summer 10-day average, spring 3-day average, summer

Fig. 16.4 Lead used to produce petrol and average blood-lead levelsSource: USEPA (1985).

relationship between lead in petrol and blood-lead levels (Figure 16.4). From 1976 to 1980, mean blood-lead levels in children in US cities fell by 37 per cent (Sinha et al., 1989) while petrol-lead emissions were reduced 60 per cent (USEPA, 1985). In virtually all developing countries, leaded petrol still dominates. With increased use of petrol, lead is rising in the blood of people living in industrial areas of developing countries: over an eight-year period blood levels rose 24 per cent in women, to 73 μg/1. Overall, it has been estimated that 60 per cent of children in developing countries have blood-lead content of 25 μg/dl or

439 higher, as compared to the WHO guideline of 20 μg/dl (the recommended maximum blood-lead level in the USA is 30 μg/dl). However, in WHO/UNEP (1988), of ten cities participating in a blood-lead survey (five in developing countries (Mexico City, Lima, Ahmedabad, Beijing, and Zagreb) and five in developed countries), only individuals in Mexico City exceeded WHO blood-lead guidelines in 1981, and this may be caused, in part, by drinking-water delivered in lead pipes. Still, the survey was very small in scope and apparently did not focus on children. WHO/ UNEP (1988) estimates that one-third of people living in cities are exposed to unacceptable lead concentrations. Further growth in vehicle use is likely to cause corresponding increases in blood-lead levels. Indoor concentrations. Smith (1988) estimates that SPM concentrations indoors in developing countries are 25 per cent higher than ambient levels because of solid fuels used in cooking, and tobacco smoke. However, as countries develop, both their indoor and outdoor SPM levels fall as cleaner fuels are substituted for dirtier ones. Krupnick (1989) notes that in four typical Chinese cities, where coal is used for cooking, respirable particulates (RSP) can reach 650 μg/m3; SO2, 860 μg/m3; and CO 14, μg/m3.

16.2.4 Evidence of health effects Data are generally insufficient to causally or even statistically link air pollution to health in developing countries. However, there is a large body of evidence linking air pollution and certain types of health effects in developed countries which can be applied to developing countries. In general, using this literature in a developing-country context is likely to understate the health effects in two plausible cases: (i) dose–response functions are non-linear and increasing with dose and baseline concentrations are larger in developing than developed countries; (ii) the marginal effect of dose on response is inversely related to health status and health status is poorer in developing countries. To summarize this literature, the linkages between exposures to ambient ozone and SO2/particulate (particularly fine particulates) on the one hand and acute health effects, on the other, are the most well established. Epidemiological studies have consistently found links between certain measures of particulates and chronic respiratory effects. A controversial literature has also linked particulates (such as sulphates as a proxy for acid aerosols) to mortality risks.70 CO at ambient levels may raise the probability of experiencing angina for an estimated 5–7 million people in the USA who are at risk (a prevalence rate of 3 per cent). However, according to Lippmann (1989), in spite of much effort expended to identify health effects of this pollutant at

70

These macro-epidemiological studies regress group mortality or incidence rates (e.g. total mortality rates) for various cities on their average annual pollution concentrations. Because the analysis is conducted at such an aggregate level, scientists are reluctant to infer such findings establish a health–air pollution link.

440 ambient levels, too little information is available to derive exposure–response functions. NOx effects are generally insignificant, except as they work through secondary pollutants (ozone or acid aerosols). There is also general agreement that children with ‘high-typical’ blood-lead levels suffer learning disabilities and recent studies link ambient lead exposure to high blood pressure. With this link, lead becomes a risk factor for hypertension, heart attacks, and strokes, particularly in men. Finally, many of the constituents of vehicle exhausts have been identified as known or potential carcinogens. However, there is some question about whether doses are high enough to represent a significant cancer risk to the population. Because data limitations in developing countries often preclude estimating health effects from these dose–response functions, comparing data on the incidence of air pollution-related health-effects in developing and developed countries may reveal whether serious problems exist in the former. There is recent and well-documented evidence that mortality from chronic obstructive pulmonary disease (COPD) and acute respiratory infections (ARI) and morbidity from respiratory-related causes are high in developing countries. From Lopez (1989), crude COPD mortality rates are estimated (with much uncertainty)—62/100,000 for developing countries compared to only about 34/100,000 for industrialized market countries.71 From the Ministry of Public Health, China (1988), COPD was the leading cause of death in China in 1988, with a death rate of 162.6 per 100,000, or 26 per cent of all deaths.72 The rate for cities in China was much lower than in rural areas, the city rate being ‘only’ 86/100,000. When standardized and compared to the USA, the rate in China is over five times greater (Bumgarner and Speizer, 1989)—105/100,000 in China, 19/100,000 in USA. The differences are even more dramatic in the 55–74 age group—600–700/100,000 in China, compared to 80/100,000 for females and 200/100,000 for males in the USA. In addition, actual mortality rates in both countries are likely to be much higher, but particularly in China, because of underreporting and diagnostic difficulties. Mortality caused by acute respiratory infections (ARI) is surprisingly high in developing countries. Age-adjusted, the rate is 141/l00,000 versus only 29/100,000 for developed countries: 25–30 per cent of this mortality is in children under 5 years old, about 5 million cases in 1985. While between 25 per cent (Stansfield and Shepard, 1990) and 50 per cent (Bulatao and Stephens, 1989) of ARI are ‘vaccine preventable’, the rest could conceivably be related to air pollution. Morbidity rates are also quite high in urban areas of developing countries compared to those in rural areas. The Chinese Ministry of Public Health (china, 1988) found chronic bronchitis ten times higher in cities polluted by

71

Bulatao and Stephens (1989) estimate a larger gap in COPD rates: 54/100,000 for developing countries, 18/100,000 for industrialized countries.

72

Survey of eighty-one countries.

441 coal emissions than in control areas; furthermore, children were found to have lower health status if they lived in coalburning households than if they lived in a household where another fuel was burned. Prevalence of chronic bronchitis in people age 60 and above in Chinese cities was 9.4 per cent. In India, Bumgarner and Speizer (1989) found chronic bronchitis rates of 12 per cent. In Nepal, with rates of chronic bronchitis above 18 per cent, cooking and heating emissions could be an important risk factor. For comparison, in the USA, even combining chronic bronchitis and emphysema, prevalence is only 4.5 per cent. The above high rates of respiratory-related disease is not, in itself, convincing evidence for an air-pollution problem. Most important, smoking is on the rise in the developing world, with this habit likely to dominate air pollution as a cause of health effects (Bumgarner and Speizer, 1989). Concerning cancer, the role of toxic air pollutants cannot be very large. Lung-cancer rates are very low in both rural and, to a slightly lesser extent, urban areas (Barnum and Greenberg, 1989). The crude mortality rate is only 9.2/ 100,000 for males, and 3.1/100,000 for females living in developing countries, while it is 65.3/100,000 for males and 16.3/100,000 for females in developed countries. The low cancer rate in developing countries is thought to be primarily a result of diet and other lifestyle factors, although with smoking on the rise, lung cancer may increase dramatically. Clearly, toxic pollutants that have been linked with cancer are not yet having major effects on health in the developing world.

16.2.5 Conclusion Based on the foregoing discussion, industrial SO2 and particulate emissions, diesel particulates, and cooking and heating emissions appear to be of current concern to urban populations in developing countries. Car and two-wheel vehicle emissions (VOCs, NOx) are of future concern, primarily because of their ozone-forming potential. Air toxic concentrations, aside from lead, are not generally measured in either developed or developing countries; although risks from lead and risks from pollutants emitted by diesel engines and indoor fuel-burning (PAH) deserve current attention, risks from auto toxics (benzene and 1,3 butadiene) are an emerging problem. Future growth in the demand for electric power in urban areas may also cause this sector to be a major source of SO2, particulates, and NOx.73 These conclusions are reached while acknowledging that the scientific chain of causation from emissions to health effects contains many gaps, and

73

However, appropriate placement of new plants away from urban areas, use of tall stacks, and installation and operation of particulate pollution-control devices should mitigate urban effects, at the expense of regional scale SO2 -based acid-rain problems. And even here, atmospheric, soil, and lake buffering capacities may be sufficient in many areas to neutralize, or substantially mitigate the regional scale effects (NAPAP, 1990).

442 information from developing countries is particularly lacking. An important information problem is the reliance of developing countries on SO2 and SPM monitoring while in developed countries monitoring of fine particulates and aerosols is becoming more important.

16.3 POLICY RESPONSE Developing countries are now particularly receptive to new ideas for pollution-control policies, because of high levels of urban air pollution, increased worldwide concern about global warming and sustainability, and increasing interest from donors in tying aid to structural and environmental policy reforms. The purpose of this section is to identify attractive (e.g. efficient) policies for controlling the urban air-pollution problems caused by point, mobile, and indoor sources. This is done by matching the requirements of economic incentive (EI) and command and control (CAC) policies on the one hand, with the unique economic, institutional, and environmental features of developing countries (what are termed ‘stylized facts’) on the other. Because developing economies are characterized, in part, by significant market distortions, some attention is given to sectoral adjustment policies as well.74,75

16.3.1 Policy types Command and control policies, such as mandating the use of certain pollution-control technologies, are those that limit discretion of the polluter in taking actions to reduce pollution. EI policies, such as emissions fees, leave the polluter with a large measure of discretion and feature either a carrot (a financial gain) or a stick (financial loss) as incentive to choose the least-cost means of reducing pollution. Sectoral policies, such as increasing the price of energy inputs, may be considered here an economic incentive approach. The presumption of this discussion is that, ceteris paribus, the discretion embodied in EI approaches makes them more efficient than CAC approaches in the short run and that, in the long run, the EI approaches create continuing incentives to innovate and cut costs that are absent in a CAC system.

74

Even if input and output prices were set competitively, there would still be market failure from pollution externalities. While, in practice, commodity-pricing reforms are probably a necessary condition for obtaining the optimal rate of pollution reductions, they are not sufficient. Internalization of the pollution externalities by instruments designed for this purpose is needed.

75

Although this section is focused on policies for developing countries, no attempt is made to comprehensively characterize their current pollution-control policies, in part because a comprehensive analysis of these policies, analogous to the recent OECD study of developed countries (OECD, 1989), does not exist. This omission may not be particularly important in any event because few air-pollution control policies are in place in developing countries, and those that are, with some exceptions, are rudimentary and poorly enforced. For instance, air-pollution control has suffered in Mexico City because of poorly drawn laws, a lack of regulations, and fragmented administrative responsibilities (DuMars and Beltran Del Rio, 1988).

443 Both classes of policies require regulations, monitoring, and enforcement, and, in practice, best operate from a set of emissions standards and information on actual emissions. CAC policies for air-pollution control typically involve mandating how these standards are to be met, such as requiring installation of particular equipment (scrubbers on electric power-plants, catalytic converters on vehicles) or require that all polluters in an industry sector meet the same emissions (discharge) standards without regard for each firm's special circumstances (e.g. one firm's higher cost of control than another firm's). The most thoroughly analysed incentive approaches, i.e. policies which convey much discretion in reducing emissions to the polluter, include emissions fees, subsidies, and tradable permits.76 Emissions fees may be distinguished from permit fees or user-charges in that the former are set higher than the marginal abatement costs for at least some polluters to induce emissions-reduction behaviour. The best example is the German system of effluent fees with rates on water pollutants based on toxicity (Brown and Johnson, 1983; Opschoor and Vos, 1989). There are few examples of what economists consider to be an emissions fee system on air pollution (and none in the USA). Proposals for such fees on CO2 emissions have recently been advanced in the USA and elsewhere (D’Arge and Spash, 1990).77 A close cousin of effluent fees—charges on polluting products or pollutant content—is found in Norway, where a tax is levied on the sulphur content of coal burned (OECD, 1989); there are also differential taxes on leaded and unleaded petrol in many European countries and on car emissions characteristics. Subsidies may be in the form of grants, soft loans, and tax allowances. Many countries provide grants to stimulate the development and implementation of new pollution reduction technology. China makes low-interest loans for pollution control from a fund established by revenues from its compliance penalty system (Qu Geping, 1989).78Tradable emissions permits involve the establishment of a desired quantity of emissions, the allocation of these emissions permits by the Government to polluters (either by seignorage or auction), establishment of emissions standards, and the sale of any excess

76

These policies have been extensively elaborated upon (e.g. a system of spatially differentiated charges, or a mix of fees and subsidies in the case of polluting monopolies). See Tietenberg (1985), Bohm and Russell (1985), Baumol and Oates (1975), Krupnick, Oates, and Van de Verg (1983), and Cropper and Oates (1990). These studies also contrast economic incentive approaches to CAC approaches and do so better than can be done here. Rather, this chapter examines to what extent the special circumstances of developing countries alters or reinforces points made in this literature.

77

France sets fees on SO2 emissions but they are too low to change behaviour and, therefore, may be considered simply a revenue-raising device (OECD, 1989, Table 5.1).

78

Subsidies for pollution abatement or other environmentally beneficial activities are given little attention in this chapter because of their inferiority to taxes and tradable permits on several grounds, including the need for public revenues, the perverse incentive to pollute to obtain a subsidy, and the improper price signal (i.e. the tendency for subsidies to reduce rather than increase the price of pollution-intensive commodities).

444 emission reductions by polluters to other polluters who find it cheaper to buy emissions permits than to reduce their wastes to meet the standard. The major examples of emissions trading are the US lead trading and Emissions Reduction Credit Program (Tietenberg, 1985) and the US Clean Air Act Amendments of 1990 which permit trading of SO2 emissions allowances between power plants (and other sources) (see Quarles and Lewis (1990) for a readable summary of the Act). In the rest of this section, point-source pollution control policies are reviewed first and in some depth. Then policies for controlling mobile source and indoor air pollution are more briefly examined.

16.3.2 Point-source pollution control policies First, some stylized facts about developing versus developed countries are offered and then the implications of these ‘facts’ for the choice of pollution control policy are examined. Some stylized facts. Of the many differences between developed and developing countries, there are eight of particular relevance to air-pollution control policy. In developing countries: (i) concerns about minimizing control costs dominate concerns about reducing air pollution to a greater extent than in developed countries; (ii) capital is more scarce and labour is cheaper; (iii) much less attention is paid to operation and maintenance of equipment; (iv) the baseline level of emissions control is lower; (v) market distortions, in the form of energy, water, and materials subsidies, state monopolies, and other competitive restrictions are far more pervasive; (vi) enforcement of all types of policies is lacking and hampered by corruption; (vii) institutions for the conduct of the economy and pollution control are weaker, including the legal framework, data collection and monitoring systems, regulations, lines of authority, and expertise; and (viii) revenue needs of local/state governments are more pressing. Most of these differences are self-evident. However, at least one deserves some comment. Capital scarcity has unclear environmental consequences. On the one hand, such scarcity implies that output and emissions will be low. On the other, the capital that is in hand will tend to be older, less reliable, and more polluting, than in developed countries. The consequences of unreliability are particularly interesting. In Lagos and Nigeria, capital scarcity in the electric-power industry has led to downtime and unreliable power service. As a result, some companies have built small power-plants as back-up power supply, typically with lower stacks, less sophisticated or absent air-pollution control equipment, and at locations closer to residential areas. Thus, even if emissions from generating power through backup generators equals that from power-stations, the net effect of reliance on back-up power is to raise pollution exposures (Krupnick and Harrington, 1990).

445

16.3.3 Implications for policy (i) Concern over costs. Probably the most potent and long-standing argument against economic incentives is the one advanced for delaying or scaling down any environmental control policy, i.e. that costs of production will go up without corresponding increases in ‘productivity’ which will cause a slow-down in growth and a reduction in international competitiveness (causing a decrease in exports and/or an increase in imports). This argument still has many adherents even though study after study, at least in the USA, has shown only a small negative effect of environmental policies on GNP (Conrad and Morrison, 1985) and the competitiveness argument has not found empirical support (Tobey, 1990). Suffice it to say, however, that a bias towards economic growth and, its corollary, reducing pollution-control costs, at the expense of not obtaining environmental goals is more compelling for a developing than a developed country. A major reason is that the benefits of economic growth are likely to be much larger at the margin than in developed countries. And, such growth is needed to generate capital for later and larger environmental improvements. Further, in spite of the legitimate concern that development will suffer if resources are diverted to environmental protection (and the more sophisticated argument that development is a precondition to the capital formation needed to finance pollution-control activities), there are a number of considerations that make this trade-off less severe for developing countries. Where inefficiency permits reductions in material and energy throughput, improvements in air quality can deliver economic, as well as environmental gains. These inefficiencies are likely to be higher in developing countries because of the widespread practice of subsidizing energy, materials, and water use. In addition, the environmental improvements deliver economic benefits in the form of increased labour productivity, reduced need for medical care, and direct increases in agricultural yields. While measures of the welfare improvements of air-pollution reductions (including visibility benefits and non-productivity enhancing health improvements, for instance) are not available, they may be quite large, further reducing the trade-off between economic growth and environmental protection. Does the preference for cost savings over environmental protection favour one class of policy over another? Other things equal, economic-incentive approaches should be favoured over command and control approaches even more in developing countries because the former promises cost savings over the latter to meet given environmental goals, at least from US studies, of from 20 to 90 per cent (see Tietenberg, 1985 for a summary). Following the work of Weitzman (1974) and others, developing countries matching stylized fact (i) would favour emissions fees over tradable permits, because the former constrains marginal costs while the latter constrains marginal damages.79

79

Alternatively, a growing economy concerned about keeping environmental quality or emissions constant would favour tradable permits over fees because the effect of economic growth on the environment is alleviated by rising permit prices rather than by letting pollution increase.

446 (ii) Scarce capital, cheap labour. The tight-capital constraints and low costs of labour found in developing countries have rather straightforward implications for air-pollution control policy. One oft-touted strength of CAC policies is that they can raise the rate of adoption of new technologies, e.g. by mandating use of best available technologies (or pollutionreduction performance criteria consistent with these technologies) in new plants (e.g. through New Source Performance Standards in the USA, or through the Chinese requirement that 7 per cent of investment capital in a new plant be set aside for investments in pollution-control equipment). However, these types of policies de-emphasize investments in low-polluting processes or product lines, or plant designs that lower the costs of operating and maintaining pollution control equipment. Capital shortages also imply that industry scale will be smaller in developing countries (except for state-supported activities). In this case, by following the market model, economic incentives are favoured because they are presumed to work well when a large number of decentralized sources with heterogeneous marginal abatement costs80 are being regulated. Where sources are few, large, and more homogeneous (for instance, the case of the state-run chemical industry in China which has only eighty firms), more direct forms of regulation may be reasonably efficient. (iii) Poor operations and maintenance. Because operations and maintenance activities and spare-part availabilities are low in many developing countries, fairly modest incentives (such as might be provided by a modest emissions fee) could bring large pollution reductions at low cost. Low-cost options could include: keeping up with regular maintenance schedules, purchasing spare parts well in advance, and training workers better. CAC approaches, on the other hand, are ill-suited to improving operations and maintenance because of the enormous range of existing conditions at firms and the broad scope of steps that could be taken to reduce pollution through tighter plant-level management. (iv) Low baseline pollution control. With marginal costs of air-pollution abatement generally growing with increasing pollution reductions, developing countries, with lower baseline pollution-abatement activity, are likely to face relatively low marginal abatement costs. This point reinforces the idea noted above that relatively modest emissions fees could bring forth significant improvements in pollution-reduction activities. However, a related implication of being on the flat section of the marginal cost curve is that, operations and maintenance activities aside, reasonably cost-effective options for pollution

80

Abatement in the broadest sense, e.g. including process changes that would reduce material/energy throughout.

447 reduction can be identified relatively easily. Thus, in this case, the efficiency loss from a CAC system may not be large.81 (v) Market distortions. Under this stylized fact, we consider price distortions, tax distortions, and distortions arising from uncompetitive firms. Developing countries, irrespective of their degree of market orientation, pervasively distort market prices of products; in particular, prices of energy are often kept below market rates. Kosmo (1989) cites petrol and natural gas prices in Egypt of one-third and from 10 to 20 per cent of world levels, respectively. In the electric-power sector in Turkey, lignite prices (in 1983) were held to only 68 per cent of the coal-equivalent border prices. In China, the existence of a two-tier price structure for coal had led to ‘in-plan’ coal in Beijing selling for about $50/ton while the ‘out-plan’, or negotiated, coal price was over double this figure (Wang, 1988). In energy-using sectors, these distortions result in inefficient and excessive use of energy inputs and excessive emissions per unit output. In the energy-producing sectors, these distortions result in economic losses and an associated lack of investment in new plant and equipment as well as pollution-control devices; both of which tend to result in higher emissions per unit output. The existence of such distortions creates an obvious target for policy reform—simply raise energy prices by fiat (in a planned economy) or permit them to find their market level. In addition to the demonstrated reluctance of many Governments to implement such reforms, the environmental problem with this approach is that latent demands for products with controlled prices may be large. Their decontrol may lead in short order to increases in consumption that may wipe out many of the emissions reductions associated with improved efficiency. The use of relatively more distorting forms of taxation is also a key divergence from the purely competitive economy, one that is present in market and planned economies alike. Developing countries rely more heavily on commodity taxation than developed countries, however. By taxing different classes of commodities differently, more distortions in consumer choices and the allocation of resources are introduced than when an income tax or even a sales tax is used. These distortions give emissions fees an advantage over other pollution-control policies. Terkla (1984) examines the welfare gains in the USA from substituting pollution taxes for more distortionary ways of raising revenues (such as the income and sales tax). By shifting the tax burden to the environment, Terkla finds not only that welfare is improved through the efficient pricing of the environmental services, but that deadweight losses associated with the traditional taxes are greatly reduced. Indeed, the major benefit of such

81

It is equally the case, at least in a static sense, that when nearly everything that can be done to reduce pollution levels has been done, the gap between EI and CAC policies would again be narrow or non-existent since the remaining control options, while very expensive, are fairly obvious.

448 substitution appears to be the latter. It follows then, that in countries where revenues are raised by more onerous, distortionary forms of taxation than in the USA, pollution taxes may bring even larger (as a percentage of GNP) improvements in welfare if they are substituted for other forms of taxation. Unfortunately, there is little evidence that revenue-short Governments in either developed or developing countries are inclined to such substitution. Rather, emissions fees are more likely to be seen as a way of raising additional revenues, not as a substitute. In this case, the case for emission fees is weakened. Indeed, in economies where prices are distorted in a major way, considerations of the second best come to the fore and one cannot confidently predict that emissions fees will raise or lower national welfare. In this second-best case, Buchanan (1969) showed that an effluent tax on a monopolist could conceivably reduce welfare because of the effect of the tax on further restricting output. However, following the same logic, an effluent tax on a subsidized commodity would be likely to raise welfare, both on environmental and economic grounds. And Oates and Strassmann (1984) show that the effects of market imperfections may be swamped by allocative gains from reduced pollution. Finally, at least at first glance, it might seem that EI approaches would be a poor choice for economies or sectors of economies with major centralized planning components, e.g. China, India. After all, where budget constraints are soft, accountability is lacking, profit retention is limited, and prices are fixed, the scope for incentives appears to be quite limited. Nevertheless, with the fall of centralized planning systems world-wide, these economies have been accelerating their movement to more market-orientated systems, even in China. The World Bank (forthcoming) reports on a wide array of economic-incentive approaches to environmental pollution control being used in China, including: (1) the ‘responsibility system’, which provides workers and managers with large bonuses and awards for meeting contractual obligations for pollution control as well as gaining for their firm the right to be considered a ‘well-performing’ or ‘outstanding’ enterprise, with control over its own foreign exchange and more discretion over the fate of accumulated profits; (2) a non-compliance penalty system which places a fee per unit on those emissions exceeding the emissions standards and recycles the collected funds for pollution-control investments. (vi) Poor enforcement. A credible enforcement effort is needed for the conduct of any pollution-control policy. However, a lack of enforcement capability, whether because of a shortage of staff, too few monitors, or weak regulatory and legal institutions may give a slight edge to CAC approaches—at least according to the perceptions of Governments who reason that the flexibility provided by EI policies requires more costly and vigilant monitoring and enforcement efforts. In reality, the details of regulatory design have more to do with such burdens than the type of policy.

449 The enforcement burdens of a fee system, for instance, can be minimized by following an approach taken by both the Germans and the Chinese. This involves basing fee payments on the discharge permit and making several unannounced visits to the plant per year to check to see if the permitted emissions are being exceeded. If they are, fees are doubled or even tripled. To counter the disincentive of this system to improve upon permitted levels between inspections, the German system also provides for steep discounts in fee rates if the measured emissions are well below the permitted emissions level. Beyond tinkering with the design of an EI system, one can take the ‘blunt instrument’ approach—sacrificing some efficiency for the gain in enforcement and administrative ease by taxing inputs or outputs of production rather than the emissions themselves. A tax on the sulphur content of coal82 can reduce SO2 emissions and is easy to document, requiring little enforcement or monitoring activity. Of course, such an approach only provides incentive to reduce purchase of the input, not necessarily to find the least-cost means of reducing pollution and, if the input is not chosen with a careful eye towards substitutes, there is no guarantee that emissions will fall. Nevertheless, on balance, this approach seems to be a reasonable second-best policy. One way to encourage enforcement effort, an approach also taken by Germany, is to tie the budget of the enforcement agency to revenue collections. The principal problems with this approach are that it can encourage overzealous enforcement as well as focus attention on increasing revenues rather than on increasing pollution cut-backs by firms trying to avoid paying revenues. What about tradable permits? The conditions for reasonable operation of this type of system—well-defined markets, up-to-date and rapidly communicated information on permit prices, and continuous monitoring (if trades are frequent)—would seem to be absent in developing countries. However, policies that permit individual firms to meet emissions standards for a plant rather than stack by stack (called the ‘bubble’ approach in the USA) are well within the capabilities of any economy and can deliver large cost reductions over stack-by-stack regulations (Tietenberg, 1985). (vii) Weak institutions. This stylized fact covers existing institutions, data shortages, lack of government expertise, and a lack of firm management expertise. Although institutions in developing countries may be weak, experience in the developed countries suggests that successful introduction of economic incentive approaches is aided by building on whatever existing systems are in place, rather than trying to start afresh. It is difficult to generalize about these existing systems. China has a formidable array of environmental laws and regulations, over 50,000 national, provincial, municipal, and county workers

82

This policy has been implemented in Europe and Yugoslavia (Kosmo, 1989).

450 in the environmental protection agencies.83 In addition, China already has a complicated non-compliance penalty system. Therefore, further modifications of this system to make it more efficient may be more appropriate than adoption of a new system, such as marketable permits. Monitored data on emissions and data on costs of reducing pollution are typically in short supply. Although economic incentive approaches economize on the regulator's costs of acquiring and processing information (relative to CAC) by requiring the polluter to choose a response to the regulations, these approaches generally require much better data on baseline emissions than a CAC approach. For instance, to distribute emissions permits under a tradable permit system, data is required on current baseline emissions as well as on post-control emissions. With technology standards, demonstration that the technology is in place and capable of operating is all that is required to start the policy.84 Even with emissions standards, information on current baseline emissions levels is not absolutely unnecessary. Still, one expects that the more expensive and more uncertain data are those associated with pollution-reduction costs rather than simply current emissions. Indeed, in most Western countries, current emissions of major pollutants are recorded on emissions permits with records kept by the Government. For major US industries with relatively few plants, such as electric utilities, historical data series on emissions of SO2 and particulates are codified and available on all plants. A major issue is to what extent expertise of government officials is lacking, relevant to the conduct of environmental programmes and policy design. Recently, assistance from foreign agencies, donors, and academic groups, plus a vast technical and popular literature have augmented these resources considerably. To the extent that CAC policies are easier to understand and implement, they would be favoured on this point. However, the easiest improvement of all is to permit prices of polluting inputs to rise. This takes no monitoring, planning, regulation writing, or enforcement but, by itself, can lead to material and energy conservation and a consequent reduction in air pollution. Of course, this strategy can only go so far, and cannot lead to internalization of the pollution externalities. The most important reason that EI approaches tend to be favoured on efficiency grounds over CAC approaches lies with the assumption that expertise in reducing pollution lies with the firm; the Government can never hope to learn all it needs to give appropriate commands for the installation of technology, particularly those that are cost-minimizing (since the Government per se does not bear the costs). Particularly in a dynamic economy, keeping up with the latest technological developments is an impossible job for an administrative body.

83

Plus tens of thousands more in the sector ministries at all levels of government and within the state enterprises (China, 1988).

84

But, to obtain an equivalent emissions reduction from the polluters under each system, post-control emissions monitoring would be needed.

451 Whether the information held by a firm in a developing country is greatly superior to that held by the Government is an open question. Such economies will have a large proportion of new entrepreneurs and small enterprises, and therefore are likely to have little expertise in pollution control. At the same time, their undercapitalization coupled with lack of enforcement of pollution laws for small firms, make these firms potential heavy polluters per unit of output. While technical support, i.e. research and development (R & D), is clearly within any government's set of legitimate functions, it is less clear that government should ever take the lead role in implementing or choosing technologies for enterprises. In the long run, enterprises that are given the incentives to learn about or develop appropriate technologies can be expected to do so themselves or hire those who can. R & D funds may better be spent on management training. (viii) Revenue needs. Some EI approaches have one additional advantage that CAC approaches do not: effluent fees and auctions of tradable permits have the added advantage of generating revenue for the Government (although a fee high enough to cause pollution reductions diminishes the revenue base). Symmetrically, industry may favour a CAC approach because it wants to avoid the potentially large transfers of revenues to the Government of a properly designed fee system, transfers that for some firms may far exceed their outlay for pollution-control equipment under a CAC programme that would attain the same environmental standard (see Tietenberg, 1985, for a summary of studies).85

16.3.4 Conclusions for point-source control policies The foregoing discussion suggests that the efficiency case for economic incentives is even stronger for developing than developed countries and that, on balance, a system of emission fees, along with price reforms on subsidized polluting production inputs, are attractive approaches for the control of urban air pollution from industrial sources. It was also suggested that relatively modest fees could encourage significant emissions reductions. Implicit in this suggestion is a serious departure from optimality in fee design, as one cannot say, a priori, whether the fee on a particular pollutant should be large or small.86,87

85

Actually, the literature shows that CAC policies, such as uniform emissions-reduction policy, result in substantial overcontrol. Only if damage functions have thresholds at this point do such additional pollution reductions have no value. Therefore, the gap in net benefits between a CAC system and an EI system may be narrower than it appears from the studies in the literature.

86

Bohm and Russell (1985) conclude that rankings of options, even in static efficiency terms, depend on the type of emission, the strictness of the ambient standard being contemplated (i.e. with a very strict or very loose standard economic incentives are not needed), and the characteristics of the economy and environment. Such a mixed system implied by this prescription seems overly complicated, particularly when suboptimal policies are being implemented, as would be the case in developing countries.

87

In addition, no attention has been paid in the above discussion to the spatial differentiation of fees and the linkage to ambient concentrations, as well as a host of other issues that would necessarily be a part of an optimal fee-setting exercise.

452 Indeed, with market distortions so pervasive, it would be particularly challenging to estimate the opportunity costs of emissions reductions as an input to an optimal fee-setting analysis. With air-pollution problems serious and likely to worsen in urban areas of developing countries, it is imperative to put a reasonable, workable economic incentive system in place now.

16.3.5 Mobile-source control policies The policy menu is much more varied for controlling mobile-source pollution than point-source pollution. This is so in part because the link between congestion and emissions as well as trips and emissions (through the ‘cold start’ phenomenon)88 makes transportation-control measures useful for reducing emissions and in part because policies can be applied at so many different points: fuel and car producers, consumers, fuel prices, vehicle prices, vehicle registrations, parking, bus fares, etc. As for point sources, we consider several stylized facts relevant to mobile sources. Developing countries compared to developed countries: (i) use leaded petrol and diesel fuels; (ii) feature a higher proportion of buses (including jitneys and other multi-passenger vehicles) and commercial vehicles in many urban areas (UN, 1989b), and (iii) have poorer maintenance of vehicles. The capital-poor nature of developing countries is also important to keep in mind. Leaded fuels. One area deserving immediate attention of policy-makers is lead in transportation fuels. Developing countries, with some exceptions (such as Mexico; see DuMars and Beltran Del Rio, 1988), rely on leaded petrol and leaded diesel fuel. One of the first priorities should be to follow developed countries in reducing and eventually eliminating lead from motor fuel. Cost-benefit analyses in the USA show overwhelming net benefits from such a policy (USEPA, 1985). Ambient lead levels will immediately fall, bringing commensurate reductions in health and other effects of this highly toxic pollutant. As unleaded fuel is more expensive to produce than leaded fuel, fuel prices will rise, which will discourage use of cars (effectively internalizing some of the externalities). Also, new vehicles could be equipped with modern pollution-control devices (since the catalytic converters would no longer be poisoned by leaded petrol), which would at least reduce the growth in emissions. The USA phased out lead by using an emissions trading approach, but this would only be applicable to countries with a large and diverse refinery industry. Command and control approaches might be just as effective for countries with state-owned refineries. High proportion of buses’ and trucks’ maintenance. As buses and trucks are such an important transport mode in developing countries and, with some

88

Over 60 per cent of the hydrocarbon exhaust emissions during a five-mile trip are emitted before the car is warmed up. This percentage only falls to 50 per cent when the miles travelled double.

453 exceptions, these vehicles are old and poorly maintained, more attention needs to be paid to their emissions than now (where emissions are generally ignored). Inspection and maintenance programmes specifically directed to these types of vehicles are needed. Such monitoring and enforcement efforts are necessary components of any pollution-reduction policy. As they are so poorly maintained now, such programmes would be more cost-effective than in the USA. Cars and two-wheeled vehicles also need to come under these programmes. Because of the large share of commuting by bus and the chronic under-capacity of this system in many cities, scope for increasing use of buses is limited, unless capacity is increased. Privatization of bus service and expansion of private jitney services are approaches judged to have been successful (e.g. in Calcutta, Mexico City, and Bogotá (Rallis, 1988)). If in public hands, investment in new buses with better pollution-control equipment could be financed from proceeds of emission taxes. Greater use of buses will probably reduce emissions loads from transportation below what they otherwise would have been had cars been used instead.89 However, because the set of pollutants created by diesel engines is different from that created by petrol engines, increasing bus use may involve increasing (highly uncertain) mortality effects from carcinogenic emissions while reducing (much more certain) respiratory distress from car-related ambient ozone exposures. Analysis of the relative costs and benefits of reducing pollution from diesel versus petrol engines is needed to focus policy-making efforts better. In developing countries, the low share of cars in the transportation mix means that there are opportunities to develop transportation-control strategies that might be less acceptable once car commuting habits are formed. The low value of time (relative to capital) in developing countries also suggests that the focus of pollution-reduction policies should be on transportation controls rather than on new investment in vehicles or (low-polluting, but expensive) rail systems. Such controls promise reductions in congestion, an increase in capacity utilization of the vehicle stock, through switches to public transport or car pooling, and possibly a reduction in total trips, all of which will reduce pollution. These controls can be of the CAC variety, such as banning vehicles from city-centre areas, or economic incentives, such as congestion tolls, and increases in petrol taxes or parking fees. One qualified success story directed at cars is Singapore's Area Licence Scheme, which requires purchase of a sticker to enter the city during the

89

Emissions per person-mile are lower than those from the corresponding number of cars, assuming up to three persons per car. However, Rallis (1988) shows that, in terms of energy use, a small car with one passenger uses about 9 litres of fuel per 100 passenger kilometres, while a bus in the city uses 3 litres. With four passengers, energy use of the small car falls below that of buses. Here, car-pooling becomes a pollution-reducing strategy even with buses as the alternative transportation mode.

454 morning rush hour (current price = $2.60/day).90 Parking fees in the city and establishment of a park-and-ride service complemented the sticker plan. The result has been a 75 per cent decrease in vehicles entering the area during the morning rush hour, a 20 per cent increase in bus commuters, a doubling of car pools (Wilson, 1988), and substantial reductions in downtown air pollution (UN, 1989b). On the negative side, the evening rush hour did not improve much and a sizable minority reported an increase in travel time (because of the switch to buses, the switch of trip start times, and the increase in (unregulated) truck traffic). Wilson (1988) estimated that social welfare actually fell as a result of the Scheme because the fees were set too high, resulting in underutilization of the road network during rush hour, overutilization of buses, and increased travel time on net. Pollution effects were not considered in his calculation. An experiment in Hong Kong (The Economist, 18 February 1989) shows the potential for moving towards congestion tolls of a more optimal nature. By fitting cars with licence plates capable of being scanned by computers set up at key arteries within the city, vehicles were charged roughly by the amount of driving they did. The plan was ultimately rejected because of concern over what would happen to the collected revenues. However, advances in such technologies and their merger with technologies being tried out in California to measure CO emissions from vehicles waiting to enter Los Angeles freeways (Stedman, 1989) could make it possible to charge vehicles for their contribution to congestion and pollution.

16.3.6 Indoor air-pollution control policies Reducing indoor air pollution from cooking can be accomplished by inducing a switch to less-polluting fuels or moreefficient stoves. Reducing domestic heating emissions can be accomplished by fuel-switching, improved fuel (coal) quality, expansion of centralized heating facilities, and improving heating efficiencies (more insulation, better heattransfer mechanisms, more flexible control of temperature). Urban (as opposed to rural) indoor air-quality problems are primarily a problem of poverty. Use of low-quality cooking fuels is closely correlated with income. This implies that a long-run strategy for improving indoor air quality is to promote economic growth and an equitable distribution of income. However, as shown by Macauley et al. (1988), the income elasticity of demand for poorer-quality fuel is not strongly negative, implying that significant increases in income are required to induce fuel-switching. For example, a doubling of average urban household income in India is predicted to result in only a 10 per cent decline in fuelwood use. Programmes to switch the urban population to various types of cooking gas are ongoing in some countries, such as China. In China, only about 13 per

90

This can be thought of as a congestion toll.

455 cent of urban households use gas, with the rest using coal briquettes (which can be smokeless and have low SO2 emissions) and raw coal (Krupnick, 1989). However, a pricing paradox is evident here. To encourage more use of gas, its price is held down by the Government. But low prices impede development of gas reserves, gasification equipment, and distribution systems. If, as several surveys suggest, such fuel is the fuel of choice, then consumers may be willing to pay for better-quality fuels. Whether or not fuel switching is a viable strategy, more efficient, less polluting stoves are needed—an almost universal policy prescription (see Macauley et al., 1988; Dunkerley et al., 1981; and Smith, 1987). Stove designs and pilot programmes have led to reductions in energy and pollution, but both technical and implementation challenges remain (Smith, 1987). Still, given the high doses of pollution from inefficient stoves, and the prospect of dramatically reducing them for $5 to $20 per stove, stove design and dissemination should continue to be a high priority. Improved stove designs and dissemination could also address some space-heating requirements (Smith, 1987), while reducing pollution inside homes without central heating. In homes with central heating, the problem is not indoor air pollution, but high ambient levels of pollution in residential areas. Price adjustments on fuel used in central boilers could lead to improved efficiencies, and, in the longer term, availability of better-quality fuel. Reducing the remaining emissions, which could still be substantial, is a particular challenge for policy design. In China, whose northern latitude makes domestic heating emissions a major problem in winter, such decisions are made at the state and municipal level. The emphasis at the World Bank has been on choosing cost-effective strategies for reducing such emissions, considering such options as increasing central heating, district heating with co-generation, increased insulation of buildings, etc.

16.3.7 Towards a social-welfare-improving mix of pollution reductions Policies for improving social welfare (maximizing net benefits) through urban air-pollution reductions need to consider and then balance reductions in emissions obtained from indoor, outdoor, and mobile sources. The net benefits of controlling emissions from any of these sources may differ because of differences in the mix and quantity of emissions types, their location and effect on ambient quality, their proximity to people and therefore, exposures, the subsequent health effects, the values people place on avoiding such effects, and the costs of reducing emissions. Ultimately, maximizing net benefits may mean putting more effort into reducing indoor air pollution from burning poor-quality fuels in inefficient stoves than on reducing air pollution from power-plants, for instance. The issue of striking just such a balance between control of indoor and outdoor pollution sources is taken up by Smith and Ramakrishna (1986) for

456 India. They find that pollution exposures are roughly 500 times greater per kilogram of coal burned for domestic cooking relative to that from power-plants. Costs of control were assumed to be about $20 for a smokeless stove and standard costs for an electrostatic precipitator (ESP) on a power-plant were used. They estimated that to reduce exposure of the cook by 75 per cent it would be nine times cheaper through the stove controls.91 They further suggested use of a bubble approach, where power-plants would pay for new stoves rather than install ESPs in order to keep exposures constant after the installation of a new coal-fired power-plant with only ‘first-generation’ air-pollution controls (85 per cent removal). 16.4 CONCLUSIONS Rapid growth of urban areas of developing countries has led to major violations of ambient air-quality standards and is probably responsible for significant degradation of health. The future, with leaps in urban population and incomes, is even less promising for the urban environment without policies in place to address the major current problems of industrial, diesel vehicle, and cooking (plus heating) emissions, and the emerging problems associated with emissions from petrol vehicles. Economic incentive policies, particularly modest emission fees, offer advantages to developing countries in controlling industrial sources. Transportation controls, whether mandated or based on congestion fees, are attractive means for controlling future mobile-source emissions, but infusion of funds (perhaps through fare increases and/or privatization or through foreign donors) to maintain, upgrade, inspect, and convert urban diesel-bus fleets is needed.92 Finally, improved cooking-stoves for the urban poor, possibly paid for by industry as part of an urban emissions or ‘exposure’ bubble policy, should be an important component of an integrated plan for improving the urban air and the health of the urban population in developing countries.

REFERENCES BARNES, DOUGLAS F. (1990), Population Growth, Wood Fuels, and Resource Problems in Sub-Saharan Africa (Washington, DC: World Bank). BARNUM, HOWARD, and ROBERT GREENBERG (1989), ‘Cancer in the Developing World (Draft)’, Ch. 19 in Dean T. Jamison and W. Henry Mosley (eds.), Evolving Health Priorities in Developing Countries (Washington, DC: World Bank).

91 92

The damages from other pollutants produced from power-plants (SO2 and acid rain, for instance) and visibility were not considered. The World Bank recently funded the retrofitting of urban buses in Mexico to natural gas (personal communication with Gunnar Eskeland, World Bank).

457 BAUMOL, WILLIAM J., and WALLACE E. OATES (1975), The Theory of Environmental Policy: Externalities, Public Outlays, and the Quality of Life (Englewood Cliffs, NJ: Prentice-Hall). BOHM, PETER, and CLIFFORD S. RUSSELL (1985), ‘Comparative Analysis of Alternative Policy Instruments’, in Allen V. Kneese and James L. Sweeney (eds.), Handbook of Natural Resource and Energy Economics, i (Amsterdam: NorthHolland). BROWN, GARDNER, and RALPH JOHNSON (1983), The Effluent Charge System in the Federal Republic of Germany, Report to US Environmental Protection Agency and the German Marshall Fund. BUCHANAN, JAMES M. (1969), ‘External Diseconomies, Corrective Taxes, and Market Structure’, American Economic Review, 59/1. BULATAO, RODOLFO A., and PATIENCE W. STEPHENS (1989), ‘Estimates and Projections of Mortality by Cause: A Global Overview, 1970–2015 (Draft)’, Ch. 2.1 and Annex A in Dean T. Jamison and W. Henry Mosley (eds.), Evolving Health Sector Priorities in Developing Countries (Washington, DC: World Bank). BUMGARNER, J. RICHARD, and FRANK E. SPEIZER (1989), ‘Chronic Obstructive Pulmonary Disease (Draft)’, Ch. 23 in Dean T. Jamison, and W. Henry Mosley (eds.), Evolving Health Sector Priorities in Developing Countries (Washington, DC: World Bank). CHINA MINISTRYOF PUBLIC HEALTH, PRC (1988), Health Statistics in China. CONRAD, K., and C. J. MORRISON (1985), The Impact of Pollution Abatement Investment on Productivity Change: An Empirical Comparison of the US, Germany, and Canada, Working Paper 1763 (Cambridge, Mass: NBER). CROPPER, MAUREEN L., and WALLACE E. OATES (1990), ‘Environmental Economics: A Survey’, Discussion Paper QE90–12, Quality of the Environment Division (Washington, DC: Resources for the Future); also, Journal of Economic Literature (forthcoming). D’ARGE, RALPH C, and CLIVE L. SPASH (1990), ‘Economic Strategies for Mitigating the Impacts of Climate Change on Future Generations’, Presented at 65th Annual WEA International Conference, San Diego, California, 29 June–3 July. DUMARS, CHARLES T., and SALVADOR BELTRAN DEL RIO M. (1988), ‘A Survey of the Air and Water Quality Laws of Mexico’, Natural Resources Journal, 28.(autumn). DUNKERLEY, JOY, WILLIAM RAMSAY, LINCOLN GORDON, and ELIZABETH CECELSKI (1981), Energy Strategies for Developing Nations (Baltimore: Johns Hopkins University Press for Resources for the Future). Economist, The (1989), ‘Traffic Jams: The City, the Commuter and the Car’, 18 February. FAIZ, ASIF, KUMARES SINHA, MICHAEL WALSH, and AMIY VARMA (1990), ‘Automotive Air Pollution: Issues and Options for Developing Countries’, Infrastructure and Urban Development Department, Policy, Planning and Research Staff (Washington, DC: World Bank). GRAHAM, JOHN D. (1990), ‘Air Toxics and Public Policy’, Seminar given at Resources for the Future, Washington, DC, May. HAHN, ROBERT W. (forthcoming), ‘The Political Economy of Environmental Regulation: Towards a Unifying framework’, Public Choice. KOSMO, MARK (1989), Economic Incentives and Industrial Pollution in Developing Countries, Policy and Research Division Working Paper 1989–2, Environment Dept. (Washington, DC: World Bank).

458 KRUPNICK, ALAN J. (1989), ‘Air Pollution in Beijing, China: A Case Study’, Working Paper, Environment Dept. (Washington, DC: World Bank), forthcoming. KRUPNICK, ALAN J. and WINSTON HARRINGTON (1990), ‘Infrastructure and The Environment: Problems and Policies’, Report to Infrastructure and Urban Development Dept. (Washington, DC: World Bank). KRUPNICK, ALAN J. WALLACE E. OATES and ERICVAN DE VERG (1983), ‘On Marketable Air-Pollution Permits: The Case for a System of Pollution Offsets’, Journal of Environmental Economics and Management, 10. KRUPNICK, ALAN J. DOUGLAS R. BOHI, and DALLAS BURTRAW (1990), ‘Emissions-Trading in the Electric Utility Industry’, Resources, summer. LIPPMANN, M. (1989), ‘Background on Health Effects of Acid Aerosols’, Environmental Health Perspectives, 79, (Washington, DC: US Dept. of Health and Human Services, Public Health Service, National Institutes of Health). LOPEZ, ALAN D. (1989), ‘Causes of Death: An Assessment of Global and Regional Patterns of Mortality Around 1985 (Draft)’, Ch. 2.2 in Dean T. Jamison and W Henry Mosley (eds.), Evolving Health Sector Priorities in Developing Countries (Washington, DC: World Bank). MACAULEY, M., M. NAIMUDDIN, P. C. AGARWAL, and J. DUNKERLEY (1988), ‘Fuelwood Use in Urban Areas: A Case Study of Raipur, India’, The Energy Journal, 10/3. NATIONAL ACID PRECIPITATION ASSESSMENT PROGRAM (1990), NAPAP 1989 Annual Report (Washington, DC: US Government Printing Office). OATES, WALLACE E., and DIANA L. STRASSMANN (1984), ‘Effluent Fees and Market Structure’, Journal of Public Economics, 24/1. OPSCHOOR, J. B., and H. B. VOS (1989), The Application of Economic Instruments for Environmental Protection (Paris: OECD). ORGANIZATIONFOR ECONOMIC COOPERATIONAND DEVELOPMENT (OECD) (1989), Economic Instruments for Environmental Protection (Paris: OECD). PEOPLEyS REPUBLICOF CHINA (Various issues), China Statistical Yearbook. Qu GEPING (1989), China's Strategic Option in Environmental Policy for the I990's, translated manuscript. QUARLES, JOHN, and WILLIAM H. LEWIS, JR. (1990), The New Clean Air Act: A Guide to the Clean Air Program as Amended in 1990 (Washington, DC: Morgan, Lewis, & Bockius). RALLIS, TOM (1988), City Transport in Developed and Developing Countries (Basingstoke: Macmillan). SINHA, KUMARES, AMIY VARMA, JAMES SOUBA, and ASIF FAIZ (1989), Environmental and Ecological Considerations in Land Transport: A Resource Guide, Report INU41, Policy Planning and Research Staff, Infrastructure and Urban Development Department (Washington, DC: World Bank). SMITH, KIRK R. (1987), Biofuels, Air Pollution, and Health: A Global View (New York: Plenum). SMITH, KIRK R. (1988) ‘Air Pollution: Assessing Total Exposure in Developing Countries’, Environment, 30/10. SMITH, KIRK R. and J. RAMAKRISHNA (1986), Traditional Fuels and Health: Social, Economic and Technical Links, ERG Monograph 98, International Development Research Centre and United Nations University, Ottawa (New Delhi: Wiley Eastern Ltd.). STANSFIELD, SALLY K., and DONALD SHEPARD (1990), ‘Acute Respiratory Infections (Draft)’, Ch. 3 in Dean T. Jamison and W. Henry Mosley (eds.), Evolving Health Sector Priorities in Developing Countries (Washington, DC: World Bank).

459 STEDMAN, DONALD H. (1989), ‘Automobile Carbon Monoxide Emissions’, Environmental Science and Technology, 23/2. TERKLA, DAVID (1984), ‘The Efficiency Value of Effluent Tax Revenues’, Journal of Environmental Economics and Management, 11/2. TIETENBERG, T. (1985), Emissions Trading: An Exercise in Reforming Pollution Policy (Washington, DC: Resources for the Future). TOBEY, JAMES A. (1990), ‘The Effects of Domestic Environmental Policies on Patterns of World Trade: An Empirical Test’, Kyklos, 43/2. UNITED NATIONS (1989a) Environmental Data Report, prepared for UN Environment Programme by the GEMS Monitoring and Assessment Research Centre, London, in co-operation with the World Resources Institute, Washington, DC, and UK Dept. of the Environment, London. UNITED NATIONS (1989b), Urban Transport Development with Particular Reference to Developing Countries (NewYork: UN Dept. of International Economic and Social Affairs). US ENVIRONMENTAL PROTECTION AGENCY (1985), Costs and Benefits of Reducing Lead in Gasoline Final Regulatory Impact Analysis, EPA-230–05–85006, Office of Policy Analysis (Washington, DC: US Environmental Protection Agency). US ENVIRONMENTAL PROTECTION AGENCY (1989), National Air Pollutant Emission Estimates 1940–1987, EPA-450/ 4–88–022, Office of Air Quality Planning and Standards, Technical Support Division, National Air Data Branch (Research Triangle Park, NC: US Environmental Protection Agency). US ENVIRONMENTAL PROTECTION AGENCY (1990), Health Assessment Document for Diesel Emissions. Workshop Review Draft. Sections 1–10, EPA/600/8–90/057A, Office of Health and Environmental Assessment (Washington, DC: US Environmental Protection Agency). WANG, YANXIANG (1988), Coal Prices in Beijing, manuscript. WEITZMAN, MARTIN L. (1974), ‘Prices vs. Quantities’, Revue of Economic Studies, 41/4. WILSON, PAUL W. (1988), ‘Welfare Effects of Congestion Pricing in Singapore’, Transportation, 15. WORLD BANK (forthcoming), China Environmental Strategy Paper, report from the Asia Department. WORLD HEALTH ORGANIZATION and UNITED NATIONS ENVIRONMENT PROGRAMME (UNEP) (1988), Assessment of Urban Air Quality, Global Environment Monitoring System, prepared in co-operation with the Monitoring and Assessment Research Centre, London.

This page intentionally left blank

PART VI Macroeconomic Policies and Environmental Resource-Use

This page intentionally left blank

17 Macroeconomic Policies and Deforestation 17.1 THE FOREST SECTOR, THE BROADER ECONOMY, AND MACROECONOMIC POLICY In developing regions of Asia, Africa, and Latin America, forests and woodlands cover 20–25 per cent of the total land area. Thus, forces affecting the demand for agricultural, industrial, or residential land will probably affect forested area. Since agricultural land, including crop land and pasture, accounts for an additional 20 per cent of total land area, the margin between agricultural and forestry land uses is particularly important. By far the greatest share of deforested land has gone into agricultural production. In broad terms, changes in demand for agricultural land depend on the growth of demand for agricultural output, and on shifts in the mix of inputs used in agricultural production. In turn, changes in domestic agricultural demand depend on growth in population, in per capita income, and on the distribution of income. All three, of course, are strongly interrelated and depend on macroeconomic policies. Countries in which population and per capita income are growing rapidly with reasonably even distribution are undoubtedly experiencing rapidly rising domestic agricultural demands. With improving per capita income, demand for cereals increases more slowly than for animal products and other foods, with further implications for agricultural land use. On the supply side, the proportion with which land, labour, and capital are used in agricultural production depends basically on their relative scarcities. Sparsely populated, low-income countries have typically adopted extensive agricultural production systems, including free grazing of livestock, shifting cultivation of food crops, and production of plantation crops (Binswanger and Pingali, 1984). Since most remaining forests are in sparsely populated regions, these closely competing agricultural production systems account for most forest conversion. As long as adequate reserves of grasslands and forests can be converted to agricultural production at roughly constant costs as demands and rural labour supplies increase, little change in production systems or factor proportions occurs (Boserup, 1981). Agricultural expansion is at the extensive

464 margin, as was the case throughout the nineteenth century, up to the Second World War in most developing countries, and is still the case in sparsely populated regions. The supply of land at the frontier of cultivation usually involves privatizing forests and grasslands held under communal tenures or nominally controlled by national Governments. Because national Governments rarely enforce the traditional land rights of forest-dwelling communities and typically encourage or acquiesce in land acquisition by outside settlers and entrepreneurs, land at the frontier is largely an open-access resource. Neither traditional users nor government agencies can effectively exclude immigrants. Because of this market failure, the opportunity costs of land conversion are inadequately reflected in forested land's supply price. Land rights and tenurial institutions strongly influence the rate of deforestation. Even with rising costs of land conversion, changes in factor scarcities in agriculture are greatly influenced by macroeconomic conditions. Land scarcities at the extensive margin are affected by the overall distribution of agricultural landholdings. In many countries, highly concentrated landownership leads to less intensive use of the most productive agricultural land, and a greater demand for land on the agricultural frontier. Favourable tax treatment of agricultural land or income, and agricultural credit and input subsidies, can lead to the consolidation of landholdings. Rural labour supply depends not only on underlying population growth rates but also on the rate of growth of urban labour demand, which depends on the rate and pattern of industrialization (Ranis and Fei, 1961). Movements of agricultural labour to the frontier are affected by changes in urban labour absorption, and by capital–labour substitutions within the agricultural sector. Capital availability in agriculture depends not only on the distribution of income and terms of trade between agriculture and industry, but also on capital and product market distortions that artificially alter relative rates of return to investments in the two sectors (Mellor, 1976). Also, complementary public investments in rural infrastructure and agricultural research influence the level of private agricultural investment. The effects of macroeconomic policy on the agricultural land-use frontier are thus strong and varied. But, on the forested side of the frontier, policy influences are also at work. Domestic demand for forest products increases with population and income growth. Wood for fuel, pulp, construction, and industrial uses is the principal forest product, although a very wide variety of non-wood products are also harvested. The income elasticity of demand for pulp and industrial wood is high in rapidly urbanizing developing countries, but fuelwood, including charcoal, still accounts for the greatest volume of wood extracted from open forests in semi-arid areas and from forest fringes. Fuelwood demand depends also on the relative cost of substitute fuels, such as kerosene, electricity, and other biomass residues.

465 The market failure arising from insecure tenurial rights dominates the supply of most forest products in developing countries. Where current demands for wood exceed annual forest growth, they are met by cutting into the stock at rates that implicitly discount future values heavily. Insecure tenures discourage conservation of forest stocks for future use, whether the insecurity arises from the breakdown of communal resource-management traditions (Jodha, 1990), lack of enforcement of government restrictions on the use of public forests, or irregularities in the administration of timber-harvesting rights (Repetto and Gillis, 1988). Even in countries where natural forests are severely depleted, private investment in reforestation is depressed, because market prices rarely reflect the user costs of wood harvested from open-access natural forests. Industrial timber is typically harvested either under long-term concessions from the national Government (in Asia and Africa) or from privately owned forest lands (in Central and South America). Since timber is a long-lived asset, industrial wood supply is greatly influenced not only by insecurities of tenure but also by capital-market conditions. Most industrial timber is still harvested in developing countries from old-growth natural forests, although in an increasing number of countries these stocks are rapidly being exhausted. The timber owner's opportunity cost of capital influences his decision whether to delay the initial harvest, to leave adequate stocks for regeneration, to manage the forest for a subsequent rotation, or to allow sufficient time for regeneration before re-entering a stand (Chang, 1981). In few developing countries are capital markets open to, or reasonably integrated with, world capital markets. Fluctuating real exchange rates, trade and investment restrictions, legal and institutional barriers restrict international capital flows. Moreover, macroeconomic and sectoral policies strongly influence the relative rates of return to investments in various sectors, usually to inflate industrial profits at the expense of the primary sector. Consequently, large-scale holders of timber rights with profitable investment opportunities in industry or elsewhere may have high opportunity costs of capital, and hence strong incentives to withdraw from mining the forest for short-term profits. This incentive is reinforced by other timber supply conditions. First, forest proprietors cannot capture the ecological benefits old-growth forests produce in the form of wildlife habitat, hydrological cycling, and climate regulation. This market failure biases timber-management decisions toward shorter rotations, smaller residual stands, and lower levels of investment in forest management (Hartman, 1976). Second, in countries where logging by private firms on public lands is carried out under concession agreements and licences, failure by government to capture the resource rent from old-growth timber makes the forest a common-pool resource. Since royalties and taxes collect only a fraction of the stumpage value, logging firms engage in rent-seeking behaviour, accumulating

466 large concession areas, lest they be awarded to others, and then selectively harvesting them for the most valuable logs. Restrictions against stockpiling concessions and short concessions periods reinforce loggers' incentives to maximize short-term rents. In countries where old-growth forests are transferred from the public domain to private ownership, laws that award land titles on evidence that forest land has been improved, usually by clearance and conversion to other uses, have a similar effect on forest-management decisions. This discussion of the domestic demand and supply conditions for agricultural and forest land, although suggestive of the many macroeconomic policy linkages, still neglects the role of international trade. Countries at early stages of development have exported mostly primary commodities, with little processing. Logs have been (and still are) important exports from many forest-rich countries. With development, the composition of exports changes to include more value added in the form of processed commodities and labour-intensive manufactures. Over time, the skill and capital content of exports from successful developing countries increases. The experience of such successful, outwardlooking economies as Korea and Thailand follow this pattern. Outward-looking trade policies can create strong export demands for resource-based commodities, including forest products and competing land-intensive plantation crops and animal products. Unless export or other taxes capture the rents from these infra-marginal exports, supply can expand rapidly, leading to rapid depletion of forest stocks and conversion of forest land to agricultural uses. Protectionist policies designed to promote local timber-processing and forest-based industrialization can build up a large, technically inefficient milling industry with large raw-material demands. Market failures surrounding forest conversion, combined with strong external demand, promote rapid deforestation (Repetto and Holmes, 1983). In time, in outward-looking economies, shifting comparative advantage toward manufactures and rising supply costs of land-intensive exports tend to reduce the pressure of external demand on remaining forest stocks. In the short run, inward-looking trade regimes that heavily protect domestic import-substituting industries and discourage exports may reduce external pressures on forests. Overvalued exchange rates and negative effective rates of protection reduce the rents from primary product exports. Over the longer run, however, inward-looking trade regimes may actually result in greater pressures on forest resources. Adverse incentives prevent the development of manufactured exports, so that countries following such policies remain heavily dependent for foreign-exchange earnings on resource-based commodities with strong comparative advantage and then to exploit them heavily. Heavy industrial protection, by inflating industrial profit rates in the short run, raises the opportunity cost of capital tied up in forest stocks. Moreover, while the slower aggregate growth rates associated with inward-looking

467 policies limit demands for agricultural and forest products, the lower growth rates of the industrial sector also limit the transfer of labour from agriculture into urban occupations. Rural underemployment accelerates migration to agricultural frontiers. Macroeconomic policies have pervasive influences on the use and conservation of forest resources. A set of concentric circles centred on the forest is one way of visualizing these influences. At the hub are policies directly affecting timber and forest management, such as forest revenue structures, tenurial institutions governing privatization of forest land and enforcement of traditional use rights, and administration of timber-harvesting concessions. In the next circle are policies directly influencing the demand for forest products, such as trade and investment incentives to promote wood-using industries, and energy-pricing policies toward fuelwood substitutes. In a third circle might fall agricultural policies directly affecting movements of the agricultural frontier and the rate of conversion of forested land. These include credit, tax, and pricing incentives for land-intensive plantations and ranches. They include all policies that lead to supply expansion at the extensive rather than the intensive margin, such as those that increase the concentration of landholdings, or lower the marginal returns to labour and capital in agriculture, or direct public infrastructure spending toward frontier expansion. In a fourth broad outer circle are macroeconomic policies that at first seem unrelated to forest management but actually have powerful impacts. In this circle are policies that retard the demographic transition by perpetuating poverty and discrimination against women. Also here are trade and investment policies that retard the flow of rural to urban migration by preventing the rapid growth of labour-intensive industries and ancillary employment opportunities. No less important are policies that shorten investors' time-horizons by distorting capital markets or by encouraging rent-seeking behaviour. These policy influences can now only be discussed qualitatively and piecemeal. There are no macroeconomic models or analytical frameworks that effectively link forest exploitation and forest land conversion to this array of sectoral and macroeconomic policies. A start has been made in modifying conventional macroeconomic accounting frameworks to incorporate changes in natural-resource stocks in an economically satisfactory way (Repetto et al., 1989). However, natural-resource accounts compatible with the national-income accounting systems are being empirically estimated in only a very few countries in the developing world. Obviously, an appropriate set of accounts would provide a useful foundation on which analytical frameworks linking macroeconomic policy and deforestation could be constructed. There is little empirical research that tests the linkages sketched out in the preceding pages. It would be overly ambitious at this stage to attempt an overall assessment of the whole range of macroeconomic policy influences simultaneously. Rather, well-focused research studies illuminating subsets of these

468 policy linkages would be a valuable building block. The preceding pages, in effect, sketch out an extensive research agenda. Now that national and international development agencies are coming to grips with the problems of deforestation and trying to formulate policy changes to control them, the results of such studies would be timely and useful. The research must, of course, refer to individual countries, not only because national policy-makers are interested in information about their own options but also because despite broad similarities policy linkages differ from country to country. For this reason an international network or collaborative research would be helpful in stimulating and implementing research on this agenda. What follows in Section 17.2 is an empirical counterpart to the foregoing conceptual sketch of policy linkages. It draws on a wide variety of sources from many developing countries to illustrate policy linkages and their effects. The material is suggestive and heuristic, and certainly neither conclusive nor definitive. In that spirit, a good introduction to the next section is provided by Brazil's experience between 1987 and 1989, after the Government suspended tax incentives for competing land uses, adopted laws that did not condition land-titling on ‘improving’ of forested land, and began to enforce penalties for illegal forest-burning. The estimated rate of deforestation in the Brazilian Amazon region fell by more than two million hectares.

17.2 POLICY INFLUENCES ON TROPICAL FORESTS Tropical forests are now being destroyed much more extensively than they were a decade ago, unless rates of deforestation for that period were greatly underestimated. Compared to the most careful estimates of forest disturbance and clearance available in the early 1980s, more recent estimates for a number of countries based on satellite imaging and ground surveys show significantly higher rates of deforestation. In India, for example, studies by the National Remote Sensing Agency resulted in an increase in the estimated deforestation rate during the early 1980s to 1.3 million hectares per year, nine times the earlier Food and Agricultural Organization (FAO) estimate of 0.147 million hectares. It was found that large areas legally designated as forest land were virtually treeless. In Brazil, the FAO estimate of 2.5 million hectares annually for the early 1980s, which was based on partial Landsat surveys and other sources, increased to 8 million hectares per year for 1985–8, based on more recent interpretations of satellite data (Malingreau and Tucker, 1988). In 1987, 20 million hectares in the Amazon (much of it grassland and pasture), were in flames. The 1989 estimate for deforestation in Brazil was 4.6 million hectares. Deforestation at this rate poses extreme risks to natural systems. The consequent release of carbon to the atmosphere is estimated to contribute 15 to 30 per cent of annual global carbon emissions, a substantial contribution to

469 the build-up of greenhouse gases (Woodwell et al., 1983; Detwiler and Hall, 1988). Moreover, loss of tropical forests is rapidly eliminating the habitat of large numbers of plant and animal species. It is estimated that half of the world's plant and animal species inhabit tropical forests, and in ten biotically rich and severely threatened regions totalling just 3.5 per cent of the remaining tropical forest area, 7 per cent of all plant species will probably go extinct by the end of the century, if current trends continue (Myers, 1988). The dismay of scientists and environmentalists over this destruction is now shared by leaders of the world's Governments. The Paris summit meeting of heads of Governments in the Group of Seven countries in July 1989 declared that ‘Preserving the tropical forests is an urgent need for the world as a whole …We express our readiness to assist the efforts of nations with tropical forests through financial and technical co-operation, and in international organizations.’ Deforestation has reached the agenda of summit meetings on the global economy. A sense of crisis is also emerging in the tropics, as once-ample forest resources disappear. An asset capable of yielding rich returns indefinitely has been badly depleted. In Thailand, commercial logging was recently banned over the protests of influential concession-holders when surveys showed that forest cover had declined from 29 to 19 per cent of the land area between 1985 and 1988, and landslides from deforested hillsides cost 40,000 people their homes. In the Philippines, undistributed dipterocarp forests have shrunk from 16 million hectares in 1960 to less than a million hectares still standing in remote hill regions. Logging has been suspended in most provinces, and mills in the Philippines, are closing or importing logs from Sabah and Sarawak. Mills in the once rich Indonesian production centres of Sumatra and Kalimantan are also experiencing shortages of accessible high-quality timber and importing logs from Sabah, Sarawak, and Irian Jaya. Indonesia's ambitious timber-development plans are now faced with potential future resource shortages. Sabah and Sarawak, currently the major sources of logs in Asia, are harvesting almost twice the sustained yield of their forests, and will also be rapidly depleted (Burgess, 1988). In West Africa, Central America, and China as well, the loss of most mature forests has depressed incomes, foreign-exchange earnings, and employment from forest-based industries. Although the issues are complicated by global environmental concerns voiced in developed countries, most tropical country Governments are coming to the realization that to them rapid deforestation represents a severe economic loss, a waste of valuable resources. In the Ivory Coast, for example, where forest cover has decreased by 75 per cent since 1960, an estimated 200 million cubic metres of commercial timber has simply been burned to clear the land, a loss of perhaps $5 billion. The Forest Department estimates that in Ghana, where 80 per cent of the forests have disappeared, only 15 per cent of the forests were harvested before the land was cleared. In Brazil, where little of the timber is extracted before the forest is burned, the resulting loss in

470 commercial timber is approximately $2.5 billion annually (Schmidt, 1989). That is a quarter of Brazil's annual net debtservicing payment on its external debt. Burning valuable timber while clearing forests is only one obvious kind of economic wastage. Other losses spring from the extremely short time-horizons with which tropical forests are exploited. Loggers in tropical forests destroy enormous quantities of valuable timber through careless use of equipment, failure to cut away vines before felling, and other practices. If loggers extract 10 per cent of the timber, selecting mature trees of the most valuable species, they typically destroy at least half the remaining stock, including immature trees of the same species as well as harvestable stock of less well-known varieties. Harvesting costs are thus reduced, but the residual stand and subsequent harvests are impoverished. Loggers often re-enter logged-over areas to extract more timber before stands have recovered, inflicting heavy damage on residual trees each time and making regeneration impossible. In Ghana and the Ivory Coast, stands have been reentered as often as three times in 10–15 years, when concessionaires obtained sales contracts for logs of lesser-known species. Diptercarp forests in the Philippines have been exploited on a clearly unsustainable cutting-cycle of 6–8 years. This also reflects an extremely high time-discount rate by concessionaries with a high opportunity cost of capital or little confidence that they will enjoy the benefits of future harvests, or both. A recent study commissioned by the International Tropical Timber Organization (ITTO) found that not even 0.1 per cent of remaining tropical forests are being actively managed for sustained productivity. Destructive logging practices on short-cutting cycles result in severely depleted timber stands. Moreover, production forests in most countries are left virtually unprotected after the harvest from encroachment by shifting cultivators, and are thus exposed to burning and clearing. This is additional evidence of loggers' high discount rates. Surveys in the Amazon demonstrate that deforestation is rapid where roads for logging or other purposes have opened up a region, but minimal elsewhere. This biological degradation of tropical forests carries a high and increasing economic price tag. The timber cost alone has been unexpectedly large, since tropical timber prices have bucked the general downward trend of declining commodity prices (Figure 17.1) and many previously uncommercial species now find ready markets. In West Africa, for example, the price of previously neglected aningeria logs is now on a level with sapele and other prime species. Countries that have previously extracted as few as two or three trees per hectare, destroying the rest as uncommercial, now regret their shortsightedness. The upward trend in tropical timber prices is likely to continue, as supplies are depleted in Asia, Central America, and West Africa over the next decade. Consequently, the timber in the Amazon basin, which is now being recklessly burned, will become increasingly valuable. For old-growth timber,

471

Fig. 17.1 Price trends: timber vs. thirty-three commoditiesSource: World Resources Report, 1988–9, World Resources Institute, Washington, DC.

an asset with a low rate of biological appreciation, expected price increases are necessary if stocks are to be held for future use. But potential timber values are by no means the only economic losses that deforested countries suffer. Probably 70 per cent of wood harvested in tropical countries is used locally, and as forests recede, fuelwood shortages increase. Other forest products become unavailable to local residents, including bushmeat, fruits, oils, nuts, sweeteners, resins, tannins, fibres, construction materials, a wide range of medicinal compounds, and such saleable products as skins, feathers, and live animals. In Indonesia, the value of only those non-timber forest products that reached the export market reached $US123 million by 1986 (HIID, 1988). Many such non-timber products are exploited as open-access resources. None the less, recent studies have shown that the capitalized value of the income derived from such non-timber forest products, which can be extracted sustainably, may greatly exceed that of the timber harvest (Peters, Gentry, and Mendelsohn, 1989). Moreover, the incomes so derived are the subsistence of local residents, while the profits from timber exploitation are typically captured by distant élites or foreign corporations, so that timber operations have sparked violent protests by indigenous communities in Sarawak, the Philippines, and other countries. Were these nontimber values adequately reflected in timber-management decisions, initial harvests would be delayed, residual stands would be protected, and re-entry would be deferred to a greater extent than is now common. Deforestation often has severe environmental impacts on soils, water quality, and even local climate. Shallow, easily leached, soils are damaged by heavy equipment, and when exposed to heavy tropical rains can quickly lose remaining nutrients or erode. Studies in Ghana showed that eliminating savannah forest raised soil erosion rates from less than 1 to more than 100 tons per hectare, with a nutrient loss 40 per cent higher than the average annual

472 chemical-fertilizer application (World Bank, 1988). Plentiful riverine fisheries have been damaged by increased sedimentation, or deforestation in floodplains which provide critical seasonal habitat. Large-scale tropical deforestation interrupts moisture recycling, reducing rainfall and raising soil temperatures, perhaps leading to long-term ecological changes (Salati and Vose, 1983). These environmental values are also external to the loggers' decisions, but make forest conservations a more rational economic strategy. Moreover, deforestation has often accompanied shifts to economically and environmentally inferior land uses, such as cattle-ranching and inappropriate modes of agriculture. For example, given the quick loss of productivity and lowcarrying capacity of pastures in the rapidly deforesting Brazilian State of Acre, the net present value of revenues per hectare from collecting wild rubber and brazil nuts was found to be four times that of cattle-ranching. In Guatemala, studies have shown that sustained forest management for non-timber and timber production is economically superior to slash-and-burn agriculture. However, tax and credit incentives, as tenurial rules that require alteration of the natural forest in order to obtain or confirm private land titles, and land speculation in inflation-prone countries, promote such inferior land uses. Both experiences and analysis reinforces the argument that deforestation has not been a path to development, but in most tropical countries is a costly drain of increasingly valuable resources. Recognition is also growing that deforestation is not inevitable in developing countries, but is largely the consequence of poor stewardship, inappropriate policies, and inattention to problems outside the forest sector.

17.3 EXAMPLE OF FORESTRY POLICY AND FOREST-MANAGEMENT IMPACTS Above all, developing countries' Governments that are the proprietors of at least 80 per cent of the closed tropical forests have not put an adequate value on the resource. As proprietors, they could capture the entire resource value except for the cost of labour and capital employed in managing and harvesting, by charging sufficient royalties and taxes or selling harvesting rights to the highest bidders. Instead, with very few exceptions, Governments have allowed most of these resource rents to flow to timber concessionaires and speculators, often linked to foreign enterprises. For example, in the Philippines in 1987, if the Government had been able to collect the full resource value of the roughly three million cubic metres of timber harvested, its timber revenues would have exceeded $250 million—more than six times the $39 million actually collected (De Los Angeles, 1989). Low royalties and taxes, combined with widespread log-smuggling and tax evasion, left much of the excess profits in the hands of timber concession-holders, mill-owners, and

473 timber traders. The Asian Development Bank estimated that profits of at least $4,500 were available for each hectare harvested. Governments have created these windfalls by keeping royalties and fees charged to timber concession-holders low, reducing export taxes on processed timber to stimulate domestic industry, and even granting income tax holidays to logging companies. Moreover, Governments have failed even to enforce the official charges effectively. In Indonesia between 1979 and 1984, 125 million hectares were harvested but taxes and royalties were collected on only 86 million. Consequently, only few of those tropical countries for which data exist have succeeded in limiting timber exploiters to a normal rate of profit and capturing the value of the forest resource for the public treasury. (The same is true, incidentally, in many temperate countries, including the USA, Canada, and Australia.) In Indonesia between 1979 and 1982, for example, the Government captured only $US1.6 billion from a potential timber resource value of over $US4.9 billion, leaving more than $US800 million per year in potential excess profits at the disposal of logging concessionaires and mill-owners. This has had most unfortunate consequences. It has sparked timber booms throughout the tropics, drawing both domestic and foreign entrepreneurs—many with little forestry experience—into the search for quick fortunes. Under this pressure, Governments have awarded timber concessions covering areas far greater than they could effectively supervise or manage, sometimes overlapping protected areas and national parks. In the Ivory Coast, concessions were let for two-thirds of the nation's production forests in just seven years. Of 755 politically favoured concessionaires, only fifty-one actually work their holdings. The others merely sell their cutting rights, profiting as middlemen. In Indonesia, Thailand, and the Philippines, the areas under concession exceed the total area of production forest (Burgess, 1988). Moreover, it has attracted both politicians and businessmen to the opportunities for private gain. In Thailand, Sarawak, Sabah, Philippines, and other countries, cabinet ministers, senators, or other senior politicians are involved in the timber industry. In the Philippines, for example, the principal opposition leader holds extensive timber concessions acquired under the Marcos regime. In Indonesia, most of the 544 concession-holders are retired military or government officials, and readily bring pressure in Jakarta to halt investigations into violations of forestry regulations (Burgess, 1988). Under such conditions, effective supervision by relatively low-ranking forestry department personnel is virtually impossible. Even in countries with large public-sector deficits, Governments are failing to collect a potentially important revenue source, despite adverse economic and environmental consequences. While sacrificing enormous sums in potential forest revenues, Governments in the tropics are failing to invest enough in stewardship and management of the

474 resource. In Indonesia, nearly half of all trained foresters work in Jakarta, hundreds of miles by sea from the forests, while those in the field are dependent on concession-holders for shelter and transportation while inside the concessions. In Ghana, a recent World Bank study found that 66 per cent of government posts for professional foresters, 54 per cent of the posts for junior professionals, and 43 per cent of technical-grade positions were vacant (World Bank, 1988). In Gabon, forestry staff are numerous, but they have no means of doing their jobs in the field, since the department budget fell by 75 per cent between 1984 and 1988. As a result, while in many countries forestry codes and stipulations in concession agreements seem adequate to ensure sustained productivity over at least several cycles, given the evidence of experimental research (Hadley, 1988) natural forests are almost nowhere being managed to achieve that goal. Ineffective government supervision is compounded by the perverse incentives created for timber companies by the terms of concession agreements, which discourage any possible interest they might have in sustained yield management. Most agreements run for twenty years or less, and many for five years or less, although intervals of twenty-five to thirty-five years are prescribed between successive harvests in selective cutting systems and longer intervals in monocyclic systems. Concession-holders have little reason to care whether productivity is maintained for future harvests. While longer concession periods are undoubtedly not sufficient to lower loggers' time discount rates, they may be necessary. Forest revenue systems based on relatively undifferentiated fees levied on the volume of wood extracted, encourage loggers to extract the highest-value logs at minimum cost, and this leads to ‘high-grading’ the timber over large areas and to extensive damage of residual stands. Trees with stumpage value less than the royalty rate are worthless to the concession-holder, and can be destroyed with impunity. In Sabah, Indonesia, and the Philippines, 45–75 per cent of residual trees are destroyed or seriously damaged during harvesting operations (Repetto and Gillis, 1988). Royalties based on the size of the concession and the total merchantable timber it contains encourage more complete utilization of the timber resource within a smaller harvesting area. Ad valorem royalties also encourage fuller utilization. Distorted incentives also affect the efficiency of wood-processing industries. Log-producing countries have had to provide strong incentives to local mills to overcome high rates of protection against processed-wood imports into Japan and Europe. Extreme measures, such as log-export bans and log-export quotas based on volumes processed domestically, have created inefficient local industries, sometimes set up only to preserve valuable log-export rights. In the Ivory Coast, such quotas have created a large processing industry that requires 30 per cent more logs than efficient mills use to produce the same output, but is supported by rights (worth up to $US15 per cubic metre on the open market) to export high-valued logs. Similarly, in Zaire,

475 where concessionaires must process 70 per cent of their harvest, the restriction has increased timber-cutting because profitable export sales of prized species support inefficient saw-milling that dumps lower-valued output domestically at prices approximately 30 per cent below production costs. Such extreme protection can create powerful local industries able to resist limitations on the supply of logs, for ecological or other reasons. Indonesia has successfully captured 70–80 per cent of the world hardwood–plywood market by manning log exports and providing generous industrial incentives, but rapidly expanding industry now requires 35 million cubic metres of logs annually, more than previous peak exports, and current plans call for a doubling of capacity during the 1990s. By flooding the Japanese market, it has driven prices down by 20 per cent in 1988. As a price-setter, Indonesia could rationally consider an optimal export tax on processed tropical wood products. Were policy co-ordination between Indonesia, Malaysia, and Papua New Guinea feasible, a joint profit-maximizing export tax would exceed Indonesia's individual optimal tax level. Although trying to increase local value added and employment, countries sheltering inefficient processing industries can incur heavy economic and fiscal losses. In the Philippines, for example, log exported as plywood is worth $US100–110 per cubic metre less than it would be if exported without processing or as sawn timber, while the Government sacrifices more than $US20 million annually in export taxes to encourage plywood exports. In 1987, the Philippines dissipated more than two-thirds of the potential resource rents from the tropical timber harvest in inefficient processing induced by excessive protection. Industrial countries have contributed to, and profited from, these forest policy problems in the tropics. Japan is the largest importer of tropical timber, accounting for 29 per cent of world trade in 1986, roughly the same share as the European Economic Commission (see Figure 17.2). Imports, which unlike those of the EEC, are mostly of unprocessed logs, were 30 per cent higher in 1987, largely because of the construction boom. Most tropical hardwood imports are used in Japan for construction plywood, primarily as disposable forms for moulding concrete, for which tropical woods were adopted because of their cheapness. European and US companies have held interests in logging and processing enterprises, especially in tropical Africa and Latin America, but Japanese business now heavily outweighs its rivals in the tropical timber trade. The large Japanese trading companies are involved in all stages of exploitation, as partners and financiers of logging concessionaires, as exporters and importers, and as processors and distributors. Japanese firms shifted their attention from the Philippines to Indonesia to Sabah and Sarawak as log supplies were successively depleted, and they are now interested in Amazon forests as a potential future source of raw material. They have shown little interest in sustained management of their holdings. Instead, highly leveraged operators

476

Fig. 17.2 1987 shares of tropical wood importsSource: Timber from the South Seas, World Wildlife Fund, 1989.

harvested as much and as fast as possible to pay down their financing charges. Significant levels of effective protection of timber-processing industries in Japan are partly responsible for distorting timber exploitation and processing industries.

17.4 AGRICULTURAL POLICIES AND TROPICAL FOREST CONVERSION In many countries agricultural policies encourage conversion of tropical forests to other uses that, in some cases, would be uneconomic without such inducements. Tenurial rules in many States, such as Sabah, allow private parties to obtain title to forested land on evidence of ‘improving’ it, by clearing away the trees, for example. In the Philippines, Brazil, and elsewhere, recognized rights of occupancy or possession are awarded on the basis of the area of land cleared (Mahar, 1988). Such provisions often become a mechanism to privatize lands from the public forest estate. Those who obtain possessory right soon sell out to larger capitalists who consolidate the land into private ranches and speculative holdings. These alternative uses would in many cases be uneconomic without heavy government subsidies. In the Brazilian Amazon, road-building projects financed by the federal Government and multinational development banks have fuelled land speculation. In addition, over 600 cattle ranches, averaging more than 20,000 hectares each, were supported by subsidized long-term loans, tax credits covering most of the investment costs, tax holidays, and writeoffs. Subsequent analysis showed that the ranches themselves were uneconomic, typically losing more than half their invested capital over a fifteen-year horizon. Indeed, field sample surveys showed that output averaged only 9 per cent

477 of that projected, and that many were repeatedly reorganized and resold, yielding nothing but tax shelters. In that respect, they were unquestionably productive at the private level, generating returns up to 250 per cent of their owners' actual equity input (Repetto, 1988). The Brazilian Government has suspended incentives for new cattle ranches in Amazonian forests, but subsidized ranches covering 12 million hectares have already cost the Government over $US2.5 billion. More general agricultural policies also indirectly contribute to deforestation. In Latin America and the Philippines, the concentration of the better agricultural land into large, generally underutilized estates, pushes the growing rural population into forested frontiers and upper watersheds. This extreme concentration of landholdings is supported by very low agricultural taxes that make farms and ranches attractive investments for those in upper-income brackets, and make it virtually costless to keep extensive holdings that generate relatively little income (Binswanger, 1987). Subsidized rural credit programmes also promote land concentration, since interest-rate ceilings inevitably lead to credit rationing in favour of large landholders with ample collateral and secure titles. Especially in inflationary settings in which land provides security, large landholders with access to virtually free credit can easily buy out small farmers who cannot finance investments to raise agricultural productivity. Many of the recent migrants into Rodonia and Acre in the Brazilian Amazon are small farmers and farm-labourers displaced from Parana by large-scale mechanized cultivation. In many countries, deforestation has been in effect a temporary escape valve, a respite from development pressures that can be dealt with effectively only outside the forest sector. In the Philippines, population growth rates in the forested uplands are even higher than the national average of 2.5 per cent per year, resulting in high rates of deforestation and soil erosion. The Philippines Government has been reluctant to address directly the challenge of reducing population growth rates, or to attack highly skewed patterns of landholding in the lowlands. In Indonesia, the Government's ambitious ‘transmigration’ programme, which has so far resettled about a million families from crowded Java to the Outer Islands (80 per cent of them to sites cleared in primary or secondary forests), was largely an attempt to provide employment and livelihoods. However, at a cost of $US10,000 per household in a country that invests only $US125 per capita annually, transmigration could obviously not compensate for slow employment growth on Java itself, and was sharply curtailed in the budget cuts necessitated by lower petroleum prices. Since that time, extensive policy changes to deregulate industrial investment and to promote labour-intensive manufactures has given a sharp boost to employment-creation and non-resource-based exports. In this respect, rapid deforestation in the tropics during the 1980s is linked to the exceptionally difficult international economic conditions most tropical

478 countries have faced. Indonesia's drive to export tropical timber products was a conscious effort to offset its lower petroleum earnings and protect its development programme from further cutbacks. The group of most heavily debtburdened countries are coincidentally those with most of the remaining tropical forests. During the 1980s, for the first period in forty years, economic growth in those countries failed to outpace labour-force increases. Employment in the organized, urban sector stagnated or declined, and real wages in the informal urban labour-market plummeted. Instead of the usual pattern of rural–urban migration out of agriculture, labour piled up in agriculture. In Brazil, for example, the agriculture labour force grew by 4 per cent annually between 1981 and 1984, compared to a growth rate of 0.6 per cent during the period 1971–6. Agricultural wages fell almost 40 per cent in real terms between 1981 and 1985. With no alternative, given the concentration of agricultural land and absence of jobs, rural households migrated to the frontier in increasing numbers. Migration to Rondonia, for example, increased from 60,000 annually in 1980 to 167,000 six years later. More favourable macroeconomic conditions could reduce the pressure of unemployment, poverty, and population growth on the remaining tropical forests. Fortunately, there are many indications of a new policy approach toward tropical forests that reflect the increasing awareness of their national and global significance. Many developing-country Governments are taking steps to capture the resource rents that have motivated the despoliation of tropical forests. The Philippines Government has imposed partial logging bans, cracked down on illegal logging, and raised timber royalties. It plans to increase timber taxes further, and assign future harvesting rights on the basis of competitive bids. The Government of the Ivory Coast also plans to allocate harvesting rights by competitive bidding to capture resource rents more fully. The Indonesian Government has also raised timber taxes substantially, by $US25–35 on most species. The Government of Ghana, with World Bank assistance, has doubled average timber royalties to an average of 12 per cent of export value, with plans for a further 50 per cent increase by 1992. A number of countries are now strengthening their forest-management capabilities, with the help of development assistance agencies. The World Bank and Asian Development Bank now have loans for forest-management improvement in the pipeline for a dozen countries. Most of these loans support forest-policy reform as well as institutional strengthening. Through the Tropical Forestry Action Plan more than fifty tropical countries are now preparing national action plans to conserve and manage their forests. Although varied in coverage and sophistication, many of these plans address programmes for conservation, community forestry, industry forestry, research and training, and are supported by international consortia of bilateral and multinational development institutions willing to assist in their implementation.

479 International interest in the tropical forests has blossomed, accompanied for the first time by a willingness to contribute to their maintenance. Voluntary organizations in a number of developed countries have raised money for ‘debt-for-nature swaps’, to buy up discounted external debt of a tropical country and exchange it for a local-currency fund (usually to be managed by a local voluntary agency) that will finance forest-conservation programmes. Some business groups have also taken an active interest. The Tropical Timber Traders’ Associations of the Netherlands and the United Kingdom have proposed that all importing countries levy a surcharge on tropical timber imports to create a fund to be used for forest conservation. None the less, a great deal remains that the world outside the tropics might do. Consumption patterns such as the use of tropical hardwoods for disposable concrete moulds contribute to deforestation. Some businesses in industrial countries are still contributing to forest destruction. Barclays Bank was recently found to be majority owner through its Brazilian subsidiary of two huge Amazon cattle-ranches that have burned half a million acres of forest to create pasture. Barclays’ chairman, on learning of this involvement, declared, ‘Being personally an extremely keen gardener and botanist… I was extremely cross’ (Sunday Times, London, 25 June, 1989). Development assistance agencies are still financing projects destructive to tropical forests. The African Development Bank has recently agreed to a project that will run a road through one of the Ivory Coast's few remaining tracts of rainforest and mangrove habitat, and another project to develop saw-milling capacity that will affect over 800,000 hectares of virgin forest in the Congo, although there is no forest-management capability in the country and the area is the home of forest-inhabiting pygmy communities. Such projects should be supplanted by others designed to improve forest management and to expand the pace of reforestation efforts. The scope for international co-operation to halt the destruction of tropical forests is large. The Tropical Forest Action Plan provides one framework. The World Bank has recently proposed a Global Environmental Facility, which would allocate additional contributions from donor members to high-priority projects that address international environmental problems, including the protection of climate, atmosphere, oceans, forests, and biodiversity. These are mechanisms through which northern countries with interests in the preservation of the global environment can share in the costs of actions taken by developing countries. The international convention now under discussion on global climate change could also be a powerful mechanism for international co-operation. Fees levied on the use of CFCs, as adopted by the US Government, and taxes on fossil fuels and other greenhouse gases in the industrial countries would not only help reduce their emissions but also provide the funds needed to implement national programmes for climate protection in developing countries, including reforestation and forest conservation. Alternatively, international

480 trading of greenhouse-gas emissions credits within a comprehensive framework of monitoring sources and sinks might lead agencies in developed countries to finance forest conservation and reforestation in the tropics as the least expensive way to curb net greenhouse-gas emissions. New forms of international co-operation would reflect our growing awareness that disappearing tropical forests represent national treasures and essential elements of the biosphere on which we all depend.

REFERENCES BINSWANGER, H. (1987), ‘Fiscal and Legal Incentives with Environmental Effects on the Brazilian Amazon’, Discussion Paper, Environment Dept. (Washington, DC: World Bank). BINSWANGER, H., and PINGALI, P. (1984), ‘Population Density and Agricultural Intensification: A Study of the Evolution of Technologies in Tropical Agriculture’, Discussion Paper, Agriculture and Rural Development Dept. (Washington, DC: World Bank). BOSERUP, E. (1981), Population and Technological Change (Chicago: Chicago University Press). BURGESS, P. F. (1988), ‘Natural Forest Management for Sustainable Timber Production: The Asia Pacific Region’, (London: International Institute for Environment and Development), mimeo. CHANG, S. J. (1981), ‘Determination of the Optimal Growing Stock and Cutting Cycle for an Uneven-aged Stand’, Forest Sciences, 27. DE LOS ANGELES, M. S. (1989), ‘Economic Rental from the Sale of Logs in the Philippines’, (Washington, DC: World Bank), mimeo. DETWILER, R. P., and Hall, C. (1988), ‘Tropical Forests and the Global Carbon Cycle, Science, 239. FAO (1987), The Tropical Forestry Action Plan (Rome: Food and Agriculture Organization). HADLEY, M. (1988), ‘Rain Forest Regeneration and Management’, Biology International (Special Issue), 18. HARTMAN, R. (1976), ‘The Harvesting Decision when the Standing Forest has Value’, Economic Inquiry, 14. HIID (1988), The Case for Multiple Use Management of Tropical Hardwood Forests (Cambridge, Mass.: Harvard Institute for International Development). JODHA, N. S. (1990), ‘Sustainable Agriculture in Fragile Resource Zones: Technological Imperatives’, (Nepal: International Centre for Mountain Development), mimeo. MAHAR, D. (1988), Government Policies and Deforestation in Brazil's Amazon Region (Washington, DC: World Bank). MALINGREAU, J.-P., and C. J. TUCKER (1988), ‘Large-Scale Deforestation in the Southeastern Amazon Basin of Brazil’, Ambio, 17. MELLOR, J. (1976), The New Economics of Growth (Ithaca, NY: Cornell University Press).

481 MYERS, N. (1988), ‘Threatened Biotas: “Hotspots” in Tropical Forests’, The Environmentalist, 8. PETERS, C. M., A. GENTRY, and R. MENDELSOHN, (1989), ‘Valuation of an Amazonian Rainforest’, Nature, 339. RANIS, G., and J. C. H. FEI (1961), ‘The Theory of Economic Development’, American Economic Review, 52. REPETTO, R. (1988), The Forest for the Trees? Government Policies and the Misuse of Forest Resources (Washington, DC: World Resources Institute). REPETTO, R. et al. (1989), Wasting Assets: Natural Resources in the National Income Accounts (Washington, DC: World Resources Institute). REPETTO, R., and GILLIS, M. (eds.) (1988), Public Policies and the Misuse of Forest Resources (Washington, DC: World Resources Institute). REPETTO, R., and HOLMES, T. (1983), ‘The Role of Population in Resource Depletion’, Population and Development Review, 9. SALATI, E., and P. G. VOSE (1983), ‘Depletion of Tropical Rain Forests’, Ambio, 12. SCHMIDT, R. C. (1989), ‘Management of Tropical Moist Forests in Brazil’, (Rome: Food and Agriculture Organization), mimeo. WOODWELL, G. M., et al. (1983), ‘Global Deforestation: Contribution to Atmospheric Carbon Dioxide’, Science, 222. WORLD BANK (1988), ‘Forest Resource Management Project: Ghana’, Discussion Paper (Washington, DC: World Bank).

18 Microeconomic Responses to Macroeconomic Reforms: The Optimal Control of Soil Erosion 18.1 INTRODUCTION To both classical and neoclassical economists, land was inert; it did not depreciate with use, and it was not something in which an economy could invest or disinvest. (Ricardo, for example, referred to ‘the original and indestructible powers of the soil’.) To classical economists, land was special because of its fixed supply and importance in the economy. (Malthus wrote: ‘a fertile soil gives at once the greatest natural capability of wealth that a country can possibly possess.’) But to neoclassical economists, land ceased to be special because of substitution possibilities and technical progress, and because of the declining importance of agriculture in industrial economies; land was a limiting factor, but the limits which land imposed on growth were not worth worrying about. (Harrod (1948: 20), for example, omitted land from his theory of growth on the basis that ‘its influence may be quantitatively unimportant’.)93 Schultz (1951: 725) summed up this view thirty-nine years ago: Clearly, in particular countries, land is no longer the limitational factor it once was; for instance, in such technically advanced communities as the United Kingdom and the United States and also in many others, the economy has freed itself from the severe restrictions formerly imposed by land. This achievement is the result of new and better production possibilities and of the path of community choice in relation to these gains. This achievement has diminished greatly the economic dependency of people on land; it has reduced the income claims of this factor to an ever-smaller fraction of the national income; and it has given rise to profound changes in the existing forms of income-producing property. The underlying economic development has modified in an important way and relaxed substantially the earlier iron grip of the niggardliness of Nature.

Conservationists contend that land is as much a constraint on the growth potential of many low-income countries today as it was on the economies of

93

Land receives not even a mention in Solow's (1970) famous book on growth theory. An important exception in this literature is Meade's (1961) treatise on economic growth, in which land plays an important role as a fixed factor.

483 Western Europe at the time the classical economists were writing. But the nature of the constraint is believed to be very different; it is not so much the availability of land that seems to limit growth in these countries as the depreciation of the asset due to soil erosion. The neoclassical model seems to have held up in the middle- and high-income countries where land depreciation has been more than compensated for by substitution and technical progress.94 But in the lowincome countries, substitution and technical progress have not always been sufficient to overcome depreciation of the land asset. The ‘iron grip’ of nature has yet to loosen its hold on these countries, and conservationists fear that soil depletion will strengthen the grip. Brown (1984) states the case plainly: The depletion of oil reserves, and its effect on world oil prices, is the most immediate threat to world economic stability, but the depletion of soil resources by erosion might be the most serious long-term threat. The unprecedented doubling of world food supplies over the last generation was achieved in part by adopting agricultural practices that led to excessive soil erosion, erosion that is draining the land of its productivity.

The World Bank's World Development Report 1986 lays the blame for the poor performance of agriculture in lowincome countries on macroeconomic and sectoral policies like overvalued exchange rates and agricultural output taxes which alter the incentives facing farmers.95 The Bank does not deny the importance of soil depletion, but it does not recognize soil depletion as requiring special policies (World Bank, 1986: 79): Protection of the environment is a task that has recently attracted much attention, especially because of the erosion of arable land in SubSaharan Africa. Although it is not often realized, the pricing policies that developing countries follow can be important from this point of view also. When farming becomes unprofitable, farmers lose the incentive to care for their land.

An important question is whether these policies are related to the observed soil-depletion. Is soil conservation harmed by the macroeconomic and sectoral policies pursued by low-income countries? The literature offers two views about the relationship between output prices and soil conservation. One maintains that low output prices have encouraged soil depletion. Repetto (1987: 45), for example, has argued that as a consequence of keeping agricultural output prices artificially low in developing countries, returns on investment in farmland development and conservation are depressed. Farmers are discouraged from levelling, terracing, draining, irrigating, or otherwise improving their land. The loss of land productivity through erosion, salinization, or

94

But this is not to say that the effects of erosion are unimportant even in these countries. Statistical analysis by Crosson and Stout (1983: 57) of crop yields in the USA showed that ‘although the effect of erosion on productivity growth was small, it was significant, at least for corn and soybeans in the major producing areas for those crops.’

95

But see the rebuttals by Lipton (1987) and Cleaver (1988). See also Streeten (1987).

484 nutrient depletion is less costly relative to other values in the economy. In general, depressing agricultural prices depresses farmer incentives for soil conservation.

The other view, articulated by Lipton (1987: 209), argues that higher prices will encourage yet more depletion: as a rule, ‘the environment’ responds badly to the normally advised, and otherwise often desirable, price reforms. Better farm prices now, if they work as intended, will encourage ‘soil mining’ for quick, big crops now. … The corrections that pricists advocate, while generally justified, will not produce earthly or even environmental paradise, but will normally have damaging environmental side-effects, requiring preventive or corrective action by the state.

The purpose of this chapter is to determine precisely how macroeconomic and sectoral policies influence soil conservation. I shall show that in general we can't predict how policy reforms will affect soil conservation. Policy reforms could improve conservation, or worsen it, or have no effect at all. To predict how the reforms will affect soil conservation, one would need to know the technical details of agricultural production, and these may well vary widely. The World Bank's contention that policy reforms will conserve soil is a hypothesis that awaits empirical testing.96 Though this conclusion may seem unsatisfying, it compels us to rethink the premiss upon which previous discussions of these matters has been based: namely that policy reforms are good if they conserve soil and bad if they don't. The flaw in this view is that our concern should not lie with soil per se. Unlike many other environmental resources like rainforests and mountain gorillas, soil does not contribute to well-being directly. Its value is as an input in agricultural production. Even if the reforms cause further depletion, the ability of the land to yield an income may still be greatly enhanced. This conclusion relates only to the on-site impacts of soil erosion. If the prices of all inputs and outputs must reflect their true costs, as the World Bank argues, then so too must the price of soil. If farmers possess full title to their land and if eroded soil causes no harm to others, then we should not want to manipulate these prices to encourage soil conservation. But these assumptions often fail to hold. Eroded soil often causes damage downstream, and farmers should be given incentives to reduce this damage.97 Hence, as regards soil conservation, our concern should lie less with the prices of marketed outputs and inputs than with the shadow prices associated with the off-site

96

These conclusions are similar to those obtained in an earlier paper (Barrett, 1991). This chapter generalizes the soil-erosion model used by Barrett (1991), and also explores the role of discount rates. The earlier paper, however, also considers soil fertility as a separate issue; soil fertility is not addressed in this chapter.

97

In developed countries, the greatest external impact of soil erosion is non-point source water-pollution (on this, see Shortle and Miranowski (1987), who show that private farmers will not necessarily conserve too little soil by ignoring water-pollution damages). In developing countries, it is probably sedimentation of hydroelectric dams (see Southgate, Hitzhusen, and Macgregor, 1984). It should be added that in some cases upstream erosion yields downstream benefits.

485 consequences of soil erosion. Correcting these prices is trickier because the shadow prices are not directly observable and will be location-specific. But that is precisely where our efforts at improving price signals for soil conservation should concentrate.

18.2 MACROECONOMIC AND SECTORAL POLICIES Macroeconomic and sectoral policies can alter the incentives facing farmers to conserve soil by altering the prices farmers receive for outputs and pay for inputs, the latter including the cost of money invested in the farming enterprise. Table 18.1 summarizes the World Bank's analysis of the effects of macro- and sectoral policies on these prices. Table 18.1 The effect of macro- and sectoral policies on prices Policy Protection to industry Exchange rate Parastatal margins for exports Export taxes Parastatal prices for import substitutes Input subsidies Credit subsidies Land tenure

Output price [—] [—]

Input price + ±

Interest rate

[—] [—] [—] [—] +

Policies intended to promote industrial growth have both direct and indirect effects on the prices of farm inputs. The direct effect is to increase the cost of protected farm inputs. The indirect effect is to change the internal terms of trade in favour of industry. As a consequence, labour is attracted away from agriculture, and the real wage of farm labour increases. Both of these effects tend to increase input prices. Overvalued exchange rates, sustained by exchange controls and quotas protecting industry, lower the prices of agricultural exports and import substitutes. Imported farm inputs may become cheaper, but migration of labour out of agriculture will raise real labour costs. The World Bank argues that in Sub-Saharan Africa the effect of macroeconomic policies on inputs would exacerbate the effect on output prices alone. Hence, for Sub-Saharan Africa at least the effect of exchange-rate policies on input prices is positive. The prices of agricultural exports are lowered by export taxes and high margins charged by public-sector marketing boards. Import substitutes are often purchased (compulsorily) by parastatal agencies for a fraction of the

486 import parity price. Subsidies on inputs like fertilizers, pesticides, machinery, and irrigated water lower input prices, but the effect is restricted as the subsidies are typically rationed to larger farms. Credit subsidies would seem to lower the cost of borrowing, but because low-interest loans are rationed, lenders may require more of the borrower—like additional collateral—effectively raising the cost of borrowing above the subsidized rate. The World Bank argues that credit subsidies tend to exclude small farmers and raise the effective cost of subsidized credit to that charged by moneylenders in informal markets. Finally, where land tenure is insecure, the cost of borrowing is higher, largely because land can be used as collateral (see, for example, Feder and Noronha, 1987; and Chalamwong and Feder, 1988). Land-tenure laws may cause other problems for soil conservation and environmental protection, but in this chapter I shall concentrate on the effect transmitted through the discount rate. Taken together, the effect on output prices is unambiguous: the macro- and sectoral policies pursued by developing countries have lowered output prices. Interest rates are likely to be higher than they otherwise would be where land tenure is insecure, even if credit subsidies are available. Input prices could be higher or lower than they would be in the absence of policy distortions, but in the poorest areas where modern farm inputs are not widely used, the effect of policy distortions may well be to raise input prices. The remainder of the chapter is devoted to determining the effect these distortions will have on the farmer's decision to conserve soil. The chapter does not consider the inter-sectoral effects; these are analysed by López and Niklitschek (1990).

18.3 THE OPTIMAL CONTROL MODEL The control of erosion represents a classic problem of balancing the immediate gains of an action with the associated long-term losses. Soil depth has a positive effect on output because in deeper soils there is more room for plant roots to take hold, and more nutrients available for plant growth. But conservation of soil nearly always requires sacrifices in output in the short run. Agricultural production can be increased in the near term by clearing and cultivating on hillsides. But unless terraces are built, such gains will be short-lived, for the soil will be quickly eroded away. Similarly, wind breaks, strip-cropping, and conservation tillage can extend soil productivity in the long run, but only at the expense of forgone near-term output. How should these gains and losses be balanced? Soil is a stock, measured in millimetres. Soil dynamics depend on natural conditions and the actions of farmers. If land is left undisturbed, soil accumulates; according to Myers (1988: 6), ‘under normal conditions, soil forms at rates of 0.01–0.5 mm per year’. This natural rate of soil formation is

487 site-specific; it depends on the slope of the land, wind speeds, rainfall, and the nature of the vegetative cover. Farmers influence soil dynamics by choosing which crops to plant, a tillage technique, the extent and nature of crop rotation, and terracing. McConnell (1983) collapses all these actions into a single decision variable, Rt, which can be taken to represent the intensity of cultivation, and describes soil dynamics by the equation:

where M is a constant representing naturally occurring additions to topsoil. The more intensive the cultivation (the greater is Rt), the greater is soil loss. Interspersing row crops with less-valued or zero-valued grasses, a practice known as strip-cropping, reduces Rt, and hence conserves more soil. Similarly, terracing reduces the area of land that can be cultivated, and hence results in less intensive agriculture, but it also reduces soil loss by altering the slope of the land. Most practices that conserve soil do not entail direct costs; their cost is largely felt in a reduction in short-run output. However, some soil-conservation measures do involve direct costs. While terracing and wind breaks reduce cultivation intensity, they are also costly to construct and maintain.98 It seems more reasonable, then, to rewrite the differential equation for soil depth as(18.1)

with

where Ct is the conservation input. According to (18.1), a farmer can maintain a constant soil depth even while practising a more intensive form of cultivation, provided use of the conservation input is increased. For example, a farmer could put more land into production (increase Rt), provided more resources were also put into terrace construction and maintenance (increase Ct). Obviously, (18.1) reduces to McConnell's specification if no direct conservation inputs are employed. Farm output depends on soil depth, the cultivation practice, and non-soil inputs like labour, fertilizer, and irrigated water. Denote these last inputs by Nt.99 Output is then given by the function Q(Rt, St, Nt). It is assumed that a single crop is produced, or that the mix of crops and crop rotation is fixed. The implication of weakening this assumption will be discussed later in the chapter. Ct does not affect current production; however, current expenditure on soil conservation increases future soil depth, all else being equal, and hence increases future output. While it is very likely that a point will be reached

98

Barbier (1988) gives some estimates on the costs of building terraces in the uplands of Java.

99

Nt may be a vector.

488 where additional soil, more intensive cultivation, and additional non-soil inputs have no effect on output (Crosson and Stout, 1983), to sharpen the results of the model it is assumed that Q is increasing, twice differentiable, and strictly concave. Let p denote the fixed price of the farm output, u the price of non-soil inputs, v the price of the conservation input, and r the rate of discount. To the farmer, p is the price paid by the parastatal agency or the border price (less transport costs) converted at the official exchange rate. Similarly, u and v are the prices paid by the farmer for inputs, inclusive of subsidies, r is the market rate of interest, which may be influenced by credit subsidies and land-tenure laws. All of these parameters are assumed to be fixed; in exploring the impact of changes in these parameters, it is assumed that such changes are unanticipated and permanent. Our interest lies in the long-run influence of policy on the soil-conservation decision, and hence the analysis focuses on steady states. The farmer's problem is

subject to (18.1) and Rt, Nt, Cv and St ≥ 0, and S0 > 0 given. Assuming an interior solution, it is easy to show (see the appendix) that the steady state must satisfy four conditions:(18.2)

(18.3)

(18.4)

(18.5)

These equations are easily interpreted. If the farmer increases R* by one unit over a short time-interval and then sets R* = M/F(C*) thereafter, profits increase by pQR. However, the one-unit deviation reduces soil depth by an amount F(C*). This reduction in soil depth reduces instantaneous profits permanently by an amount pQs. The present-value cost associated with the one unit deviation in R* is , or pQSF(C*)/r. Hence, (18.2) requires that the marginal cost and marginal benefit of cultivation intensity be equal in the steady state. Equation (18.4) has a similar interpretation to (18.2). Hold R* fixed. Now increase C* by one unit for a short interval of time. Soil depth then increases permanently by − R*F′(C*). Instantaneous profits therefore rise by − pQSR*F′(C*). The present-value benefit of the one-period, one-unit increase in C* is therefore the RHS of (18.4). The cost is the LHS of (18.4). As is true of R*, conservation inputs must be employed up to the point where the marginal cost of increasing the use of such inputs just equals the associated benefit.

489 Eq. (18.3) is the familiar condition for the employment of variable inputs, and says that the value of the marginal product of every non-soil input must be equal to its price. Eq. (18.5) must hold in the steady state by definition.

18.4 THE EFFECT OF PRICE CHANGES It is immediate from (18.3) and (18.4) that if input and output prices change by the same percentage, then the steadystate values for R*, S*, N*, and C* will remain unchanged for any functions Q and F. This observation is important, because across-the-board price reforms could, in some cases, raise both output and input prices. However, (18.2) and (18.5) indicate that even when the ratios of input prices to the output price are not fixed, changes in these ratios may not affect S*. This is true even if such changes do affect R*, N*, and C*. Proposition 1. Changes in the output price or input prices will have no effect on optimal soil depth in the steady state if QRR/QS is independent of R and N. If this condition is not satisfied, S* will be affected, but the direction of change will in general be indeterminant. QRR/QS will be independent of R and N for some plausible production functions. If Q(R, S, N) = ARαSβNγ, then QRR/QS = αS/β and, using (18.2), S* = βM/αr. Likewise, if Q is a two-level production function with an unrestricted constant elasticity of substitution (CES) between N and G(R, S), where G is an unrestricted Cobb–Douglas unction, i. e. if

where σ ≥ 0 is the CES and 1 > π > 0, then S* = βM/αr if G(R, S) = ARαSβ. The two-level form has a certain intuitive appeal. F may be thought of as ‘utilized soil services’, or the amount of soil made available to production.100 Output is then seen to depend on how these soil services are combined with non-soil inputs. The intuition behind Proposition 1 is as follows. First, consider changes in the output price only. If C and N are fixed and not controllable, then (18.3) and (18.4) drop out. The remaining two equations, (18.2) and (18.5), are independent of p. An increase in the output price increases the benefit of more intensive cultivation, but it increases the benefit of additional soil conservation by the same amount. Hence, the steady-state optimal values for S and R are unaffected by the price change. If S* is to be affected, it must be indirectly affected—through changes in N* and C*. Suppose that C is fixed but N is free to be chosen. Then, by (18.5), R* is also fixed. If S* is fixed at the level that was optimal before the price change, then

100

Berndt and Wood (1979) use the same functional form to combine capital and energy.

490 (18.3) tells us that the rise in output price will induce farmers to increase their use of non-soil inputs. S* can only be affected if QR/QS changes as a consequence of the change in N*. This can only happen if the change in N raises (lowers) the marginal benefit of conservation by more (or less) than the marginal cost. Consider a very slight deviation from the original equilibrium. Suppose R is increased very slightly for a very short period and then once again set equal to , where is the fixed conservation input. Soil depth is then reduced permanently by, say, one millimetre, but present-value profits are left unchanged compared to the original equilibrium. Now add an extra unit of a non-soil input (say, fertilizer) to the farm in the original position and the farm that deviated slightly from this position. If, as a consequence of adding the fertilizer, the profits earned in both instances rise by precisely the same amount, then the price change would not affect the soil-conservation decision. If instead presentvalue profits fall (rise) upon deviating from the original position, then an increase in output price will induce additional soil conservation (erosion). Although the indirect effect of price changes on soil conservation could go either way, some evidence suggests that an increase in the output price will encourage soil conservation. Crosson and Stout (1983: 56) report: ‘The yield response to fertilizer on some heavily eroded soils is smaller than on less eroded soils, suggesting that on these soils the response of farmers to erosion may be to put on less fertilizer, not more. ‘In the context of our discussion this implies that farmers may well respond to the application of more fertilizer (made economic by an increase in the output price, all else being equal) by conserving more soil. All erosion-control measures with the exception of conservation tillage reduce the quantity of land in row crops, and hence the short-term profitability of the farming enterprise (Crosson and Stout, 1983: 67–8). Interspersing row crops with less-valued or zero-valued grasses (strip-cropping), for example, protects the soil but reduces profits in the near term. Conservation tillage, a practice whereby crop residue is left on the soil surface, need not reduce the fraction of land in row crops but it will reduce current yields unless additional herbicides are added to suppress the weed growth normally eliminated by tilling. If herbicides are not available or are not economic, an increase in the output price will not alter the farmer's decision to adopt conservation tillage. But if herbicides are economic, an increase in the output price may (indirectly) affect additional conservation. The price increase would lead farmers to add more herbicides. But the increment in herbicide use is likely to raise output by more on plots adopting conservation tillage. Hence, conservation tillage will probably appear more attractive after the price rise. Conservation tillage is practised widely in the USA. Though not as popular in poor countries, use of conservation tillage does hold a promise for higher yields in Africa (Brown and Wolf, 1985).

491 If C and N are both free, then it is not obvious that an increase in the output price will lead to an increase in N. Suppose QRR/QS is independent of R and N. Then S* is not affected by the price change. From (18.3) we see that N* must rise if p rises, unless R* falls. It is possible that (18.3) and (18.4) together may require that R* fall and that N* either remain unchanged or even fall. Similarly, the effect of a change in p on R* and C* is not clear from conditions (18.2)–(18.5). It is possible that an increase in p could lead to an increase in C* and R*, or a decrease. The appendix presents a simple example which illustrates the sensitivity of the comparative statics to different functional forms and parameter values. The analysis of changes in input prices is similar. There exist both direct and indirect responses, and the total effect on S* depends on the sum of these responses. The appendix provides examples where S* is unaffected by input price changes. Interestingly, these examples also show that even dN*/du and dC*/du may vary in sign, depending on the functional forms and parameter values. In short, the effect of price changes on the farmer's soil-conservation decision depends on the technical details of production as summarized by the production functions Q and F. Plausible functional forms suggest that price changes will not affect soil conservation at all. However, the effect of price changes on soil conservation cannot be reliably deduced without empirical knowledge of the production relationships; and it is likely that these relationships will often be location-specific. In other words, there may not exist a general rule for the effect of price changes on steady-state soil conservation.

18.5 THE EFFECT OF CHANGES IN THE DISCOUNT RATE Conclusions regarding the effect of changes in the discount rate on soil conservation are similar to those obtained above for price changes. Proposition 2. An increase (decrease) in the discount rate will decrease (increase) optimal soil depth in the steady state if QRR/QS is independent of R and N or if F′(C) = 0. If neither of these conditions is satisfied, S* will be affected by changes in the discount rate, but the direction of change will in general be indeterminant. If QRR/QS is independent of R and N, then (18.2) indicates that S* is directly related to r; and our conditions on Q guarantee that dS*/dr < 0. If F′(C) = 0, then (18.4) drops out and by (18.5), dR* = 0. Totally differentiating (18.2) and (18.3) and substituting yields(18.6)

Strict concavity of Q implies that the coefficient on dS* is negative. Since the coefficient on dr is positive, dS*/dr < 0 if F′(C) = 0.

492 Proof of the final part of Proposition 2 is messy and hence left to the appendix. The important point to note is that the ambiguity in the effect of changes in r on S* arises from there being two means of controlling S*. If C were fixed, the farmer would want to reduce S* (by increasing R temporarily) somewhat in response to an unanticipated permanent increase in the discount rate. Similarly, if R* were fixed, the farmer would want to reduce S* (by reducing C temporarily) if r were increased. With both R and C free, we require that R* and C* satisfy (18.5). Since R* and C* are positively related by (18.5) (if F′(C) < 0), it is not possible for R* to increase and C* decrease in the steady state (if F′(C) < 0). R* and C* could both rise or both fall. But then, of course, the effect on S* would be ambiguous. Barbier (1988) employs a similar model but finds that a rise in the discount rate will unambiguously lower equilibrium soil depth. The reason seems to be his assumption that an increase in C* increases soil loss attributable to an increase in R*. Barbier makes this assumption upon observing that farmers often adopt conservation measures only after they switch to producing more erosive (and valuable) crops. However, this adjustment for crop choice is ad hoc. We need to separate the effect a change in prices or the discount rate will have on the nature of cultivation from the effect such a change will have on the decision of which crop to produce. This latter decision is discussed below.

18.6 PRICING POLICY, CROP CHOICE, AND SOIL CONSERVATION The conclusions reached thus far assume that the choice of which crop (or crop rotation) to plant is fixed. But if the relative prices of different crops change, then farmers may want to adjust their crop mix. The empirical evidence indicates that substantial switching does indeed occur as a result of such price changes (see Bond, 1983). And the World Bank (1986: 79) has argued that pricing policies can worsen soil erosion by encouraging farmers to plant less environmentally benign crops: ‘different crops have different effects on soil conservation, and pricing policies may exacerbate soil erosion by inducing farmers to choose the wrong crops.’ Lipton (1987: 209) is correct in saying that ‘pricing policies can just as well induce environmentally “right” crops’. Price reforms will not necessarily benefit soil conservation by providing incentives for farmers to grow less-erosive crops (see also Repetto, 1987). However, implicit in both views is the belief that crops entailing more erosion are in some sense ‘wrong’, or that farmers will blindly switch crops without considering the implications of such a change for soil conservation. If farmers find it attractive to switch to a crop that is more destructive to the soil, then the relation governing optimal soil conservation will have to be recomputed; but it will not be forgotten. The alternative crop may demand less soil conservation in the steady state, but this

493 will cause no loss in efficiency provided the prices of both crops reflect their true opportunity costs. Our concern should not lie with soil per se but with the land's ability to yield a stream of net social benefits.

18.7 COMMENTARY ON THE EMPIRICAL EVIDENCE To my knowledge, no empirical analysis has established the relationship between prices and soil conservation. However, research has shown that the aggregate production impact of price policy changes in many low-income countries is slight.101 Bond (1983) estimated the relationship between aggregate output and price for nine Sub-Saharan countries, and found that in only two of these countries was the relationship significant (see Table 18.2). Perhaps just as important is her finding that the long-run response to a price change is the same as, or not much greater than, the short-run response. If soil conservation responds positively to price increases one would expect the long-run elasticities to be significantly greater than the short-run elasticities. Of course the fact that they are not does not necessarily mean that soil conservation is unresponsive to price; many other factors are involved.102 But the Table 18.2 Aggregate supply elasticities for nine African countries Country Ghana Kenya Ivory Coast Liberia Madagascar Senegal Tanzania Uganda Upper Volta

Short-run elasticity 0.20a 0.10b 0.13 0.10 0.10 0.17 0.03 0.05 0.22

a

Significantly different from zero at the 99 per cent level.

b

Significantly different from zero at the 95 per cent level.

c

Significantly different from zero at the 90 per cent level.

Long-run elasticity 0.34b 0.16c 0.13 0.11 0.14 0.17 0.03 0.07 0.24

Years 1963–81 1966–80 1969–78 1966–80 1968–81 1970–9 1972–81 1968–78 1964–80

Source: Bond (1983, Table 4).

101

Chhibber (1988) provides estimates showing that the short- and long-run aggregate production elasticities are about three times greater in developed countries than in developing countries.

102

Such factors include: (i) non-soil inputs such as fertilizer, and infrastructure such as research and extension, are often available in only limited supply; (ii) a substantial proportion of agricultural output in poor countries is produced by subsistence farmers; (iii) products often trade at prices that differ markedly from their official levels; (iv) household farms comprise a substantial proportion of the total, and may respond less vigorously to a price rise than commercial farms; and (v) a great many farmers and farm labourers are undernourished and may not be capable of working harder after a price rise.

494 evidence at least suggests that reform of agricultural pricing policies alone will be inadequate to the task of significantly raising agricultural output in Sub-Saharan countries. A number of recent empirical analyses purport to estimate the ‘cost of soil erosion’. For example, Magrath and Arens (1989) estimate the present-value cost of one year's on-site soil erosion in Java to be $323 million, or just below half of one per cent of GDP. Bishop and Allen (1989) estimate the present-value cost of one year's on-site soil erosion in Mali to be $31–123 million, or 4–16 per cent of agricultural GDP. Although these studies are not addressing the issue of policy reforms and soil conservation, their analyses should be consistent with the models presented here. They are not, and this raises the question of whether the results of these empirical studies are meaningful for policy. Both studies proceed by estimating (i) soil loss as a function of natural conditions and cultivation practices; (ii) the reduction in yield associated with the soil loss; and (iii) the reduction in farm profit associated with the yield reduction. In estimating soil loss, Bishop and Allen employ the universal soil loss equation (USLE), while Magrath and Arens simply take into account many of the factors included in the USLE—like slope and rainfall erosivity. Bishop and Allen recognize that the USLE ignores soil deposition from upstream erosion, and try to correct for this omission. Magrath and Arens (1989: 17) ignore the matter altogether: ‘Because it is generally accepted that sawah production is not subject to appreciable erosion and in fact benefits from the deposition of nutrients from erosion upstream, no attempt has been made to calculate a cost estimate for sawah areas.’ Quite apart from ignoring the off-site benefits of upstream erosion, Magrath and Arens seem to take the view that erosion is a problem only where much soil is eroded. But the implication of soil deposition from upstream erosion is a higher value for the constant M in (18.1). The implication is not that soil erosion is not a problem for such areas. Soil erosion could still be excessive given the higher value for M. In estimating the reduction in yield associated with soil erosion, both studies compare yields on eroded soils with the yields that would have been observed had the erosion not occurred, holding all else constant. But of course all else cannot be held constant. If yields were higher with less erosion, there would be no incentive for farmers to deplete soil in the first place. If farmers are to conserve soil they will have to sacrifice output in the short run or employ inputs that conserve soil directly. Both studies fail to recognize that profit may be lower in the near term if soil is conserved. Bishop and Allen estimate the costs of employing inputs to conserve soil, but their estimates of the ‘costs of erosion’ are not net of these costs. Both studies also ignore the potential for farmers to mitigate erosion-induced yield losses by employing more non-soil inputs like fertilizers. This is

495 an important omission given the importance of non-soil inputs in production, at least in Java (Magrath and Arens, 1989: 24): Over the last 15 years yields of major dryland crops have consistently risen despite ongoing erosion. However, these yield increases have only been possible through the continued intensification of farming practices. For example, over the period 1972–83 upland rice, maize and cassava yields on Java increased 4.3, 4.7, and 2.8 per cent per year, respectively. … However, fertilizer inputs rose in the case of maize from 38 kg/ha. to nearly 106 kg/ha. and for cassava from 8 kg/ha. to more than 16 kg/ha. … The release and rapid adoption of high yielding maize varieties … may also have masked declines in the productivity of the resource base.

Farmers will add non-soil inputs up to the point where the marginal cost of adding the inputs just equals the marginal benefit (see (18.3)). This equation is ignored in these empirical studies. In estimating the lost profits associated with the predicted yield decline, both studies estimate the loss by subtracting the reduction in costs associated with smaller harvests from the corresponding reduction in revenue. It is not clear from the analyses whether the price and cost data pertain to farmers or to society at large. If the price and cost data are at the farm level, the estimates would reflect the ‘cost’ to farmers of on-site soil erosion; and if these were positive, the question of why would need to be addressed. If the data reflect the true opportunity costs of inputs and outputs to society, the estimates would reflect a part of the ‘cost’ to society of on-site soil erosion (consumers’ surplus would need to be added). Then the question would arise as to whether the source of the problem was that farmers faced the wrong incentives (the issue addressed in this chapter) or whether they conserved too little soil even when facing the right incentives. Because the analyses are not clear on these points, it is not clear how the information provided by these studies might be used for policy.

18.8 CONCLUDING REMARKS The relationship between policy reforms and soil conservation is more complicated than is often asserted. The claim that higher output prices will encourage conservation is blind to the incentives created for farmers to seek bigger gains now by depleting soil. The claim that higher output prices will lead farmers to deplete soil is similarly blind to the incentives to build up soil depth and fertility so that bigger harvests can be reaped in the future. The economic problem is not, as Gray (1913) claims, ‘the balancing of present expenditures against future benefits’; it is the balancing (at the margin) of present costs—including forgone near-term profits—against the sum of future benefits, appropriately discounted. What is more, the effect of reforms on soil conservation will often be indirect—by making soil conservation appear more (or less) attractive when additional non-soil inputs or soil-conserving inputs

496 are employed. Even the effect of a change in the discount rate on soil conservation cannot always be predicted with certainty. The ambivalence of this conclusion is exceeded only by its importance. The suggested reforms should not be defended—or attacked—on the basis of their implication for soil conservation. If wheat farmers in Ethiopia receive half as much per bushel as the country pays for imports, then we can be sure that Ethiopian farmers will produce too little wheat. Whether these farmers will conserve more soil or less when facing the right prices is incidental to the objective of the reforms: economic growth. However, the reforms should equally recognize that it is not just non-soil inputs and outputs that are improperly priced. Soil, too, carries a price; and where there are important off-site impacts, the role for policy should be to ensure that the economic consequences of these impacts are borne by the soil depleters.

Appendix 1. MAXIMUM PRINCIPLE FORMULATION OF OPTIMAL-CONTROL PROBLEM The current-value Hamiltonian is

where λt is the current-value shadow price of soil depth. Assuming that the non-negativity constraints are non-binding, the first-order conditions for a maximum (dropping time subscripts) are(A.1)

(A.2) (A.3) (A.4) plus eq. (18.1). Eq. (A.1) says that the profit obtained by a one-unit increase in cultivation intensity must at all times equal the (shadow) value of the soil that is lost as a consequence of that increase. Eq. (A.2) is the familiar optimality condition for the employment of inputs, and says that non-soil inputs should be employed at all times such that marginal value product equals input price. Eq. (A.3) says that conservation inputs should be employed up to the point where the cost of increasing C by one unit just equals the (shadow) value of the soil saved as a consequence. Finally, substituting (A.1) into (A.4) yields(A.4′)

is the capital gain on S, and QSF(C)/QR is the rate of return on S (holding λ constant). Hence, the LHS of (A.4′) is the return which the farmer earns by holding on to 5. Eq. (A.4′) says that this return must equal the return that the farmer could earn on an alternative investment—the market rate of interest, r. Under our assumptions, the Hamiltonian is concave. Hence (18.1) and (A.1)–(A.4) will be sufficient for a maximum provided the following transversality conditions are satisfied (see Arrow, 1968):

In the steady state, . Eqs. (18.2)–(18.5) are found by setting λ. It is assumed that the optimal steady state is stable.

in (18.1) and (A.1)–(A.4), and substituting for

498

2. EXAMPLE Suppose Q = ARαSβNγ. Then we know(A.5) Consider two alternative functional forms for F(C). In Case 1, let F(C) = e−ωc. In Case 2, let F(C) = (C + 1)−ρ. For Case 1, substitution into (18.2)–(18.5) yields(A.6)

with C* = ln(R*/M)/ω, where X = (αβ+γ−1ωγ−1/γγAββMβ)1/α. For Case 2 we obtain(A.7)

With R* = M(C* + 1)ρ, where Y = (αβ+γ+1/ββAMα+βγγρ1−γ)1/(αρ+γ−1) and Z = γY/αρ. Table A. 1 summarizes the comparative statics for the two cases. Table 18.A1 Comparative statics for Cases 1 and 2 Case 1: F(C) = e−ωc du 0 − + +

dv 0 + + +

dr − 0 + +

Case 2: F(C) = (C + 1)−ρ with αρ + γ < 1 dp dS* 0 dN* + dR* + dC* +

du 0 − − −

dv 0 − − −

dr − − − −

Case 2: F(C) = (C + 1)−ρ with αρ + γ > 1 dp dS* 0 dN* − dR* − dC* −

du 0 + + +

dv 0 + + +

Dr − + + +

dS* dN* dR* dC*

dp 0 0 − −

499

3. PROOF OF PROPOSITION 2 Totally differentiating (18.2)–(18.5) and substituting yields:(A.8)

(A.9)

It is clear from both (A.8) and (A.9) that if dC* = 0, dS*/dr < 0 by strict concavity of Q. Further substitution yields(A.10)

It is clear by inspection that neither side of (A.10) can be signed without detailed information about Q and F.

REFERENCES ARROW, K. J. (1968), ‘Applications of Control Theory to Economic Growth’, in Lectures in Applied Mathematics 12, Mathematics of the Decision Sciences Part 2 (Providence, RI: American Mathematical Society). BARBIER, E. B. (1988), ‘The Economics of Farm-Level Adoption of Soil Conservation Measures in the Uplands of Java’, Environment Dept. Working Paper 11 (Washington, DC: World Bank).

500 BARRETT, S. (1991), ‘Optimal Soil Conservation and the Reform of Agricultural Pricing Policies’, Journal of Development Economics, 36: 167–87. BERNDT, E. R., and D. W. WOOD (1979), ‘Engineering and Econometric Interpretations of Energy-Capital Complementarity’, American Economic Review, 69: 342–54. BISHOP, J., and J. ALLEN (1989), ‘The On-Site Costs of Soil Erosion in Mali’, Environment Dept. Working Paper 21 (Washington DC: World Bank). BOND, M. E. (1983), ‘Agricultural Responses to Prices in Sub-Saharan African Countries’, IMF Staff Papers, 30: 703–26. BROWN, L. R. (1984), ‘Overview’, in L. R. Brown, et al (eds.), State of the World 1984 (New York: Norton). BROWN, L. R. and E. C. WOLF (1985), Reversing Africa’s Decling, Worldwatch Paper 65 (Washington, DC: Worldwatch Institute). CHALAMWONG, Y., and G. FEDER (1988), ‘The Impact of Landownership Security: Theory and Evidence from Thailand’, World Bank Economic Review, 2: 187–204. CHHIBBER, A. (1988), ‘Raising Agricultural Output: Price and Nonprice Factors’, Finance and Development (June), 44–7. CLEAVER, K. M. (1988), ‘The Use of Price Policy to Stimulate Agricultural Growth in Sub-Saharan Africa’, presented at the World Bank's Eighth Agricultural Sector Symposium on Trade, Aid and Policy Reform in Agriculture, Washington, DC, January. COLLINS, R. A., and J. C. HEADLEY (1983), ‘Optimal Investment to Reduce the Decay Rate of an Income Stream: The Case of Soil Conservation’, Journal of Environmental Economics and Management, 10: 60–71. CROSSON, P. R., and A. T. STOUT (1983), Productivity Effects of Cropland Erosion in the United States (Washington, DC: Resources for the Future). FEDER, G., and R. NORONHA (1987), ‘Land Rights Systems and Agricultural Development in Sub-Saharan Africa’, World Bank Research Observer, 2: 143–69. GRAY, L. C. (1913), ‘The Economic Possibilities of Conservation’, Quarterly Journal of Economics, 27: 497–519. HARROD, R. F. (1948), Towards a Dynamic Economics (London: Macmillan). LIPTON, M. (1987), ‘Limits of Price Policy for Agriculture: Which Way for the World Bank?’ Policy Development Review, 5: 197–215. LóPEZ, R., and M. NIKLITSCHEK (1990), ‘Dual Economic Growth in Poor Tropical Areas’, World Bank and University of Maryland, mimeo. MCCONNELL, K. E. (1983), ‘An Economic Model of Soil Conservation’, American Journal of Agricultural Economics, 65: 83–9. MAGRATH, W., and P. ARENS (1989), ‘The Costs of Soil Erosion on Java: A Natural Resource Accounting Approach’, Environment Dept. Working Paper, 18 (Washington, DC: World Bank). MEADE, J. E. (1961), A Neo-Classical Theory of Economic Growth (London: Allen & Un win). MYERS, N. (1988), ‘Natural Resource Systems and Human Exploitation Systems: Physiobiotic and Ecological Linkages’, Environment Dept. Working Paper, 12 (Washington, DC: World Bank). REPETTO, R. (1987), ‘Economic Incentives for Sustainable Production’, Annals of Regional Science, 21: 44–59. SCHULTZ, T. W. (1951), ‘The Declining Importance of Agricultural Land’, Economic Journal, 61: 725–40.

501 SHORTLE, J. S., and J. A. MIRANOWSKI (1987), Intertemporal Soil Resource Use: Is it Socially Excessive?’ Journal of Environmental Economics and Management, 14: 99–111. SOLOW, R. M. (1970), Growth Theory: An Exposition (Oxford: Oxford University Press). SOUTHGATE, D., F. HITZHUSEN, and R. MACGREGOR (1984), ‘Remedying Third World Soil Erosion Problems’, American Journal of Agricultural Economics, 66: 879–84. STREETEN, P. (1987), What Price Food? (London: Macmillan). UNGER, P. W. (1984), Tillage Systems for Soil and Water Conservation, FAO Soils Bulletin, 54 (Rome: Food and Agriculture Organization). WORLD BANK (1986), World Development Report 1986 (Oxford: Oxford University Press).

This page intentionally left blank

PART VII Valuation and Management

19 Valuation of Tropical Forests

505

19.1 INTRODUCTION Tropical forests provide a wide variety of services to humankind. Yet, as documented in countless books, articles, and other media presentations, the forests are under threat. Repetto (1988: 2–15) observes that, since the Second World War, deforestation has shifted from temperate to tropical forests and that, in most developing countries today, deforestation is accelerating. Table 19.1 shows that, at 1981–5 annual rates of deforestation, there are a number of countries where forests will disappear within thirty years. Others, having larger reserves, are losing vast areas every year. A question that naturally arises is, given the value of the tropical-forest resource, why is it being destroyed? The answer, it seems to us, is that a very substantial part of the value simply does not get counted, either because it is hard to measure or because it is not Table 19.1 Tropical deforestation Country With high rates of deforestation (≥3.0% annually) Ivory Coast Paraguay Nigeria Costa Rica Nepal Haiti El Salvador With large absolute losses (≥ 500,000 hectares annually Brazil Colombia Indonesia Mexico Source: Repetto (1988: 7–8).

Closed forest area, 1980 (‘000 ha.)

Annual rate of deforestation 1981–5(%)

Area deforested annually (‘000 ha.)

4,907 4,100 7,583 1,664 2,128 58 155

5.9 4.6 4.0 3.9 3.9 3.4 3.2

290 190 300 65 84 2 5

396, 030 47, 351 123,235 47,840

0.4 1.7 0.5 1.2

1, 480 820 600 595

506 captured by those who make the decisions on deforestation. It is probably true, as suggested by Zylic (1990), that the latter reason is more important. But it has also been addressed at length elsewhere, by Repetto and others. Our charge here is instead to deal with the issue of measurement, or rather lack of measurement, by providing a framework for a more complete valuation of tropical forests. We begin in the next section with a discussion of the major uses of tropical forests, paying particular attention to the relationships among uses. For example, are they compatible with forest preservation? Are they sustainable? Section 19.3 provides the elements of a framework for valuation, taking account of the varied uses. The time-dimension will be important here. One issue is, of course, sustainability. Another is feasibility of a sequential pattern of use; livestock ranching may follow the clearing of land for a timber harvest but not vice versa. Also, the values associated with different uses may grow at different rates. Finally, as we shall see, the present value of a tract of land will depend on how uncertainty about future values is resolved. Section 19.4 is about issues that arise in empirically estimating values or benefits, and Section 19.5 offers some concluding thoughts.

19.2 THE USES OF TROPICAL FORESTS 19.2.1 Uses and utilitarianism: a caveat When we talk about uses of the forest, we have in mind human uses. This is an important distinction, since some would argue that human uses and the values to which they give rise are not deserving of any special consideration when it comes to a decision on whether to preserve a tropical forest. According to one interpretation of this view, nature has rights; to exploit nature is just as wrong as to exploit people (Nash, 1989). Another interpretation is that non-human species are intrinsically valuable, independent of any use they may be to humans (Callicott, 1986). We would prefer not to take issue directly with this view. Rather, we would observe that economics is about the human use and valuation of resources. As such, it is embedded in utilitarianism. In the larger philosophical universe, utilitarianism is, of course, only one of many possible approaches to questions of ethics and choice. Advocates of conservation for its own sake are presumably appealing to an alternative philosophical utilitarianism. In this chapter, we confine our focus to what we understand to be the subject-matter of economics—the uses and values of resources to humans. At the same time, we recognize that decisions, especially public decisions, affecting tropical forests may be made on the basis of a variety of other considerations as well—including, perhaps, inherent rights or intrinsic values. There is an important point to note in this connection. Often in environmental economics, we speak of intrinsic or ‘non-use values’, referring to the

507 benefits some people derive from the mere existence of a natural environment (such as, for example, the Amazon rainforest) even though they make no use of it. In our judgement these benefits are likely to be quite significant for many environmental resources and are legitimately included in our notion of economic value. However, as Batie (1989) points out, this is still a utilitarian view in that the resources, although not used, have value in relation to human welfare. Taking into account this extension of the notion of economic value, a better title for this section of the chapter might be: ‘The Goods and Services Provided by Tropical Forests’, with the understanding that among these services is the existence of the forests, apart from any use to which they may be put by humans. There is a further, and equally important, point to be made here. We shall very shortly be talking about local and global environmental services provided by standing tropical forests. These environmental services are, as we shall see, quite tangible and, indeed, impinge quite directly on human activities. Existence value, as just defined, does not. It is derived from the knowledge that the forests or other environmental resources are alive and well, again, apart from any human activity affected by them.

19.2.2 Uses compatible with preservation Several kinds of human activities in and around the forests appear to be reasonably compatible with preservation: hunting and fishing; gathering of food such as nuts and fruits; gathering of forest products such as rubber, oils, and medicines; and even traditional shifting agriculture. By definition, the setting up of parks and preserves also falls within this category. We observe in passing that all of these uses are sustainable, almost by definition. It may be objected that shifting cultivation is, in fact, a major cause of deforestation and, as such, hardly qualifies as compatible with preservation or even as sustainable. A study by the National Academy of Sciences (1982: 13), for example, concludes that at least half of current deforestation results from shifting cultivation. But by traditional shifting agriculture, we have in mind the kind of activity that involves little disturbance to the forest cover and root systems outside the small plot under cultivation, and that allows the plot to regenerate for twenty to thirty years before a new round of cutting and burning. As noted by Gradwohl and Greenberg (1988: 102), many forested areas once considered ‘virgin’ are now believed to have been occupied for centuries by people practising shifting agriculture. The difficulty arises when population pressures—and perverse incentives as, for example, the linking of ownership rights to the clearing of land—result in the cutting of what had been protective buffer zones and a shortening or even elimination of the fallow period. It is this ‘non-traditional’ agriculture that is implicated in deforestation. Standing tropical forests are also associated with the provision of environmental services, as distinguished from the uses just noted. There are no doubt

508 a number of ways in which these services can be classified, but one that in our judgement will be helpful in discussing valuation issues is the local and global one. What we are calling local environmental services are, perhaps, best understood by considering some of the consequences of deforestation. For example, the loss of forest cover leads to soil erosion which, in turn, aggravates flooding and contributes to premature silting of reservoirs for irrigation and electric-power production. Though local, these impacts are not trivial. It is estimated that revenue losses from sedimentation behind just one dam in Costa Rica have reached a level of $133–274 million (Postel and Heise, 1988: 92). At a global level, tropical deforestation appears to be related to what may well be the gravest environmental issues of our time: the ‘greenhouse effect’ and the wholesale extinction of species. As is well known, the build-up of several trace gases in the atmosphere (most importantly, carbon dioxide) is expected to lead to a substantial warming over the next several decades with an attendant rise in sea-level and change in patterns of precipitation. Potential consequences, to coastal settlements, to agriculture, and to other activities, have been discussed at length in many places (for a brief overview, see Brown and Flavin, 1988). What is important to note here is that deforestation, almost entirely tropical deforestation, is estimated to account currently for a very substantial fraction of global carbon emissions—between one-fifth and one-half as much as the burning of fossil fuels (Postel and Heise, 1988: 94). The second global environmental issue we noted is the threatened loss of species. Although this is the popular perception of the issue, it would be more accurate to speak of the threatened loss of biodiversity. The point of the distinction is that biodiversity, as well as being the source of potentially valuable individual species, is an input to such ecological processes as nutrient and water-cycling, soil generation, erosion control, pest control, and climate regulation—all essential to human survival (Reid and Miller, 1989: 88). With respect to individual species, wild relatives of economically important crops, trees, and livestock often carry unique genes that can be used to improve the characteristics of the domesticated stocks or just help them survive changes in the environment. Plants, animals, and micro-organisms found in the wild are also major sources of medicines and industrial substances. Reid and Miller note that tropical species have been particularly important sources of medicines because many active medical compounds are derived from the toxins that they have evolved to combat predation (ibid.: 27). More generally, tropical forests are important to the conservation of biodiversity because it is believed that they contain more than half of the world's species, though only 7 per cent of the land surface. About half of all vertebrates and vascular plant species occur in tropical forests, and recent discoveries of great insect-species richness there suggest tropical forests may account for as much as 90 per cent of all of the world's species (Erwin, 1982). Although one cannot predict with a high degree of confidence that a particular tract of tropical forest land will be the source of a cure for cancer, or a liquid hydrocarbon, or a key crop pest

509 control, the chances of finding any or all of these are surely greater, the greater the preservation of tropical forests generally.

19.2.3 Commercial forestry Particularly in Africa and South-east Asia, the first step in the conversion of tropical forests is typically opening an area to logging. Commercial forestry covers a variety of activities—including selective culling of highly valued woods; clearcutting for timber or pulp production; and plantation harvesting of an introduced, non-native species. Of course, there is also cutting for fuel, but this is more prevalent in relatively arid areas as opposed to moist tropical forests (Gradwohl and Greenberg, 1988: 37). The difficulty with any of these activities is that they may not be sustainable. Without intensive (and probably also expensive) management, the forests may not regenerate the harvested species successfully. The chief problem is the loss of nutrients once the trees are cut since, in tropical forests, the soil is relatively poor, with most of the nutrients stored in the vegetation (ibid. 31). Another reason why it may be difficult to practise a sustainable forestry is that, especially without intensive management, the harvested species do not regenerate quickly enough to compete with other uses of the cleared forestland. As Gradwohl and Greenberg put it, ‘if the timbering system is only marginally profitable, or becomes unprofitable, it simply sets the stage for a more intense form of forest destruction’ (ibid. 31). Of course, this is a matter of choice—or economics—rather than a physical constraint on the system of the sort imposed by loss of nutrients.

19.2.4 Commercial agriculture One example of the ‘more intense form of forest destruction’ might be commercial agriculture, which involves presumably irreversible conversion of forests. Commercial agriculture includes both plantation farming (of such crops as bananas, sugar cane, rubber, and pineapple) and livestock production, especially (in the Amazon and other tropical American forests) beef-cattle ranching. To these activities, one might add intensive subsistence agriculture, involving both shifting and continuous cultivation (the latter, primarily irrigated paddy rice). Like commercial forestry, large-scale or intensive agriculture may not be sustainable. Long-term, continuous cultivation or grazing leads to soil erosion and loss of nutrients and, at least in the case of cultivation, tends also to involve heavy application of fertilizers and pesticides. The build-up and dispersal of these substances, in turn, interferes with the provision of local environmental services. As with forestry, (costly) management inputs can make an agricultural operation relatively sustainable. Mulching, the use of careful cultivating techniques, long fallow periods, and avoidance of poorer soils can all contribute to this objective (Gradwohl and Greenberg, 1988: 32).

510

19.2.5 Other extractive activities: mining, water-resource development, and transportation To some extent, extractive activities are just an extension of the hunting and gathering that is consistent with forest preservation. For example, medicinal substances, meat, skins, plumage, and even live animals may be taken for export rather than subsistence. Additionally, however, fairly large areas may be affected by mining, water-resource, and transportation projects. Of all of the uses discussed thus far, these are probably the most disruptive of the forest ecosystem and their consequences almost certainly the most difficult to reverse. By definition, a mining project cannot be sustainable, though it can, of course, produce great wealth over the life of the mine. Water impoundments (the construction of large dams for irrigation or hydroelectric power) will also have finite lives as reservoirs silt up over several decades. Moreover, as we have seen, the silting process is accelerated by deforestation and resulting soil erosion.

19.3 A FRAMEWORK FOR VALUATION We start by making a distinction between valuing the specific services provided by a rainforest (such as those described on the preceding pages) and valuing the forest itself, viewed as an asset generating a stream of services over time. The methodological issues associated with valuing specific services provided by a rainforest will be discussed in the following section. Here, we focus on the valuation of the forest. Mapping from the valuation of service flows to the valuation of the asset raises two important issues. The first is discounting and the way in which current and future values are counted. This is, of course, an old issue in welfare economics, and we simply observe that it remains controversial, both in theory and application. One point worth noting here, perhaps, is that environmentalists and advocates of sustainable development have recently joined the debate, generally to argue for a low or even a zero discount rate, on the grounds that this favours resource conservation and, more broadly, the interests of future generations. What might be called the ‘environmentalist critique of discounting’ is presented—and discussed—at some length in the recent volume on sustainable development by Pearce, Barbier, and Markandya (1990). A similar discussion is beyond the scope of the present chapter, but we note the conclusions of Pearce et al. With respect to the choice of a social discount rate different from private market rates, (a) calculating the appropriate rate is difficult; (b) lowering the overall rate will stimulate resource-using investment (and thus be counter-productive from the environmentalist's standpoint); and (c) a selective lowering of the rate for environmental projects is inefficient, cumbersome, and difficult. Instead of adjusting discount rates, they therefore

511 recommend that efforts to take account of environmental concerns be concentrated on (a) improving valuation techniques (including future costs and benefits more carefully), (b) integrating environmental considerations into economic decisions, and (c) incorporating a sustainability constraint, as developed in their study. We might note that all of these conclusions except the last are consistent with earlier discussions by environmental economists (see, for example, Krutilla and Fisher, 1975). The last, the sustainability constraint, is novel. The second issue is the allocation of forest land among alternative uses. As indicated in the preceding discussion, there are a great many different kinds of goods and services provided by the forest, not all of them compatible with each other. In the circumstances, a choice among them is required, and this choice will dictate the value of the forest. In effect, the forest can be regarded not as a single asset but rather as a portfolio of assets, whose composition can be varied over time (subject to some constraints). Thus, the forest cannot be valued without regard to future choices about how it will be managed: valuation cannot be divorced from decision-making. The issue of choice is particularly relevant in the tropical-forest setting, given the wide range of uses and activities relative to those supported by temperate forests in developed countries. In this section we lay out a framework for valuing a tract of tropical forest land, allowing for different choices about the uses of the forest and the role of time discounting and taking into account constraints on the sequencing of uses. We are deliberately vague about the size of the tract: it may be anything from the one hectare sample of Amazon rainforest considered in a recent study by Peters, Gentry, and Mendelsohn (1989) to some much larger area. About the only restriction is that it is not so large that choice among uses is not meaningful because, on a large enough tract, one might reasonably expect to find a little of everything. Our framework, in contrast, is designed to exhibit the consequences, for the value of the tract, of a particular set of choices (for example, indigenous gathering, followed by logging, followed, in turn, by beef-cattle ranching). Of course, in applying this framework to an appropriately delimited tract, the analyst would need to know (or assume) something about what is going on elsewhere in the forest, as well. Spatial relationships may be quite important here. For example, the benefits of preservation will be a non-concave function of area if there is some critical minimum habitat size. Also, as noted earlier in the discussion of shifting cultivation, preservation benefits will be affected by the intensity of activities in adjacent tracts and by the configuration of the tracts. In what follows, we assume that information of this sort can be developed in an empirical case-study or policy analysis and indicate how it might be fitted into a larger framework—one that is readily adapted to show the consequences of different choices and sequences of uses and assumptions about such things as time-discounting, sustainability, and the benefits of particular uses in particular periods.

512 Our point of departure is the work on choices between two alternative uses of a natural environment: development and preservation. Work begun in the Natural Environments Program at Resources for the Future (RFF) in the 1970s (see, for example, Fisher, Krutilla, and Cicchetti, 1972; Krutilla and Fisher, 1975, 1985). The focus of that work was on methods of estimating time-profiles of benefits of the alternative uses, strategies for choosing between uses when information about benefits (especially the future benefits of preservation) is unavailable, and implications for efficient choices when one of the alternatives (development) is irreversible. Another aspect of the analysis of the choice between development and preservation was (and continues to be) an explicit treatment of the implications of irreversibility coupled with uncertainty in the form of a Bayesian information structure in which information about future benefits improves with the passage of time (see, for example, Arrow and Fisher, 1974; Henry, 1974; Hanemann, 1989; and Fisher and Hanemann, 1986, 1987, 1990). The focus of this chapter is on laying out a broader framework for valuation and decision, drawing on results in the earlier literature where relevant. One important way in which the current framework is broadened is by consideration of more than two alternative uses of the land. In the preceding section, we distinguished uses compatible with preservation, commercial forestry, commercial agriculture, and other extractive activities. To make the conceptual transition from just two uses (one irreversible) to several, it will be sufficient to specify three generic uses with appropriate constraints on feasible sequences. Thus, we consider preservation, P; development, D; and an intermediate use, M. We assume that it is possible to go from P to P, M, or D; from M to M or D; and that D is a trapping state. The relationship of the generic uses to those discussed in the preceding section would need to be specified in a particular empirical setting. For example, indigenous gathering (a use compatible with preservation) could be P, commercial timber-harvesting could be M, and large-scale beef-cattle ranching or mining could be D. In some settings, it might be helpful to specify more than one intermediate state as, for example, M1 for low-intensity shifting cultivation and M2 for high-intensity fuelwood gathering. However, for purposes of discussion here, the three states should suffice. Another way in which we broaden the focus of the earlier work is in giving explicit attention to the sustainability of alternative land uses. For example, it will be important to indicate, in our valuation framework, that (say) beef-cattle ranching might not be expected to be sustainable beyond a single planning period of one, two, or five years. Our object here will be not to take a stand on this issue (i.e. to argue for or against a sustainability criterion, as suggested by Pearce et al (1990).), but rather to lay out a framework for valuing sequences of uses of a tropical forest that can accommodate differing views—as well, of course, as differing assumptions about the feasibility of moving from one state to another, discounting, and rates of growth in benefit streams.

513

19.3.1 An illustration The framework is best understood in the context of an illustration, or example, involving the three uses of a tropical forest tract: P, M, and D. The relationships among them can be represented in a decision-tree format as in Figure 19.1. To keep the figure from becoming too cumbersome, we specify just three periods. These may be individual years, conventional five-year economic planning periods, or even something like (optimal) timber-rotation periods in the problem at hand. We note in passing that determination of an optimal rotation cycle is not a concern of ours in the present chapter. The original Faustmann solution is well known, and a comprehensive theory of extensions to account for the influence of non-timber uses has been provided most recently by Bowes and Krutilla (1989). Associated with each use, or state, of the tract in each period is a figure of merit, the benefit of the use, indicated in Table 19.2. We shall initially assume that the benefits in periods 1 and 2 are discounted back to period 0 at a rate of 10 per cent. This is simply a convenient number, and its use here should not be taken as a considered judgement about an appropriate social discount rate. We also assume, for now, that the (expected) benefits of each use in future periods (1 and 2) are known and that no information that would lead to a change in expectations will be forthcoming. An implication of this assumption is that it is possible to value any feasible three-period sequence on a once-and-for-all basis. Later on, we shall relax this ‘open-loop’ assumption and Fig. 19.1 Three-period decision problems

514 Table 19.2 Values of alternative uses and sequences Discounted at 10%

Undiscounted

Use P

Period 1 80

Period 2 98

Period 3 120

Sum over periods 298

M D P M D

90 100 80 90 100

82 100 108 90 100

74 100 145 90 121

246 300 333 270 331

consider a more realistic, though more complex, ‘closed-loop’ format in which information is forthcoming over time and valuation depends on a sequence of choices about uses made on the basis of the new information. With this (open-loop) framework, and the associated benefit figures, we can readily calculate the value of each of the feasible sequences of uses. To avoid cluttering the discussion, we shall consider here just of few of these. Suppose that the development alternative is the extraction of an exhaustible resource; in the simplest case, theory tells us that the discounted (marginal) benefit should be the same in each period. This is represented in the specification of Dt = 100, all t, where Dt is the benefit of D in period t (and, similarly, for P and M). Let us assume that the benefits of uses associated with preservation, P, are growing over time, even relative to the benefits of development, as some of the early RFF work suggested (though not specifically with reference to tropical forests). Thus, we have Pt growing at a rate of 20 per cent to 25 per cent. With these assumptions, development exhibits the greatest value in the first period, but its advantage diminishes over time as, say, the local and global benefits of preserving the forest environment loom larger. It is still true that the development sequence yields the greatest discounted value over all periods (ΣDt = 300 vs∙ΣPt = 298), but this would cease to be true if our time-horizon were extended by just one period and relative rates of growth of benefits in P and D persist. It would also cease to be true even in the three-period case if benefits were not discounted; then, as shown in Table 19.2, ΣDt = 331 and ΣPt = 333. Note that the potential problem of unbounded values is dealt with in this example by ignoring benefits that accrue after the third or fourth period, in effect discounting them at an infinite rate. Finally, note that, if development were postponed beyond the first period, the value of the development sequence would fall. That is, the sequences P0, D1, D2, and M0, D1, D2 both exhibit values below ΣDt = 300 and, indeed, below ΣPt = 298. We have thus illustrated in a simple but plausible example how valuation of a tract of tropical forest land depends on (a) constraints on feasible sequences of uses (D is a trapping state, and one can only get to P from P), (b) the timehorizon and the choice of a discount rate or, indeed, of whether or not to discount, and (c) relative rates of growth of benefits of the alternative uses as

515

Fig. 19.2 A modified decision problem

well as current-period benefits. In the next section, we shall discuss methods of current-period benefit estimation. The point here is simply that fine-tuning the estimation may be less important than laying out a framework for valuation that takes into account constraints and parameters of the decision problem and, indeed, allows decision-makers to explore the consequences of differing assumptions about them. In this spirit, let us change one other key assumption in the example: that development is sustainable. Accordingly, we assume that the mine will be exhausted after the second period, leaving a wasteland not suited to the production of any other valued goods and services. Up to this point, we have said nothing about the intermediate use, M. Suppose that this is some sort of sustainable forestry—the selective cutting (and replanting) of trees for a commercial lumber operation. This is consistent with the specification, in Table 19.2, of an even flow of (undiscounted) benefits from M. Of course, if D is not sustainable, then even the present discounted value of the preservation sequence exceeds that of the development sequence which is now just D0 + D1 = 200. But development benefits are now exceeded also by those of the intermediate activity (the sustainable timber harvest) which displays discounted benefits of M0 + Ml + M2 = 246. The possibility of some intermediate state can play a still more important role in a valuation exercise. Suppose, in addition to being sustainable, that the harvest in our example is conducted with minimal disruption of the surrounding forest environment. As we have noted earlier, this is not likely

516 where the avoidance of disruption is expensive and the benefits of avoidance are not captured by the harvester. From a social point of view, though, the benefits are certainly relevant; the question is, are they worth the cost? Let us further suppose that, if the harvest is conducted in this minimally disruptive fashion, something like the original forest ecosystem can be regenerated, perhaps with a lag of one or more periods after the harvest is stopped. To this end, let us change the example a bit, as in Figure 19.2, where the relevant sequences are traced. In the modified example, a harvest takes place in the first period, yielding somewhat lower net benefits—say, M0 = 80—to reflect the costs of environmental controls. Now we assume that the forest is allowed to regenerate in the second period (indicated on the figure as R1 where R1 = 0) and that, in consequence, preservation-related benefits (including local and global environmental services) can be obtained in the next and succeeding periods. As shown in Figure 19: 2, the discounted present-value of this sequence, over four periods, is 347 as compared to just 314 over four ‘normal’ harvests. What this example suggests is the importance, in an empirical application, of exploring the technical and economic feasibility of a sequence involving recovery from an extractive activity to the point where preservationrelated benefits can, again, be obtained.

19.3.2 The value of exibility To this point, our analysis has been based on the assumption that no further information about future benefits is forthcoming. A hypothetical decision-maker would simply look, as we have done, at the expected benefits over a feasible sequence with all choices (of P, M, or D—or R) specified at the outset. Thus, for example, the maximum expected present-value associated with putting the forest tract to the preservation use in the first period is computed as(19.1a)

where the expectation is with respect to the information set available in the first period. Similarly, the discounted present-value associated with intermediate and development uses are(19.1b)

(19.1c)

In these formulas, while it is recognized that the discounted present-value associated with a current use depends partly on decisions about future uses, the current anticipation of these decisions is based entirely on current information about future benefits and costs.

517 However, this overlooks the possibility that better information about future benefits and costs will be forthcoming in such a way as to influence the future decisions about the uses of the forest tract. Let us now make the more realistic assumption that such information is forthcoming. Specifically, we assume that, at the start of each period, the decisionmaker learns what the benefits of each of the alternative uses of the tract will be in that period (though not in future periods) and then chooses the highest-yielding alternative. This affects how one computes the present values associated with the various uses; it corresponds to a closed-loop type of control rule. Under this control scenario, the maximum expected present-value associated with preservation in the first period is computed as(19.2a)

Similarly, the present values associated with the intermediate and development uses are:(19.2b)

and(19.2c)

Observe that, in the case of the development use, there is no difference between the values associated with the two information scenarios: . For the other two uses, however, there is a difference, given by(19.3)

and(19.4)

By making repeated use of the convexity of the maximum operator and Jensen's inequality, it can be shown (see appendix) that these expressions are non-negative:(19.5)

That is, the present value associated with the preservation or intermediate uses is larger when one recognizes the prospect of being able to use better information in making future decisions than when one disregards this prospect. The difference is what is known in decision theory as the expected value of information; that is, measures the expected value of future information conditional on allocating the forest tract to a preservation use in period zero. Similarly, is the expected value of information conditional on intermediate use. With regard to development, the conditional expected value of

518 information, is zero because allocating the tract to development at time zero eliminates all options with respect to alternative future uses of the forest and thus deprives the decision-maker of the freedom to take advantage of any future information. That is why the information has no economic value. In the terminology of the literature on environmental valuation, the quantities and represent the quasioption value associated with preservation and intermediate uses in period zero. They measure the value of these uses’ flexibility with respect to exploiting new information in later decisions. There is another related, but distinct, element of flexibility: part of the benefit associated with preservation or intermediate uses arises from the breadth of choice that these uses permit in future decisions. Intuitively, preservation affords more flexibility than intermediate uses—the reason being that it bequeaths a larger choice set to decision-makers in periods 1 and 2. This is true under both the open- and closed-loop controls; from (19.1 a, b, c) and (19.2a, b, c), we have(19.6)

and(19.7)

By way of proof, observe that the first inequality in (19.6) yields(19.8)

while the first inequality in (19.7) yields(19.9)

The result follows because the right-hand side of (19.8) takes the form E[max{X, Y, Z} − max{Y, Z}] ≥ 0, while (19.9) takes the form max{E[X], E[Y], E[Z]} − max{E[Y], E[Z]} ≥ 0. The basic principle is: the greater the number of elements in a maximization, the greater the maximum. Thus, in terms of impact on the breadth of future choices, preservation in period zero outranks intermediate use (and development). Does the same ranking apply to the value of information associated with these two uses? In other words, what is the relationship between the two kinds of flexibility: does the prospect of a larger choice set make information more valuable so that Perhaps contrary to one's intuition, a simple counter-example shows that this is not true in general. Consider, first, two alternatives (Y and Z) and two states of nature (S1 and S2), each with a probability of occurring of one-half. Suppose that the benefits of Y and Z

519 are distributed over the states as follows: Y = 5 in S1 and 15 in S2, and Z = 10 in S1 and 12 in S2. Then max{E[Y], E[Z]} and E[max{Y, Z}] are readily computed as

and

respectively. Now add a third alternative, X, where the benefit of X is 9 in S1 and 14 in S2. Clearly, E[max {X, Y, Z}] = E[max {Y, Z}], since the maximum benefit obtainable in S1 and S2 is unchanged. However, max{E[X], E[Y], E[Z]} > max{E[Y], E[Z]} > since E[X] = 115. In this example, having a larger choice set raises V* more than it raises so that the conditional value of information is lowered. Of course, in a particular empirical application, it may turn out that the use which bequeaths the larger future-choice set does have the larger quasi-option value. We have simply shown that this need not be so (see also Hilton, 1981). Also, we do not mean to suggest that the optimal initial choice can never be M or D. We have argued that P and M both provide more flexibility than D with regard to both the breadth of future choice sets and the value of future information and that P outranks M by at least the first of these criteria. But M or D might still be the optimal action in period zero, depending on the relative magnitudes of P0, M0, and D0.

19.4 EMPIRICAL ISSUES It should be noted at the outset that the empirical techniques for valuing the alternative uses of tropical forest land have been developed and applied almost exclusively in the industrialized countries: placing an economic value on the natural environment has so far been a pastime of the rich. Clearly, however, it is highly relevant to developing countries since, as suggested earlier, one reason for deforestation in these countries is that a substantial part of the tropical forests’ value is being overlooked when forest land-use decisions are made. The goods and services generated by a tropical forest may be viewed as intermediate goods (e. g. timber, watershed protection) or as final goods for some set of people (e. g. fuelwood, fruit, recreation, intrinsic values). The contribution of the tropical-forest use may be seen as making available something that would otherwise be unavailable or improving the supply (lowering the cost or raising the quality) of an existing commodity. To the extent that marketed commodities are involved, these benefits can be measured using standard techniques based on shifts in demand and supply functions or related

520 concepts (value of the marginal product, avoided cost, preventive expenditures saved, etc.). There may be practical problems in modelling the market correctly—for example, marketing channels that impose constraints on the seller's ability to dispose of an increased supply of tropical fruit and nuts, or price effects that spill over to other markets and call for a general-equilibrium analysis. Also, if there are distortions arising from government actions or imperfect competition, shadow prices will be needed to correct for divergences from true opportunity cost or willingness to pay. In general, though, these marketed services of tropical forest raise no new conceptual issues. However, many of the services provided by a tropical forest are not supplied through a market. These would include most of the environmental services such as protection of habitat, promotion of genetic diversity, protection against the greenhouse effect, provision of parks and wilderness preserves, etc. This is because these aspects of the natural environment are, to a large degree, public goods (or public inputs) that cannot be divided up and sold. The absence of markets poses a challenge to conventional valuation techniques. In response, two approaches have been adopted. One approach is to identify commodities that are marketed and whose consumption is related in some manner to the enjoyment of the natural environment—for example, commodities that are complements or substitutes for the natural environment. The classic example is the travel-cost method of valuing the recreational use of the environment; the hedonic property value, hedonic wage, and hedonic travel-cost models are other examples. In these cases one uses conventional techniques to recover individuals’ preferences for the market goods from their observed market demand behaviour; and, since their enjoyment of the natural environment is bound up with their enjoyment of these marketed goods, their preferences for the natural environment are recovered at the same time. These ‘indirect’ techniques of valuing non-market goods are, by definition, subject to two limitations. First, for some environmental attributes, there simply may be no substitute or complementary market goods that can serve to reveal a person's preferences for the natural environment. Second, to the extent that substitute or complementary market goods do exist, there can be no assurance that these capture all of the person's preferences for the natural environment: in addition to caring for nature in connection with his use of the related market commodities, a person may also care about nature for reasons unconnected with them. For example, a hunter may want to protect the forest because it provides habitat for the animals that he hunts; but he may also wish to see the forest protected for motives that have nothing to do with his own or others’ hunting. This additional component of a person's preferences may not be reflected in his demand function for any market commodities and, thus, it cannot be recovered by the indirect valuation techniques. The other approach is ‘direct’ valuation using surveys to elicit from respondents measures of their willingness to accept or willingness to pay for the services provided by a tropical forest (the contingent valuation approach).

521 This approach has attracted much interest recently and is the subject of much current research with regard to its statistical and survey research aspects. By construction, it offers the prospect of recovering those components of preferences that elude the indirect measurement techniques. The recent books by Smith and Desvousges (1986), Johannson (1987), and Mitchell and Carson (1989) provide excellent introductions to the various valuation techniques; some applications to developing countries are described by Hufschmidt et al. (1983). Rather than giving more details here, we propose to comment on some of the lessons to be learned from experiences with non-market valuation to date. First, it must be emphasized that framing can be very important to the success of the exercise—how one conceptualizes the consequences of a change in the flow of services from a tropical forest greatly affects the form of the subsequent analysis. Whether the natural environment is seen as an input or as a final good, or whether the change is seen as primarily entailing a change in income, a change in choice sets, or a change in prices, inevitably shapes the economic analysis to be performed. The framing is inherently a subjective decision on the part of the analyst; asking the right questions is a key to her success. Our second point concerns the influence of the availability of tools and data on what gets measured. A few years ago, there was a popular song with lyrics that ran: ‘If you can't be with the one you love, then love the one you’re with.’ It seems to us that economists too often embrace a similarly pragmatic morality: if you can't measure what you want, then be satisfied with what you can measure. Data limitations obviously matter; but an effective analyst must demonstrate a good sense of what aspects of the tropical forest are important even if they cannot readily be quantified. The substance of the issues, not the techniques, should drive the analysis. Third, before proceeding to the technical details, the analyst should start with a balance sheet listing the various groups of people (including future as well as present generations) that may be affected by a change in uses of tropical forests and indicating the nature of this impact in physical (non-monetary) terms. This is an essential prelude to the economic valuation exercise. In addition to providing an overall perspective, the balance sheet delineates the distinct groups that have standing in the analysis and need to be considered from a distributional point of view. Many of the recent applications of non-market valuation techniques in the USA have been relatively unconcerned with distributional issues and have concentrated instead on whether the aggregate benefits of, say, a proposed regulatory action, outweigh the aggregate costs. The focus on aggregate benefits stems from two sources, one philosophical and the other political. The philosophical source is the Kaldor–Hicks potential compensation criterion for assessing welfare changes which implicitly de-emphasizes questions such as how benefits differ among distinct sub-groups of the population

522 and concentrates, instead, on the overall population mean. The political source, at least in the USA, is a distrust of estimates of regional impacts and a belief that agencies such as the Army Corps of Engineers have abused them in the past in order to justify unwarranted projects. However, both the emphasis on aggregate benefits and the Kaldor–Hicks criterion may be inappropriate when applied to the valuation of tropical forests. For some forest services, the benefits clearly transcend national and temporal boundaries, e.g. prevention of the greenhouse effect, protection of species diversity. To ensure a proper accounting, it is necessary to look beyond the current preferences of individuals in countries in which the tropical forests are located and include the values that people in other parts of the world, and in future times, would place on these services. In that case, however, it would be meaningless to summarize the results in terms of an average per capita benefit. For example, with respect to current benefits, one would surely want to break the total down into benefits accruing to residents of the country where the tropical forest is located, benefits accruing to residents of industrialized countries, and benefits accruing to residents of other Third World countries. Moreover, since regional impacts are important in the economic development process, one cannot avoid paying explicit attention to them. When the benefits involve the natural environment as a final good, there is an additional reason for wanting to identify the values associated with distinct sub-groups of the population—namely the diversity of people's preferences for the natural environment. To be sure, observed differences in monetary values attached to environmental resources can be linked in part to differences in income; protecting the environment is likely to be quite income-elastic not only within countries but also among them. Beyond this, there appear to be genuine differences in tastes with regard to both use and non-use values for the natural environment. Differences in tastes—not differences in income or prices—are surely the key factor determining participation versus non-participation in outdoor recreation. Similarly, there is much more variation in the willingness to pay values elicited by contingent valuation surveys in the USA than can be explained by income alone. Different people clearly have different interests: some people—perhaps a small number—place a very high monetary value on the given environmental resource; other people care for the resource, but not as passionately and with lower monetary values; and there are some people who place no value at all on the resource. Therefore, for the population as a whole, the aggregate value can be thought of as depending on (a) the fraction of the population falling into each distinct preference group, including the zero-value group and (b) the typical value (e.g. median, mean) associated with that group. Approaching the aggregate value in this way is useful when it comes to extrapolating from the responses to a contingent valuation survey (or an outdoor recreation survey) to the overall population, since the distribution over preference groups in the sample

523 may be different from that in the population. It may also be useful when dealing with the crucial but awesome task of projecting the values that future generations place on the environmental resource. To the extent that future generations have entirely different preferences from the present generation, there is no way to predict them. But, to the extent that future generations are composed of the same preference groups as the current generation, albeit in different proportions, one has some hope of making a prediction by using information about the values currently associated with distinct preference groups combined with projections of their future population shares. Projecting future population shares, as opposed to future preferences, may be associated with a manageable degree of uncertainty, since we currently have information about the relationship between preferences for the environment, on the one hand, and readily measured and projected variables such as income and education levels on the other. Predicting future benefits is probably the single most challenging aspect of valuing tropical forests. More than for some other resources, current decisions about managing tropical forests can have significant long-term impacts. Dealing with these in a sensible manner is crucial to the success of the valuation exercise. Unfortunately, little guidance can be obtained by looking at the experience with environmental valuation in the industrial countries. Almost all of these exercises have been static in nature. They employ data—whether housing-market data or data from contingent valuation or recreation surveys—that are collected at a single point in time and convey no information about secular trends in environmental behaviour or attitudes. Of course, collecting time-series data on environmental behaviour or attitudes requires a greater commitment of resources and takes much more time than cross-section data, which is why such databases are scarce. But, having data from a single point in time greatly limits one's ability to make projections about the future. In fact, this has not been seriously attempted in most recent valuation exercises: current per capita values are projected to future populations. Our suggestion above about projecting sub-group values separately from their population shares is intended as an improvement, but it is by no means a complete solution. Krutilla and Fisher (1975) have stressed the need for ‘secondbest’ approaches to estimating time-profiles of future benefits in the absence of good information, for example, by using current estimates of preservation benefits and postulating a future growth rate for the ratio of these benefits to those associated with the alternative uses of the resource. In both cases, there is a substantial degree of uncertainty associated with the projections of future population shares or future growth rates in benefits. This could be handled by developing alternative scenarios for these future outcomes and attaching probabilities to them. Such an exercise would certainly be subjective and ‘soft’, but that cannot be avoided. The alternative—to treat the future as known with certainty and a replication of the present—is unacceptable.

524

19.5 CONCLUDING REMARKS Throughout this chapter, we have sought to emphasize the link between the valuation of tropical forests and decisions about their uses. We firmly believe that the valuation exercise cannot be designed effectively without reference to the types of decisions that are being made. The framing of the decisions determines the valuation strategy. To this end, we have reviewed alternative uses with a view to identifying feasible sequences that, in turn, affect the value of some initial choice of use or activity. Beyond the link between valuation and decision, perhaps always important, in the case of tropical forests in particular it seems to us that key features of the valuation problem are the long time-horizon and the great uncertainties associated with the future consequences of current management decisions. As our illustrative example shows, it may be more important to take account of feasible sequences, the sustainability of a given use, the choice of discount rate, and the planning horizon, than to fine-tune the estimates of current benefits. The economic literature contains a number of treatments of tropical forest management which de-emphasize the uncertainty and treat the future costs and benefits of alternative forest uses as known with certainty. (A recent example in a leading journal of environmental and resource economics frames the problem of managing a tropical forest as a deterministic optimal depletion problem; at what rate should the tropical forest land be converted to agricultural use so as to maximize the discounted stream of net benefits, when the benefit functions themselves are taken as known and stationary over time?) We believe that this is an inappropriate way to frame the problem, even as a first approximation. For an economic analysis to be useful, it must find a strategy for coming to grips with uncertainty with regard to both how one approaches the decision problem and how one approaches the valuation of alternative uses. The decision problem has to be seen as one of stochastic control, in which information acquisition and flexibility rank more highly than nicely determining the allocation of land based solely on current estimates of benefits and costs. The valuation problem becomes one of guessing how the future may be different from the present and identifying blind spots as much as fine-tuning the estimates of what is known. Both as a means of eliminating gaps in the analysis and also as an aid to predicting how the future may be different from the present, we have suggested that the analyst develop a ‘balance sheet’ of affected parties, both present and future. The balance sheet should take account of all significant impacts, including those that cannot readily be quantified. Valuation in monetary terms is desirable in order to ensure a common yardstick—but it must yield to the goal of comprehensiveness with regard to covering the things that matter in the real world. The balance sheet forces the analyst to be explicit about who has standing and what are the distributional implications of alternative uses

525 of the tropical forest—both of which tend to be treated with some skittishness by researchers in the industrialized countries. Finally, the balance sheet provides a framework for extrapolating future values. We have suggested that changes in people's preferences may be a powerful force affecting the value of environmental resources over time, and one way to project this is to identify the distinct preference groups, or ‘market segments’, in the current population and then project changes in their future population shares. We concede that this type of approach is somewhat fuzzy but judge that the alternative—assuming the future to be the same as the present—is spuriously precise.

Appendix: Proof that Define Et/t − 1 as an expectation held in period t, based on observation through period t − 1. We begin by noting that

from the convexity of the maximum operator and Jensen's inequality. Thus,

Similarly, one can prove that

Let

Since

it follows that

Thus,

527

REFERENCES ARROW, K. J., and A. C. FISHER (1974), ‘Environmental Preservation, Uncertainty, and Irreversibility’, Quarterly Journal of Economics, 88. BATIE, S. S. (1989), ‘Sustainable Development: Challenges to the Profession of Agricultural Economics’, American Journal of Agricultural Economics, 71. BOWES M. D., and J. V. KRUTILLA (1989), Multiple Use Management: The Economics of Public Forestlands (Washington, DC: Resources for the Future). BROWN, L. R., and C. FLAVIN (1988), ‘The Earth's Vital Signs’, in State of the World (NewYork: W. W. Norton & Co.). CALLICOTT, J. B. (1986), ‘On the Intrinsic Value of Nonhuman Species’, in B. G. Norton (ed.), The Preservation of Species: The Value of Biological Diversity (Princeton: Princeton University Press). ERWIN, T. L. (1982), ‘Tropical Forests: Their Richness in Coleoptera and Other Arthropod Species’, Coleopterists Bulletin, 36. FISHER, A. C, and W. M. HANEMANN (1986), ‘Option Value and the Extinction of Species’, inV. K. Smith (ed.). Advances in Applied Micro-Economics (Greenwich, Conn.: JAI Press). FISHER, A. C, and W. M. HANEMANN (1987), ‘Quasi-Option Value: Some Misconceptions Dispelled’, Journal of Environmental Economics and Management, 14. FISHER, A. C, and W. M. HANEMANN (1990), ‘Information and the Dynamics of Environmental Protection’, Scandinavian Journal of Economics, 92. FISHER, A. C, J. V. KRUTILLA, and C. J. CICCHETTI (1972), ‘The Economics of Environmental Preservation: A Theoretical and Empirical Analysis’, American Economic Review, 62. GRADWOHL, J., and R. GREENBERG (1988), Saving the Tropical Forests (Washington, DC: Island Press). HANEMANN, W. M. (1989), ‘Information and the Concept of Option Value’, Journal of Environmental Economics and Management, 16. HENRY, C. (1974), ‘Investment Decisions under Uncertainty: The Irreversibility Effect’, American Economic Review, 64. HILTON, R. W. (1981), ‘The Determinants of Information Value: Synthesizing Some General Results’, Management Science, 27. HUFSCHMIDT, M. M., D. E. JAMES, A. D. MEISTER, B. T BOWER, and J. A. DIXON (1983), Environment, Natural Systems, and Development (Baltimore: Johns Hopkins University Press). JOHANNSON, P. (1987), The Economic Theory and Measurement of Environmental Benefits (NewYork: Cambridge University Press). KRUTILLA, J. V, and A. C. FISHER (1975), The Economics of Natural Environments: Studies in the Valuation of Commodity and Amenity Resources (Baltimore: Johns Hopkins University Press). KRUTILLA, J. V, and A. C. FISHER (1985), The Economics of Natural Environments: Studies in the Valuation of Commodity and Amenity Resources, 2nd edn. (Baltimore: Johns Hopkins University Press). MITCHELL, R. C, and R. T. CARSON (1989), Using Surveys to Value Public Goods: The Contingent Valuation Method (Washington, DC: Resources for the Future). NASH, R. F. (1989), The Rights of Nature: A History of Environmental Ethics (Madison: University of Wisconsin Press).

528 NATIONAL ACADEMYOF SCIENCES (1982), Ecological Aspects of Development in the Humid Tropics (Washington, DC: National Academy Press). PEARCE, D. W., E. B. BARBIER, and A. MARKANDYA (1990), Sustainable Development: Economics and Environment in the Third World (Cheltenham: Edward Elgar Publishing Ltd.). PETERS, C. M., A. H. GENTRY, and R. O. MENDELSOHN (1989), ‘Valuation of an Amazonian Rainforest’, Nature, 339. POSTEL, S., and L. HEISE (1988), ‘Reforesting the Earth’, in State of the World (WorldWatch Institute, NY: W. W. Norton & Co.). REID, W. V., and K. R. MILLER (1989), Keeping Options Alive: The Scientific Basis for Conserving Biodiversity (Washington, DC: World Resources Institute). REPETTO, R. (1988), ‘Overview’, in R. Repetto and M. Gillis (eds.), Public Policies and the Misuse of Forest Resources (Cambridge: Cambridge University Press). SMITH, V. K., and W. H. DESVOUSGES (1986), Measuring Water Quality Benefits (Boston: Kluwer-Nijhoff). ZYLIC, T. (1990), ‘Mismanagement or Optimal Extinction: A Comment on Fisher and Hanemann’, UNV, WIDER Conference on ‘The Environment and Emerging Development Issues’, Helsinki, Finland.

20 The Management of Drylands 20.1 INTRODUCTION The most appropriate strategies for the improved management of dryland areas in situations where budgetary resources are limited are by no means self-evident for three main reasons. First, the physical nature and the extent of what has been termed ‘desertification’ is not well understood, partly because of the elusiveness of the concept itself, but partly because of weak data. Second, the underlying causes of land degradation in dryland areas, which might appear clear at first sight, often become less clear when probed more deeply. Third, the impact of changes in public policy aimed at improving land management are not easy to predict and seem to be quite dependent on both the geographic location and the stage of economic development. This chapter explores the physical processes, the adaptive strategies of individuals faced with the challenge of dry areas, the economic environment within which these strategies have evolved, and some possible, but by no means proven, policy responses to achieve a sustainable development. The chapter is based predominantly on the African experience and is aimed at the developing-country situation.

20.2 DEFINITIONS We define drylands as the hyperarid, arid, semi-arid, and sub-humid zones of the world as described by UNESCO (1979). The upper end of the sub-humid zone reaches areas where the mean annual precipitation is less than or equal to 75 per cent of the mean annual potential evapotranspiration. There is an invisible, but highly significant, boundary, lying somewhere between the semi-arid and sub-humid zones, depending on climatic factors. Often, in Africa, it occurs around the 800 mm rainfall isohyet. Above this rainfall boundary science has found many technologies that not only succeed in a technical sense, but are profitable to farmers; below this boundary science has found very little that is both profitable and sufficiently riskless for the low-income rainfed farmer.

530 Generally, in tackling dryland degradation with a focus on sustainable development, we are less concerned with the true desert areas which, unless the location of an irrigation investment, will remain true deserts and are seldom capable of supporting significant numbers of people at economic investment levels. We are more concerned with those intermediate dryland areas where rain fed cropping or pastoral activities are possible albeit marginal, which do have to support significant numbers of people and which may suffer degradation due to the burden on the carrying capacity. This chapter will try to avoid the term ‘desertification’ because its widely varying use by different authors has devalued it to the extent that, paradoxically, the term itself has become, in a sense, ‘desertified’. However, we offer the following (Nelson, 1989) as a definition because it is a useful first step towards exploring the issues, also because many readers will already have their own image of what they mean by ‘desertification’ and it is important to reach a common frame of reference: Desertification is a process of sustained land (soil and vegetation) degradation in arid, semi-arid and sub-humid areas, caused at least partly by man. It reduces land resilience and productive potential to an extent which cannot be readily reversed either by removing the cause or by making substantial investment.

The first point to be noted about such a definition is that it seeks to differentiate the reversible from the irreversible. Given the long cycles of as much as ten to fifteen years of below-average rainfall in dryland areas, it is extremely difficult to separate the temporary from the permanent. As Dregne and Tucker (1988) note, since annual oscillations of as much as 200 km in the southern vegetational boundary of the Sahara appear normal, it would take about thirty to forty years of data to identify the permanent average shifts of 5–6 km per year often said to have occurred. The second point to be noted in the definition is the concept of resilience (Warren and Agnew, 1988). Given the inevitable massive and persistent rainfall fluctuations that one must expect in such areas, an important objective in any management or policy interventions must be to retain resilience, that is the ability of a system to bounce back to its previous level of productivity when more normal conditions return. Resilience is an important strategic objective not simply with respect to land but, as we return to later, with respect to people (Mortimore, 1989).

20.3 THE PHYSICAL PROCESSES Contrary to the popular image, the limited evidence available does not suggest that the world is being engulfed by advancing sand at an alarming rate. (One reads in quite reputable magazines such absurd statements as: ‘the Sahara and the Kalahari deserts are closing at the rate of 100 km per year’.) It is true that there are significant areas where sand movement is a serious problem, but this is not the main dryland problem in terms of the global impact on

531 either production or people. The main problem is the more insidious, less dramatic, but far more widespread, problem of deteriorating soil fertility or structure, and deteriorating vegetational cover or complexity, in areas of increasing population pressure where previously adequate management systems have not adapted fast enough. Such areas seldom turn into moving sand. The sequences in the process leading to land degradation in dryland areas can be too easily oversimplified. However, in cropping areas one quite common sequence has been described as the following (Newcombe, 1984): 1. Increasing population pressure leading to reduced fallow periods in farming areas and increased stocking rates in pastoral areas. 2. Loss of soil fertility and destruction of vegetation. Trees are cleared for farming, harvested for fuelwood, overgrazed by livestock or cut for livestock fencing. This last has been shown to take very large quantities of wood, although it is invariably the less desirable species which are cut (pastoralists know their species well). 3. Increased use of animal manure for fuel. In the Sahel the human carrying capacity that can be sustained by the fuelwood supply is lower in all agroclimatic zones than that which can be sustained by the crop/livestock production capacity, so fuelwood is the overriding constraint (Gorse and Steeds, 1987). 4. Difficult-to-reverse damage to the land resource as the system collapses, usually triggered by a severe drought. Sometimes accompanied by out-migration of people. In pastoral areas there are a number of possible degradation sequences. One common sequence is a vegetational degradation which, paradoxically, can lead to dense bush. The sequence begins with overgrazing of grasses by cattle and loss of grasses in a mixed grass-shrub subclimax. This results in less frequent, more patchy, and less hot burns by fire because of lack of grass to carry the burn (in most rangeland environments fire has been a dominant influence). This leads to a build-up of bush since fire controls the woody-bush shrubs. This increased bush competes out the remaining grasses exacerbating the process. Once the resultant bush has become dense, and because it is generally unpalatable, most livestock and herders will not penetrate. While the resulting unproductive dense bush may protect soil reasonably well, the loss of productivity is inefficient and may increase livestock pressures on neighbouring areas. While the described sequences are not uncommon, there are many variations in degradation sequences. Furthermore, there are enough observed exceptions to the commonly accepted scenarios to raise questions about the inevitability of such sequences. The following are some examples of those exceptions. First, there seem to be examples in some locations of surprisingly high populations being carried, apparently sustainably, in dryland situations, for example in the Kano Close Settlement Area in Nigeria (Mortimore, 1989).

532 Second, there are examples of degradation arising following reductions of rural populations, for example in Yemen where terraces are now collapsing due to lack of labour arising from outmigration. Third, questions have been raised about the impact of increasing livestock numbers on the land resource. The conventional wisdom for almost 100 years in Africa has been that livestock numbers have been way beyond sustainable levels. But as Sandford (1976) has pointed out, this does not lie well with the fact that all this time livestock numbers, human population numbers, and total livestock production have been increasing, albeit with the usual drought fluctuation. Even at a more localized level the conventional wisdom about the prevalence of degradation around bore-holes has been recently questioned (Hanan, 1989 unpublished, quoted in IUCN, 1989) by a study that failed to find a correlation between biomass and distance from bore-holes. In Russia, in the Kara Kum desert, problems due to lack of animal impact have arisen from the build-up of a desert moss on the surface which, with insufficient animal hoof action to break it up, has reduced moisture percolation to the main vegetation growing in the dune troughs. With particular relevance to these last two examples, Savory (1988) suggests that many dryland areas of the world are, in reality, overgrazed but understocked. He argues that range management specialists, while always noticing the negative impacts of overgrazing by animals, have failed to understand fully two positive impacts: first, the soil disturbance of hoof action which breaks surface crusts and encourages moisture penetration and germination, and, second, the removal of growth-inhibiting dead plant material. He argues that intensive, but carefully managed, animal impact is what is needed, rather than long periods of rest. In this way both greater productivity and greater soil protection can be achieved. There are very interesting issues here that call for further research and testing under a range of different conditions. Given the mixed evidence it is risky to oversimplify and argue merely that the dryland areas of the world are being destroyed by too many animals pursuing a limited grazing resource or by too many people trying to farm too little marginal land. This may be part of the story, but it is certainly not the whole. There are enough questions about the validity of the conventional wisdom to warrant at least being open-minded to the possibility of myths being exploded in pursuing public strategies for improved dryland management. The technical relationships are often complicated and highly location-specific. Furthermore, the most appropriate management action may vary widely depending on the sequences of climatic events both in the past and, unfortunately for the decision-maker, in the future too.

20.4 THE CAUSES OF THE PHYSICAL PROCESSES It has been argued above that the physical process may not be quite as obvious as it might seem. But one can go further and say that knowing the facts of the

533 physical process (i.e. the symptoms) may tell a very misleading story about the true underlying reasons (i.e. the cause). Perhaps the most damaging interventions in terms of wasted resources have been those which have superficially identified the true cause of degradation as being simply ‘loss of trees’ and have then proceeded, with a massive public planting programme, to plant trees, without any attempt to understand why they were lost in the first place, or what might prevent them being lost a second time, or indeed, whether the people in the area even see trees as a high priority. Deep probing for underlying cases and historical perspective, both on the past and the future, are needed in designing sensible responses. Sandford (1976) identifies four broad, global views on the causes of desertification: (a) the structural argument, laying the blame on the social and economic structures; (b) the natural events argument, laying the blame on climatic events; (c) the human fallibility argument, laying the blame on the shortsightedness of farmers, pastoralists, governments, donors, etc., and (d) the population argument laying the blame on human and animal population growth. Few would argue that any of these have been entirely absent as a cause, but the analysts view of the balance of causation here is important for the design of strategies. Causes (a) and (c) are susceptible to policy change, although social structures will generally have to change themselves from within. Cause (b) is not susceptible to much influence at present, although more evidence on climatic feedback linkages (loss of vegetation reducing rainfall due to surface reflectivity and evaporative changes) would alter that, although only by throwing the area of focus back on to one of the other causes. Cause (d) is only susceptible to change over a very long period. Typically pastoralists in Africa attribute causes to drought, although many will add the cause of pressure from ‘outsiders’ on previously group-managed tribal grazing areas; in effect a deterioration from a common property situation, with some agreed group rules about forage use, towards an open-access, ‘free-for-all’, situation. A historical perspective on the causes is important to understand how one arrived at a situation. It may not provide clues to responses to the current situation. To take the Sahel as an example, there appear to have been considerable changes over periods of centuries (not all necessarily bad). Some of the most important changes have been the following: (a) loss of trees due to charcoal-making by the trans-Sahel trade caravans; (b) increased localized degradation due to an increase in settlements and towns composed of permanent structures (rather than tents), which require large amounts of wood for both buildings and fuel and which face the inevitable problems of permanent structures in a mobile landscape; (c) loss of wildlife due to the advent of sophisticated firearms, leading to reduced dispersal of those tree and shrub seeds which are mainly dispersed by wildlife, and also leading to greater willingness of herders to enter and graze forest and thickets due to the reduced threat from

534 carnivores; (d) agricultural expansion northwards into previously hostile areas which are marginal for agriculture, partly due to the cessation of slavery and then the pacification of the Sahel early in the twentieth century; (e) increase in cattle numbers as cattle people from the south moved northward bringing their cattle traditions with them into areas less suited to cattle, sometimes encouraged by considerable periods of above-average rainfall; and, (f) veterinary and welldrilling programmes which helped to greatly increase cattle numbers and expand their range. However, while these factors may, in one sense, be causes, they do not give a lot of clues for policy responses now. Can one remove the towns and settlements? Can one ban firearms? Can one encourage less ‘pacification’ of the Sahel? Can one remove veterinary knowledge? Obviously these are not realistic solutions. Changes are not always negative. In another continent, parts of the Simpson Desert, and surrounding areas, in Australia now look better than for at least 100 years due to a short and highly unusual sequence of high rainfall in the 1970s which was enough to generate a strong episode of bush germination, survival, and establishment. This bush will mature, age, and die over the next seventy years or so, giving what may be a very atypical vegetation for over half a century. Such substantial and quite persistent changes are part of the management challenge of dry areas. If one cannot tell what the next decades will bring it is difficult to do anything other than follow an opportunistic strategy, guided by what little science can offer from relatively shorter-term experiments. As a wise old Australian farmer once said of these areas, ‘You can learn enough in one year to make a fool of yourself the next.’

20.5 ADAPTATIONS TO DRYLAND CONDITIONS Farm and pastoral families in dry areas have evolved a wide range of adaptive practices to attempt to reduce the riskiness of production. Any improvements in technology offered by research will usually need to build on these existing mechanisms. They can be divided into ex ante strategies, aimed at reducing risk before the drought hits, and ex post strategies to adopt and make the best of the situation after the drought has hit. With respect to ex ante strategies, the following are common approaches to avoiding risk (Matlon, 1990): diversification of crops, cultivars, or plot location; intercropping, in which, if one crop in the mixture fails the other crop takes over, and which also draws moisture and nutrients from different parts of the soil profile; water-harvesting; land-type diversification, in which the farmer seeks at least a small amount of lower-risk land; delayed fertilizer application to reduce weed competition and to avoid waste if the crop should die; low plant densities; planting larger areas than can ultimately be weeded to maximize yield in the good year while minimizing in the bad year when parts of the crop fail, and, finally, of course, storage of substantial quantities of grain from previous harvests.

535 In designing research and assistance strategies for dryland areas it is important to understand that yield stability is not necessarily the same thing as income stability. Income is essentially yield multiplied by price less costs. In a very dry year, when yield falls, food grain prices on local markets may rise to very high levels, particularly if imports are not getting in. A farm family with surplus to their consumption needs can exploit these high prices. However, where a family is close to subsistence, the opportunity cost of the food they retain that year for the family, probably everything they grew, goes up also, so they lose in their costs of family subsistence what the price rise might have given them in terms of greater potential net farm income. The converse is also a problem in dryland areas. The extreme variability of rainfall often means tremendous gluts and very low prices during a good year, especially where transport is not functioning well. Such a year of low prices may be the only year when a poor family have a grain surplus to exchange for their cash needs. Although risk-avoiding cultivation strategies may be more important at the drier end of the rainfall range, there are severe limitations in the availability of adaptive farming strategies in the very dry areas. Matlon (1990) found that the degree of crop diversification was actually lowest in the very dry Sahel zone simply because of a lack of options and it increased directly with zonal production potential. As rainfall increases, more and more flexibility in the production system becomes technically possible. With respect to adaptive strategies after the failure of harvest, which can been termed ex post strategies, i.e. strategies to help after the cropping activities have failed, spatial and activity mobility are the main responses. With spatial mobility, this is usually temporary movement of some family members to other areas, sometimes along with livestock. (Knowledge of this type of response should prompt the policy-maker to ask whether there is anything that Government might do to help such temporary movement, through perhaps employment information or investments in receiving locations, tempered, however, by an assessment of the environmental risks of altered population densities in both the departure and receiving areas.) With respect to activity mobility, family members often switch to other income-earning activities. For example, Mortimore (1989) found mat-making to be a major activity in times of drought in northern Nigeria. He found the productive systems there surprisingly resilient. Somehow, through the droughts of thirteen years, the village of Dagaceri, which he studied, survived. While manuring was reduced, artificial fertilizer partly compensated; while farmers lost the groundnut market they gained a fast-maturing cowpea; while they lost their ploughs, which they were forced to sell to make money for more immediate consumption needs, they gained a new appropriate technology, the long-handled cutting hoe; while they lost most of their animals for transport, they gained motor transport; and while they lost access to cheap food and fuel, due partly to increased connections with the outside world, they gained the benefits in terms of increased markets in the outside world which came with those improved

536 connections. A major focus of government policy must be at least to avoid hindering people's adaptive strategies and, if possible, to actively support them. In looking at priority locations for assistance in times of drought, perhaps to initiate food-for-work programmes to improve infrastructure while ensuring food security, the location of most need may not necessarily be the location of greatest crop failure. In the 1984 drought in the Sahel it was found that despite the fact that the Sudan zone and the drier Sahel zone exhibited identical production failure in terms of the percentage of minimum family energy requirements harvested (only 29 per cent of requirements), consumption in the Sahel households actually exceeded requirements, whereas consumption in the Sudan-zone households was 18 per cent below requirements. Households in the former zone had more convertible assets and other adaptive strategies to fall back on and thus proved more resilient than those in the less dry zone.

20.6 DO THE TECHNOLOGIES EXIST TO IMPROVE PRODUCTION OR REDUCE DEGRADATION? The common view is that there are plenty of improved technologies for dryland areas sitting on the shelf waiting to be taken off and applied. But is this really true? A viable technology can be defined as being ‘a changed farming practice which is perceived by a significant number of farmers as being profitable’. Often, the examples quoted of successes in dryland technology are examples of one successful farmer or one successful village. Technologies need to be replicable with respect to the level of extension investment, with respect to the volume the market for the product can absorb, and the level of subsidy, if any. Many apparently successful programmes have had to pay farmers subsidies to participate, for example to plant acacia albida trees in Chad (Kirmse and Norton, 1984) or to dig water-harvesting structures in Kenya. While subsidies may sometimes have a place to catalyse the initial adoption of a new technology, success in those circumstances is not evidence of replicability. The problem with subsidies for a predominantly agricultural economy is that the majority cannot very well subsidize itself. The challenge in offering new technology is getting the timing right with respect to the stage of evolution of the farming system. Rural sectors typically pass through stages in which, as population builds, intensification becomes increasingly necessary. Indeed, it has been argued that it is population growth itself that drives this push for technologies (Boserup, 1965). The land/labour ratio is one important ratio to calculate in projecting responses. For example, high level of labour availability relative to land results, in dry areas, in such high labour-demanding practices as water-harvesting. However, within the economy, alongside the agricultural intensification will usually be running, in parallel, an industrial and services sector expansion and these other sectors of

537 the economy can be expected to grow relative to agriculture. At some point this is likely to lead to rural outmigration, and labour-intensive practices in rural areas can be expected to be abandoned, perhaps replaced by mechanization. Such outmigration may occur first in the less hospitable, more risky, dryland areas. Thus, what is technically appropriate as an innovation due to the land/labour ratio at one point in the farm-system evolution may be quite inappropriate to another. Another important ratio relevant to the timing issue is the labour/capital ratio. This is most simply expressed as the ratio between the cost of a day's wages and the cost of a litre of fuel, the latter being a proxy for mechanization costs. It should not be assumed that a technology that ‘works’ in a technical sense in a rich country will be appropriate in a poor country with a similar dryland environment. Examples of promising machinery used in the drylands of Australia, USA, or Russia, for example pitters and pitter-seeders, which make intermittent troughs or indentations to concentrate rainfall for forage establishment, are not necessarily appropriate just because they are technically efficient. The ratio between the cost of a day's labour and the cost of a litre of fuel is vastly different between the USA and, say, Somalia. In the former it is about 200 : 1, in Somalia it is about 2 : 1. It would be surprising if the same technology was optimal in both. An important research implication arising from the greater uncertainties of dryland areas is that farmers should be given a much wider range of technological options to choose from. In dryland research the conditions under which the researcher did his experiments may not even recur for some years, although the variability of rainfall across scattered off-station experimental sites may partly substitute for the variability of average rainfall between years. Chambers (1988), in advocating this menu approach, differentiates between what he calls the new ‘Toyotaist’ approach (cars offered with many different options and colours) as opposed to the old ‘Fordist’ approach (Henry Ford offered the consumer one type of car and any colour he wanted as long as it was black). Such a menu approach can be fostered partly through community participation in proposing village-level research and development strategies, perhaps by using the increasingly well-researched participatory procedures of rapid rural appraisal (McCracken, Pretty, and Conway, 1988) which involve structured but flexible data collection and discussions of development options between villagers and technical and community development staff—a form of controlled, group brain-storming which has been found quite productive.

20.7 ECONOMIC FACTORS AND POLICY DISTORTIONS LEADING TO DEGRADATION Behind observed processes of land degradation there will be an array of economic factors or policy distortions that dictates the behaviour of individuals

538 and which may appear to lead to poor land management. While it may not be easy to alter these, a first step is to understand them. Policies can really only be blamed as the ‘cause’ of degradation to the extent that some feasible policy alternative would be available to reverse it. To give the issues more immediacy, this section describes a particular situation of dryland degradation which exhibits many of the most common policy issues. The example is taken from Sudan and is mainly related to the mechanized farming sector. There are a number of complex efficiency, equity, and sustainability conflicts in this case which require more careful analysis. But what appear to be the main problems are these. The mechanized farmer in Sudan over the last forty or so years has been given a strong incentive to practise what are becoming increasingly unsustainable techniques of land management. He did so for the following reasons: 1. The extensive opening up of new land, which is increasingly competing with the needs of small farmers and pastoralists, has given higher returns to the mechanized farmer than more intensive and sustainable use of the existing land in spite of what seem to be considerable externality costs to society (both present and future generations) in terms of lost fuelwood and forage. These costs are not borne by the mechanized farmer himself, but by society as a whole. Quite strict leasehold conditions do, in fact, exist to protect society but they have not been adequately enforced. The leasehold fee has been, for many years, less than $1 per ha., which is below what land changes hands at between farmers. Therefore the Government has lost a great deal of the ‘economic rent’ that could have been put to improving services and policing the legislation. Farmers, for their part, are disinclined to pay higher leasehold fees because they represent a strong lobby and they argue that government does not provide adequate services in these remote farming areas. 2. Twenty-five-year leases, once expired, are now only renewed annually; there is therefore no incentive to husband land carefully, since the farmer cannot be sure how long his lease will continue to be renewed. A short lease may give Governments a greater sense of control, but they leave no incentive to farmers to husband land for a long-term objective. 3. The fact that land is often opened up illegally means that there is an incentive to get rid of trees rapidly by wasteful burning rather than to call in charcoalers to use the wood productively, which would be a slower process and thus increase the chances of being caught. Converting the wood to charcoal would at least remove pressure from other areas facing degradation through supplying fuel to urban markets. 4. In cases where land has been obtained for mechanized farming from a local leader (sheikh) (who may or may not adequately represent the interests of the local community), prevention of tree or bush regeneration by the farmer may be essential to prevent the local community from re-establishing its rights to the land, even if the land is exhausted and efficiency dictates the need for a

539

5.

6.

7.

8. 9.

10.

bush-fallow land recovery period. Thus, if at all, only the very shortest of fallows can be risked by a mechanized farmer where rights over the land are felt to be fragile. The incentive to ‘grab’ further new land has been raised by a recent government decision to legalize previously illegally cultivated land. This results in still further competition with pastoralists. Rangeland taken for mechanized farming would have been grazed by pastoralists following reasonably sustainable traditional practices. Some pastoralists would be as happy to have their stock graze a crop residue as the poor natural bushland which it replaced, but pastoralists’ stock-movement routes have become increasingly blocked by the growing crops during the cropping season. Disputes are common. Large-scale mechanized farmers receive a tractor fuel allocation at a very favourable exchange rate related to the area cultivated. It is often profitable, therefore, for the farmer to continue to make even totally exhausted land appear, to the inspector, to be cultivated. This calls for it to be free of any regeneration of soil-improving shrubs or trees in order to claim the fuel allocation. If the fuel is not used for crop cultivation, it can be sold on the unofficial market at several times its official price. Technical innovations being offered by the extension service for intensification, such as fertilizer, improved seed drills, intermittent deep-soil ripping, etc., may be appropriate if there were less incentive for the extensive cultivation option, but they are of less interest in the present circumstances where, because of the policy environment, extensive practices are still profitable. Other related policies have also impinged. For example, for some years improved equipment could not be imported because it had to be first tested and approved by the designated authorities. Even if it was eventually approved for importation, foreign-exchange restrictions often still prevailed. Thus, the more intensive and more sustainable technologies being developed, even when seen to be profitable by some farmers, have been difficult to adopt because of lack of availability of equipment. Royalties for fuelwood-cutting had not been increased for many decades. Therefore, there was no longer any reflection of the social opportunity cost in the cutting of fuelwood, for example, for making charcoal for the urban market in Khartoum. (Recently, royalties have been raised.) The main crop of the drier areas, gum arabic (Acacia Senegal), a water-soluble gum which is tapped from trees, has been taxed in the past to excessive levels given the benefits of the tree for improved land management, contributing to soil fertility, reducing wind erosion, and contributing to fuel-wood supplies. (Although the fact that Sudan is a major player in the world market has to be considered in determining optimal national income from the crop and hence optimal taxation policy.) Another dryland smallholder cash crop, hand picked select quality groundnuts, a substantial foreign-exchange earner, has been partly wasted

540 through the perverse effects of exchange-rate distortions. The overvalued exchange rate plus import and foreign-exchange restrictions have resulted in high-quality nuts being crushed, within the country, for oil instead of exporting them for the confectionary trade to earn foreign exchange and using a portion of those foreignexchange earnings to import an equivalent quantity of lower-quality oils as a substitute. In the past environment described above, there are a range of policies that seem to require attention: the leasehold agreements, leasehold enforcement, leasehold fees, fuelwood royalties, empowering of local communities in land and forest management, maintaining or reintroduction of stock-movement routes, removal of distorting differential fuel prices and fuel allocations, coordination of research strategies with anticipated land policy, exchange-rate policy (although this cannot be dictated by only one sector), pricing of gum arabic, etc. The following sections address some of the main generic issues in the quite difficult process of improving dryland policies.

20.8 POLICIES FOR IMPROVING LAND MANAGEMENT IN DRYLAND AREAS Policies are the tools which Governments have at their disposal to influence the way the millions of individuals in the economy utilize the resources of the economy. In reviewing them generically, they can be divided into four types: investment priorities, legislation, land tenure, and prices (including exchange rate, taxes, subsidies, etc.). To utilize policy change as a tool, a Government first of all must have a clear view of the broad national or zonal objectives. Most Governments will include some balance of efficiency, equity, and sustainability as broad objectives. However, since there will always be gainers and losers from any policy change, the relative strengths and the detailed characteristics of these three objectives need to be specified more precisely. Perhaps the most fundamental element in the development of strategies for dry areas is to understand the agroclimatic gradation and the trade-off that this may pose arising from the very different human- and animal-carrying capacities in different zones. Oversimplified, this is often exhibited as gradual change from the higher rainfall, upload, more heavily vegetated areas, down through the intermediate plateau areas, modestly vegetated, to the lower, drier, sparsely vegetated, or even desert, areas. There are many links between such areas, not least of which is the movement of people and livestock through temporary or permanent migration. In many cases in the last century, this appears to have been a movement of people pushed out from higher into lower potential areas due to population pressure. This has resulted in changes over time in land/labour ratios in different zones. There are instances where expected land/ labour ratios have been complicated by human and animal disease factors, for example some recent movements into previously sparsely

541 populated onchocerciasis-freed areas in West Africa or movements in East Africa related to the degree of tsetse fly challenge to domestic livestock. Decisions on investment priorities, legislation, land tenure, and prices may all have an impact across the range of zones present and these impacts may be different in different places. A sound strategy, therefore, cannot be simply single-zone-focused; it must address the linkages across the agroclimatic gradation and include also the urban area.

20.9 INVESTMENT PRIORITIES With respect to the efficiency objective, although the question may be heretical in a chapter on drylands, the most fundamental issue is whether the dryland areas are worth investing in at all. However, it is better rephrased into the question: how much investment should be put into dry areas relative to higher potential areas? One set of indirect evidence on the comparative returns to investment in dry compared to higher potential areas is contained in the 1982 FAO Carrying Capacity Study (FAO, 1982). The data for the warm tropic zones of Africa appears to be representative. It shows that human-carrying capacity expands very rapidly as one moves from the zone with only a 100-day length of growing season carrying less than 0.5 persons per ha. at the assumed intermediate level of input use up to about 6 persons per ha. in the higher potential areas. More significantly, the potential gains in carrying capacity when one moves from low to high levels of technology are very small in the drier areas, but substantial gains of the order of four to eight times the present carrying capacity are projected as possible in the wetter areas from making such technology improvements. While one would need to know the marginal costs of those technologies in each case, it seems unlikely that the dry areas would have as high a return to investment as the wetter areas, given the rainfall variability, the lower absolute yield levels, and the scale advantages for supporting infrastructure investment in areas of higher population density. A similar picture seems to emerge when one looks at rangelands, a picture of rather modest returns to investment. For example, estimates indicate that maximum gains through improved technology in the dry 250-mm to 500-mm rainfall zones of Africa, using legumes and some phosphate, could be expected to be of the order of 600 Mcal of metabolizable energy per ha. whereas gains in the higher potential zones would be of the order of 25,000 Mcal metabolizable energy per ha., about a 40:1 ratio in terms of the potential gain. However, in Burkina Faso, at Yatenga, quite high economic rates of return of about 40 per cent were found in one successful dryland moisture-conservation programme managed by Oxfam, but the levels of investment per unit of land area were very modest compared to projects in higher-potential areas. The appropriate conclusion seems to be not that dry areas should receive no investment at all,

542 but that one should be circumspect and selective about the level of resources one puts into these dry areas. Mellor (1990) and Binswanger (1989) both argue for not putting too much investment into marginal areas. They argue for country-investment strategies that predominantly go for growth as the route to dealing with poverty on an adequate scale, with the main emphasis in agriculture on the higher-potential areas. In a number of predominantly dry countries, there would be few if any high-potential rain fed areas, but always there will be choices between relatively higher- and lower-potential zones and the economics of the alternatives needs to be understood, provided that the ultimate decision is also tempered with consideration of the equity impact of the strategy. It has sometimes been argued that the dry areas of the developing world, particularly in Africa, have been neglected by the development community. While it is difficult to separate out investments by zone, a casual scanning of the figures for total external aid per capita for African countries with predominantly low rainfall compared with African countries with predominantly high rainfall does not show any obvious bias away from dry countries. Indeed, if there is any bias evident in the figures of recent years, it seems to go somewhat in favour of the drier countries. Whatever the strategy indicated by efficiency objectives, there would always be the need to investigate the case for an equity justification, perhaps for selected targeted investments aimed at poverty alleviation in the dry areas. Here there is a common myth that the drier the area, the poorer the people. But as noted earlier, in Burkina Faso, there is evidence that it is in the intermediate, more humid, Sudanian zone that the greatest food insecurity exists. Identifying who are the real poor is therefore important and may not be clearly indicated by climatic data. With respect to sustainability, the policy-maker must appreciate that environmental impacts often have far-reaching effects and that what investment is made or not made in a dry zone may have an impact elsewhere. For example, a decision not to invest in more than the minimum services in the dry zones of a country may lead to outmigration to higher-potential areas. Whether this is desirable or not from an economic point of view will be a matter for locationspecific analysis looking at the costs of adjustment for society as people move and also at the likely land-management impacts in the departure and the receiving areas. In West Africa, now that onchocerciasis (river-blindness) is being controlled, substantial areas of high-potential land are becoming more habitable. Should a substantial movement into those areas be encouraged to relieve the pressure on the drier areas? What will this do to the departure and receiving areas with respect to long-term environmental impact? Frequently, even when the balance of efficiency, equity, and sustainability objectives has been fairly clearly established politically, what are offered by the responsible ministries as public investment ‘strategies’ for improved dryland management are better described as shopping lists. Often the total cost of the

543 proposed programme is way beyond any feasible level of sustainable sectoral expenditure. Experience suggests that usually the priority should be to concentrate mainly on the catalytic investments needed to encourage millions of small farmers to do such things as planting trees and managing natural woodland regeneration themselves. Government cannot afford to grow trees for people through the usual high-cost plantation approach on any significant scale and, even if it could, it would have to tax the consumer to find the money to do it. In other words, the strategy should focus predominantly on defining the means rather than the end. A nation-wide target-based strategy indicating, for example, numbers of trees to be planted by zone and by year is generally indicative of a strategy not fully thought through. In this respect local NGOs can be extremely valuable as motivators and organizers to develop local initiatives without relying on costly and not very efficient government machinery. Increasingly the evaluation of priorities in public-investment programmes will have to incorporate an estimate of changes in land values to reflect changes in the productive capacity of land. For example, the benefits of a project to rehabilitate land should reflect not only the direct gains in productivity during the project period but, to the extent that these are not reflected in the direct productivity changes, the changed value of the land asset. This should be not just with respect to the land within the project area, but with respect to other resources that may be impacted by these changes, for example damaging sedimentation to a dam lower in the catchment. At present, adopting this approach presents problems because, unless all projects in a country are analysed in the same way, one cannot make the comparisons which one should be making in project selection.

20.10 LEGISLATION In many countries facing dryland degradation, legislation will need review. There are four particular areas of legislation which are frequently relevant: land tenure, institutional legislation, social-organization legislation, and forest codes. Land tenure is dealt with separately in the next section. Institutional legislation will often need review to clarify national institutional responsibilities or to give some degree of autonomy to improve the ability of an institution to manage drylands. For example, there may be a case for establishing a co-ordinating body for environmental management and for imposing certain rules or responsibilities on other institutions with respect to clearance and reporting. But generally this will be a nation-wide issue not dryland-specific. Social organizations’ legislation has been found to be quite inhibiting in some countries making it difficult, or even illegal, for groups of people to get together to manage their nearby resources. Frequently, as groups of people

544 start to plan the management of, say, a common-property forest area, they realize the need to raise funds, essentially to tax themselves or outsiders, as users of the resource, or to impose local by-law penalties on misuse. However, in many countries they find that it is illegal for them, as a village unit, to collect taxes, or impose rules on themselves. Forest codes and gazetted royalty levels are frequently many decades out of date and may require complete revision, if only to collect the same real level of royalty that was collected when they were established. However, usually it will not simply be a matter of catching up with inflation because the marginal social opportunity cost of felling a tree will have changed with the reduced stock of trees and, furthermore, an imaginative policy might now attempt to have differential royalty rates for areas of different degrees of fragility. Perhaps the most pervasive and damaging aspect of forest codes is the frequent presence of regulations that make it illegal for a farmer to cut down any tree, even one he planted himself, thus destroying any incentive that might have existed to plant trees. A number of countries are now in the process of removing such rules from the books, but often persistent traditions of plantation forestry management within forest departments, as well as the real practical problem of differentiating between the origin of wood found on a person, hinder such changes.

20.11 LAND TENURE Land-tenure systems may need adjustment to improve land management, but before any decisions on such adjustments are made it is essential to understand in detail the system that exists. The following are the three most frequent questions facing policy-makers dealing with land tenure in dryland areas: (a) Will more secure tenure given to pastoral groups improve land-management practices? (b) Is security of tenure in dryland cropping areas preventing a longer-term, socially more sustainable, attitude to land management—for example, does it prevent such investments as tree planting? (c) If forest/woodland ownership is transferred to local groups will the woodland be better managed than under government management or will it be managed as well and at lower cost? These three questions are discussed in the following paragraphs. Pastoral areas in many countries have been facing a land ‘grab’ of unprecedented proportions over the last two decades. Quite apart from the encroachment of cropping, which may well be an economic necessity in a land-short country, often either outsiders, or powerful individuals within the pastoral groups, are securing private freehold to what previously was held under some form of common-property arrangement. How much should Governments intervene in what, in the long run, may be an inevitable process as land scarcity bites? Almost certainly, on equity grounds, and probably on

545 both sustainability and efficiency grounds, they should intervene quite strongly, although this needs location-specific study. Pastoralists did, at one time, have reasonably adequate, although probably not entirely environmentally benign, systems of land management which have been shown to compare more than favourably with respect to efficiency with their beef-producing ranch counterparts, mainly because their systems utilize milk and other products. Outside incursion is now threatening pastoral systems with extinction. Privatization of ranch holdings is usually highly inequitable because minimum-sized holdings for a target income, or minimum biological units which incorporate a sufficient spread of wet- and dry-season grazing area, result in land allocations well above the average land available per family. Thus, under that type of land distribution, significant numbers of people become dispossessed entirely, or are forced to put increased pressure on a diminishing area of common land, exacerbating the breakdown of traditional rules of land use. What is needed, to ensure local commitment to sustainable management, is to enable a defined body of people to have rights over a defined body of land so that some control can be reasserted by the group on itself over whose animals eat what forage and at what time of the year. Historically, pastoral groups in Africa, prior to the recent advances in the control of human and animal diseases, managed rangelands quite well; if they had not they would no longer exist. These systems, now under increased pressure, are highly complex. For example, in Turkana pastoral areas in Kenya, while most of the grazing areas are common property, managed under traditional rules which, to a considerable extent, control misuse, well-established individual family land-tenure rights exist over certain higher-potential riverain areas where dry-season feed is available from the pods of Acacia tortilis. Valuable trees are rarely cut. However, while limited individual land-tenure rights exist, trees identified as being ‘important’ require permission from the elders to be cut. There is an extensive knowledge of the value of different species. Particular types of utilization have been recorded for 222 species (Barrow, 1988). Another example of complex ownership patterns can be found in Sudan. In gum arabic (Acacia Senegal) areas, herders browse and graze the forage on the land, farmers collect dead wood for fuel, merchants may purchase gum-collection rights from tree-owners, while the land itself may be communally owned by settled agriculturists. It is often difficult to improve on such complex but pragmatically evolved land-tenure systems. In dryland cropping areas it has been argued (IFAD Report, 1986) that farmer's short-time preference (i.e. his inability to look more than a few years ahead due to the pressure of poverty and other factors) has tended to override any lack of tenure security. It has also been argued that, in any case, traditional usufruct rights have been quite secure. These are questions amenable to research before any precipitate action is taken to change the tenure system. Changing the tenure systems can be an extremely costly exercise if full cadastral surveys etc. are required and it may have unpredicted effects.

546 In forested areas policy-makers are faced with a dilemma. On the one hand, they may see the advantages in handing over forest reserve land for management by local communities. On the other hand, the experience is not by any means all good, and there may be overriding environmental issues to be considered, such as the loss of a unique forest animal if the trees should be lost through even a temporary breakdown of management. A major problem with management by local groups is the wide range of interests of different users. For example, pastoralists may want to lop trees for fodder, but, for half the year they may not even be physically present in the area to agree on any rules of utilization or to defend their interests; cropping people may be interested in fuelwood; hunters will be interested in game which may compete with livestock and they may call for the frequent use of fire; and the landless will want to encroach on the woodland or at least get rights to a ‘taungya’-type of system in which they can crop newly felled areas for a year or two before trees are replanted or regenerate. Achieving sustainable management by local people in such circumstances is extremely difficult, particularly where there may even be different tribal or caste affiliations. It is essentially an exercise in conflict resolution and a number of projects around the world are trying to develop conflict-resolution procedures to deal with this type of increasingly common situation. In conclusion, it is worth recalling that, while economic theory might be read as suggesting that individual tenure would result in better land management, the evidence, while possibly tending in that direction, is by no means conclusive. There have been plenty of examples of the misuse of the land under all systems of tenure (Dixon, James, and Sherman 1989).

20.12 PRICE INCENTIVES Pricing is the traditional weapon of the economist and it is tempting to see ‘getting the prices right’ as the panacea for all dryland problems. However, in real-world cases, it is not always easy to either implement price, subsidy, or tax changes to reflect better the costs of an action to society, or to forecast the results. Indeed, the evidence from research so far suggests that the influence of price changes on land management often cannot even be given a sign let alone a magnitude. Given the limited state of present knowledge, price and subsidy interventions in dryland areas should certainly be seen as potentially useful instruments, but still something of a loose cannon in practice. It is rather rare to find a case where the outcome seems likely to be unambiguous, or wholly self-evident from economic theory. The following are seven of the most important issues to understand in the development of price-related policies. The first issue is the well-known ‘common-property’ issue (Hanak, 1987). The costs to society of cutting down a tree impinge only insignificantly on the

547 cutter because they are costs, such as reduced seed stock and increased erosion, which are carried by the whole group. The same issue arises with common-property grazing. The main strategy here usually will be to encourage the reestablishment of land rights and group rules which support the imposition of social pressures on antisocial behaviour so that costs are indeed brought to bear on the individual. The second issue is the technical interrelationships aspect, that one tree alone does not do much to hold soil or reduce wind erosion or crop desiccation. Significant benefits can only be achieved if people agree to protect or plant trees in substantial numbers, and even then, perhaps, only if the trees are set out in a certain pattern. In both these last two cases the individual responding to his own private incentives is likely to undervalue the existence of the tree or the forage relative to its benefits to society as a whole. In economic terms the marginal private cost curve lies below the marginal social cost curve. In theory, at the demand end, the gap can be closed to equate the private with the social cost by either a tax or a quantity limitation on the cutting of the tree or, at least partially, by changing to private ownership so that more of the costs, if not all of them, are felt by the individual. At the supply end, to encourage the planting of trees, as opposed to discouraging the removal of trees, such a tax would, of course, become a subsidy. The economic principles are the same. The problem with the direction that the theory points to is that, in practice, the imposition and management of taxes, controls, or subsidies in remote rural areas is very difficult. For example, taxing cutting at source requires large numbers of field staff in remote areas. However, such ‘at source’ royalty collection is an avenue that needs to be explored and if it can become community managed may not be too costly. The third issue is the high discounting of future benefits usually observed in the survival strategies of poor farm households. Often their decisions appear to be not inconsistent with discount rates on future earnings as high as 50 per cent (although there is considerable debate about what level of discounting is exhibited and debate also on the methodology of measurement). Thus they will tend to mine soil for this year's crop because later-crop income is given substantially less weighting. This is by no means irrational given the high risks of the dry areas and the urgent immediate needs of the typical rural family; better to survive this year and let next year take care of itself. Theory suggests that a wellfunctioning credit market should help to reduce this high rate of discounting of future benefits, but given the problems of collateral, the remoteness and the riskiness of these dry areas, providing publicly subsidized credit is seldom an efficient solution. The fourth issue is that price adjustment is a rather blunt instrument. It has impacts that are usually nation-wide, if not wider, through cross-border trade. For example, it as been suggested that in some dryland areas a phosphate subsidy would be environmentally positive because, in the absence of proven sustainable farming systems, with a phosphate subsidy one might at least be able

548 to maintain some modest level of soil fertility, and hence protective plant ground-cover, until such time as technologies are found to support the expanding population on the poor land-base. However, the problem with such a proposal, even if it were efficient for the dryland zone, is that subsidized phosphate will find its way everywhere, not simply within the target dryland area, but to other areas in the country where it may not be needed and even to neighbouring countries, especially where exchange-rate imbalances at the border exist. The fifth issue with respect to adjusting price incentives is that in most cases there are counteracting economic impacts, the net impact of which is not easy to predict. For example, it has often been argued that higher output prices for crops or livestock will raise the incentive to manage land sustainably by raising its value and thus indirectly raising the value of any improvements to maintain its productive capacity. But there is another counteracting effect, particularly relevant to the dry areas. In the case of Sudan, for example, higher sorghum prices would not only raise the value of land and the marginal return to a given level of soil-conservation treatment, it would also raise the incentive for the mechanized farmers to move out into the yet more marginal areas, which might have a number of damaging environmental effects given the fragility of such areas. Higher sorghum prices would also bring forward the date at which it would be profitable for a farmer to bring his land out of the fallow phase back into cultivation, raising the risk of soil erosion that might have its most damaging effect outside his particular farm (an externality). The sixth issue is that future expectations come strongly into play when a producer reacts to a price change, especially in countries with a history of unstable pricing policies. The producer is likely to think as follows: How long will this better price last? Would it not be wise, given what has happened to prices in the past, to rapidly expand my cropped area and immediately bring back barely recovered, and therefore erosion-prone, land into cultivation to make the most of this windfall price increase, especially since in these dry areas this year could be the last good rainfall for five years? After all, even from the environmental point of view, the evidence in dry areas, with their highly variable, but occasionally very heavy, rainfall, is that really serious erosion occurs only intermittently through a very limited number of extreme events. In some locations four massive events over twenty-five years have accounted for 75 per cent of the erosion. Even the farmer with a strong incentive to show concern for the long-term health of his land might be forgiven for asking whether it is not worth the risk to take the short-term view and farm to the limit while the price lasts. The expectation that a relative price improvement is permanent only comes slowly, particularly in countries with a history of perverse and volatile price policies. The seventh issue is the livestock supply-response problem. Livestock may exhibit even more complex and difficult to predict price responses than crops. With respect to short-term prices it is known that higher livestock prices can

549 sometimes give an initially negative supply response, although there is conflicting evidence on the existence of this phenomenon in dryland pastoral situations in Africa. This negative response—animals not regarded as of great value being marketed the higher the price—is the effect caused by livestock owners initially holding back animals from sale to build up herds to take more advantage subsequently of the raised prices. In dryland situations much will depend on the state of the season as to how the owners will react and, again, will depend on future price expectations. However, their longer-term response can be expected to be more normal. Therefore, the commonly voiced proposal that, in order to encourage increased livestock off-take rates to lower the pressure on overgrazed rangelands, one should raise livestock prices, is completely contrary to what economics would predict, whatever may be the perverse shorter-term responses. In the long term one would expect a higher price to put increased stocking pressure on the rangelands, although the extent of this will depend on the type of landownership and, in communal systems, on the extent to which group pressures for sustainable rangeland-use function. In conclusion, while this section has tried to give some pointers, it is risky to make a priori assumptions about likely responses to price changes, including taxes, subsidies, exchange-rate changes, etc., without rather careful modelling of the particular situation and without tracing the ‘knock-on’ effects at both a technical and economic level. There are few, if any, rules of thumb.

20.13 INSTITUTIONAL AND RESEARCH ISSUES Many of the strategies touched on above for dealing with the highly uncertain and inhospitable environment of dryland areas require a somewhat different focus to those for dealing with more predictable higher-potential areas. Certain research initiatives may be needed. Furthermore, consistent with a focus on the means rather than the end, the first step towards a rational drylands strategy may be to adjust the type, direction, training, and staffing of the institutions involved. The following are twelve high-priority institutional and research needs that arise from the discussion in this chapter. 1. The central planning and finance ministry needs to address the relative economics of alternative patterns of investment in the different agroclimatic zones and to establish some order of magnitude for what budgetary allocation can be justified over the long term in each of these areas. 2. A national monitoring capacity for the monitoring of changes in productive capacity and resilience may need to be established or strengthened. This should not have costs way out of proportion to the likely benefits, therefore remote sensing with judicious verification at ground level through indicator species (species which readily indicate changed productive potential or resilience) is likely to be the core of the technique. Methodologies for this are

550

3.

4.

5.

6. 7.

being tested in Kenya and elsewhere. Since the need is to monitor over long periods (at least twenty to thirty years) a financially sustainable and efficient level of investment must be sought. This work should be coordinated with economic studies to attempt to get approximate estimates of the costs to the country of degradation. For example, quite simple techniques, combining remote sensing with zonal farm-budget types of analysis, can give indications of the costs to a country of soil loss over time and, even if inaccurate in an absolute sense, can indicate relative priority areas to focus on (Magrath, 1989). A co-ordinating institution, such as a department of environment, may be necessary given the broad range of issues that need to be addressed in dealing with development and natural-resource sustainability. However, the planner should be aware that co-ordination has a cost as well as a benefit. For specific tasks, such as soil conservation, a lead agency may need to be appointed, although there are some situations where practitioners in the field have found advantages from initially having a measure of competition between agencies. Close links with NGOs will often ensure much-improved participation by local people. Training programmes for staff need to be established which provide guidance on: rapid rural appraisal and other participatory techniques of local development and also on new low-cost and low-risk technologies that draw not only from research but from local experience (it will often be necessary to have local farmers and pastoralists involved in the development of the training to inject this local experience; without this it is easy for the academically trained agriculturist to be far off-track with appropriate and acceptable technologies). An institution with experience on land tenure may need to investigate the existing land-tenure situation and develop a long-term strategy for each area, particularly to establish an efficient, equitable, and sustainable future land-tenure strategy for rangeland areas. This may be urgent in some countries to prevent the foreclosing of options which may be happening very fast. Studies related to such things as forest codes, farm leasehold conditions and enforcement, and social organization legislation (rules for local tax-raising etc.) often will need to review what legislation exists and propose changes that enhance the likelihood of sustainable management of dryland areas. Research will often need to be re-focused, The following are some of the highest priorities: (a) a general shift towards low-cost, high immediate payoff, risk-reducing technologies; (b) offering farmers and pastoralists a much wider menu of choices so that they themselves can make judgements about what suits their particular agroclimatic and family circumstances; (c) for some countries, a small amount of research on the longer-term, more sophisticated technological directions such as the use of dryland areas as solar- or wind-energy power sources, mariculture using seawater for fish production associated with use of the effluent on salt-tolerant crops (there are about 30,000 km of coastline around the world bordering arid areas), possibly harnessing

551 photosynthesis in phyto-plankton types of system based on genetically engineered organisms, etc.; (d) research on land-management technologies using low-cost vegetative techniques such as the planting of Vetiveria zizanioides (in India, khus grass), or alternative, more drought-tolerant species as a vegetative barrier on the contour to replace costly and ineffective earth-bound technology, and, second, comparative studies of the effectiveness and productivity of alternate crop, pasture, tree, and shrub land-cover in dryland areas. (The advantage of forest cover in higher-potential areas for reducing erosion and improving infiltration appears to derive almost entirely from the understory of herbs, shrubs, and litter, yet in some dryland areas trees compete strongly with the understory due to the limited moisture availability); (e) research on the nutrient and energy balance of high population-density dry areas to understand how sustainable they may be, given the inflow and outflow; (f) climate research, for example, on ‘response farming’ (Stewart, 1990), a technique by which farmers are given forecasts on the basis of early rains of the likely type of season ahead of them and react accordingly in terms of cultivar, planting dates, etc.; (g) research on dryland trees for a range of uses, on the cloning and provenance testing of superior tree individuals, and on tree-associated symbiotic root micro-organisms (Gorse and Steeds, 1987), and also a need for adaptive research on leguminous trees for forage purposes; (h) continued work on dryland cultivars with high yields under adverse conditions, and, hopefully, with the advent of biotechnology, the development, first, of food-grain cultivars that are more drought-resistant, moving beyond the present improved cultivars which have mainly been drought-evading (by simply maturing quickly to avoid running into the dry period) and, second, cultivars that have better efficiency and produce more energy with the same nutrient availability (so far plant-breeding has not made significant improvements in efficiency—the cultivars have simply had the potential to exploit higher levels of nutrients); (i) research on low-cost seeddispersal techniques such as wadi-head planting (planting at the top of watercourses) to get widespread and efficient seed dispersal by flood water down basins at low cost (NRC, 1984); and, (j) research with a range of domestic and wild species on animal impact in relation to plant growth and soil conditions (Savory, 1988) and, within this work, more attention to game-ranching/game-cropping and possibly the use of some dryland areas for game-hunting. Interesting results are starting to emerge on the relative impacts of domestic stock and wildlife on plant biodiversity with indications that, with sensible management of domestic stock, a mixture of domestic and wildlife species may be best for biodiversity. 8. A review of the historical and expected future demographics of the dryland areas will be needed in many cases to get a clear picture of where the present problems came from, to the extent that they have a population link at all, and what direction they are likely to take in the future. Some countries have been taken by surprise by a rural outmigration from dryland areas

552

9.

10.

11. 12.

calling for a whole new set of strategies. The use of historians may be important for understanding the longerterm trends in relation to society. A migration-support programme may warrant consideration, coordinated through a number of institutions and NGOs, to support the temporary migration to alternative income sources which has enabled so many people in these areas to be resilient and to adapt to the ravages of drought. Usually improving such flexibility will be not only good for the beneficiary but good for the environment. A programme of development orientated (i. e. offering recommendations for action) socio-economic studies should be supported, perhaps through the universities, fully to understand the problems of people in dryland areas, their adaptations to drought, the true underlying causes of the situation (i.e. beyond the simplistic and misleading ‘loss-of-trees’ type of diagnosis), the locations of the areas most severely affected by degradation, the linkages to other sectors, the problems faced by women, and particularly any differential impact on women that may be arising from de facto or de jure changes in land tenure, and the emerging types of land and natural-resource conflict at the village level. A study of the dryland areas of particular value with respect to bio-adversity may be needed. Institutions dealing with household energy will need to play a key role in extending technologies to improve the efficiency of energy use through improved stoves etc. in order to reduce fuelwood demand, since, as noted earlier, the human-carrying capacity of dry areas is generally constrained more by fuelwood supplies than by food supplies.

20.14 CONCLUSION This chapter has raised a number of difficult issues faced by decision-makers dealing with dryland areas and with problems of ‘desertification’. It has offered few clear solutions because the situations are complex and quite locationspecific. Higher risk is the main difference from higher-potential areas. Perhaps the single most important conclusion is that policy, legislation, and investment decisions made for dryland areas cannot be seen in isolation from the agroclimatic zoning and demographic characteristics of the country as a whole, nor can they be seen in isolation from the historic and projected future trends of rural and urban development.

553

REFERENCES BARROW, EDMUND G. C. (1988), Trees, People and the Dry Lands: The Role of Local Knowledge, Paper presented to the Second Kenya National Seminar on Agro-forestry, Nairobi, Kenya. BINSWANGER, HANS (1989), The Policy Response of Agriculture, Proceedings of the Annual World Bank Conference on Development Economics 1989, Supplement to World Bank Economic Review and World Bank Research Observer. BOSERUP, ESTER (1965), The Conditions of Agriculture Growth (London: Allen & Un win). CHAMBERS, ROBERT (1989), A New Administration: Beyond Fordism and the Self-deceiving State, Paper for the IDS Retreat, 11–12 Dec. 1989, Institute of Development Studies, University of Sussex. DIXON, JOHN, DAVID A. JAMES, and PAUL B. SHERMAN (1989), The Economics of Dryland Management (London: Earthscan Publications). DREGNE, H. E., and C. J. TUCKER (1988), ‘Desert Encroachment’, Desertification Control Bulletin, 16.(Nairobi: UNEP). FAO (1982), Potential Population Supporting Capacity of Lands in the Developing World, FAO Technical Report FPA/INT/ 513 (Rome: FAO). GORSE, JEAN E., and DAVID R. STEEDS (1987), Desertification in the Sahelian and Sudanian Zones of West Africa, Technical Paper, 61 (Washington, DC: World Bank). HANAK, ELLEN E. (1987), The Economics of Government-sponsored Afforestation Exercises: Implications of Theory in an African Setting, Dept. of Economics, University of Maryland. IFAD (1986), Soil and Water Conservation in Sub-Saharan Africa: Issues and Options, prepared for IFAD by Centre for Development Cooperation Services, Free University, Amsterdam. IUCN (1989), The IUCN Sahel Studies, IUCN, Nairobi, Kenya. KIRMSE, ROBERT D., and BRIAN E. NORTON (1984), ‘The Potential of Acacia Albida for Desertification Control and Increased Productivity in Chad’, Biological Conservation, 29: 121–41. MCCRACKEN, JENNIFER, JULES PRETTY, and GORDON CONWAY (1988), An Introduction to Rapid Rural Appraisal for Agriculture Development, International Institute for Environment and Development, Sustainable Agriculture Programme, London, UK. MALTON, PETER J. (1990), Farmer Risk Management Strategies: The Case of the West African Semi-Arid Tropics, Tenth Annual Agriculture Symposium: Risk in Agriculture, World Bank Symposium, January 1990 (Washington, DC: World Bank). MELLOR, JOHN (1988), Agriculture Development Opportunities for the 1990s: The Role of Research. Address presented at the International Centres Week of CGIAR, Washington DC, 4 Nov. 1988. MORTIMORE, MICHAEL (1989), Adapting to Drought: Farmers, Famines and Desertification in West Africa (Cambridge: Cambridge University Press). NELSON, R. (1990), Dryland Management: The ‘Desertification’ Problem, World Bank Technical Paper, 116 (Washington, DC: World Bank). NEWCOMBE, KENNETH (1984), An Economic Justification for Rural Afforestation: The Case of Ethiopia, Energy Dept. Paper 16 (Washington, DC: World Bank).

554 SANDFORD, STEPHEN (1976), ‘Pastoralism under Pressure’ ODI Review, 2: 45–68. SAVORY, ALAN (1988), Holistic Resource Management (Washington, DC: Island Press). STEWART, IAN J. (1990), Managing Climatic Risk in Agriculture, Tenth Agriculture Symposium: Risk in Agriculture (Washington, DC: World Bank). UNESCO (1979), Map of the World Distribution of Arid Regions, MAB Technical Notes 7 (Paris: UNESCO). WARREN, ANDREW, and CLIVE AGNEW, (1988), An Assessment of Desertification and Land Degradation in Arid and SemiArid Areas, Ecology and Conservation Unit, University College, London.

21 Management of Wildlife and Habitat in Developing Countries 21.1 INTRODUCTION The growing prominence of environmental concern in world affairs is a welcome development, particularly for natural-resource economists who have worked in this area for one or more decades. Many events have prompted this concern. Deforestation of the Amazon is occurring at an alarming pace. Here and elsewhere species are becoming extinct at a rate some believe exceeds one per day. In the USA, concern about the possible extinction of the spotted owl has led to a prohibition of logging hundreds of thousands of acres of old-growth Douglas fir (owl habitat) worth hundreds of millions of dollars. Use of the heat-absorbing capacity of the atmosphere has prompted its decline since the turn of the century. Some scientists and certainly the popular press have concluded that the phenomenon of global warming can no longer be disregarded. The issue is perceived to be serious enough that former Prime Minister Thatcher called a conference on global warming and advocated such a politically novel economic solution as a carbon tax. Domestic and international fishery stocks have declined, some like the blue whale to the point where International Commissions have advocated, not always successfully, the cessation of harvesting. On the land, the precipitous decline of the elephant population in Africa led recently to a ban on the international trade of ivory. What can economics contribute to these issues? We have tried to explain why and in what circumstances the phenomenon of stocks of natural resources diminishing at a precipitous rate is not, by itself, cause for any great professional concern. Natural resources are part of a nation's and the world's endowment of capital stock. Our growth models inform us that when certain forms of capital stock are too plentiful, there is an optimal rate to draw down capital until a steady-state level is reached for renewable natural resources. A dramatic rate of exploitation need not worry economists. Technical and economic conditions may prescribe it inasmuch as the optimal rate of

556 extraction in excess of steady-state levels depends importantly on the excess stock of resource capital, and the marginal rate of profit. Indeed, a constant rate of profit is the condition driving the bang-bang control models of Arrow and Kurz and others, in which an economy tries to approach the steady state by reducing its natural-resource stocks as fast as it can. There are surprisingly few studies of resource-extraction paths but there are two instructive examples. Johnson and Libecap showed that the pace of forestry depletion in the Great Lakes Region of the USA during the four decades around 1900 earned resource-owners only a competitive rate of return. Berck showed that private timber-owners of old growth in the Northwest reduced their capital timber stocks at rates predicted by competitive capital markets. Even extraction is not prima facie evidence of an economic error. When natural-resource capital cannot earn as much as rival capital of equivalent riskiness, at any level of the stock, some of our models prescribe extirpation. Resource exhaustion is not a surprise for those familiar with the non-renewable resource literature, including Hotelling, and for those who, like Ricardo, introduce a quality dimension of natural resources and expect to see the highest-quality resources exhausted first. Resource exhaustion is not a concern for those familiar with the growth literature which includes natural resources, at least those prepared to believe in robust natural-resource saving technology or in benign elasticities of substitution which enable relatively painless substitution out of natural resources and into other factor inputs as the natural endowment grows increasingly scarce. The provision of wildlife and related habitat is optimal only fortuitously for the reasons I spell out in the next paragraphs. Because the sources of ‘market failure’ in the realm of species and habitat management are plentiful, it is natural to expect substantial public involvement which I describe in Section 21.4. Nevertheless, the private sector has important niches to fill and these are discussed in Section 21.5 where particular attention is devoted to the prospects for private provision of non-consumptive wildlife resources in mixed game and cattle enterprises. In Section 21.6 some of the economists’ tools of valuation are applied to parks and wildlife in a developing-economy setting.

21.2 SOURCES OF DEPARTURE FROM OPTIMAL USE OF WILDLIFE AND HABITAT Many natural resources have the misfortune of not being able to become private property in a fashion that the owner readily can capture the benefits obtained from the use and enjoyment of the resource. Fish, fowl, and other species have very little respect for the political boundaries of countries. All too frequently these resources reflect the tragedy of the commons where the lack of proper economic incentives and institutional

557 framework prompts premature and excessive rates of harvest. The mammal I don't harvest today is yours for the taking and not available for me in the future. Second, the habitat of some species is large or complex in a particular jurisdiction making it economically costly both to keep one's property separated from others and to keep out natural and human predators. Third, custom and tradition are largely responsible for the lack of charges or other economic policies to ration use and to capture and indicate resource value to policy-makers. Fish might be but typically aren't subject to landing taxes. Charges to enter parks, reserves, and other types of recreation sites either don't exist or are determined by administrative criteria and rarely bear much resemblance to standard market-pricing principles. When charges don't exist, values necessary for making resource-allocation decisions have to be found by conceptually valid but tedious, fairly expensive non-market valuation schemes such as various forms of travel-cost analysis, hedonic techniques, or discrete-choice (random utility) models. In short, non-market resource-use benefits are difficult to estimate. Many natural resources are unique so valuing them by using values from other circumstances invites great and perhaps grave errors. Fourth, many will treasure the existence of natural resources such as the elephant or aged redwoods even if they never intend to visit them. Value may arise because people are uncertain if they want to use the resource in the future and would pay to keep the option of visiting available. Still others may value knowing that a species will exist so that their children and others can enjoy it or them in the future. By their nature, these non-consumptive values, ably discussed by Krutilla, Weisbrod, and others, cannot be accurately estimated directly or indirectly by behaviour or by changes in behaviour in response to price variations. Resort must be made to what has come to be known as contingent-valuation techniques discussed by Mitchell and Carson and Cummings et al. Fifth, when some resources are ‘used’, they are not consumed and do not deny others the ‘use’ of the resource. This phenomenon of undepletable externalities (Baumol and Oates) is evident in such straightforward activities as viewing wildlife or hiking until congestion sets in but it is particularly pronounced for resources with scientific or genetic value. Since genetic resources are the linchpins in research and development programmes, the literature on market failure in the context of R&D has immediate relevance (see Dasgupta and Stiglitz). Knowledge is a non-depletable public good. When we acquire some new knowledge from nature—a new seed variety, the template for a new drug—it costs virtually nothing to transmit the knowledge to others. The economic problem is evident. If the social marginal cost of genetic information is charged, there will be too little R&D (information gathered). The right to discover particular forms of genetic information can be transferred to a firm, in order to make R&D more profitable. This also can and

558 often does provide the firm with market power. When there is market power, the larger the market, the greater is the production over which the fixed costs of R&D are spread. This is just another way of saying that there are increasing returns to the use of information which makes efficient market solutions more problematical. The resulting social undervaluation of genetic information by all private market structures translates to the undervaluation of, and reduced demand for, the genetic resources in which the information originally is stored. Sixth, there seems to be an unusual amount of technical complexity and interdependence between one natural resource and the rest of the ecosystem. This creates technical externalities with its counterpart implications for economic efficiency. As we know, well-functioning markets handle pecuniary externalities but technical externalities are the classic basis for market failure. When there are many interdependent species of flora and fauna, each with its own biological rhythm dictated by non-parametric birth-rates and death-rates, designing an optimal set of intertemporal actual or shadow prices or rates of harvest is as daunting a task as managing a socialist economy without markets. When wildlife management agencies are given jurisdiction over one or a few species and there are ineffective or no stewards for other interdependent species, the weak management sub-system can be expected to deteriorate and parts may become extinct. This evident ecological complexity gives rise to the view that habitats, not single species are the appropriate management unit. The difficulties of ecological complexity are exacerbated all too frequently by inadequate knowledge. Because many resources are renewable, a mistake today has intertemporal repercussions. Indeed some mistakes are irreversible if they lead to extinction as in the case of the passenger pigeon and sea cow. In these situations, if the arrival of future information will reduce uncertainty about the value of alternative uses of resources, the optimal pace of extraction is not, for most decision-makers, a routine intuitive solution. Optimal policy rules must be derived from techniques such as dynamic programming (Arrow and Fisher). Finally, there is what might be termed a political externality, when a public policy designed to benefit one segment of the economy has repercussions, often unanticipated and negative, on an environmental asset. Subsidies to reduce the cost of preparing land for agricultural use decrease the acreage in wetlands which decreases waterfowl populations and other species dependent on wetlands. Slightly more circumlocutory are personal and corporate income-tax-liability exemptions when equivalent amounts of money are invested in approved development projects. Such a subsidy is responsible for destroying nearly one-third of the rainforest in Brazil's Amazon region (Browder, 1988). Economists have a particularly important role to play in determining the appropriate use of wildlife and their habitats because they have a comparative advantage in addressing the underlying issues discussed above.

559

21.3 USING PROPERTY RIGHTS FOR CONSERVATION Individuals who do not benefit directly or indirectly from the continued viability of wildlife cannot be expected to shape their behaviour in a manner that preserves wildlife, particularly when the preservation options have private opportunity costs. One therefore wants to design a property-rights system that rewards (punishes) those who control the destiny of wildlife and their habitats. If natural-resource assets are to be managed properly, the ‘owner’ must be able to exclude others from taking the resource and must be able to capture the pay-off from sound management practices. Some natural and social circumstances are easier to manage than others. In the case of fisheries, annual catch quotas often have replaced open access. This is a policy improvement because it helps to preserve the species in question but it merely shifts open access from the complete fish stock to the quantity in the annual or seasonal quota. The actual results are predictable. Economic rents from the fishing are dissipated by free entry, the attraction of profit, and a zero price on the natural resource. Not only are there too many entrants but non-optimal technologies are used. Boats are built to have a bigger capacity in order to increase the number of days at sea; larger engines decrease the time off the grounds; sophisticated and duplicate electronic equipment increase accuracy and reduce the chance of losses due to equipment failure at crucial moments, often measured in minutes or hours during openings dramatically limited by the amassing of so much fishing power. Gradually a new policy of individual tradable quotas (which sum to the total quota) is evolving. Thus while the individual harvester does not receive a property right to ‘own’ specific fish in the sea, the harvester can acquire the right to harvest a given amount and then is free to choose the optimal seasonal harvest pattern. Excludability and capturability are key elements in this management policy. Some of the marketable rights for salmon fishing in Alaska sell for more than $300,000. It is hard to imagine a much more controversial topic in natural-resource management then that which surrounds East African elephant and ivory resources. The subject is vast and complex. Discussion here is limited to highlighting a few issues bearing on property rights. In recent years, hundreds of thousands of East African elephants have been destroyed by local and foreign poachers who sell the ivory tusks. This tragedy was made possible by ineffective ownership rights of the elephant. In years past, when hunting was permitted, poaching was not a serious problem at all in Kenya, for example, because it was in the interests of the professional hunters to protect the elephants from poachers. Introducing a ban on big-game hunting removed the private interest in conserving the elephant population. For years, many have been concerned that remarkably little revenue, earned from visitors on safari to game parks and reserves, accrued to the inhabitants around the parks and reserves on whose land the elephants

560 foraged as they migrated. All too often the elephants’ foraging behaviour was competitive with the domestic cattle operations of the Africans. If hotels and camps were located on lands outside the parks and if, in other ways, the economic fortunes of those controlling these contiguous lands were tied directly to the health of the elephant population, then there would be strong economic incentives for the inhabitants contiguous to the parks to dissuade poaching. In recent years new hotels and camps have been located outside Kenya's parks and reserves such as the Masai Mara. Additionally, the Kenya Wildlife Services have made a commitment to distribute up to 25 per cent of annual park revenues to the bordering communities. Clearly the connection between the disbursement of park revenues earned from providing a complex joint product to tourists is more tenuous than specific ownership rights to a multidimensional flow of services from elephants. Nevertheless, the revenue-disbursement policy is an important first step in internalizing the externalities the migratory game cause to the neighbouring villagers and it will contribute to poacher control as well. In contrast to Kenya, some countries such as South Africa and Zimbabwe view ivory as a valuable resource and manage it accordingly. The revenues from ivory sales are an important source of general funds or funds available for game management. Zimbabwe earned $63–78 per kg of ivory harvested from its culling activities to maintain a herd size compatible with available carrying capacity. In contrast, harvest in Zaire earned 10 per cent of this amount in a recent year. The other 90 per cent, equivalent to economic rent, was captured by foreign traders and others (Swanson, 1989). An average harvest of ivory from contemporary elephants is worth about $500 in contrast to a game scout's annual income of $150 per year, according to Swanson. Real earnings from ivory sales can support substantial and effective enforcement activities and leave residual revenues for productive activities. Zimbabwe and South Africa have designed game-management institutions in order to exclude others from capturing the existing rent. This is not an easy task to accomplish. In some countries poaching is successful because key government officials share the profits from poaching in return for protecting poachers from being apprehended. Little progress in designing effective incentive systems and institutional structure can be expected until these disruptive public officials pursuing private gains can be neutralized by enlightenment and discipline from above, if there is an above, or by sufficiently increased exposure from the outside. However, one conclusion is apparent. There will be less collusion by public officials with poachers and less poaching if the property-rights structure permits specific private or non-governmental entities to capture the economic rents and these rights have security of tenure. When the intertemporal returns from a renewable or non-renewable resource are assured, the ‘owners’ of the stream of value have no incentive to behave

561 myopically, holding too few stocks unless, of course, they are encouraged to do so by other market imperfections.103

21.4 PUBLIC PROVISION OF WILDLIFE AND PROTECTED AREAS 104

Most people in North America will think of national and provincial parks when they form mental images about areas reserved for scenic beauty, opportunities for solitude and for engaging in a wide variety of wilderness-related recreation activities. Parks are the source of a substantial fraction of recreation activity on public lands. Natural and provincial parks have been in existence for over 100 years in North America but a preponderance of the new national parks established in recent years are in the developing economies, according to Malik. The International Union for the Conservation of Nature (IUCN) lists seven other kinds of protected areas in addition to parks, reproduced below in Table 21.1. Accessibility of people to the resource increases as one goes down the categories. It is natural to expect a rich assortment of institutional forms and management structures to accommodate the complex set of demands placed on resources providing a variegated array of substitute and complementary goods. The number and area of protected areas throughout the world have been growing quite rapidly in recent years, by more than a factor of five since 1950 (see Figure 21.1). Presently more than 400 million ha. are under protection, more than one-half of the areas are in national parks and a bit under 50 per cent are located in developing economies.

21.5 PRIVATE PROVISION OF WILDLIFE Because it is difficult to capture the value of wildlife from those who enjoy it, public ownership and management of critical habitats and species can be expected to continue and to be dominant throughout the world. However, private ownership can play an important conservation role, should not be overlooked, and might readily form a ninth category in the IUCN typology of protected areas presented earlier. Of the many private forms of stewardship which might be considered I will focus on two which are of particular interest: commercial wild-species rearing and mixed cattle and wild-game ranching.

103

These other imperfections would induce second-best outcomes under any alternative property-rights structure governing ivory unless the property-rights structure itself depended on market imperfections external to ivory.

104

For this section I have drawn liberally from Dixon (1990).

562 Table 21.1 Categories and management objectives of protected areas 1.

2.

3.

4.

5.

6.

7.

8.

Scientific reserve/strict nature reserve. To protect nature and maintain natural processes in an undisturbed state in order to have ecologically representative examples of the natural environment available for scientific study, environmental monitoring and education, and the maintenance of genetic resources in a dynamic and evolutionary state. Examples include the Yala Strict Nature Reserve in Sri Lanka, the island of Barro Colorado in Panama, and the Gombe Stream National Park in Tanzania. National park. To protect relatively large natural and scenic areas of national or international significance for scientific, educational, and recreational use, under management by the highest competent authority of a nation. Examples include the Royal Chitwan National Park in Nepal, the Etosha National Park in Namibia, the Iguazu National Parks in Argentina and Brazil, and Volcan Poas National Park in Costa Rica. Natural monument/natural landmark. To protect and preserve nationally significant natural features because of their special interest or unique characteristics. Good examples include Angkor Wat National Park in Kampuchea, the Petrified Forests Nature Monument in Argentina, and Gedi National Monument in Kenya. Managed nature reserve/wildlife sanctuary. To ensure the natural conditions necessary to protect nationally significant species, groups of species, biotic communities, or physical features of the environment when these require specific human manipulation for their perpetuation.* Examples include Manas Wildlife Sanctuary in India. Most of the national reserves in Kenya also fall in this category, as do the biotope reserves in Guatemala. Protected landscapes. To maintain nationally significant natural landscapes characteristic of the harmonious interaction of man and land, while providing opportunities for public enjoyment through recreation and tourism within the normal lifestyle and economic activity of these areas. Examples include Pululahua Geobotanical Reserve in Ecuador and Machu Picchu Historic Sanctuary in Peru. The national parks of England are also classified under this category. Resource reserve. To protect the natural resources of the area for future use and to prevent or contain development activities that could affect the resource pending the establishment of objectives based on appropriate knowledge and planning. Few countries have yet applied this category, but several resource reserves exist in Kenya, including Kora and South Turkana National Reserves. Other examples include Brazil's Forest Reserves, and Tahuamanu Protected Forest, Bolivia. Natural biotic area/anthropological reserve. To allow the way of life of societies living in harmony with the environment to continue undisturbed by modern technology. The Gunung Lorentz Nature Reserve of Indonesia, Xingu Indigenous Park of Brazil, and Central Kalahari Game Reserve of Botswana are all occupied by indigenous people and are classified as Category 7 areas. Many protected areas in the South Pacific islands also fall into this category. Multiple-use management area/managed resource area. To provide for the sustained production of water, timber, wildlife, pasture, and outdoor recreation, with the conservation of nature primarily orientated to the support of the economic activities (although specific zones can also be designed within these areas to achieve specific conservation objectives). The most famous example is the Ngorongoro Conservation Area of Tanzania. Other examples are the Kutai National Park of Indonesia, Jamari and Tapajos National Forests of Brazil, and Von Humboldt National Forest of Peru.

* To this list should be added marine reserves such as the one located in the Galapagas (see Broadus, 1987). Source: IUCN (1984), Mackinnon et al (1986).

563 Fig. 21.1 World-wide growth in number and area of protected areas, 1870–1983 Source: IUCH (1985)

21.5.1 Commercial wild-species rearing There are many instances where private enterprise has raised wild species and profited from the sale of related commercial products. Smith et al. provide a good survey of these experiments. Examples include eider-duck husbandry in Scandinavia, extensive aquaculture activities in Japan, the rearing of alligators in Florida and crocodiles in Zimbabwe and Papua New Guinea, turtles in the Cayman Islands, and rare cactus plants in the USA. Captive rearing, when successful, can make its greatest ecological contribution by preventing the extinction of particular species. It can provide a period of grace in which effective regulations can be devised to protect species in the wild from subsequent extinction. One would therefore expect conservation organizations to support captive-rearing proposals. Often they don't and it is of some interest to understand the economic roots of their opposition. The general argument for commercial rearing is that entry reduces the market or illicit price of a species, or products derived from them, by shifting out

564 the supply curve over the relevant range. Pressures on the natural population are reduced because the higher-cost harvesters are driven out of business. If harvesting is illegal and the expected penalties for getting caught are not altered, poaching should decline as product price drops. This reasoning may be faulty on three grounds as R. Johnson makes clear. First, if there are prohibitions on the trade of naturally produced wildlife products such as ivory or hides, permitting the trade of commercially raised substitutes (which created the incentive for entrepreneurship in the first place) raises the cost of enforcement because it often makes it more difficult to identify the illegal products. This can make illicit trade relatively cheaper and can lead to greater pressure on the natural stocks of wildlife. Second, commercial rearing may have its own set of environmental costs. This is vividly illustrated by aquaculture which transforms natural bays into fish-farming operations; denies access to sport fishermen and other recreationists; reduces scenic beauty; promotes elevated levels of natural waste discharge; and may transmit exotic diseases to natural stocks against which they are uniquely poorly selected to defend. The third reason is a bit less transparent as it involves intertemporal dynamics and political economy. Were commercial rearing permitted and product price to fall as a result, exit would occur from the natural harvesting sector, and the natural population would begin to grow back to higher natural levels of stock and higher levels of steady-state harvest. Regulations introduced when a resource such as a fishery has been over-exploited might permit long-run sustained harvest levels at (i) rates higher than open access levels; and (ii) market equilibrium prices lower than the cost of commercial rearing making it economically non-viable. This argument developed by Anderson and Berck and Perloff resembles the use of excess capacity by the monopolist, in the industrial organization literature, to keep out potential entrants. In this case, a resource industry such as a fishery, could use the public regulatory authority for its own advantage against the potential commercial-rearing rival. However, competition in the political market-place can complicate matters. Harvesters may be homogeneous in simple renewable resource models but heterogeneous harvesters is more descriptively accurate. Those earning rents in the open-access state may be chary about any changes which could threaten profits.105 Future regulations, when steady-state harvests are higher, might discriminate against those most clever when stocks and harvests are low. This sub-group of fishermen and the captive-rearing sector may not have the political muscle to block lobbying efforts by other, more numerous fishermen to increase the quota as stocks recover.

105

Salmon fishermen in Alaska, where limited-entry regulations have created highly valued property rights, successfully have blocked pen-rearing of salmon in Alaska. Johnson argues that Norway's 60 per cent share of the world's pen-reared salmon was not stymied initially because the natural salmon fishery was very small.

565

21.5.2 Mixed game and cattle In the last section I discussed commercial rearing of wild species with an implicit emphasis on the idea of using the commercial activity to preserve a species. More generally, there are a variety of motives for wildlife ranching and cropping which I will summarize before turning to the idea of a mixed game and cattle enterprise. Throughout the world there has been an interest in big-game hunting. Deer hunting or stalking is a source of substantial revenues for private landowners in Scotland, Wales, and England, and for ranchers in the south-west USA. The right to hunt species of trophy dimensions is sold by eastern European countries such as Hungary and Czechoslovakia. Big-game hunters, formally the source of revenues and foreign exchange for many African nations, now takes place in selected countries including South Africa. Some wildlife are legally cropped for the products they provide. Musk has been extracted from living deer in China on a renewable regime for several decades (Eltringham, 1984). Illegal harvest of ivory and rhino horns is a well-publicized activity but several countries, including Zimbabwe and South Africa, have legally cropped ivory on a sustained yield basis, earning revenues which are used to manage game parks and reserves. A number of projects sponsored by the United Nations, the Tanzania Game Department, and others have been launched to raise and crop African ungulates (zebra, wildebeest, gazelle, and antelope) on a commercial basis. There seems to be some controversy about how many, if any, of these experiments have been economically successful. They have not been a smashing success (Eltringham) for a number of reasons. Conservationists have staunchly opposed slaughtering wildlife. While there are important exceptions, slaughtering wildlife in a fashion both economical and humane is difficult to achieve. Additionally, it does not appear that sufficient care has been taken to develop a proper marketing strategy which suitably finesses opposition from the conventional meat trade. A particularly attractive prospect, in my judgement, is mixed game and cattle ranching, where the game can provide both non-consumptive viewing values and directly productive values through cropping to preserve steady-state populations. For all of Eltringham's pessimism about profitable, specialized private wildlife management, he does cite economically successful cases of mixed game and cattle ranching in South Africa (Eltringham, 1984: 122, 218). What key features would make mixed game and cattle profitable? Clearly economic interaction between game and cattle is necessary for an interesting story. Game and cattle can interact on the demand or supply side. The demand side is straightforward and is mentioned only in passing. Assume that the experience of being on a ranch and viewing game is more valuable than viewing game separately. The simplest way to get interdependence on the supply side is to suppose some costs like marginal enforcement costs fall with aggregate activity. If

566 cattle could be expanded to capture the economies of size, there would be no gains from introducing wildlife. This prospect won't occur if the revenue function for cattle slopes downward at a congenial rate. Let total enforcement costs be: EC(G + B), where EC1 = EC2 < 0, G = game, B = cattle, and where the subscript refers to the argument with which the derivative has been taken. For concreteness:

Let R(B) = revenue for cattle, R′(B) > 0,R″(B) < 0, and V(G) = revenue for game,106V′ (G) > 0, V″ (G) < 0. I assume here that value arises from viewing the stock of game, not from a harvest rate. It is also assumed that all value of the game is capturable, an assumption which is not very objectionable unless the resulting optimal values of game are sufficiently small that non-use and typically non-capturable marginal values begin to influence the socially optimal values for the game. The build-up to the steady-state stock level is not featured here. Such an omission is innocuous as long as own rates of return dominate market rates of interest. Grazing costs are given by:

Specifically, C(αG + B) = αG + B, a > 1. There is no need to complicate this function except perhaps to acknowledge that game eat more than cattle eat, after paying due regard to units of measurement. This explains α > 1. On the other hand, game are not always competitive with cattle. Some wild species are specialized to browse on species of grass and other food sources which cattle spurn (Eltringham, 1984: 123). Ranchers are assumed to maximize profit π(G,B),

The necessary conditions for a maximum are(21.1)

(21.2)

assuming an interior solution. If the discrepancy between the social and private value is substantial, the maximization problem will have to be amended in a fashion which induces a choice of the socially optimum G. For example, a suitable tax-subsidy policy can be devised when the decentralized decision-making context is to be preserved. Figure 21.2 illustrates the solution when no game are raised.

106

We could write V (G,B ) where V12 > 0 to indicate favourable interaction between game and cattle on the demand side. People who choose certain kinds of safaris enjoy seeing cattle and wildlife integrated.

567 Fig. 21.2 No game raised

It is apparent from (21.1) that if the marginal profit of game, excluding enforcement costs, is positive for some G, then we should expect to see multiple use. There should be more enthusiasm for game: (i) the smaller is the relative grazing costs of game; i.e. the smaller is α. The aggregate marginal gross profit curve shifts out in Figure 21.3. Let and (ii) the greater are the economies of size. Interestingly, a consequence of producing game in this special model is to increase the production of cattle which could be of national interest if cattle earn scarce foreign exchange. This is confirmed by noting in (21.1), that as G becomes positive, since enforcement costs are falling in aggregate activity, cattle (B) must increase to drive the Ihs of (1.1) down to its new equilibrium level (see Figure 21.4). The phenomenon of joint-species production owes its existence to economies of size which were embedded in the enforcement-cost function in this model. Naturally, the same qualitative result can be obtained by any Fig. 21.3 Mixed domestic and wildlife solution

568 Fig. 21.4 Introducing game increase cattle production

economic phenomenon which gives rise to falling costs. Another potential cause could be the falling unit cost of acreage as the number of acres purchased increases. This economic phenomenon would have to be explained by an imperfect capital market, or diminishing returns to size combined with the plausible assumption of increasing costs of complexity and asymmetric information, otherwise we would expect to see small ranchers co-operating to capture the scale economies. If true, it would account for another concern. Why do we see, or I expect we see, joint production on large but not small ranches? This solution also would be expected if game populations such as elephants are not well behaved, but rather violate divisibility assumptions, creating essentially a threshhold level of population below which they are too costly to raise on ranches, on a routine basis. If the threshhold is large enough, then only the large-scale ranches can exploit the reward of a falling marginal enforcement cost function. The ranchers are assumed to maximize profits from game and cattle. If increasing game on these ranches contributes to biodiversity or to existence value, then a case can be made for subsidies which would increase the number of game on ranches and relatively decrease the number of cattle. Public policies which discriminate against large landownership must be recognized as having a deleterious impact on the viability of mixed cattle and game ranches, if this model is descriptively accurate. There is another possible intriguing interdependence between game and cattle. It may be that the tastes and size of elephants and other animals make them part-time factors in the production of cattle. That is, elephants may, by removing taller shrubs today, convert land into more agreeable habitat for cattle in the future. Whether this or other dynamic supply-side interactions make mixed-species ranching attractive remains to be studied. On the demand side one wonders if safaris to mixed-species ranches is complementary or competitive with safaris to national parks and reserves in Africa. If the economic relationship is complementary, a case can be made for stimulating mixed game and cattle enterprises.

569

21.6 VALUING WILDLIFE AND HABITAT IN DEVELOPING ECONOMIES Applied economists can be very helpful by providing an economic perspective on the social benefits and costs of wildlife and habitat preservation. Not all habitats and species should be preserved because the opportunity costs will be too great in some cases. On the other hand, social benefit-cost analysis may well support a preservation decision when revenue-cost analysis would conclude otherwise. The following cases illustrate how some of our tools can be applied. I have drawn heavily from Dixon (1990) and Dixon and Sherman (1990a, 1990b).

21.6.1 Lumpinee Park, Bangkok, Thailand Lumpinee Park provides open space for recreation in the centre of Bangkok, illustrating that the importance of undeveloped space extends to the urban core. The economic value of urban recreation space can be demonstrated by using travel-cost analysis, a method which exploits the fact that people who live different distances from the park and incur different opportunity costs of time and travel can be expected to participate at different rates. The results of a travel-cost study of Lumpinee Park indicate that the consumer's surplus of park-users is about $630,000 per year. A contingent-valuation study of the park also was conducted. Park-users and non-users were asked their willingness to pay to preserve the park rather than have it converted to other purposes. The contingent-valuation study indicated that people were willing to pay about $4.6 million per year to preserve the park as open space. Interestingly, users asked the contingent-valuation hypothetical question reported the same value as was derived from the behaviour-based travel-cost study. One could also use hedonic techniques to estimate the use value of the park if property values are observable. People ought to be willing to pay more for property closer to the park because it saves on travel costs. Estimating the rent gradient as a function of distance to the park, among the other economically important characteristics of the site, provides the basis for valuing the open space.

21.6.2 Korup National Park, Cameroon Cameroon and WWF-UK were the instrumental forces in securing an economic valuation of a project to develop the 126,000 ha. Korup National Park and a buffer zone more than twice as large. The buffer zone would mitigate unsustainable forest and farming practices. The project, summarized by Dixon and Sherman (1990a), illustrates an attempt by Ruitenbeck to estimate a number of important non-market values.

570 The largest benefit flows from a sustainable forestry plan (see Table 21.2). The next largest benefit, £3.8 million, stems from reducing erosion damage to the watershed due to improper forestry and farming practices, which, in turn through technological externalities, preserves a sustainable fishery. Avoiding poor timber-harvesting practices by providing alternative employment options for local inhabitants creates a benefit of £1.6 million in the form of reduced flood damage to thousands of people in the watershed. Estimated tourism benefits of £1.4 million are a small share of the project pay-off. Ruitenbeck took a stab at estimating the genetic value of this project, recognizing that plants and animals in the area may be the fundamental source of better seed varieties, new or improved drugs, or new or improved industrial products. The basis for the £500,000 valuation is an estimate of the value of a patent, £5,000, obtained from research by Schankerman and Pakes. Other benefits stem from subsistence production and soil-fertility benefits. Indirect benefits from the project arise from increased agricultural productivity and new lands devoted to forestry. Ruitenbeck makes three instructive adjustments to the benefit calculations. Some of the watershed benefits would accrue to Nigeria and are netted out here because the analysis is for Cameroon. Similarly it is believed that most of the genetic values would not be captured by Cameroon but would accrue to others outside the country. Finally, the project earns a substantial foreign-exchange benefit because the external funds would not have been forthcoming Table 21.2 Korup National Park, Cameroon: benefit-cost analysis (£’000s) Benefits and costs (Present value) Direct benefits: Sustained forest use Watershed protection of fisheries Control of flood risk Tourism Genetic value Other Indirect benefits: Total benefits: Accounting stance adjustments: Adjusted total benefits: Direct costs: Opportunity costs: Total costs: Net benefits: Benefit-cost ratio:

3,300 3,800 1,600 1,400 500 1,500 4,300 16,400 6,500 22,900 11,900 3,300 15,200 7,700 1.51

571 in the absence of this project. These accounting stance values sum to £6.5 million. Calculation of the direct costs of the project was based on an 8 per cent discount rate (5 per cent was used in other cases) and a shadow wage rate of 50 per cent on the grounds that the official wage rate is not an equilibrium one. (This adjustment amounts to £2.6 million.) Ruitenbeck estimated the opportunity costs of giving up the deforestation and unsustainable agriculture revenues at £3.3 million. The point of this summary is not that the project has positive net benefits but rather that (i) more subtle forms of benefits like watershed protection drive the performance of this project not the obvious benefits, tourism, which is under 10 per cent; (ii) the value of preserving genetic resources needs to be acknowledged even if the exact calculations are speculative; and (iii) sustainable forestry appears to be superior, even net, of the opportunity cost of myopic, unregulated harvesting.

21.6.3 Viewing value of elephants The first case-study of Lumpinee Park illustrated how contingent valuation and travel-cost methods could be used to value a park. There are policy issues more particularistic than an all-or-none question of whether, on economic grounds, a park should be preserved. Illustrative recent controversies include the economic value of preserving from extinction the Californian condor, the furbish lousewort, the snail-darter, and the African rhinoceros. Less ecologically precipitous is the recent effort by conservation groups and others to arrest the decline of African elephants by urging successfully a ban on the international trade of raw ivory and ivory products. Surprisingly, conservation groups believed that economic analysis would support their case. To this end I estimated the viewing value of elephants. Tourists spend hundreds of millions of dollars annually in Kenya. While an important figure for the Chamber of Commerce, it does not inform us about the rental value of animals that produce this value. Since there is no market for wildlife, non-market valuation techniques are required. Both the travel-cost and contingent-valuation methods discussed earlier were used. In this case, the market is international and origins used in the travel-cost analysis are continents. The surprising result is that the estimated consumer's surplus for a safari is not substantially less than the cost of travel. The value attributed to elephant viewing is about $25 million annually, based on respondents reporting that about 13 per cent of their pleasure was derived from viewing and photographing elephants. The point of professional interest is how to value one component of a joint product. Proportionality rules may do at a pinch and were used in this study, but they are suspect to economists who are marginalists at heart. In a data-rich world, one would do hedonic analysis using observed, measurable characteristics of different quality safaris and their

572 respective prices, then compute the value of changing the numbers or types of elephants viewed. Alas, no such data exist. In the contingent-valuation portion of the study, a sample of those on safari were asked if they would be willing to pay a $100 fee added on to their safari costs if the money would be used successfully to maintain elephant populations at their current levels. Their maximum willingness to pay also was asked. The average value (WTP) was $89 per respondent; the median value was $100. Using the median value produced a value of $25–30 million annually for keeping elephants at their current levels in Kenya. The range depends on the number of tourists used. The value is remarkably, perhaps fortuitously, similar to the value obtained from the travel-cost method. The estimate of $25 million is annually available to support the maintenance of elephants in Kenya, assuming no further growth in safari demand. It is an amount of money more than adequate to preserve the elephants and it dwarfs the value of the ivory trade prior to the ban on trade.

REFERENCES ANDERSON, J. L. (1985), ‘Market Interactions Between Aquaculture and the Common-Property Commercial Fishery’, Marine Resources Economics, 2: 1–24. ARROW, K. J., and M. KURZ, Public Investment, the Rate of Return and Optimal Fiscal Policy (Baltimore: Johns Hopkins University Press). BAUMOL, W., and W. OATES (1988) The Theory of Environmental Policy,2nd edn. (Cambridge: Cambridge University Press). BERCK, P., and J. M. PERLOFF, ‘The Commons as a Natural Barrier to Entry: Why There are so Few Fish Farms’, American Journal of Agricultural Economics, 67: 360–3. BROADUS, JAMES (1987), ‘The Galapagos Marine Resources Reserve and Tourism Development’, Oceaning, 30/2: 9–18. Summer 1987, 9–18. BROWDER, J. O. (1988), ‘Public Policy and Deforestation in the Brazilian Amazon’, in R. Repetto and M. Gillis (eds.), Public Policies and the Misuse of Forest Resources (Cambridge: Cambridge University Press). BROWN, GARDNER (1989), ‘The Viewing Value of Elephants’, Report prepared for the Ivory Trade Review Group. CUMMINGS, R., D. BROOKSHIRE, and W. SCHULZE (eds.) (1986), Valuing Environmental Goods: A State of the Arts Assessment of the Contingent Method (Totowa, NJ: Rowman &Allanheld). DASGUPTA, P., and J. STIGLITZ (1980), ‘Industrial Structure and the Nature of Innovative Activity’, Economic Journal 90: 266–93. DIXON, JOHN (1990), ‘Valuation of Protected Areas in Developing Countries’, unpublished manuscript. DIXON, JOHN and PAUL SHERMAN (1990a), Economics and Protected Areas: A New Look at Benefits and Costs (Washington, DC: Island Press). DIXON, JOHN and PAUL SHERMAN (1990b), ‘Economics of Protected Areas’, Ambio, forthcoming.

573 ELTRINGHAM, S. K. (1984), Wildlife Resources and Economic Development (John Wiley &Sons). HOTELLING, H. (1931), ‘The Economics of Exhaustible Resources’, Journal of Political Economy, 39: 137–75. IUCN (1984), ‘Categories, Objectives and Criteria for Protected Areas’, in J. A. NcNeely and K. R. Miller (eds.), National Parks, Conservation and Development (Washington, DC: Smithsonian Institution Press). JOHNSON, R. ‘Commercial Wild Species Rearing: Competing Groups and Regulation’, Journal of Environmental Economics and Management. JOHNSON, R. and G. LIBECAP (1982), ‘Contracting Problems and Regulation: The Case of the Fishery’, American Economic Review, 72: 1005–22. KRUTILLA, J. (1967), ‘Conservation Reconsidered’, American Economic Review, 57: 777–86. MACKINNON, J., K. MACKINNON, G. CHILD, and J. THORSELL (1986), Managing Protected Areas in the Tropics (Gland, Switzerland: IUCN). MALIK, A. (1984), ‘Protected Areas and Political Reality’, in J. A. McNeely and K. R. Miller (eds.), National Parks, Conservation and Development (Washington, DC: Smithsonian Institution Press). MITCHELL, R., and R. CARSON, Using Surveys to Value Public Goods: The Contingent Valuation Method (Washington, DC: Resources for the Future). RUITENBECK, H. J. (1989), Social Cost-Benefit Analysis of the Korup Project, Cameroon (London: Worldwide Fund for Nature). SCHANKERMAN, M., and A. PAKES (1986), ‘Estimates of the Value of Patent Rights in European Countries during the Post-1950 Period’, Economic Journal, 96: 1052–76. SMITH, R., J. GOLDSTEIN, and R. DAVIS (1983), ‘Economic Incentives as a Conservation Strategy for Nongame and Endangered Species of Wildlife’, in K. Sabol (ed.), Transactions of the Forty-Eighth North American Wildlife and Natural Resources Conference (Washington, DC: Wildlife Management Institute). SWANSON, T. (1989), ‘Policy Options for the Regulation of the Ivory Trade’, in S. Cobb (ed.), The Ivory Trade and the Future of the African Elephant (Oxford: Ivory Trade Review Group). WEISBROD, B. (1964), ‘Collective Consumption Services of Individual Consumption Goods’, Quarterly Journal of Economics, 77: 71–7.

22 Public Policy toward Social Overhead Capital: the Capitalization Externality The provision of social overhead capital and development of land-based resources are among the most important issues in public policy today. They have become increasingly important over time as we come to realize that the earth is a finite resource that can no longer be treated as a free good and must be managed carefully, as with any scarce resource. Aside from a few countries (and the number seems to be shrinking every week) where the the Government is directly involved in substantial production activities, the important issues for public policy are if and how to regulate private development, and what public services to provide in conjunction. Brazil must decide how much clearing to allow in the Amazon basin and how many roads to build; the USA must decide how much drilling to allow on Alaska's North Slope; California must decide how many condominiums to allow on its Pacific coastline, and what water resources to channel to southern California; Mexico must decide whether to allow Mexico City to become any larger and if so, what public utilities to provide, and so on. Even in ‘free market’ economies it has been recognized for some time that some regulation is required when private operations generate externalities (such as environmental pollution).107 However, once these are taken into account it is frequently argued that no further interference is justified: the profit motive can be counted on to provide incentives to undertake the socially desirable projects but not the undesirable ones. The process of allowing developers to compete for use of land is alleged to generate correct valuation of land: land will go to the developer who can put it to best use and its value will rise accordingly. Indeed, it has been further argued that developers need to be allowed to reap capital gains on the land they develop to induce them to provide the associated real social benefits.108

107

There are a wide range of externalities that are relevant and important in the contexts considered here. While not wanting to minimize these, we choose to concentrate on some problems that are not as well exposited or understood. See Arrow and Fisher (1974), Baumol and Oates (1975), Dasgupta and Mäler (1997), Ehrlich et al. (1977), Johda (1986), and Johansson (1990) for discussion of the rich range of relevant externalities.

108

The capitalization argument has been around in various forms for some time. See Starrett (1988), ch. 11 and 13, for a summary and references to earlier literature.

575 This argument that land values ‘capitalize’ the benefits of associated projects is obviously important to evaluate. For example, it has been reported that a large percentage of the (intertemporal) profits made by developers in the Amazon basin comes from capital gains on land rather than profit from sales of outputs: indeed, many make losses in terms of real cash flow.109 If such capital gains really are allowable as benefits, then the argument against such development must rely on the existence of somewhat tenuous externalities associated with loss of the rainforest, whereas otherwise they are clearly undesirable. We will argue in this chapter that while capitalization is a real benefit in some circumstances, it is virtually never attributable to private development: rather it is due to some concomitant improvement in social overhead capital. Not only should land profits be disallowed to developers, but in some cases they should be made to pay some share of the cost of these improvements. Further, we will show that in many other instances there are ‘hidden’ externalities associated with land values, externalities that sometimes reverse the capitalization logic. Indeed, there are cases when the fact of increased land values is ‘proof that the project is imposing undesirable social costs: developers should actually be taxed rather than rewarded in such situations. The implications for social policy are extremely important. We will argue that the provision of social overhead capital in relatively undeveloped areas provides perverse incentives to overuse the corresponding natural resources unless the private developers are taxed appropriately. Further, we will show that the incentives are even more distorted in urban areas where apparently ‘neutral’ policies lead to overconcentration in too few urban areas. These incentives are particularly damaging in political environments typically encountered in the Third World and may go a long way towards explaining the obvious overcrowding in cities like Rio and Mexico City. The chapter is organized as follows. In the following section, we lay out a model of the economy that is structured to focus on policy issues with respect to land and land values. Then, we develop and explain the traditional view in welfare economics that changes in land values (or any other price for that matter) should have no net effect on social welfare: intuitively, gains to sellers’ welfare are just offset by losses to buyers. It follows, that in most situations, changes in land values should be counted only to the extent that they reflect benefits and costs associated with non-market variables such as environmental quality. Consequently, in Section 22.3 we explore systematically the relationship between environmental quality and land values. Surprisingly, perhaps, we find that this relationship may be either positive or negative. Section 22.4 looks at the capitalization argument, deriving conditions under which it holds, and showing why capitalization argument, deriving conditions under which it holds, and showing why capitalization cannot be attributable

109

See Binswanger (1989), Mahar (1988), and Repetto (1988) for detailed discussion of the Brazilian case.

576 to private development, but must represent benefits from parallel public improvements. The following section then develops a correct calculus for evaluating development projects when they are undertaken in conjunction with concomitant public improvements. Section 22.6 develops the argument for ‘negative capitalization’.

22.1 THE ECONOMIC FRAMEWORK We focus here almost exclusively on policy issues relating to environmental quality, public goods, and the treatment of land and land values. Thus in our modelling we abstract from many other issues in project evaluation. In particular, we ignore many standard sources of market failure (monopoly power, imperfect or missing markets, and the like) to concentrate on some particularly associated with the treatment of land and social overhead capital. We will aggregate all private goods except land and labour into a single numeraire (thus, we will not be concerned with how policy might affect the relative prices of such goods). Aside from this restriction, our modelling of consumers is relatively general. We do assume that they have no influence over prices, but we can allow them relatively more or less choice in their purchases on the labour and land markets. On the land market, consumers may have a choice over location and amount of land at that location. In all cases, we will think of location as a discrete choice among a finite set of alternatives (so everyone will have a single place of residence). When full freedom of choice is allowed, it will be convenient to think of households as solving a two-stage problem. First, conditional on being in a particular location they solve the problem of choosing the continuous consumption variables (numeraire, land, labour). Then, they optimize over the discrete choice of location. The outcome of this process is naturally the same as if all choice variables are contemplated simultaneously, but our sequencing is mathematically convenient. The household first-stage problem typically takes the form:

subject to

where c,n and ℓ represent consumption, labour supply, and land demand respectively, w and r are the market prices of labour and land (at the specified location), y is exogenous income (to be discussed further below) and α stands for a vector that may include public goods, amenities, and environmental quality variables. This problem naturally defines an indirect utility function of the form V(α,w,r,y). At times we will consider restrictions on the choice of lot size or labour supply to see what difference (if any) it would make.

577 When the household has a choice over a set of locations indexed by {i} all of the above parameters will be correspondingly superscripted and the second-stage problem may be represented as

(y is not indexed, since exogenous income must be independent of location choice.) Note that if similar individuals end up residing at different locations they must get the same level of utility at each of these locations. The second major restriction we place on our model is to treat households as being alike in all respects and to count them equally in our welfare formulation. This will mean both that all have the same utility function, and that all own equal shares of exogenous income (derived from land and profits). Such restrictions obviously are strong and require some justification. Our first reason for making them is that they lead to a very clean benchmark theory of differential rent: rent differences between locations must exactly reflect the differences in worth of those locations to the representative household. When people differ with respect to tastes or endowments, they will typically sort themselves out by location and the theory of rent is considerably more complicated.110 However, we believe that many of the conclusions reached in the sequel will be qualitatively unchanged in such a model. We provide some justification for this claim in the concluding section. The other reason for assuming identical agents is to abstract from some fairly obvious equity considerations associated with land values. If we consider the users of land to be a more deserving group than the owners, then it is obvious that there will be a welfare loss associated with an increase in land values (ceteris paribus). While not denying that this might be a relevant consideration, we prefer to work in a model where the presumption is that price changes will have no net welfare impact in order to isolate particular features of the land market that generate counter-examples. Of course, our modelling choice implies that we abstract from all equity issues, not just those associated with land use. There are some very powerful equity arguments for conserving the environment associated with guaranteeing survival to future generations, and equally important equity considerations in reconciling ‘North-South’ development interests.111 Again, we do not mean to minimize these concerns but rather to focus on an orthogonal set of issues. Our final restriction is to consider only an essentially static model. Again, we acknowledge here that there are very important dynamic issues involving conservation of resources over time, social discounting, and the like, but we cannot deal with them effectively here. The only sequential element in our model will be of the ex ante/ex post type: we assume that there is an ex ante status quo

110

For a discussion of relevant considerations involving heterogeneity, see Berglas (1976), Kanemoto (1980), and Starrett (1981).

111

See Dasgupta (1990) and Dasgupta and Maler (1990: vol. 1) for a discussion of these issues.

578 (before projects are initiated) and an ex post equilibrium which will dependence on private development and public policy choices.

22.2 LAND VALUES AND WELFARE: THE CASE FOR NEUTRALITY Conventional wisdom would have it that (in the absence of equity considerations) price changes have no net welfare implications. If a price goes up, sellers are better off but buyers are worse off by an equal and opposite amount. Here we exposit the sense in which this view is correct as applied to land. The neutrality result will always hold when the total amount of land is fixed, the model is closed (so that there are no outside owners or users of land), and all nonmarket welfare impacts are correctly measured.112

22.2.1 Income cancellation: the intuition For illustrative purposes, we demonstrate this result first in a special case with the following features: a project (parametrized by g) is contemplated that will generate net profit (in terms of the private-good numeraire) and change land values, but will have no other direct effects on the economy. In particular, it will not affect amenity levels or the wage rate (which are assumed fixed at levels α* and w* respectively). The project ‘owns’ land ℓ(g) which it rents out to households and produces profit equal to

where Π1(g) stands for profits directly attributable to the development. Lot sizes are fixed at ℓ* and households find consumption and leisure to be perfect substitutes at rate of exchange w*. Thus, they have preferences of the form U = U(α*,ℓ*,c − w*n). All exogenous income is held by residents in equal shares. Here, there are two sources of this income: project profits and land. Assuming a fixed population of size N and fixed land in the amount L, the income of a representative individual is

We see right away that it is completely irrelevant to project evaluation whether or not the developer owns land: everything in the economy is owned directly or indirectly by the households. A representative household chooses consumption and labour supply to

112

For an exposition of this point and some of its general implications in a development context, see Little and Mirrlees (1969, 1974).

579 subject to

Due to the special structure we have imposed, it is a trivial matter to evaluate indirect utility from the project. The budget constraint implies

the last equality following from the fact that total land supply (L) must equal total land demand (Nl*) in our closed model. Notice that potential effects through rents have cancelled again. Income from ownership cancels against rental costs. We are left with indirect utility V(g) = U(α*,ℓ*,Π1(g)/N. The project enters welfare calculations only through the direct profits it generates and all reference to rental changes disappears.

22.2.2 First-order welfare calculus Now we drop most of the special assumptions made above. However, we continue to assume that the wage is determined outside the model. This is done for notational convenience only. The reader should see that if we modelled the labour market as closed with the wage determined endogenously, wage effects would cancel just as rent effects do below. (See the appendix for a general demonstration of this fact.) The project is as described above except that we allow for the possibility that it is a user of land; let ℓ1(g),ℓ2(g) stand for land owned by and used by the project respectively. Consequently, profit is now written as

Our representative consumer problem now takes the more standard form:

subject to

Where

Optimal choices imply an indirect utility function of the form: V = V(α(g),r(g),y(g)). Although we can no longer evaluate this function directly, we can compute its derivatives with respect to g: that is, we can measure the effect on welfare of a small change in the project ‘level’. This is done by taking a total derivative of V(.):

580 and using consumer duality theory to evaluate its partial derivatives. Normalizing by the ‘marginal utility of income’, multiplying by N (to convert to an aggregate measure), and letting Ωα stand for the unobservable marginal rate of substitution between amenities and income, we have

Again, we see that terms involving dr/dg cancel because the equilibrium demand for land must equal the total supply minus that amount used by the project, so we have finally:

Our more general approach leads to two modifications of the results in Section 22.2.1 above. Any non-market amenities created by the project should be counted as benefits and land used by the project should be counted as costs; but induced capital gains or losses still play no role. At the cost of notational complexity, we could further generalize this result on the welfare neutrality of rent changes. It continues to hold when 1 and is heterogeneous (simply interpret all land-use variables above as vectors over land types and the rent variable as a corresponding vector of prices (see the appendix for details). It even holds with heterogeneous households, as long as dollars of income are counted equally (regardless of who gets them) in the welfare aggregation (as they will be in a social-welfare function formulation when lump-sum transfers are allowed).113 Despite these facts, there are two reasons why changes in land values may legitimately enter welfare measures. First, if some land is supplied or demanded by agents ‘outside’ the model, then the land market is not closed in the sense required above and neutrality generally will not hold. More importantly, land-value changes may be a perfect proxy for unobservable changes in amenity values. Since we just saw that such changes ought to be included in the welfare calculus, capital gains and losses on land may be relevant even in a closed model. Thus, we need to examine the relationship between amenities and land values.

22.3 ENVIRONMENTAL QUALITY AND LAND VALUES Since environmental quality has a direct influence on the intrinsic value of the associated land, it is natural to expect that land values and environmental quality would be positively related.114 And as we just suggested, this is one

113

For further discussion of first-order welfare calculus with heterogeneous households, see Starrett (1988, ch. 9).

114

Anderson and Crocker (1971), Mäler (1974), and Polinsky and Shavell (1976) were among the first to explore carefully the relationship between environmental quality and land values. See Mäler (1971), Freeman (1979), and Johansson (1987) for general discussion of methods for estimating the value of amenities from market data.

581 possible justification for incorporating induced changes in land values into measures of project benefits. In this section, we examine this relationship systematically and find that the issue is not so clear-cut. Indeed, there are times when an inverse relationship holds, and these will be important for policy implications discussed later. There are two main reasons for positive land rents. Homogeneous land will command a positive rent if the aggregate demand for it exceeds supply at a zero price. However, even if homogeneous land were to be a free good, heterogeneity in land quality will lead to differential rent, with high-quality land commanding a rent premium over lower-quality land. We think that the second type of rent is more important in our present context. For one thing, the lowest-quality land is nearly a free good in many parts of the world. Further, we doubt that the aggregate demand for land is much affected by environmental quality: people will use the land regardless of its quality since there is no viable alternative. Consequently, in what follows, we will assign a zero value to ‘marginal’ land and concentrate on differential rent. The reader should note that the same qualitative results will emerge if we assign a positive exogenously fixed value to such land.

22.3.1 Absentee ownership We begin (for simplicity) with the partial-equilibrium case where exogenous income (in particular land) is owned by households ‘outside’ the model and wages are unaffected by amenity levels, so that the only economic parameters that can respond to changes in amenities are land values. Even in this case, we will see that the relationship need not be positive. Since everybody is assumed alike, the welfare level achieved will be independent of location (label the common level V*). Consequently, for all occupied parcels (i), we find(22.1)

whereas on unoccupied land (u),

We define marginal land as those parcels where equality holds above at a zero land rent. Suppose first that marginal land is unaffected by a change in amenities: that is, a change in amenities has no direct or indirect effects on the welfare of agents living on marginal land. This situation would apply, for example, if amenities involved urban pollution that never reaches the countryside. In such situations, the common welfare level will be unaffected by changes in α and any welfare change through amenity levels must be exactly compensated for by a corresponding change in rents (if it weren't, people would be moving in or out from marginal land). In particular, implicitly differentiating (22.1) and using the assumption that wages are unaffected, we find

582

where the inequality follows from our convention that amenities are positively valued and the fact that rent increases are bad for land users. Thus, as long as marginal land is unaffected by amenity levels, we do indeed find a positive relationship between environmental quality and the corresponding rental values, and this relationship will hold not only in the aggregate, but at each separate location. However, let us now consider the possibility that marginal locations are affected. Let g stand for the policy variable that is affecting amenity levels at the various locations, so that αi = αi(g). Any change in welfare that does occur still must be the same on all occupied locations. Consequently, for each location i and any marginal location (m), we must have the relationship:

Differentiating with respect to g and rearranging yields115

Now we see that rents move at location i in response to the differential movement in amenities. Consequently we may now find situations where aggregate rents and environmental quality move in opposite directions. For example, suppose the amenity of concern is congestion (or lack thereof). An increase in congestion typically has a greater effect on those on the outskirts of a city, who have to commute a long distance, than on those who live closer to the centre (so that the ‘marginal’ term above dominates the ‘intramarginal’ one). Such an increase (which obviously lowers environmental quality) will increase the rent gradient, thus raising aggregate land rent. These relationships will naturally be more complicated when we incorporate general-equilibrium considerations, but some (albeit less precise) associations are likely to remain. We defer further discussion on this to the appendix.

22.4 INCREASES IN LAND VALUES AS REAL BENEFITS: THE CAPITALIZATION ARGUMENT The ‘capitalization’ view stands in stark contrast to our ‘conventional wisdom’ of Section 22.2. According to this view, increases in land value measure real benefits of associated projects and should be counted as such in the calculus of project evaluation. The logic behind this view is based on the following

115

Actually the derived relationship holds exactly only if a location that was marginal in the status quo remains marginal after sufficiently small changes in g ; otherwise slight modifications are required.

583 modifications of assumptions made in the last section. Suppose that users of land can always escape from the costs of increased land rents if they want to (for example by moving elsewhere). Then, if they choose to stay, we have revealed evidence that they are not worse off. Since the owners of land are actually better off, the rent increase must indirectly reflect a real increase in welfare. Frequently, we will find that this increase in welfare is associated with an improvement in amenities or environmental quality. We start with a simple example to illustrate the possibilities. There are three types of agents: developers who hire workers and produce a private-good numeraire; workers who rent land and work for developers; and landlords who own the land and purchase the private good. Workers are alike everywhere and they always have a fixed outside option where rents are zero and wages w*. We assume that this wage is competitive in that it represents the value of the private good these workers would produce in the outside option. Developers offer a wage w′ and rents near the development adjust to a level r′. Assuming that workers are freely mobile, this level must be such that workers are indifferent among locations; that is, V(w′,r′) = V(w*,0). Suppose now that L workers move and are hired by a new development. We seek a correct measure of social profit from their operation. Since everything can be measured in terms of the single private consumption good, we can compute net benefits in these consumption units. Because there is no change in amount of labour supplied or amenities, net benefits are simply the net increase in output. This will be the extra output of the project (y) minus the lost output elsewhere (w*L). Thus, net social profit (NSP) is measured as

where Π = y − w′L is the actual profit earned by the developer. But now, from the fact that workers were indifferent between options, it follows that the extra wages earned in the new project are exactly equal to the extra rents paid. Consequently, it is legitimate to write:

We see that net social profit is indeed equal to private profit (as usually measured) plus the capital gain on the land (which had zero value in the ex ante situation). However, even if we accept the mobility assumptions, the example is not very convincing as it stands. We should ask why it is that the developer paid wage w′ instead of w*. If workers really were willing to come on any terms that left them indifferent (as we must assume to get our result) then there is no reason for wages to be bid up. But if w* is offered, rents are not bid up and the capital gain disappears. To get capitalization in a consistent competitive model, one needs to have developers providing some real amenities not sold through the market (α′). The presence of such amenities will lead to a rent increase even if there is no

584 wage premium: worker indifference now requires V(α′, w*, y′) = V(0, w*, 0), so if amenities are desirable, r′ will have to be positive to compensate. Assuming that workers are the only ones to benefit from amenities, the above analysis carries over exactly and net social profit again includes capital gains. Here, these capital gains reflect the real nonmarket benefit from amenities. However, unless the developer owns the land, it is still not clear that any correction to private profit will be required. If the developer really can vary the wage subject only to matching the worker's outside option, why not lower it to the w′ where V(α′,w′,y′) = V(0,w*,0). Then, actual profit (Π = y − w′L = y − w*L + (w* − w′)L) correctly reflects the value of amenities measured as willingness to forgo wages [(w* − w′)L] and there are no capital gains. These examples suggest that capital gains on land will rarely, if ever, be attributable to developers guided by the profit motive. The general logic of this position is as follows. Developers rarely will own all the land affected by capital gains. Thus, to the extent that project benefits are allowed to accrue as capital gains, the developer will not appropriate all of them as project profits. He is better off in such situations to save costs by offering somewhat less favourable terms to those coming to occupy the land, thereby transferring the capital gains (accruing to landowners) to direct project profit. Consequently, when we see capital gains on land at the same time that private development is taking place, we should assume that other forces are at work. Frequently, these forces will involve the public provision of amenities. We show in the appendix that in certain circumstances such provision will generate increases in land values.

22.5 PUBLIC AMENITIES AND PRIVATE DEVELOPMENT: THE SIMULTANEITY PROBLEM Most private projects are not carried out in isolation but are undertaken in conjunction with public provision of social overhead capital. For example, new farming operations in the Amazon basin and new logging operations in US national forests are linked to public road building. In such situations, there is an identification problem for project evaluation. To what extent are project profits attributable to efforts of the developer, and to what extent are they attributable to the social overhead capital? Obviously, the answer should affect appropriate public policy toward private development. We just argued in the last section that capital gains on land are not likely to be attributable to the project. In this section, we examine other aspects of this identification problem. We start with a simple example in which there are two potential uses for the land: a public use (recreation) and a private use (logging). Each of these will require a fixed piece of social overhead capital (road). Without the road the land will have zero value, while with it, and one or both of the uses, it presumably will

585 have positive value. We seek a correct benefit-cost calculus for determining what to do.

22.5.1 Mutually exclusive uses We deal first with the case where we can do logging or recreation but not both. We assume that the land is owned publicly and ask how it should be valued. Consumers own the logging company and enjoy the recreation if it is provided. Assuming that no prices of things they consume are affected, we can think of their indirect utility as being a function of amenities and net private income only. Given the discreteness of our example there are three distinct options to evaluate: 1. Do nothing. In this case the utility level will be V(0,y), where y stands for exogenous income and 0 is the (normalized) ex ante level of recreation amenities. 2. Build the road (q) for recreation. This option generates a utility level V(α(q),y − Γ(q)) where Γ(g) stands for the cost of the road. 3. Build the road for logging. Because recreation is then ruled out, the utility level will be V(0,y − Γ(q) + Π(q)), where Π(g) is the net profit from logging given that nothing is paid for the land. We seek a benefit-cost calculus that will guarantee that the right decision is made. Suppose first that recreation is not desirable, i. e.

Then we would know that logging should be carried out if and only if

or, equivalently, if Π(q) − Γ(q) ≥ 0. Thus, we see that correct accounting would have the logging operation treat the public road as a direct private cost. If the logging company builds the road directly, no further land rent charge is needed, but otherwise they should be taxed (or charged rent) to pay for it. But now suppose that recreation is desirable. How can we still be sure that logging is the best option? Let us define B by the equation

B stands for the amount of ‘money’ we would need to add to incomes in the no-recreation situation to compensate households for not having the recreation option. If recreation is desirable in the sense described above, B is positive and measures its net potential value. Now, logging will be the preferred option only if

or equivalently if Π(q) − Γ(q) − B ≥ 0. Consequently, to guarantee correct decisions, loggers must be made to treat B (in addition to Γ) as a cost. Since B represents the opportunity cost of the land, it is best thought of as a land rent.

586 Note here that land values have gone up as a consequence of improvements but the increase should still be treated as a cost, not a benefit to the private developer.

22.5.2 Potential joint use One might argue that we have stacked the deck against logging by assuming that logging operations completely rule out recreational use. Let us see how the analysis changes when we allow joint use. To keep our options simple, we now assume that recreation cannot be prevented once the road is built. Consequently, there are still just three options to evaluate: 1. No road. This option has the same valuation as it did before. 2. Recreation only. The welfare valuation of this option is now expressed as V(α0(q),y − Γ(q)), where α0(.) represents the amenities enjoyed from recreation in the absence of logging. 3. Recreation and logging. The value of this option is V(α1(q),y − Γ(q)) + Π(q)), where α1(.) represents amenities enjoyed when recreation and logging coexist. It is reasonable to assume that α1 is worth less than α0. We quantify this difference as follows: Define compensating variations B0 and B1 as solutions to

Clearly, B0 measures the value of recreation against the option of doing nothing (so, in particular, a road for recreation only would be desirable when B0 is positive), whereas B1 measures the value of a dual use road against the (fictitious) option of having logging only. Note that if B1 is positive, the road (once built) will be used for recreation, but otherwise not. Now, dual use will dominate both other options if and only if

In this situation, it is possible that logging need not pay the full cost of the road. This will be true if B1 ≥ Max {B0,0}, namely if the incremental value of adding recreation to logging (with the road already in place) exceeds the value from building a road for recreation only versus doing nothing. Indeed, when recreation is justified on its own (B0 0) and is equally valuable whether or not logging is allowed (so B1 − B0 = r(q)), then our efficiency criteria reduces to Π(q) ≥ 0, and there is no need (from a pure efficiency standpoint) for the logging company to share in road costs. The situation just described is one in which the road generates considerable surplus value and this surplus can be shared in any combination by pure profits to logging and utility surplus to recreational users/taxpayers. As long as these constituencies are equally valued (an assumption we might want to reconsider here), it does not matter who pays for the road.

587 But even if such a scenario does obtain at one point in time, it is unlikely to persist for long if loggers are allowed to reap the surplus through profits and expand their operations in response to the profit motive. As operations expand, they are sure to create negative externalities and otherwise interfere with environmental (recreational) quality. And notice from the analysis above that when logging seriously interferes with (environmental) recreational enjoyment, the logging company should pay and we may even find that it should pay a premium over the costs of the road (this will be so when B1 < B0).

22.6 DEVELOPMENT IN CROWDED AREAS: THE CAPITALIZATION ARGUMENT IN REVERSE We argued above that when capitalization is based on assumptions of costless mobility and attractive outside options, it is virtually never going to apply to private development projects. Here we want to re-examine the treatment of land rent under alternative (and perhaps more realistic) assumptions concerning the outside world. Surprisingly, perhaps, relatively small changes in our mobility assumptions completely reverse the policy implications with respect to treatment of land. In particular, we want to drop the idea that people always have an attractive outside option. Rather, we assume that people must choose among a limited set of options and will always take the best among these. We also drop the assumption that firms have control over the wage structure, and assume instead that wage rates are determined through general equilibrium forces.

22.6.1 Projects in isolation To illustrate the aforementioned reversal, we start with a simple example. There will be three types of agents: landlords who own all the land and rent it out; workers who rent the land and work for developers; and developers who hire workers and produce the numeraire good. Developers have a choice between ‘large’ and ‘small’ operations. A particular set of households (2n) can be accommodated by one large or two small operations. A small operation is sufficiently compact that all workers can live next to the ‘factory’ and no transport costs are incurred. By contrast, in a large operation, half the workers live close by, but half must live further out and incur a transport cost t. We assume that lot sizes and labour supply are fixed (and normalized to one) so that a worker's net income is measured by wage minus rent minus transport paid. Further, let the wage be exogenously given and the same in all operations. Note that rental rates cannot be the same even though all marginal land still has zero value. Those living close to a large operation will have to pay a rent of t in order to deter those living further out from wanting to move in. Thus, large operations generate positive rent, whereas small ones do not. All workers in

588 large operations have net incomes of w − t; those outside incur the extra cost as transport while those inside pay it as a rent. Suppose that developers make a net profit (in numeraire units) of π1 in two small projects and π2 in one large project (recall that these options absorb the same amount of labour). Table 22.1 summarizes the benefits to agents from two small projects versus a single large one. Assuming that all these benefits count equally in the welfare calculations, we see that the large project is preferable if and only if

Since nt measures the rent attributable to a large development, the capitalization argument has been turned on its head! Not only is there no net benefit attributable to rent increases, but there is actually a cost. Table 22.1 Net benefits from alternative projects Small Large

Workers 2nw 2n(w−t)

Developer π1 π2

Landlord 0 nt

22.6.2 A general urban model Let us consider now a setting which has an explicitly urban structure, in which households locate at varying distances from the central business district. Let s index distance from this centre. Being far away from the centre is a disadvantage in that commuting incurs transport costs (and we assume that households care about location only for this reason). Let φs stand for the commuting cost associated with location s. To illustrate the basic forces at work, we start with a situation in which there are no non-market amenities, lot sizes are fixed (and we normalize so that ℓ* = 1), and labour supply is perfectly elastic at wage w*. (For a more general treatment, see the appendix.) Conditional on being at location s, our representative household faces the problem

subject to

Looking at this problem, we see immediately that as long as households are mobile at the margin within the community, the sum of rents and transport costs must be constant over all locations (label this constant ψ). Note further the following relationship:(22.2)

589 where R, Φ, N, are total rent, transport costs, and population in the community, and Ns is the population residing in location s. Furthermore, for the same reasons we explained in Section 22.2, y − ψ is a perfect proxy for the common utility level in the community. Let us now think of a project whose size (g) affects the economic parameters and generates numeraire profit Π(g). The net effect of the project on community welfare (ΔW) can hence be measured as(22.3)

where the last equality follows by substitution from (22.1) above. We see that rent charges exactly cancel as in Section 22.2, but transport is an extra social cost. In the examples given above there was a one-for-one relationship between transportation and land rents so that capital gains on land correctly measured the ‘extra’ costs imposed by the project. Although the relationship is generally not so exact, we would always expect to see a positive relationship between transportation and rents whenever transport requirements influence the relative desirability of locations.116 To illustrate by example, we work out the relationship for a simple urban model with radial symmetry and linear transport costs. Radial symmetry will mean that rents and transport costs are the same for all locations at the same distance (s) from the centre. Let σ be the distance to the boundary of the community, so ℓσ is marginal land and rσ = 0. Linear transport costs means that φs = φs, where φ is the transport cost per unit distance. To facilitate computations, we assume here that distance is a continuous variable. Then, we compute transport costs as

Now recall that rents at location s are linked to transport costs according to the relationship:

the last equality following since rσ = 0. Integrating to get total community rents yields

Thus, we see that for this special geometry, transport costs are exactly twice rents and we can if we like write our welfare measure (22.3) in the form

116

For a general discussion of the transport externality and its implications for optimal city size, see Mirrlees (1972), Starrett (1974), and Arnott (1979).

590 Now, the external (‘congestion’) costs imposed by the project is actually twice the induced capital gain! Although the factor of two is special to a symmetric circular city, there is a strong presumption that transport varies more than one-for-one with rents in reality, and in any case certainly varies positively.117 Consequently, projects that raise land rents in urban areas impose an important external cost: private profits may be positive when social benefits are negative. Note that this is true even if developers do not count capital gains as part of profits—clearly, the distortion only gets worse if such gains are counted. We believe that this distortion goes a long way towards explaining the apparent overcrowding of large cities (London, Mexico City, Rio, etc.). The problem seems most acute in Third World countries where ‘new’ labour is readily available and financial payments from developers to politicians and civil servants (for permits and the like) are commonplace. The bottom line here is that transport costs should be thought of as an externality in exactly the same way that we think of congestion (indeed, transport costs are in some sense a type of congestion). To the extent that private development increases this congestion, it should be taxed just as with any non-market externality. Allowing profits to be made on capital gains is perverse in that it provides a subsidy in a situation where there ought to be a tax.118

117

See Starrett (1974) for a general discussion of this relationship.

118

Of course, there may be a simultaneity problem here just as there could be one before, in that the land-value increase might be attributable to public rather than private policy. However, we would argue that it is at least plausible that private development is responsible, whereas it was implausible before.

Appendix Here, we generalize the examples given at various points in the text. We continue to assume a homogeneous population (for a discussion of models with heterogeneity, see Starrett 1981), but otherwise employ a relatively general model of the closed economy. At the highest level of generality, we find that only the results of Section 22.2 hold true. Then, we show what extra assumptions are required to get ‘capitalization’ and ‘anticapitalization’ results. Our generalization allows for a private-goods vector of arbitrary length, a pre-existing private-production sector engaged in the production of these goods, and a public sector engaged in the provision of collective goods. We let c now stand for the vector of net consumption (that is, demand minus exogenous holding) of non-land private goods and p will be the corresponding vector of prices. Labour is naturally incorporated in this vector with cn = −n. The private-sector profit function now takes the form Π = Π(g,p,r), where g continues to stand for the level of some new private development project whose benefits we seek to measure and (p,r) are private-sector prices. We let q index the level of public projects which we assume are produced according to a cost function Γ = Γ(q,p,r).119 We assume that private firms are competitors on all markets and the public firm is a competitive purchaser of inputs. From these facts and the duality theory of production, it follows that

where (z, v) is the vector of net outputs (non-land, land) from the production sectors. We continue to assume that households are alike and share equally in all exogenous income. Here this will mean that they own equal shares of land and firms, and that they pay equal shares of public-goods costs (in the form of a nonshiftable tax). Consequently, a representative household at location i faces the problem

subject to

defining an indirect utility function

.

First we show that as long as direct effects of (g,q) are measured separately, private-goods price changes have no net welfare effect. To see this, let us measure the welfare impact from a small change in g (holding q fixed).120 In computing this measure, we count a marginal dollar equally no matter who gets it; since a dollar given to someone at location i is worth in utility we weight i's utility change by . Consequently, we compute marginal welfare change as . Let us compute ∇ pW. Using duality theory we have

119

Note that publicly provided social overhead capital (q ) does not directly affect the developer's profits; this restriction is made for simplicity only—such an effect can be easily incorporated.

120

The analysis for changes in q is very similar and is not included here.

592

where Ni stands for the number of people at location i. But at a market-clearing equilibrium net supply of all market goods must equal net demand, so ∇ pW=0. An exactly analogous argument assures us that . It follows that our welfare measure takes the form:

Therefore, capital gains on land can enter welfare measures only as a proxy for unmeasured amenity values. To derive exact proxies we must add two more assumptions: (i) non-land prices do not vary systematically over locations; and (ii) the marginal utility of income is the same everywhere. The first assumption will hold when goods markets are national in scope so that there is a uniform price throughout the country, and it could hold more generally as long as price differences are not correlated with marginal locations. The second assumption seems innocuous given that everyone is alike and gets the same utility level at equilibrium.121 Let us now write down the statement that utility change is the same at an arbitrary location (i) as it is at some marginal location where rents do not change (in doing so, we divide by the common value of Vy and cancel the common change in exogenous income):

Thus, as long as there are no systematic correlations of private-goods market changes across locations, we can write

Rent change at location i measures the differential value of amenity change just as we saw in special cases earlier. If amenity values in marginal locations are unaffected by the project, we get full positive capitalization, whereas if amenity values are more strongly affected in marginal than in intramarginal locations, we get some degree of negative capitalization.

REFERENCES ANDERSON, R., and T. CROCKER (1971), ‘Air Pollution and Residential Property Values’, Urban Studies, 8. ARNOTT, R. (1979), ‘Optimal City Size in a Spatial Economy’, Journal of Urban Economics, 6. ARROW, K., and A. FISHER (1974), ‘Preservation, Uncertainty and Irreversibility’, Quarterly Journal of Economics, 88/2.

121

Actually, the assumption is less innocuous than it seems due to the so called ‘Mirrlees problem’. To the extent that marginal residents have less amenities they may have lower or higher marginal valuations on private goods than others even when all get the same utility. See Mirrlees (1972) and Starrett (1988) for further discussion.

593 BAUMOL, W, and W. OATES (1975), The Theory of Environmental Policy (Englewood Cliffs, NJ: Prentice-Hall). BERGLAS, E. (1976), ‘Distribution of Tastes and Skills and the Provision of Local Public Goods’, Journal of Public Economics, 6. BINSWANGER, H. (1989), ‘Brazilian Policies that Encourage Deforestation in the Amazon’, Environment Dept., Working Paper, 16 (Washington, DC: World Bank). DASGUPTA, P. (1990), ‘Well-Being and the Extent of its Realization in Poor Countries’, Economic Journal, 100.(Suppl.). DASGUPTA, P., and K. MäLER, (1997), ‘The Environment and Emerging Development Issues’, this volume. EHRLICH, P., A. EHRLICH, and J. HOLDREN (1977), Ecoscience, Population, Resources and the Environment (NewYork: W. H. Freeman). FREEMAN, A. (1979), The Benefits of Environmental Improvement (Baltimore: John Hopkins University Press). JODHA, N. (1986), ‘Common Property Resources and Rural Poor in Dry Regions of India’, Economic and Political Weekly, 21. JOHANSSON, P. (1987), The Economic Theory and Measurement of Environmental Benefits (Cambridge: Cambridge University Press). JOHANSSON, P. (1990), ‘Valuing Environmental Damage’, Oxford Review of Economic Policy, 6/1. KANEMOTO, Y. (1980), Theories of Urban Externalities (Amsterdam: North-Holland). LITTLE, I., and J. MIRRLEES, (1969), Manual of Industrial Project Analysis in Developing Countries (London: OECD). LITTLE, I., and J. MIRRLEES, (1974), Project Appraisal and Planning for Developing Countries (New York: Heinemann). MAHAR, D. (1988), ‘Governmental Policies and Deforestation in Brazil's Amazon Region’, Environment Dept., Working Paper, 7 (Washington, DC: World Bank). MäLER, K. (1971), ‘A Method of Estimating Social Benefits from Pollution Control’, in The Economics of the Environment, P. Bohm, and A. Kneese, (eds.) (New York: Macmillan). MäLER, K. (1974), Environmental Economics, A Theoretical Enquiry (Baltimore: Johns Hopkins University Press). MIRRLEES, J. (1972), ‘The Optimum Town’, Swedish Journal of Economics, 74. POLINSKY, M., and S. SHAVELL, (1975), ‘Amenities and Property Values in a Model of an Urban Area’, Journal of Public Economics, 5. REPETTO, R. (1988), ‘Economic Policy Reform for Natural Resource Conservation’, Environment Dept., Working Paper, 4 (Washington, DC: World Bank). STARRETT, D. (1974), ‘Principles of Optimal Location in a Large Homogeneous Area’, Journal of Economic Theory, 9. STARRETT, D. (1981), ‘Land Value Capitalization in Local Public Finance’, Journal of Political Economy, 89. STARRETT, D. (1988), Foundations of Public Economics (Cambridge: Cambridge University Press).

This page intentionally left blank

Index Note: Italic numbers denote references to illustrations abatement costs and benefits 313, 315, 334; constant marginal abatement benefits 335; co-operation effects 335; identical-country model 333–4 absentee ownership 581–2 accounting , See environmental accounting accounting prices 10–11, 13–15, 129; meaning of 130 Acheson, J. M. 63 acidification ; acid rain 162; water-cycle as propagator 78–9, 79 activity mobility 535 acute respiratory infections 440 adaptation 62 Adelman, I. 142, 172 Africa ; air pollution 162; desertification 164, 165, 533–4; drylands management 542; elephant and ivory resources 599–60; energy consumption, regression analysis 175; forest-dwellers 58; grazing impact 4, 7; logging trade 470; population pressure 87; soil erosion 5; soil loss potential 377; water availability 90; see also individual countries Agarwal, A. 57 agreement , See co-operation agricultural intensification simulation 182–94 agricultural policy ; crop choice 492–3; discount rate change effects 491–2; price change effects 489–91; soil conservation and 486–9 agricultural production 463–4; forest conversion policies 476–9, 509 agricultural-development-led industrialization (ADLI) 172–4 air pollution ; ambient concentrations 430–1; Chinese cities 438; control of 149–56; developing countries 149–56, 162–3, 425–6; health effects 439–41; implications for policy 445–51; indoor concentrations 162, 439; indoor control policies 454–5; mobile-source-control policies 452–4; point-source control policies 444–52; policy types 442–4; pollutant categorization 426–7; social welfare improvement and 455–6; urban sources 427–9; US emissions 427; see also command and control policies; economic incentive policies Allais, M. 211 Allais paradox 210–12, 211 Allen, J. 378, 494 Anderson, A. B. 21 Anderson, J. L. 564 Arens, P. 378, 386, 494–5 Armington assumption 145 Arrow, K. J. 205, 556

Asheim, G. 132, 137 asset balance account 116 assurance game 295–6 atmosphere ; as environmental resource 5; see also air pollution Aumann, R. 285 Australia, Simpson Desert 534 automobiles ; air pollution sources 427–8; buses 452–3; cars 453–4; control policies 452–4 Bahia 55–6, 59, 60, 61, 64 balanced growth 171 bang-bang control models 556 Bar-Hillel, M. 234 Barbier, E. B. 492, 510 Barnes, D. F 428 Barrett, S. 25, 313, 314, 315, 320, 330, 333, 345, 365 Bartelmus, P. 101, 103, 104, 111, 112 Batie, S. S. 507 Battalio, R. 215 Becker, G. 223

596 Beckerman, W. 102 Bell, D. 225–6 benefit–cost analysis 403–12; in practice 404–6; in theory 403–4; institutional factors 411–12; property rights and 410–11; see also valuation methodology Benjamin, N. 148 benzene emissions 429 Berck, P. 556, 564 Bergen paradox 213 Bergman, L. 152 Bergson, A. 241 Bergstrom, J. C. 409 Bernoulli, Daniel 203 Bernoulli, Nicholas 202 Beteille, A. 20 biased demand 13 biased shadow prices 15 bifurcating solutions 278 Bigot, Y. 190 Binswanger, H. 154, 190, 542 Bintuni Bay, Irian Jaya 416–17 bio-diversity role in eco-system evolution 4–5 biofuels as air pollution sources 428 Bishop, J. 378, 494 Blaikie, P. M. 49 Blyth, C. 225 Bond, M. E. 493 Bovenberg, L. 147 Bowes, M. D. 513 Braunstein, M. 228 Brazil ; Bahia 55–6; commercial energy use 163; deforestation 154, 468, 469–70, 472; forest conversion 476–8 Bromley, D. 60 bronchitis, chronic 440–1 Brookfield, H. 49 Brooks, K. N. 375 Brown, L. R. 483 Buchanan, J. M. 448 Bumgarner, J. 441 Burbridge, P. R. 414 Burkina Faso 541 buses 452–3 1, 3–butadiene emissions 429 Cameroon ; forest-dwellers 58; Korup National Park 569–71 Canada, trade liberization 146 cancer 441 capital constraints 446 capital theory 8 capitalization 575, 582–4; policy reversal 587–0 captive breeding 563–4 carbon dioxide (CO2) emissions 162, 339–58; co-operative outcome 342–4; country size and 359; deforestation role

167, 468–9; dynamic game 353–8; emission taxes 149, 340–1, 346–9; equal percentage reductions 344–6; nonco-operative outcome 342–4, 360; participation in agreement modelled as two-stage game 365; social optimum 360; sources 427; tradable permits 349–53; two-period model 353–8 cardinal utility 202–6 Carrano, C. 365 carrying capacity 4 Carson, R. 557 cattle-ranching 167; mixed game and cattle 565–8, 567, 568 CFCs , See chlorofluorocarbon emissions Chanthaburi Province, Thailand 413–14, 413 chemical disturbance, water-cycle as propagator 78–9, 78, 79, 84 Chenery, H. B. 180 children ; attitudes towards 259–62; current use of 24; family decision-making model 262–7 China ; air pollutant concentrations 430, 431, 438; air pollution control policies 446–50; air pollution health effects 440–1; soil-conversion 378 chlorofluorocarbon (CFC) emissions 25–6, 308–32, 328; reduction benefits 317; see also Montreal Protocol on Substances that Deplete the Ozone Layer choice under uncertainty 201–46; Allais paradox 210–12; expected utility model 202–10; fanning out 211–15; framing effects 227–33; linearity in the probabilities 206–10; non-expected utility models 215–20; preferencereversal phenomenon 222–7; probability theory relevance 233–6; safety-based models 220–2; see also decision-making chronic obstructive pulmonary disease 440 cigarette smoking as air pollutant source 439, 441

597 climate, low-income country relationships 74 climatic change ; greenhouse effect 339; water-cycle as propagator 78–9 coal-burning as air pollution source 427, 428 Coase, R. 17, 35–47 CoaseTheorem 35–47 coastal wetlands , See wetlands management Collett, D. 58 command and control (CAC) policies 425–6; cost concerns 445; enforcement problems 448–9; implications for 445–51; urban air pollution 442–3 commercial energy use 163 commoditization of the environment 36–7 common-consequence effect 212–13 common-property resources 18–19, 59–60, 546–7; definition 49; erosion of 21–4; management in traditional societies 48–67; management of local resources 20; privatization 22–3; see also commons common-ratio effect 213–15, 214 commons 49; global 19, 25–7; local 19–25; tragedy of 19–20, 153 communal property 59–60 comparative risk-aversion 217–18, 219 computable general-equilibrium models 140–57; air pollution control 149–56; dynamics 155–6; energy analysis 148–9; environmental issue inclusion 149–56; evolution in developing countries 141–3; generic 143–5; lessons learned 143–9; market failures 153–5; natural resource analysis 148–9; production functions 151–3; public finance analysis 147–8; trade policy analysis 145–7 congestion tolls 453–4 Congo, deforestation 479 Conklin, H. C. 61 conservation ; benefit-cost analysis 404; property rights and 559–61; wetlands 403; see also soil conservation contingent valuation method (CVM) 409 conversion ; for agricultural production 463–4; forests 463–4, 466, 476–80, 509; soil 378; wetlands 403, 410–11, 419–21 cooking fuels 454–5; air pollution sources 428–9, 439 cooking stoves 455 co-operation 283–305; abatement effect 335; assurance game and 295–6; benefits 575, 315; CO2 emmissions and 342–6; experience in developing countries 296–8; model of trust 288–90; partial co-operation efficiency issues 311; participation modelled as two-stage game 365; policy implications 304–5; repeated games 284–96; South Indian study 299–304; success determinants 303 Cordell, J. C. 55–6 Costa Rica, depreciation of natural resources 12 Cramer, Gabriel 203 crop choice 492–3

cropping areas ; land degradation 531; land tenure 545 Crosson, P. R. 490 cultivation , See agricultural production current-value Hamiltonian 132–3, 137–8; sustainable wellbeing and 137–8 Dales, J. H. 26 dams ; debris dams 380; deforestation and 510; environmental impact 385–6, 387–8; Narmada River dams, India 385 Dani, A. A. 382 Dasgupta, R S. 102, 104, 131 de Fermat, Pierre 202 de Melo, J. 146 de Neufville, R. 215 Deacon, R. 154 Deaton, A. 148 debris dams 380 decision-making 236–46; delayed-resolution risk implications 242–3; fanning-out preference implications 241–2; modified decision problem 515; private sector 238–40; public 240–1; response-mode effect implications 244–5; threeperiod decision problems 513 deer stalking 565

598 defensive expenditures 101–2 defensive industries 111, 112 deforestation 166–8, 468–72; ecological impact 164–5, 471–2; tropical deforestation 505, 505; watersheds 373; see also forests DeGroot, M. 223 delayed-resolution risks 242–3 desalinization 7 desertification 83, 164–5, 529–30; causes of 533; definition 530; land-reform programmes as cause 22; see also drylands management Devarajan, S. 146, 152 developing countries ; air pollution , See air pollution; computable general-equilibrium models 141–3, 149–57; co-operation experience 296–8; deforestation 166–8; desertification 164–5; environmental problems 80–3, 81, 161–8; land degradation 164–6; mesoscale landscape approach 80–2; Montreal Protocol provisions 325–30; soil erosion 164–6; water pollution 163–4; water scarcity 82–3, 163–4 development 574–90; benefits, wetlands 410–11; capitalization argument 582–4; crowded areas 587–90; economic framework 576–8; economics, environmental issues 1–2; simultaneity problem 584–7 development strategies 168–74; agricultural intensification simulation 182–94; agricultural-development-led industrialization 172–4; balanced growth 171; energy consumption and 174–6; export-led growth 170; import substitution 169–70; integrated rural development 371; simulation experiments 176–82; staple export strategies 172 diesel engines 428 Dixon, J. 391, 569 Doolette, J. B. 373, 374 Dore, R. 298 dredging 380 Dregne, H. E. 530 Dréze, J. 1 drought , See water scarcity drylands ; adaptations to 534–6; definition of 529; degradation causes 532–4; degradation processes 530–2; Sudan example 538–40 drylands management 529–52; economic factors 537–40; institutional issues 549–52; investment priorities 541–3; land tenure 544–6; legislation 543–4; policies for management improvement 540–1; policy distortions 537–40; price incentives 546–9; research issues 549–52; sustainability 542; technology potential 536–7 Dunkerley, J. 428 duplex gambles 228, 229 ‘Dutch disease’ models 143, 148–9, 156 ecological services 2–3 ecological valuation 104, 111

economic incentive (EI) policies ; cost concerns 445; enforcement problems 448–9; implications for policy 445–51; urban air pollution 442–3 eco-systems 2–3; carrying capacity 4; evolution, role of biodiversity 4; resilience 3–4 Ecuador, mangrove valuation 415 Edwards, W. 225 effluent fees 443 Eisner, R. 97, 98 El Nido watershed, Philippines 390; competing resource-users 389–92, 392 El Serafy, S. 98, 99 elephants 559–60; ivory trade 559–60; viewing value 571–2 Ellsberg, D. 234 Eltringham, S. K. 565 emissions fees 443 emissions permits , See tradable emissions permits emissions taxes 149, 340–1, 346–9 endogenous fertility 259–74; environmental–demographic interactions 268–73; family decision-making model 262–7 energy analysis 148–9 energy consumption ; development strategy relationships 174–82; regression analysis 174–6; simulation experiments 176–82

599 Ensminger, J. 22 environment 91; change, multi-cause syndrome 91; commoditization 36–7; developmental strategy relationships 161–95; property rights to 39–40; statistics 97; value of, non-concavity 40–1 environmental accounting 96–126; asset balance account 116; defensive expenditures 101–2; defensive industries 111, 112; framework 103–17; GDP redefinition 97–102; Indonesian example 119–26; land degradation 112; natural products 111–12; natural resources 98–101; social accounting matrix 104–10, 120; use-value added 112–13, 114–15; waste products 104, 111 environmental–demographic interactions 268–73 environmental multipliers, Indonesia 123–6, 124–5 environmental problems 140, 149; air pollution 162–3; deforestation 166–8; desertification 164–5; developing countries 80–3, 87, 161–8; institutional failure as cause 8–10; land degradation 164–6; poverty as cause 8–10; soil erosion 6, 164–6; water pollution and depletion 163–4 environmental quality ; land values and 580–2; links with wealth 10; time-paths 280–1 environmental resource-base 3; degradation of 3, 12 environmental resources ; accounting for 98101; atmosphere 5; computable general-equilibrium modeling 148–9; economic analysis 1–2; forests 7–8; groundwater 7; land 5–7; management of 49–50; renewable 3; water 73–95 Ethiopia, forest-dwellers 58 expected utility model 201, 202–10; classical perspective 202–6; linearity in the probabilities 206–10; non-expected utility models 215–20; violations 237 export-led growth 170 externalities, uncertain 44–6 extraction path studies 556 family decision-making model 262–7; dynamics of 268–73 fanning out 210–15, 211, 214, 216; implications for decisionmaking 241–2 farming , See agricultural production Feder, E. 22 fertilizer pollution, water-cycle as ; propagator 78, 78 Fiji ; mangrove forest benefit-cost analysis 404–5, 408, 410–12; mangrove forest valuation 415–16 First Welfare Theorem 153 first-order welfare calculus 579–80 Fischhoff, B. 232 Fisher, A. C. 386, 511, 523 fisheries ; Bahia 55–6; competing resource use, Philippines 389–92 flexibility value 516–19 FolkTheorem 285–7 food production, water scarcity and 90 ‘Fordist’ approach 537 forest codes 544

forest-dwellers 56–9 forestry 509; logging 465–6, 469, 470, 509; timber supply 465; timber values 471, 471 forests ; as environmental resources 7–8; conversion 463–4, 466, 476–80, 509; macroeconomic policy 463–8; management impacts 472–6; policy influences on tropical forests 468–72; products 57; regeneration 133; see also deforestation; tropical forest valuation fossil fuels ; coal-burning as air pollution source 427, 428; taxes on 347, 479 fractile method 205 framing effects 227–33, 521; economic analysis and 232–3; evidence 227–31; issues 231–2; public policy implications 245–6

600 Friedman, M. 205 Fudenberg, D. 285 Gabon, forestry policy 474 game theory ; assurance game 295–6; CO2 emissions analysis 342–4, 353–8; co-operation in repeated games 284–96; Montreal Protocol and 310; subgame perfect equilibrium 366 Gentry, A. H. 511 Gentry, J. 223 geographic information-based system (GIS) model 386 Germany, effluent fees 443 Ghana, deforestation 469, 470, 471 Ghosh, P. 152, 153 global commons 19, 25–7 global efficiency problem 313; identical-country model 333–4 Global Environmental Monitoring System (GEMS) 430–1; SO2 levels 430–1, 432–3, 436; suspended particulate matter 430–1, 434–5, 437 global public bads (GPBs) 308–9, 312–13; model 312–13 Golan, Elise 183 goods , See products and services Gordon, H. S. 60, 64 gradient process 131 Gradwohl, J. 507, 509 Graham, J. D. 428 grazing ; overgrazing 532; semi-arid grasslands 45 Greenberg, R. 507, 509 greenhouse gases 162; climatic change 339; deforestation effects 468–9; emission of 26–7; taxes on 479–80; tradable emissions permits 26–7; water-cycle as propagator 79, 79; see also carbon dioxide emission; ozone layer depletion Greenley, D. A. 409 Grether, D. 223, 234 gross domestic product (GDP) redefinition proposals 97–102 groundwater 7 growth ; balanced 171; export-led 170 Guatemala, forest management 472 Gunasekara, D. 146 habitat management 555–8; optimal use, sources of departure from 556–8; valuing habitat 569–72 Hahn, R. W. 310 Hames, R. 62–3 Hamilton, L. S. 375, 376, 378, 412 Hamiltonian, current-value 132–3, 137–8; sustainable wellbeing and 137–8 Harberger, A. 147 Harris, R. 146 Harrison, A. 97, 101 Hartwick, J. M. 99, 101, 132, 137 Heady, C. 147 health, air pollutant effects 439–41

Hecht, S. 21 Hicks, J. R. 97–8 hill-climbing method 131 Hodgson, G. 391 Hoel, M. 320, 323, 324–5, 330, 365 homoclinic loop 271 Hong Kong, congestion tolls 454 Hotelling, H. 98, 556 Hotelling’s rule 99–100, 102 Howe, J. 20 Hudson, E. 152 hunting and gatherering 57, 61, 63 hydrologic cycle , See water-cycle hysteresis 278–9 identical-country model 333–4 import substitution 169–70 income, national, definition of 97–9 income cancellation 578–9 incomes approach to wetland valuation 407–8 independence axiom 209–10 India ; common-property resources 20–1, 22; deforestation rate 468; forest-dwellers 57–8, 59; milk producers’ cooperative societies 299–304; Narmada River dams 385; see also individual states indifference curves 207–8, 208; Allais paradox and 211; fanning out 211, 214, 216; non-expected 219; regret model 226, 226; risk-averter 209; risk-lover 209 Indonesia ; environmental effects model 121–3; environmental multipliers 123–6, 124–5; forestry policy impacts 473–5; logging trade 469, 471; mangrove valuation 414; social accounting matrix 119–21; transmigration programme 477

601 indoor air pollution 162; concentrations 439; control policies 454–5 industry, toxic emissions 429 information presentation, public and corporate obligations 245–6 institutional failure as cause of environmental degradation 8–10 integrated rural development 371 intrinsic worth 14 investment 12 Irian Jaya, mangrove valuation 416–17 irreversibility in use 15 irrigation 163–4, 165; salinization and 6–7 Ivory Coast ; deforestation 469, 470, 479; forestry policy 473, 474, 478 ivory trade 559–60 Japan, timber importation 475–6 Java, soil erosion 378, 386–9, 389, 494–5 Jodha, N. S. 20–1, 22, 49 Johnson, R. 556, 564 Jorgensen, D. 149, 151, 152, 155, 156 Kagel, J. 215 Kahneman, D. 213, 214–15, 230, 234 Kaldor–Hicks potential compensation criterion 521–2 Kara Kum desert, Russia 532 Karmarkar, U. 215 Keller, L. 231 Kenya 545; common-property resource privatization 22; elephant resource management 560; tourism 571 Kerala 50–1, 59; KodayarDam 51 Keur Marie, Senegal ; agricultural intensification simulation 182–94; land degradation 189; social accounting matrix 184–7 Khan, H. A. 119 King, P. N. 375, 378 Kodayar Dam, Kerala 51 Korea ; balanced growth 171; trade liberization 146 Korup National Park, Cameroon 569–71 Kosmo, M. 447 Kosrae, Micronesia, resource planning districts 418–19, 420 Krieser, G. 231 Krupnick, A. J. 439 Krutilla, J. V. 386, 511, 513, 523, 557 Kurz, M. 556 Kverndokk, S. 345 labour/capital ratio 537 Lancaster, K. J. 102 land ; as environmental resource 5–7; see also conversion land degradation ; accounting for 112; causes in drylands 532–4; cropping areas 531; developing countries 164–6; economic factors 537–40; pastoral areas 531; physical processes 530–2; policy distortions 537–40; salinization 6–7, 165–6; Senegal example 189; water-related 75; see also

deforestation; desertification land/labour ratio 536–7 land tenure, drylands 544–6; cropping areas 545; forested areas 546; pastoral areas 544–5 land values ; environmental quality and 580–2; increases as real benefits 582–4; welfare and 578–80 Landes, D. S. 173 landscape 80, 90–1; misuse matrix 91–2, 92 lead ; concentrations in blood 431–9, 438; emission sources 429 leaded fuel 429, 438–9, 452; differential taxes 443 Lesotho, common property 60 Libecap, G. 556 Lichtenstein, S. 222–3, 224 limit cycle 272 Lindman, H. 223 linearity in the probabilities 206–10; violations of 210–22 Liouzhou, China, sulphate levels 431 Lippmann, M. 439 lipton, M. 484, 492 local commons 19–25 logging 167, 465–6, 469, 470, 509; competing resource use, Philippines 389–92 longevity, pollution and 36, 41–4 Louisiana, wetlands valuation 409 Lumpinee Park, Thailand 569 Mabbut, J. A. 164

602 Macauley, M. 454 McCay, B. J. 63 McConnell, K. E. 487 McCord, M. 215 MacCrimmon, K. 212 MacDonald, D. 215 McKean, M. A. 55–6 McNicholl, G. 260 Magrath, W. B. 49, 373, 374, 378, 386, 494–5 Mahmood, K. 379, 380 Malaysia ; coastal wetland products 401–2; mangrove valuation 414; Matang Mangrove Forest Reserve 401, 414 Mäler, K.-G. 102, 104 Mali, soil erosion 378, 494 Malthusian mortality 260, 261 management ; common property resources in traditional societies 48–50, 60–7; definition of 49–50; forests 472–6; habitat 555–8; local resources 20; protected areas 562; see also drylands management; watershed management; wetlands management; wildlife management mangrove forests 399–421; benefit–cost analysis in practice 404–6; benefit–cost analysis in theory 403–4; conversion 403, 410–11, 419–21; development benefits 410–11; development pressures 402–3; Ecuador 415; Fiji 404–5, 408, 410–12, 415–16; Indonesia 414; Irian Jaya 416–17; Malaysia 414; products 402; safe minimum standards evaluation approach 418–19; Thailand 413–14, 413; valuation examples 412–17, 412, 413; see also wetlands Maragos, J. E. 414 marginal abatement benefits, constant 335 marginal abatement costs 313 marginal land 581–2 marginal rate of substitution 206 Marglin, S. A. 131 Markandya, A. 510 market distortions 447–8 markets, failure of 16 Markowitz, H. 229 Marschak, J. 223 Martin, R. 148 Maskin, E. 285 Matang Mangrove Forest Reserve, Malaysia 401, 414 Matlon, P. J. 534–5 May, K. 225 May, P. 21 Meade, J. E. 96 Mellor, J. W. 172, 542 Mendelsohn, R. O. 511 Metcalfe, C. 48 Mexico ; air pollution 150; development strategy simulation

experiments 176–82; social accounting matrix 177–9 Micronesia, resource planning districts 418–19, 420 milk producers’ co-operative societies 299–304 Miller, K. R. 508 mining 510 Mirrlees, J. A. 131, 147 Mitchell, R. 557 Mitra, P. 147–8 mixed game and cattle ranching 565–8, 567, 568 Miyashita, S. 231 mobile-source-control policies 452–4; buses and trucks 452–3; cars 453–4; leaded fuels 452 Montreal Protocol on Substances that Deplete the Ozone Layer 25, 308–32, 328, 331; 1990 revisions 332; compensation to developing countries 325–30; equilibrium number of signatories 336–7, 336; implications for future optimal treaty 323–5; intersignatory trade 322–3; non-signatory effects on cheating behaviour 321–2; as optimum treaty 315–20; shortfalls 310–11; signatory cheating behaviour 320–2 morbidity, respiratory diseases 440 Morrison, D. 211 mortality ; Malthusian 260, 261; respiratory diseases 440 Mortimore, M. 535 Moskowitz, H. 212, 231 motor vehicles , See automobiles Mowen, J. 223 Myers, N. 486

603 Narain, S. 57 Narmada River dams, India 385 national income, definition 97–9 natural products, creation of 111–12 natural resources , See environmental resources Neher, P. A. 259 Nepal, soil erosion 377–8 Nerlove, M. 273 net national product (NNP) 129–39; biases 13–14; in deterministic environment 133–7; future uncertainty 138–9; as measure of social well-being 129–31; measurement of 10–12 Netting, R. 65 neutrality 578–80 Nigeria, common property 60 nitrogen dioxide (NO2) emission sources 428 non-co-operative behaviour 342–4, 360 non-expected utility models of preferences 215–20; implications for decision-making 241 Norway, tax on coal sulphur content 443 Oates, W. E. 448 optimal control ; economics 131–3; maximum principle formulation 497–9; soil erosion model 486–9 option value 15 overgrazing 532 ozone layer depletion ; depletion potentials 319; global economy model 312–13; see also chlorofluorocarbon emissions; Montreal Protocol on Substances that Deplete the Ozone Layer Pakes, A. 570 Palawan, Philippines 390; competing resource-users within water-shed 389–92, 392; logging roads impact 373 Pandian, M. S. S. 50–1 Pareto, V. 241 particulate pollution ; concentrations 430–1, 434–5, 437; sources 428 Pascal, B. 202 pastoral areas ; land degradation 531; land tenure 544–5 Payne, J. 228 Pearce, A. J. 375, 376 Pearce, D. A. 97, 98 Pearce, D. W. 510 Pereira, H. C. 375 perfect treaty 312; model 312–13; Montreal Protocol evaluation 315–17 periodic solutions 277–8 Perloff, J. M. 564 Persson, A. 154 Peters, C. M. 511 petrol ; air pollution sources 429; differential taxes 443; lead sources 429, 431–9 Philippines ; competing resource-users within water-shed

389–92, 392; forest conversion 476, 477; forest-dwellers 57; forestry policy impacts 472–5; logging roads impact 373; logging trade 469, 470, 478 Pigou, A. C. 44, 46 Pigovian taxes 41, 44, 46–7 Pingali, P. 190 Plott, C. 223 point-source pollution control policies 444–52 pollution ; costs 41; fertilizers 78, 75; longevity and 36, 41–4; particulate pollution 428, 430–1, 434–5, 437; point-source control policies 444–52; pollutant categorization 426–7; property rights 38–40; see also air pollution; water pollution Pommerehne, W. 223 population growth ; avoidable 92–3; food demands 90; timepaths 280–1; unavoidable 93–4; under water scarcity 84–90; water availability and 86 poverty, as cause of environmental degradation 8–10 Pratt, J. 205 preference function 207 preference-reversal phenomenon 222–7; evidence 222–4; implications for decision-making 244; implications of economic worldview 225–6; implications of psychological worldview 226–7; interpretations 224 preferences , See choice under uncertainty pricing policy ; common-property issue

604 546–7; counteracting economic impacts 548; discounting of future benefits 547; drylands management and 546–9; future expectations 548; livestock supply-response problem 548–9; price adjustment as blunt instrument 547–8; soil conservation and 492–3; technical interrelationships aspect 547; underpricing 13, 14, 17 private sector decision analysis 238–40 probabilities ; linearity in 206–10; subjective 233–5; violations of linearity 210–22 probability theory 233–6, 238–9 production function approach 14 products and services ; accounting for 111–12; ecological services 2–3; forests 57; mangrove forests 402; tropical forests 519–20; wetland valuation approach 407–8; wetlands 400, 401–2 project evaluation 10–12 property rights 16–17, 43–5; assignment of 35–6; conservation use 559–61; environment 39–40; lack of 153–4; to pollute 38–40; wetlands 410 protected areas ; categories and management 562; growth of 563; public provision 561 public decision-making 240–1 public finance analysis 147–8 public policy 12, 574–90; capitalization argument 582–4; development in crowded areas 587–90; economic framework 576–8; environmental quality and land values 580–2; land values and welfare 578–80; public amenities and private development 584–7 Pyatt, G. 119, 121 Raiffa, H. 211 Ramakrishna, J. 455–6 Ramsar Convention 399 Razin, A. 273 Rees, J. A. 62 reference point 229 regression analysis, energy consumption 174–6 regret theory model 225–6, 226 regret–rejoice function 225 Reid, W. V. 508 Reilly, R. 223 renewable natural resources 3 Repetto, R. W. 97, 98, 99, 119, 123, 483, 505–6 research ; costs 557–8; drylands management issues 549–52 reservoirs, environmental impact 385–6, 387–8 resilience 3–4 resource allocation 25–7; modern theory 10 resource consumption ; accounting for 99–101; exhaustion 556; extraction path studies 556; sustainable consumption 100 resources , See environmental resources respiratory diseases 440–1 response-mode effects 224, 226–7; implications for decision-

making 244–5 revealed preferences 408–9 Richards, K. R. 310 risk attitudes ; Allais paradox 210–12; expected utility model 202–10; fanning out 211–15; global risk-aversion 219, 219; linearity in the probabilities 206–10; non-expected utility models of preferences 215–20; safety-based models 220–2 risks ; comparative risk-aversion 217–18, 219; delayedresolution 242–3 Robinson, S. 142 Rodrik, D. 146 Roland-Hoist, D. 146 Roose, E. J. 377 Round, J. 119, 121 Rubinstein, A. 285 Ruitenbeek, H. J. 416–17, 570–1 Russia, Kara Kum desert, land degradation 532 Russo, J. 231 saddle-source connection 271–2 Sadka, E. 273 safe minimum standards evaluation approach 418–19 safety principle 220, 222 safety-based models 220–2 safety-fixed principle 221, 222 Sahel 535, 536; desertification

605 533–4; soil erosion 5 St Petersburg paradox 202–3 salinization 165–6; irrigation and 6–7 Salisbury, R. F. 63 Samuelson, P. 241 Sandford, S. 532, 533 Savage, L. 205, 212 Savory, A. 532 Schankerman, M. 570 Schneider, F. 223 Schultz, T. W. 259, 482 Seabright, P. 300 sectoral policies 442; soil conservation and 485–6 sedimentation 378–81; control of 379–81 Sefton, J. A. 100 self-confirming beliefs ; with extrinsic uncertainty 290–1; without extrinsic uncertainty 291–3 semi-arid grasslands, grazing effects 4–5 Sen, A. 1, 131 Senegal, Keur Marie , See Keur Marie services , See products and services shadow prices 10–11, 13–15, 129; biased 15 Shapley, L. 17, 285 Sherman, P. 569 Shubik, M. 17 Simpson Desert, Australia 534 simultaneity problem 584–7; mutually exclusive uses 585–6; potential joint use 586–7 Singapore, Area Licence Scheme 453–4 Siniscalco, D. 365 Slovic, P. 211, 212, 222-3, 224, 228 Smith, K. R. 428–9, 439, 455–6 smoking as air pollutant source 439, 441 Snedaker, S. C. 412 social-accounting matrix 105–10, 120; Indonesian example 119–21; Mexico 177–9; Senegal, Keur Marie 184–7 social overhead capital 574–5 social well-being 10–11; Hamiltonian and 137–8; net national product as measure 129–31 soil conservation ; agricultural policy and 486–9; crop choice and 492–3; discount rate change effects 491–2; empirical evidence 493–5; policy influences 484, 485–6, 495–6; price change effects 489–91; pricing policy and 492–3 soil erosion 6, 164–6, 483–5; costs of 378, 494; deforestation effects 471–2; Java study 378, 386–9, 389; optimal control model 486–9; Universal Soil Loss Equation (USLE) 377–8, 494; watersheds 373, 376–8 soil erosion-productivity model 386 soil quality 5–6 Solorzano, R. 12 Solow, R. M. 65, 99, 132, 137 South Africa, ivory trade 560

South Korea, balanced growth 171 spatial mobility 535 Speizer, F. E. 441 Spence, A. M. 15 Spooner, B. 64 Stahmer, C. 101, 103, 104, 111, 112 staple export strategies 172 Starrett, D. A. 17, 40 state-preference model 235–6 statistics, environmental 97 steady-state solution 275–6 Stern, N. H. 1 stochastic dominance preference 217, 218–19, 228 Stone, J. R. N. 96 Stout, A. T. 490 Strassmann, D. L. 448 strict safety-first principle 221, 222 subgame perfect equilibrium 366 subjective probabilities 233–5 Sudan 51–5, 60; dryland degradation 538–40, 548 sulphur dioxide (SO2) emissions 162; concentrations 430-1, 432–3, 436; sources 427, 428 survey-based valuation techniques 409, 520–1 suspended particulate matter (SPM) 430–1, 434–5, 437; indoor concentrations 439 sustainable systems 64–5; drylands management 542–3; resource consumption 100; well-being 137–8 Sweden, CFC-use reduction 322–3 System of National Accounts (SNA) 96–7, 103 Taiwan, balanced growth 171 Tamil Nadu, milk producers’ co-operative societies 299–304 Taylor, L. 142 technological change 13 Tennessee Valley Authority (TVA) 372–3 Terkla, D. 447 Thailand ; forestry policy impacts

606 473; logging ban 469; Lumpinee Park valuation 569; mangrove valuation 413–14, 413 Thorbecke, E. 119 three-period decision problems 513 timber supply 465; values 471, 471 tobacco smoke as air pollutant source 439, 441 total economic value (TEV) 407–9 total factor productivity (TFP) 146 tourism, Palawan, Philippines 389–92 ‘Toyotaist’ approach 537 tradable emissions permits 26–7, 443–4; CO2 emissions 349–53; enforcement 449; initial distribution 26–7 trade liberization 145–6 trade policy analysis 145–7 traditional societies ; Bahia 55–6; common property resource management 48–67; definition of 50; forest dwellers 56–9; Kerala 50–1; Sudan 51–5 tragedy of the commons 19–20, 153 transaction costs 37–8, 46 transferable national rights 26 travel cost methods 409, 520 tropical forest valuation 505–25; empirical issues 519–23; framework 510–19; uses of tropical forests 506–10; see also forests Tropical Forestry Action Plan 478–9 trust ; model of 288–90; renegotiation and 295; as strategic variable 294–5 Tucker, C. J. 530 Turkey, trade liberization model 146 Turner, R. K. 406–7 Turton, D. 58 Tversky, A. 211, 212, 213, 214–15, 225, 230, 234 Tyers, R. 146 uncertainty 138–9; delayed-resolution nature 242–3; future use-values 15; see also choice under uncertainty uncooperative behaviour 308–9 underpricing 13, 14, 17 Universal Soil Loss Equation (USLE) 377–8, 494 urban air pollution , See air pollution use-value 14 use-value added 112–13, 114–15 utility functions 203–6, 216, 217, 218; risk-averter 204; risklover 204 utility models ; expected 202–10; non-expected 215–20 utility-evaluation effect 215 valuation ; ecological 104, 111; habitat 569–72; mangrove forests 412–17, 412, 413; total economic value 407–9; wetlands 406–9, 406, 407; wildlife 569–72; see also tropical forest valuation valuation methodology 406–9; contingent valuation method 409; direct methods 520–1; incomes approach 407–8; indirect techniques 520; products and services approach

407–8; revealed preferences 408–9; survey-based techniques 409, 520–1; travel-cost method 520 van Tongeren, J. 101, 103, 104, 111, 112 van Wijnbergen, S. 148 volatile organic compound (VOC) sources 427, 429 von Neumann–Morgenstern utility of wealth function 203, 215, 221; recovery of 205 Wade, R. 18, 20 Walrasian models 141, 142 Walsh, R. G. 409 waste product accounting 104, 111 water ; availability 76–7; demand 88–9; depletion 163–4; groundwater 7; as natural resource 73–95; importance of 73; population effects on availability 86; resources analysis 371; see also dams water pollution 75; developing countries 163–4 water scarcity 75; developing countries 82–3, 163–4; differences in predicament 87; food production and 90; genuine 82–3; low-income country relationships

607 74; man-induced 83; population growth implications 84–90; threats to ecological security 85–7; see also desertification water-cycle 75–80, 375, 375; as disturbance propagator 77–80, 78, 79, 84; macroscale 76–7, 76; population growth and 84–5 watershed management 371–96, 381; dams and reservoirs 385–6, 387–8; definitions 374; deforestation 373; economics of 384–92; highland–lowland relationships 382–3, 383; management responses 393–4; multi-level management 394–6; Palawan study of competing resource-users 389–92, 392; sedimentation 378–81; sociology of 382–4; soil erosion 373, 376–8, 386–9, 389 Weale, M. R. 100, 121 Webster, A. 49 Weisbrod, B. 557 Weitzman, M. L. 99, 445 welfare ; air pollution and 455–6; first-order welfare calculus 579–80; land values and 455–6 well-being , See social well-being West Africa , See Africa wetlands management 399–421; benefit valuation 410; benefit–cost analysis in practice 404–6; benefit–cost analysis in theory 403–4; conservation 403; conversion 403, 410–11, 419–21; development benefits 410–11; development pressures 402–3; institutional factors 411–12; products of 400, 401–2; property rights 410; safe minimum standards evaluation approach 417–19; valuation of 406–9, 406, 407; see also mangrove forests Wilcoxen, P. 149, 151, 152, 155, 156 wildlife management 555–72; commercial wild-species rearing 563–4; elephant resource 560; mixed game and cattle ranching 565–8, 567, 568; optimal use, sources of departure from 556–8; private provision 561–8; public provision 561; valuing wildlife 569–72 Willis, R. J. 259 Wilson, P. W. 454 women, workload of 23 Yeldan, E. 172 Young, R. A. 409 Zaire ; forest-dwellers 58; forestry policy 474–5 Zeckhauser, R. 40 Zimbabwe, ivory trade 560 Zweifel, P. 223 Zylic, T. 506