Budget Deficits and Economic Activity in Asia The growth and persistence of government budget deficits is causing incre...
29 downloads
1281 Views
1MB Size
Report
This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form
Budget Deficits and Economic Activity in Asia The growth and persistence of government budget deficits is causing increasing concern in both developed and developing countries. They have provoked extreme responses: some economists hold that they have devastating effects, others that they have no real impact at all. Budget Deficits and Economic Activity in Asia examines both of these claims in the context of the Asian economies. After testing for the feasibility of the current levels of budget deficits and therefore of the current fiscal policies, the study turns to a quantification of the effects on seigniorage, money supply, inflation, aggregate demand and interest rates. The findings for the ten countries studied are far from uniform, but neither of the extreme positions is vindicated. Budget deficits are monetized to a considerable extent, thus impairing, or at least reducing, the ability of the monetary authority to pursue an independent monetary policy. On the other hand, government expenditures in most of the countries covered unambiguously, though only partially, crowd out private expenditures. The widespread view that budget deficits are inflationary because they increase the money supply receives only partial support. The apparent effect on interest rates is more interesting. It appears to be positive and as the processes of fiscal deregulation accelerate, interest rates seem set to become even more sensitive to the behaviour of budget deficits. The countries covered include India, Indonesia, South Korea, Malaysia, Pakistan, the Philippines, Singapore, Sri Lanka, Taiwan and Thailand. Kanhaya L.Gupta is Professor of Economics at the University of Alberta, Edmonton. He has published widely in the area of Development Economics, particularly on financial issues. His previous publications include Finance and Economic Growth in Developing Countries (Routledge, 1984) and Industrialisation and Employment in Developing Countries (Routledge, 1989).
Budget Deficits and Economic Activity in Asia Kanhaya L.Gupta
London and New York
First published 1992 by Routledge 11 New Fetter Lane, London EC4P 4EE This edition published in the Taylor & Francis e-Library, 2006. “To purchase your own copy of this or any of Taylor & Francis or Routledge’s collection of thousands of eBooks please go to http://www.ebookstore.tandf.co.uk/.” Simultaneously published in the USA and Canada by Routledge a division of Routledge, Chapman and Hall, Inc. 29 West 35th Street, New York, NY 10001 © 1992 Kanhaya L.Gupta All rights reserved. No part of this book may be reprinted or reproduced or utilized in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage and retrieval system, without permission in writing from the publishers. British Library Cataloguing in Publication Data Gupta, Kanhaya L. (Kanhaya Lal) 1935– Budget deficits and economic activity in Asia. 1. Asia. Economic conditions I. Title 330.95 ISBN 0-203-02616-0 Master e-book ISBN
ISBN 0-203-14933-5 (Adobe e-Reader Format) ISBN 0-415-05540-7 (Print Edition) Library of Congress Cataloging in Publication Data Gupta, Kanhaya L. (Kanhaya Lal), 1935– Budget deficits and economic activity in Asia/by Kanhaya L.Gupta. p. cm. Includes bibliographical references and index. ISBN 0-415-05540-7 1. Budget deficits—Asia. 2. Asia— Economic conditions—1945– I. Title. HJ2151.G87 1991 339.5′23′095–dc20 91–13668 CIP
For Carol Marshall
Contents Figures
vii
Tables
viii
Acknowledgements
xii
Introduction 1 A review of the record
1 5
2 Sustainability of perpetual deficits
32
3 Deficits and seigniorage
51
4 Deficits and money supply
91
5 Deficits and inflation
102
6 Deficits and aggregate demand
125
7 Deficits and interest rates
145
8 Conclusions
169
Notes
172
Selected bibliography
174
Index
178
Figures 1.1 Nominal and real federal deficits as percentage of GDP
6
1.2 Government expenditure and taxes as percentage of GDP
11
1.3 Rates of growth of reserve money, M1, and nominal deficits
30
3.1 Estimates of seigniorage
53
4.1 Money multipliers for M1 and M3
92
5.1 Nominal deficits, growth of M1, and rate of inflation
106
7.1 Nominal and real interest rates
146
7.2 Nominal interest rates and budget deficits
149
Tables 1.1 Nominal and real deficits as percentage of GDP
7
1.2 Variances of nd and rd
8
1.3 Components of nominal deficits as percentage of GDP
10
1.4 Central government expenditure as percentage of GDP
12
1.5 Public sector revenues as percentage of GDP
15
1.6 Correlation coefficients between nd and rd
18
1.7 Monetary erosion and inflation tax on public debt
19
1.8 Government debt as percentage of GDP
23
1.9 Variability of the major variables
24
1.10 Mean values of the major variables for different periods
27
1.11 Correlation coefficients between nd, rd, and variables of main interest
29
2.1 Data for analysing the sustainability of public debt
33
2.2 Values of real interest rate minus real growth rate
38
2.3 Values of the condition for instability of the debt/ GDP ratio
40
2.4 Federal debt dynamics
42
2.5 Unit root tests for budget deficits excluding interest payments
48
2.6 Unit root tests for government debt
49
3.1 Table for S1 and S2—alternate estimates of seigniorage
51
3.2 Monetary base growth and the share of government
55
3.3 Budget deficits and the degree of monetization
59
3.4 Estimates of equations (6) and (7)
63
3.5 Increases in reserve money attributable to current and past government borrowing requirements
70
3.6 Contemporaneous correlations between seigniorage and deficit
74
3.7 Dynamic estimates for S1
78
3.8 Dynamic estimates for S2
82
3.9 Test statistics for the null hypotheses for S1
86
3.10 Test statistics for the null hypotheses for S2
87
3.11 Total short-term and long-term effects of past deficits on seigniorage S1
88
3.12 Total short-term and long-term effects of past deficits on seigniorage S2
89
4.1 Money multiplier values
93
4.2 Money growth estimates (M1): the conventional model
95
4.3 Effects of deficits on money growth, M1: the static and the dynamic models
97
4.4 Effects of deficits on money growth, M3: the static and the dynamic models
98
4.5 Comparisons of the effects of budget deficits on monetary base 99 and money growth 4.6 Effects of budget deficit on money growth in other studies for
100
LDCs 5.1 Results of causality tests for the works surveyed
104
5.2 Estimates of equations (9), (10), and (11)
111
5.3 Estimates of equation (12)
114
5.4 Summary of structural models’ results
117
5.5 Contemporaneous relationships of inflation with money growth 118 and budget deficits: equation (14) 5.6 F values for the effects of each money growth term to be zero on inflation (Inf)
119
5.7 Total short-run effect of money growth on inflation
120
5.8 F values for the effects of each deficit term to be zero on inflation
121
5.9 Total short-run effect of deficits on inflation
122
5.10 Long-term effects of deficits on inflation
123
6.1 Estimates of equation (10)
130
6.2 Estimates of the Aschauer model: Version 1
133
6.3 Values of the log-likelihood ratio test statistic: Version 1
137
6.4 Estimates of θ and log-likelihood ratio test statistic: Version 2
138
6.5 Estimates of the modified Aschauer model: Version 3
139
6.6 Values of the log-likelihood ratio test statistic for the Aschauer model: Version 3
142
7.1 India: twelve-month deposit rate
152
7.2 India: money market rate
153
7.3 India: bond rate
154
7.4 India: bond rate
155
7.5 South Korea: twelve-month deposit rate
156
7.6 Malaysia: money market rate
157
7.7 Pakistan: twelve-month deposit rate
158
7.8 Pakistan: money market rate
159
7.9 Pakistan: bond rate
160
7.10 The Philippines: twelve-month deposit rate
161
7.11 Singapore: twelve-month deposit rate
162
7.12 Sri Lanka: twelve-month deposit rate
163
7.13 Thailand: twelve-month deposit rate
164
7.14 Thailand: bond rate
165
7.15 Test statistics for restrictions on the coefficients of deficits
166
Acknowledgements Thanks are due to Daniel Lal and Zheng Wei for their research assistance, to Alan Sharpe for his computing help, to Pat Gangur for her excellent typing of the manuscript, and to Michael Fisher, Cartographic Services, for the preparation of the graphs. Partial funding for the results reported here was provided by the University of Alberta.
Introduction Just as in the developed countries, rising and persistent budget deficits have become a major cause for concern in the developing countries also. But in spite of the urgency and the importance which is being attached to this issue, it is surprising that, except for a very few attempts at the quantification of the possible economic effects of these deficits, there has been no systematic study of this area for many of the developing countries. This is particularly the case for the Asian countries. The aim of this study is to fill this gap to some extent. More specifically, this is a comparative study of the Asian countries, namely, India, Indonesia, South Korea, Malaysia, Pakistan, the Philippines, Singapore, Sri Lanka, Taiwan and Thailand. Due to lack of data, the coverage of Taiwan is not as comprehensive as that of the other nine countries. The experience of all these countries, in terms of economic growth, inflation, budget deficits, and the growth of public debt has been sufficiently diverse to merit such a comparative study. In addition, this approach should allow us to shed some light on the issue of relevance of the experience of newly industrialized countries for others in our sample. Given the vastness of the field and, even more important, the limitations of the data, the treatment is of necessity selective. I shall focus on the following aspects of the budget deficits in these countries: their sustainability; their effects on aggregate demand; the channels through which budget deficits engender these effects; the irrelevance of the mode of financing a given budget deficit; and effects on monetization, money growth, and inflation. These are the aspects which have received the most attention in the literature.1 For example, it is often asserted that in the developing countries, the budget deficits are largely monetized, which lead to higher rates of monetary growth causing higher rates of inflation. If these assertions were true, the implications of budget deficits for the conduct of monetary policy would be serious. And yet, for the developing countries, there is little evidence available on such assertions. Whenever such evidence is available, it tends to be either dated or descriptive with little attention being paid to modelling, measurement and estimation issues.2 Consequently, the aim of this study is to conduct, for each of the ten countries listed, an extensive empirical investigation into the areas identified above and analyse their policy implications as well as the implications of their differing experiences. The basic approach to analysing the above issues is to estimate single equation models using annual time series data. A distinctive feature of the study is the attention paid to the measurement issues, particularly with regards to the estimates of budget deficits, seigniorage and money supply. The importance of this aspect is highlighted below. In order to facilitate comparisons between the results for the ten countries, a uniform methodology for estimation and expectation formation is used. The estimates are used to examine a number of questions: for example, what is the extent of the monetization of the budget deficits? Does it differ across the ten countries? Is the growth in M3 less sensitive to the growth in public debt than in M1? Do public expenditures crowd out private expenditures and if so, to what extent? Is the transmission
Budget deficits and economic activity in Asia
2
mechanism for the crowding effect via changes in interest rates? If so, do the budget deficits affect short-term interest rates or the long-term interest rates or both and if so, to what extent? Have the deficits been inflationary in each of the ten countries? Are the observed levels of public debt sustainable? More details about these issues are given in the chapters to follow. A brief outline of the chapters is given below. Chapter 1 presents a descriptive picture of the behaviour of the budget deficits over the period covered and the behaviour of the variables of interest in relation to the budget deficits. It starts with an examination of the recent history of budget deficits and the public debt, followed by a study of the nominal deficits and its components, then by a treatment of the behaviour of expenditures and revenues along with their structure. This allows us to say something about the proximate causes of the observed deficits. Since nominal deficits are not always the most appropriate concept for economic analysis, estimates of real deficits are presented and the implications of correcting the nominal deficit for inflation are examined. It then presents estimate of total government borrowing requirements, the monetary erosion and the inflation tax on government liabilities. Having discussed the trends in deficits and debt, I undertake a brief review of the data on deficits in conjunction with the variables of interest identified above. This kind of graphic and descriptive analysis, while not necessarily a proof of causality between deficits and the variables concerned, can nevertheless shed interesting light on the hypotheses to be tested subsequently. Chapter 2 explores the following issue: can the governments in the countries of our sample follow a policy of perpetual deficits, even if they wanted to? In other words, is such a policy feasible or sustainable? There is no consensus in the literature on this issue. It is generally agreed, however, that if the real rate of interest at which a government borrows is in excess of an economy’s real growth rate, the ratio of the public debt to GNP will rise without bounds, so that a policy of perpetual primary deficits will not be feasible. Another way to look at the same issue is to think in terms of the implications of such a policy for the government budget constraint. If the governments, like the individuals, are subject to a present-value borrowing constraint, then following a policy of permanent primary deficits will be impossible because it will violate this constraint. We thus have two ways of testing for the feasibility of fiscal policies practised by the countries in our sample. This chapter presents evidence using both approaches. Chapter 3 starts the process of analysing the implications of budget deficits and the public debt. The issue dealt with in this chapter is that of the monetization of the budget deficits. Monetization of the budget deficits provides the governments with a source of revenue. This process of creating money to provide revenue, called seigniorage, is one of the most important and controversial aspects of deficit financing. This chapter considers four aspects of the issue. First, it provides alternative estimates of seigniorage and examines their importance as a source of revenue. Second, it considers the extent to which these revenues are accounted for by the treasury. Third, it provides estimates of the monetization of the government debt. Fourth, and finally, it presents some quantification of the effects of budget deficits on seigniorage for the countries in the sample. This is done by using alternative modelling approaches. Extensive inter- and intra-country comparisons are then carried out. Chapter 4 turns to the effects of budget deficits on the growth of money supply. The preceding chapter dealt with the effects of budget deficits on monetization or the creation
Introduction
3
of high powered money. However, there is no simple, mechanistic relationship between the growth of high powered money and the growth of money supply. Not only will the outcome depend on whether we are concerned with a narrow definition (M1) or a broader definition (say, M3) of money supply, but also on the underlying structural relationships determining the behaviour of the money multipliers. A stable relationship between high powered money and budget deficits does not necessarily mean a stable relationship between high powered money and money growth, and therefore between money growth and budget deficits. Chapter 4 first presents evidence on money multipliers. The behaviour of these multipliers is analysed to examine their stability across the ten countries. Then a number of alternate reduced form equation models are estimated to examine the effects of budget deficits on the growth of money supply. The sensitivity of the results is examined both with respect to nominal and real deficits as well as to the alternate definitions of money supply. Implications of estimates for monetary policy are then discussed. While it is generally agreed upon that increases in money supply caused by the monetization of the budget deficits lead to higher rates of inflation, it is also argued that budget deficits contributed directly to such pressures. Chapter 5 addresses this issue. Alternate reduced form models based on both structural and non-structural approaches are estimated and evaluated. The chapter also supplies descriptive evidence on the movements of the growth of money supply, deficits and inflation, which provides a background for the econometric work carried in the rest of the chapter. The chapter also provides a brief survey of the literature. Chapter 6 deals with two other contentious aspects. One is the possible crowding out effect of government expenditures on private expenditure. To the extent that such crowding out occurs, the effectiveness of fiscal policy is reduced correspondingly. The importance of the estimates of the crowding out effect is thus obvious for a proper evaluation of the usefulness of fiscal policy. The other aspect deals with the neutrality of the mode of financing such excess expenditures, i.e. budget deficits. The issue is whether the Ricardian equivalence proposition as restated by Barro holds. According to this proposition it is irrelevant whether a given budget deficit is financed by tax increases or by debt issue. This outcome, however, is based on a number of assumptions, so that an a priori resolution of this issue is not possible. Recourse must be taken to empirical estimates. The chapter also presents estimates on this issue. Since the policy implications of both the crowding out effects and the Ricardian equivalence proposition are serious, the evidence presented allows us to discuss these implications in a concrete way for the countries of our sample. The results of Chapter 6 suggest that, although a poor substitute for private expenditure, government expenditure nevertheless crowds out some private expenditure. Given this, the question arises as to the possible mechanism(s) which lead to this result. The most common route is via an increase in interest rates caused by high and persistent budget deficits. Unfortunately, not only is there ambiguity in the theoretical literature, but the empirical results are also inconclusive. Even worse, for the developing countries there is little evidence on this issue. Chapter 7 presents some evidence on this aspect of the budget deficits. The chapter first reports the movements of nominal and real interest rates. It turns out that the two rates have not remained constant over the period covered. Further, the inter-country variations are substantial. The determinants of this lack of
Budget deficits and economic activity in Asia
4
constancy and inter-country variability are then explored with special reference to the role of budget deficits. The results of Chapters 6 and 7 are compared for consistency in terms of their implications for the Ricardian equivalence proposition. The concluding Chapter 8 brings the major findings of the study together and draws broad conclusions.
1 A review of the record As a preliminary to the quantitative analysis to be undertaken in the subsequent chapters, this chapter briefly reviews the recent history of the budget deficits and the government debt in the countries of our sample. This is done by examining the recent trends in nominal deficits and its components, followed by a study of the behaviour of expenditure and revenue, along with their structure, to shed some light on the causes of the observed trends in the deficits. Since the nominal deficits are not the most appropriate concept for economic analysis, the real deficits are then examined. This is followed by a look at the history of the public debt which, of course, reflects the history of commulated budget deficits. Having discussed the trends in government deficits and debt, we undertake a brief review of the data on deficits in conjunction with the variables of major interest in this study. This kind of review, while not necessarily being a proof of the direction of causality between deficits and the variable concerned, can nevertheless shed interesting light on the hypotheses to be tested subsequently. Clearly, if certain general patterns were to emerge from this review, they will need explaining.
NOMINAL DEFICITS AND THEIR COMPONENTS Before proceeding with this review, a word is in order about the data used. The data on the deficits and their components as well as the debt are all derived from various issues of Government Finances and the rest of the variables from International Financial Statistics, both published by the International Monetary Fund. The concept of the government deficits used is the national income and accounts one. That this concept may suffer from some measurement problems is by now well recognized (see, for example, Boskin 1988, and Eisner and Pieper 1984, among others). But given the large number of countries in our sample, and the lack of availability of individual country primary sources, it is virtually impossible to make any meaningful corrections. In any event, as a first approximation, the NIAC concept is not a bad start. Similarly, the data on public debt also may not be the best measure, either. This is because the measure we use is based on the par value of the debt rather than the market value, which is the more appropriate one. But once again, such data for the countries in this study are simply not available. However, one advantage of the data which are being used is that they are measured by a uniform methodology and to that extent facilitate inter-country comparisons, which is the major focus of this study. The time period covered in this study was determined by the availability of the data on deficits, its components, and the other major variables.
Budget deficits and economic activity in Asia
6
We can now turn to a review of the record. The recent history of the nominal and the real budget deficits is shown in Figure 1.1. Concentrating on the nominal deficits first, we can see from this figure that the nine countries did not have a uniform experience during the sample period. To take the two extreme examples, Singapore and Sri Lanka; whereas Singapore experienced budget surpluses for most of the period, Sri Lanka, on the other hand, shows high and rising deficits. In between we can see less extreme though still differing experiences. In order to highlight these similarities and dissimilarities more sharply, we can look at Table 1.1 which gives the levels of nominal deficits as a percentage of GDP for different time periods. This table highlights a number of points. The first thing to note is that for India, South Korea, Malaysia, the Philippines, Sri Lanka and Thailand, the experience from the mid1970s onwards differs significantly from the period before that. More specifically, the levels of nominal deficits (again as percentages of GDP) are higher than before. And even within this decade, the deficits were higher in the first half of the 1980s than the latter half of the 1970s. For the other three countries, there are no such similarities. Thus for Indonesia the deficit falls in the late 1970s, then rises. Pakistan’s experience is just the opposite of Indonesia. Its deficit falls in the early 1970s, then rises in the late 1970s, followed by a sharp decline in the early 1980s. Singapore stands in a class by itself. Except for the period 1961–65, it shows a budget surplus, although there is considerable variability over time. Apart from these similarities and
Figure 1.1 Nominal and real federal deficits as percentage of GDP
A review of the record
7
differences in the trends, the absolute magnitudes of the percentage of the deficit also differ widely. Thus, leaving aside Singapore, during 1981–85 the figure was as low as 1.18 per cent for Indonesia and as high as 13.87 per cent for Sri Lanka. In general, using below and above 5 per cent as the dividing line between low and high deficit economies respectively, we can see that for the decade of 1976–85 Indonesia, South Korea, the Philippines, Thailand, and, of course, Singapore fall into the low deficit category and the rest into the high deficit category. A point was made above about the differing degrees of variability in the experiences of the nine countries. A better way to consider this aspect is to look at Table 1.2, which gives variances of the nominal and the real deficit rates. It is immediately clear that only Malaysia and Sri Lanka show the greatest variance in nominal deficits. In order to highlight the role of public debt in the budget deficits, it is useful to distinguish between primary deficits and interest
Table 1.1 Nominal and real deficits as percentage of GDP* India Period
nd
Indonesia rd
nd
South Korea
rd
nd
rd
1961–65
4.93
0.60
—
—
0.56
−1.31
1966–70
4.03
0.71
—
—
0.72
1.26
1971–75
3.87
0.35
2.22
—
1.76
1.99
1976–80
5.67
2.55
0.80
—
1.67
0.81
1981–85
6.65
3.20
1.18
—
2.20
1.94
Malaysia Period
nd
Pakistan rd
nd
The Philippines rd
nd
rd
1961–65
—
—
3.00
2.37
0.36
0.15
1966–70
—
—
7.32
3.81
0.98
0.44
1971–75
7.40
3.41
6.78
3.62
−0.41
0.63
1976–80
8.88
1.91
8.04
3.53
1.31
−0.10
1981–85
13.36
5.69
5.91
3.15
2.76
0.85
Singapore Period
nd
Sri Lanka rd
nd
Thailand rd
nd
rd
1961–65
3.09
1.34
5.49
4.12
0.52
0.79
1966–70
−0.97
6.57
6.64
3.20
2.14
2.10
1971–75
−1.52
4.89
6.69
0.07
0.26
0.56
1976–80
−0.40
7.55
13.26
2.08
3.35
1.42
1981–85
−2.99
8.47
13.87
−0.68
4.49
2.42
Budget deficits and economic activity in Asia
8
Note * nd and rd stand for nominal and real deficits as a percentage of real GDP, respectively. Surpluses are denoted with a negative sign
payments on public debt. Total deficits considered above, of course, equal the sum of these two components. The point is that unless the interest payments on outstanding government debt are, at least, balanced by a primary surplus, there will always be a budget deficit. In order to bring out the significance of this distinction for the nine countries, we present some recent data on these components and the total deficit. These data are given in Table 1.3. This table again shows important inter-country differences. For example, in the 1980s, budget deficits in India, Indonesia, Malaysia, Pakistan and the Philippines largely reflect interest payments on public debt. On the other hand, the deficits in Sri Lanka and Thailand were compounded both by primary deficits and interest payments. The importance of the distinction between these two
Table 1.2 Variances of nd and rd Variance of nd Full sample
Before 1974
Variance of rd Since 1974
Full sample
Before 1974
Since 1974
India
1.72
1.19
1.45
4.83
2.20
6.47
Indonesia
1.32
0.17
1.26
—
—
—
South Korea
1.38
1.59
0.57
4.83
8.22
0.99
Malaysia
15.7
3.34
17.28
12.45
14.97
12.94
Pakistan
5.6
6.10
3.29
83.75
127.03
32.46
The Philippines
2.52
1.84
2.90
1.51
0.48
2.68
Singapore
4.28
6.24
2.11
15.03
21.54
8.62
Sri Lanka
22.42
1.48
28.95
12.19
7.43
14.92
Thailand
3.34
2.64
3.17
2.72
2.39
3.29
components will be brought out further in the next chapter when we examine the sustainability issue.
NOMINAL DEFICITS, EXPENDITURE AND REVENUE The trends in nominal deficits discussed above reflect the trends in government expenditure and revenue. It is not our intention to carry out a detailed analysis of the factors underlying the deficits or to examine in detail the expenditure and the revenue
A review of the record
9
policies of the nine governments. That is a task beyond the scope of any study such as this. However, the aim is to consider the general trends in expenditures and revenues and in particular, consider evidence of some popular perceptions about the causes of the observed deficits. It is often alleged that in developing countries some of the major causes of these deficits on the expenditure side are the expenditures on ‘nondevelopmental’ items like subsidies, defence and administration. On the revenue side the main problem is the relatively small tax base and therefore relative stagnation in the share of direct taxes as a source of revenue. The discussion below is meant to shed some light on these factors. The broad trends in total expenditures and revenues over the period are presented in Figure 1.2. These figures are self-explanatory, except for the fact that once again we notice considerable differences in the experiences of the nine countries. However, one similarity is quite noticeable, namely, that for most of the period, the two move together. To get a better insight, we consider Tables 1.4 and 1.5 which provide a more detailed analysis of the two aggregates for the recent period. Considering Table 1.4 first, we concentrate on expenditure on goods and services, fixed investment, interest payments, subsidies and defence. Comparing 1974 with the latest year for which the data are available, we can draw a number of conclusions. The expenditure on goods and services rose for South Korea, Malaysia, Pakistan, Sri Lanka, and Thailand, whereas the expenditure on fixed investment declined in India, Malaysia, and Sri Lanka. Thus there is suggestive evidence of crowding out of fixed investment expenditure by government consumption expenditure in the cases of Malaysia and Sri Lanka. But in general, the expenditure on goods and services either remained constant or rose for all the countries, whereas the expenditure on fixed investment remained constant or rose for six of the nine countries. Interest payments as a percentage of GDP rose for all countries. Interestingly enough, subsidies and other transfer payments show a rising tendency for only two countries, India and South Korea. In terms of expenditure on defence, rising trends can be noticed in the cases of South Korea, Malaysia, Pakistan, Singapore, Sri Lanka and Thailand. From this brief discussion, it is clear that one of the major reasons for the observed deficits was the rising cost of servicing the government debt and in the case of six countries, also expenditures on defence. But the support for the widely held belief that subsidies and government consumption expenditures were largely responsible does not receive much support. Next we consider the revenue side. The data for the same time period as those in Table 1.4 are given in Table 1.5. Here we concentrate on the inter-country differences in the value of the ratio of direct taxes to GDP and its trend. Broadly speaking, we can observe two major differences. Countries with low ratio of direct taxes also had a relatively constant ratio, while countries with high ratios had a rising rate, the only exception being Thailand. The first category consists of India, South Korea, Pakistan, the Philippines, Sri Lanka, and Thailand, while the latter category includes Indonesia, Malaysia, and Singapore. What is thus clear is that the relatively small and constant contribution of direct taxes to government revenues, ceteris paribus, also plays a role in generalizing the observed deficits for a number of countries in the sample.
Budget deficits and economic activity in Asia
10
Table 1.3 Components of nominal deficits as percentage of GDP India
Indonesia
South Korea
Year Primary Interest Total Primary Interest Total Primary Interest Total deficit payments deficit deficit payments deficit deficit payments deficit 1974
2.92
1.38
4.30
−0.39
0.55
0.16
0.11
0.35
0.46
1980
−1.25
1.77
0.52
0.04
0.88
0.92
−1.85
1.14
−0.71
1981
−1.03
1.87
0.84
0.03
0.80
0.83
−2.23
1.21
−1.02
1982
−0.96
2.13
1.17
0.12
1.09
1.21
−0.97
1.23
0.26
1983
−0.64
2.29
1.65
−0.47
1.62
1.14
−5.12
1.13
−3.99
1984
−0.04
2.63
2.59
−3.9
1.70
−2.2
−2.30
1.24
−1.06
1985
—
—
—
−1.25
1.69
0.44
−2.08
1.31
−0.77
1986
—
—
—
—
—
—
−2.34
1.31
−1.03
Malaysia
Pakistan Total Primary deficit deficit
Year
Primary deficit
Interest payments
Interest payments
Total deficit
1974
0.43
2.17
2.60
1.30
1.74
3.04
−1.87
0.35
−1.52
1980
−0.18
2.90
2.72
−0.81
2.08
1.27
−1.46
0.87
−0.67
1981
8.41
3.55
11.96
0.86
2.03
2.89
0.32
0.80
1.12
1982
4.60
4.33
8.93
0.70
2.26
1.56
−0.13
1.05
0.93
1983
−0.27
4.98
4.68
0.60
2.93
3.53
−1.40
1.31
−0.10
1984
−3.96
5.56
1.60
−0.48
3.25
2.77
−2.64
1.95
−0.64
1985
—
—
—
0.30
3.33
3.63
−3.15
2.41
−0.70
Singapore
Interest payments
The Philippines Total Primary deficit deficit
Sri Lanka
Thailand
Year Primary Interest Total Primary Interest Total Primary Interest Total deficit payments deficit deficit payments deficit deficit payments deficit 1974
−9.49
1.42
−8.07
4.03
2.44
6.47
−2.45
1.17
−1.28
1980
−9.39
2.95
−6.39
17.73
3.42
21.15
2.93
1.41
4.34
1981
−8.97
2.62
−6.35
10.68
4.54
15.22
2.19
1.76
3.95
1982
−12.44
2.88
−9.56
11.84
5.15
16.99
4.47
1.64
6.11
1983
−13.76
4.38
−9.38
10.16
5.45
15.61
1.63
2.41
4.04
1984
−16.91
4.73 −12.18
6.41
4.38
10.79
1.35
2.50
3.85
1985
—
— −10.87
9.83
4.57
14.40
2.27
2.95
5.22
A review of the record
11
Figure 1.2 Government expenditure and taxes as percentage of GDP NOMINAL VERSUS REAL DEFICITS It is well known that nominal budget deficits do not always accurately represent the stance of fiscal policy (King and Plosser 1985 and Masera 1987, among others). A more accurate measure is the concept of real deficits. This requires adjusting nominal deficits for inflation. This can be done by using the budget constraint. The trends in real deficits are shown in Figure 1.1 and its quantitative aspects are given in Tables 1.1 and 1.2. It is clear from Figure 1.1 that real deficits display greater variability than nominal deficits. Also, in many cases deficits in real terms are lower than in nominal terms. We can see the distinction between the nominal and the real deficits more concretely in terms of Tables 1.1 and 1.2. Considering the extreme example, Sri Lanka, where the nominal deficits were the highest from 1976 to 1985, the situation in terms of the real deficits is quite different. Not only is it no longer the extreme case, but in fact in 1981–85 it enjoyed a budget surplus in real terms. The differences in the variability of the two measures are further highlighted in Table 1.2. The variance of the real deficits exceeded that of the nominal deficits for four of the eight countries for the entire sample period. For the period prior to 1974, this was the case for six countries and for the period since 1974, this happened for five countries. In terms of individual countries’ experiences, the variance of real deficits was on the low side for South Korea, the Philippines, and Thailand for the post-1974 period. Other differences can also be read from this table. Finally, we can get a
Budget deficits and economic activity in Asia
12
better feel for the relationship between the two concepts of the deficit by considering their correlation coefficient given in Table 1.6. The coefficient is very low, even negative for India, Pakistan, and the Philippines. On the other hand, it has changed remarkably over time for some countries, for example India, South Korea, Malaysia, Sri Lanka, and Thailand. In short, nominal deficits may not always reflect the situation about the real deficits. It is possible to examine the implications of correcting nominal deficits for inflation further. This can be done by following Masera (1987). In order to carry out his approach, we can rewrite the government budget constraint as follows: (1) where D is nominal government debt, d is real debt, Y and y are nominal and real GDP respectively, and π it is the rate of inflation as measured by the rate of change of the GDP implicit price deflator. In this equation, ∆D/Y represents total government borrowing requirements which equal ∆d/y plus (D−1/y)π, where the first term represents the change in real debt as a proportion of real GDP and the latter term represents monetary erosion or the loss of purchasing power of initial nominal debt as a proportion of nominal GDP. Table 1.7 presents estimates of total government borrowing requirements, the monetary erosion and the inflation tax on government liabilities. Inflation tax was calculated as the differences between interest payments on public debt minus the monetary erosion, both expressed as percentages of nominal GDP. The most interesting aspect of Table 1.7 is that none of the countries experienced high inflation tax rate. If anything, the period covered was dominated by periods of net transfers from the public sector to the rest of the economy. Further, such periods have been more the rule during the 1980s than the 1970s. In spite of these
Table 1.4 Central government expenditure as percentage of GDP 1974
1980
1981
1982
1983
1984
1985 1986
Current expenditure
11.61
12.48
12.27
12.92
12.81
14.08
—
—
Export on goods and services
(4.96) (4.06) (4.13) (4.27) (4.19) (4.40)
—
—
Interest payments
(1.38) (1.77) (1.87) (2.13) (2.29) (2.63)
—
—
Subsidies and other transfers
(5.27) (6.65) (6.27) (6.52) (6.33) (7.05)
—
—
2.31
—
—
(1.36) (0.48) (0.55) (0.56) (0.58) (0.73)
—
—
15.11
16.39
—
—
(1.17) (2.88) (2.90) (3.02) (2.90) (3.11)
—
—
India
Capital expenditure
3.50
Fixed investment Total expenditure a
General public services
1.66
14.15
1.83
14.10
1.88
14.80
2.00
14.81
A review of the record Defencea
13
(3.39) (2.80) (2.99) (3.08) (3.04) (3.14)
—
—
Indonesia Current expenditure
11.55
11.89
—
Export on goods and services
(5.65) (5.99) (5.51) (5.54) (5.18) (4.83) (5.06)
—
Interest payments
(0.55) (0.88) (0.80) (1.09) (1.62) (1.70) (1.69)
—
Subsidies and other transfers
(5.35) (5.74) (5.53) (4.57) (4.61) (4.24) (5.14)
—
Capital expenditure
5.60
Fixed investment Total expenditure a
General public services Defence
a
12.61
—
17.05
21.62
—
— (7.92) (7.11) (6.47) (7.14) (6.17) (7.55)
—
— (3.21) (3.12) (3.01) (2.61) (2.48) (2.29)
—
1980
1981
21.72
10.78
10.77
(4.34) (9.78) (8.79) (8.36) (5.97) (4.63) (6.67) 24.51
10.52
11.41
—
23.82
12.67
11.20
9.73
1974
11.21
11.84
22.19
8.42
19.19
1982 1983 1984 1985 1986
South Korea Current expenditure
10.97
14.88
14.74 15.68 12.41
Export on goods and services
(5.64)
(7.81)
(7.86) (7.94) (7.57) (6.92) (7.18) (6.92)
Interest payments
(0.35)
(1.14)
(1.21) (1.23) (1.13) (1.24) (1.31) (1.31)
Subsidies and other transfers
(4.98)
(5.93)
(5.67) (6.51) (3.71) (6.67) (6.77) (6.70)
2.63
2.43
(1.47)
(1.35)
(1.16) (1.64) (1.41) (1.30) (1.29) (1.30)
13.60
17.31
17.11 19.12 14.93
(2.82)
(1.73)
(1.89) (2.01) (2.02) (1.75) (1.86) (1.84)
(3.99)
(5.94)
(6.62) (5.98) (5.58) (5.19) (5.24) (5.05)
19.45
19.17
24.07 25.44 23.18 22.01
Capital expenditure Fixed investment Total expenditure a
General public services Defence
a
2.37
3.44
2.52
14.83 15.26 14.83
2.41
2.40
2.45
17.24 17.66 17.28
Malaysia Current expenditure Export on goods and services
(10.15) (10.84) (14.09)
—
—
—
—
—
—
—
Interest payments
(2.17)
(2.90)
(3.55) (4.33) (4.95) (5.56)
—
—
Subsidies and other transfers
(7.13)
(5.43)
(6.43)
3.95
9.90
(3.28)
(2.78)
(1.92)
23.04
Capital expenditure Fixed investment Total expenditure a
General public services Defence
a
—
—
—
—
—
15.38 10.04
8.20
5.38
—
—
—
—
—
—
29.07
39.45 35.48 31.38 27.39
—
—
(3.95)
(1.55)
(1.77)
—
—
—
—
—
(4.18)
(4.23)
(5.79)
—
—
—
—
—
—
Budget deficits and economic activity in Asia
1974 1980
1981
1982
14
1983
1984
1985
1986
Pakistan Current expenditure
14.19 14.47
17.28
—
Export on goods and services
(8.13) (8.31) (10.78) (9.40) (10.93) (10.97) (10.94)
—
Interest payments
(1.74) (2.08)
(2.03) (2.26)
(2.92)
(3.25)
(3.33)
—
Subsidies and other transfers
(4.32) (4.08)
(3.54) (2.62)
(2.50)
(3.11)
(3.01)
—
2.92
3.12
2.43
2.32
—
(1.67) (3.04)
(2.61) (2.83)
(3.07)
(2.42)
(2.31)
—
16.62 17.52
19.21 17.20
19.48
19.76
19.60
—
(1.33) (1.26)
(1.49) (1.41)
(1.56)
(1.43)
(1.54)
—
(5.70) (5.37)
(5.48) (5.76)
(6.79)
(6.39)
(6.64)
—
9.05
8.98
7.96
9.08
—
Capital expenditure
2.52
Fixed investment Total expenditure a
General public services Defence
a
3.05
16.35 14.28
2.86
16.36
17.33
The Philippines Current expenditure
9.27
9.15
8.59
Export on goods and services
(5.27) (7.43)
(7.11) (6.71)
(6.72)
(5.41)
(5.97)
—
Interest payments
(0.35) (0.87)
(0.80) (1.05)
(1.31)
(1.95)
(2.41)
—
Subsidies and other transfers
(3.37) (0.85)
(0.68) (1.29)
(0.95)
(0.60)
(0.70)
—
2.94
2.71
1.81
1.44
—
(0.99) (1.27)
(1.93) (1.31)
(1.29)
(0.90)
(0.52)
—
10.48 12.32
12.75 11.99
11.69
9.77
10.55
—
(0.93) (2.47)
(2.36) (2.59)
(2.30)
(1.56)
(1.44)
—
(1.95) (1.93)
(1.81) (1.63)
(1.59)
(1.17)
(1.25)
—
1974
1981
1983
1984
Capital expenditure
1.21
Fixed investment Total expenditure a
General public services Defence
a
3.17
1980
4.16
1982
1985 1986
Singapore Current expenditure
11.74
Export on goods and services
(9.58) (11.53) (14.57) (12.85) (11.91) (13.52)
—
—
Interest payments
(1.42)
(2.95)
(2.62)
(2.88)
(4.38)
(4.73)
—
—
Subsidies and other transfers
(0.74)
(1.10)
(1.14)
(0.96)
(0.77)
(1.23)
—
—
1.94
4.45
4.96
4.62
5.46
6.67
8.47
—
Fixed investment
(1.43)
(4.02)
(4.46)
(4.16)
(5.10)
(6.46)
—
—
Total expenditure
13.68
20.03
23.29
21.31
22.52
26.15 27.46
—
General public servicesa
(2.05)
(1.96)
(3.12)
(2.84)
(2.76)
(2.91) (3.87)
—
(4.50)
(5.09)
(5.10)
(4.92)
(4.16)
(5.25) (6.17)
—
Capital expenditure
Defence
a
15.58
18.33
16.69
17.06
19.48 18.99
—
A review of the record
15
Sri Lanka Current expenditure
17.78
24.74
19.72
17.82
18.83
17.96 20.68
—
Export on goods and services
(7.96) (12.92)
(8.35)
(7.02)
(7.36)
(7.95) (9.33)
—
Interest payments
(2.44)
(3.42)
(4.54)
(5.15)
(5.45)
(4.38) (4.57)
—
Subsidies and other transfers
(7.38)
(8.40)
(6.83)
(4.65)
(6.02)
(5.63) (6.78)
—
4.65
16.62
12.88
15.46
13.39
12.58 12.73
—
(2.91)
(7.22)
(4.23)
(4.74)
(4.55)
(3.32) (3.57)
—
22.43
41.36
32.60
33.28
32.22
30.54 33.41
—
(3.05)
(3.92)
(2.81)
(3.83)
(2.65)
(2.52) (3.20)
—
(0.72)
(0.69)
(0.56)
(0.49)
(0.81)
(0.83) (2.84)
—
Capital expenditure Fixed investment Total expenditure a
General public services Defence
a
1974
1980
1981
1982
1983
1984
1985 1986
Thailand Current expenditure
10.69
16.83
—
Export on goods and services (7.66) (10.03) (11.00) (11.81) (11.56) (12.16) (12.32)
—
Interest payments
(1.17)
(1.41)
(1.76)
(1.64)
(2.41)
(2.50)
(2.95)
—
Subsidies and other transfers
(1.86)
(2.55)
(1.52)
(2.08)
(1.63)
(1.24)
(1.56)
—
2.64
4.21
4.12
4.53
3.90
3.43
4.00
—
(2.25)
(3.32)
(3.48)
(3.82)
(3.38)
(3.00)
(3.57)
—
13.33
18.17
18.40
20.06
19.50
19.33
20.83
—
(1.57)
(1.56)
(1.49)
(1.77)
(1.79)
(1.80)
(1.89)
—
(2.62)
(3.94)
(3.71)
(3.19)
(3.78)
(3.83)
(4.21)
—
Capital expenditure Fixed investment Total expenditure a
General public services Defence
a
13.96
14.28
15.53
15.60
15.90
Note a
indicates both current and capital expenditure
Table 1.5 Public sector revenues as percentage of GDP 1974
1980
1981
1982
1983
1984
1985
1986
India Current revenue
18.74
12.48
13.03
13.44
12.97
13.62
—
—
Direct taxes
(4.58)
(4.56)
(4.98)
(4.87)
(4.51)
(4.34)
—
—
Indirect taxes
(4.78)
(5.31)
(5.23)
(5.10)
(5.46)
(5.42)
—
—
—
—
—
—
—
—
—
—
0.07
0.15
0.23
0.19
0.19
0.18
—
—
Social security contributions Capital revenue
Budget deficits and economic activity in Asia Total revenue
16
10.81
13.63
13.26
13.63
13.16
13.80
—
—
16.99
22.90
23.68
20.51
21.05
21.39
21.18
—
(11.24) (17.86) (17.16) (15.77) (15.48) (14.33) (14.01)
—
Indonesia Current revenue Direct taxes Indirect taxes
(2.16)
(1.98)
(1.86)
(2.13)
(2.18)
(2.00)
(3.40)
—
Social security contributions
—
—
—
—
—
—
—
—
Capital revenue
—
—
—
—
—
—
—
—
16.99
22.90
23.68
20.51
21.05
21.39
21.18
—
Current revenue
12.99
17.77
18.15
18.66
18.72
18.17
18.19 18.17
Direct taxes
(3.64)
(3.97)
(4.16)
(4.45)
(4.26)
(4.16)
(4.60) (4.59)
Indirect taxes
(5.98)
(8.16)
(8.11)
(8.31)
(8.56)
(8.17)
(7.86) (7.75)
Social security contributions
(0.11)
(0.19)
(0.18)
(0.21)
(0.21)
(0.23)
(0.27) (0.29)
0.15
0.25
0.15
0.20
0.20
0.13
13.14
18.02
18.30
18.86
18.92
18.30
Total revenue South Korea
Capital revenue Total revenue
1974 1980
1981
1982
1983
1984
0.24
0.14
18.43 18.31
1985
1986
Malaysia Current revenue
20.74 26.28
26.79
—
Direct taxes
(5.68) (9.87) (10.21) (9.68) (10.45) (10.03) (11.36)
—
Indirect taxes
(4.56) (4.41)
(4.23) (4.31)
(5.32)
(4.77)
(4.76)
—
Social security contributions
(0.07) (0.10)
(0.14) (0.15)
(0.14)
(0.13)
(0.16)
—
0.05
0.06
0.05
0.13
—
20.80 26.35
27.49 26.55
26.70
25.79
26.92
—
Current revenue
13.58 16.25
16.32 15.64
15.95
16.99
15.97
—
Direct taxes
(1.00) (2.24)
(2.54) (2.59)
(2.42)
(2.07)
(1.91)
—
Indirect taxes
(4.64) (5.47)
(5.41) (5.22)
(5.18)
(5.95)
(5.28)
—
Capital revenue Total revenue
0.06
0.07
27.38 26.50
0.11
26.64
25.74
Pakistan
Social security contributions
—
—
—
—
—
—
—
—
Capital revenue
—
—
—
—
—
—
—
—
13.58 16.25
16.32 15.64
15.95
16.99
15.97
—
11.97 12.90
11.63 11.06
11.79
10.44
11.25
—
Total revenue The Philippines Current revenue
A review of the record
17
Direct taxes
(2.84) (2.73)
(2.52) (2.41)
(2.27)
(2.25)
(2.99)
—
Indirect taxes
(3.06) (5.41)
(4.87) (4.53)
(4.45)
(3.49)
(4.10)
—
Social security contributions Capital revenue Total revenue
—
—
—
—
—
—
—
—
0.03
ng
ng
ng
ng
ng
ng
—
11.63 11.06
11.79
10.44
11.25
—
12.00 12.90
1974
1980
1981
Current revenue
20.62
25.37
26.58
Direct taxes Indirect taxes
1982
1983
1984
1985 1986
29.77
28.59
27.96
—
(7.14) (8.24) (9.31)
(10.18) (9.70) (8.84) (7.54)
—
(3.32) (4.01) (3.67)
(3.94) (3.99) (3.99) (3.85)
—
Singapore
Social security contributions
27.71
—
—
—
—
—
—
—
—
1.13
1.02
3.06
3.16
2.13
0.61
10.37
—
21.75
26.39
29.64
30.87
31.90
20.20
38.33
—
Current revenue
15.94
20.20
17.36
16.67
16.60
19.74
19.00
—
Direct taxes
(2.19) (3.14) (2.39)
(2.95) (2.77) (3.56) (3.60)
—
Indirect taxes
(5.32) (5.41) (5.83)
(6.52) (7.93) (9.19) (8.70)
—
Capital revenue Total revenue Sri Lanka
Social security contributions
—
—
—
—
—
—
—
—
0.02
0.01
0.02
0.02
0.01
0.01
0.01
—
15.96
20.21
17.38
16.29
16.61
19.75
19.01
—
Current revenue
14.61
13.83
14.45
13.95
15.46
15.48
15.61
—
Direct taxes
(1.74) (2.46) (2.75)
(2.89) (2.89) (3.15) (3.24)
—
Indirect taxes
(6.19) (6.37) (6.59)
(6.64) (7.73) (7.14) (6.86)
—
Capital revenue Total revenue Thailand
Social security contributions
—
—
—
—
—
—
—
—
Capital revenue
ng
ng
ng
ng
ng
ng
ng
—
14.61
13.83
14.45
13.95
15.46
15.48
15.61
—
Total revenue Note ng: negligible
Budget deficits and economic activity in Asia
18
Table 1.6 Correlation coefficients between nd and rd Country India
rnd,rd
Country
rnd,rd
The Philippines
1961–83
0.35 1961–85
0.06
1961–73
0.02 1961–73
0.22
1974–83
0.25 1974–85
−0.07
South Korea
Singapore
1961–84
0.43 1964–86
−0.29
1961–73
0.37 1964–73
−0.25
1974–84
0.72 1974–86
−0.09
Malaysia
Sri Lanka
1971–86
0.57 1961–84
0.17
1971–73
0.93 1961–73
0.05
1974–86
0.56 1974–84
0.75
Pakistan
Thailand
1961–84
−0.16 1961–85
0.69
1961–73
−0.10 1961–73
0.66
1974–84
−0.07 1974–85
0.90
qualitative similarities, the magnitude of these net transfers differed significantly; for example it was as high as 4.3 per cent in 1984 for Singapore, and as low as 0.19 per cent for Sri Lanka in the same year. From the above discussion, it is clear that correction for inflation can make significant differences to the estimates of budget deficits. However, this is not to suggest that nominal budget deficits are an irrelevant construction. We have already seen how its various components, namely the components underlying expenditure and revenue, can shed interesting light on controversies surrounding causes of deficits.
FISCAL DEFICITS AND PUBLIC DEBT It can be seen from the government budget constraint that a necessary condition for the growth of public debt over time is the existence of fiscal deficits. However, this is not a sufficient condition if the deficits are totally financed by foreign grants or monetary expansion. In such an event the debt to GDP ratio will not grow, and in fact may fall. The issue of whether the actual ratios experienced by the countries in our sample are sustainable in the long run or not is examined in the next chapter. While there are
A review of the record
19
Table 1.7 Monetary erosion and inflation tax on public debt Public sector borrowing requirement as percentage of nominal GDP
Public sector borrowing requirement as percentage of real GDP
Year
Monetary erosion of public sector liability as percentage of nominal GDP
Interest Inflation tax payments as as percentage percentage of GDP (−ve of GDP in sign)
India 1974
2.8937
−1.8226
4.5101
1.1264
−3.3837
1975
3.0319
−1.1336
4.0273
1.3071
−2.7202
1976
2.9127
3.9144
−0.8825
1.5050
2.3875
1977
6.9238
1.1034
1.8093
1.5422
−0.2671
1978
2.9575
6.0074
0.9164
1.7147
0.7983
1979
5.7114
2.2982
0.6594
1.7652
1.1058
1980
6.4347
1.3670
4.3443
1.7692
−2.5751
1981
5.0149
3.2133
3.2213
1.8707
−1.3506
1982
9.2842
2.2508
2.7641
2.1296
−0.6345
1983
4.4363
6.8149
2.4694
2.2938
−0.1756
1974
4.3823
1.7899
2.5924
0.3541
−2.2383
1975
3.8963
1.6651
2.2312
0.3492
−1.8820
1976
2.5893
0.7896
1.7998
0.4588
−1.3410
1977
2.6711
1.2153
1.4559
0.7402
−0.7157
1978
2.6036
0.8602
1.7434
0.7919
−0.9515
1979
1.7599
0.9023
1.6675
0.9137
−0.7538
1980
4.7475
2.8956
1.8519
1.1462
−0.7057
1981
4.2908
2.7701
1.5206
1.2118
−0.3088
1982
3.7726
2.9261
0.8465
1.2332
0.3867
1983
2.2494
1.6742
0.5752
1.1258
0.5506
1984
1.3056
0.7198
0.5858
1.2441
0.6583
South Korea
Budget deficits and economic activity in Asia
Public sector borrowing requirement as percentage of nominal GDP
Public sector borrowing requirement as percentage of real GDP
Year
Monetary erosion of public sector liability as percentage of nominal GDP
20
Interest Inflation tax payments as as percentage percentage of GDP (−ve of GDP in sign)
Malaysia 1974
3.6399
0.3421
3.2978
2.1743
−1.1235
1975
5.4227
6.4951
−1.0724
2.6688
3.7412
1976
5.8252
2.3127
3.5125
2.8904
−0.6221
1977
5.8319
3.7575
2.0743
2.9499
0.8756
1978
3.0750
0.1256
2.9494
2.9114
−0.0380
1979
5.4002
2.2758
3.1244
2.7904
−0.3340
1980
4.3839
2.4692
1.9148
2.9843
1.0695
1981
7.0991
6.7538
0.3453
3.5513
3.2060
1982
10.1232
9.2604
0.8628
4.3321
3.4693
1983
7.5383
5.7107
1.8275
4.9522
3.1247
1984
3.9221
1.4724
2.4497
5.5563
3.1066
1974
6.1370
−4.6759
10.8129
1.7375
−9.0754
1975
3.2559
−6.1298
9.3859
1.4952
−7.8907
1976
11.5450
6.8902
4.6548
1.6438
−3.0110
1977
8.1002
3.3185
4.7817
1.8015
−2.9802
1978
6.6262
2.5792
4.0470
1.8362
−2.2108
1979
9.9239
7.3251
2.6077
1.9543
−0.6534
1980
4.3235
−0.4299
4.7534
2.0765
−2.6769
1981
2.3457
−2.1250
4.4707
2.0258
−2.4449
1982
13.6155
10.1686
3.4469
2.5580
−0.8889
1983
5.7294
3.0763
2.6531
2.9334
0.2803
1984
6.9847
2.8128
4.1718
3.2489
−0.9229
Pakistan
A review of the record
Public sector borrowing requirement as percentage of nominal GDP
Public sector borrowing requirement as percentage of real GDP
Year
21
Monetary erosion of public sector liability as percentage of nominal GDP
Interest Inflation tax payments as as percentage percentage of GDP (−ve of GDP in sign)
The Philippines 1974
3.0339
2.5441
0.4899
0.3473
−0.1426
1975
1.7757
1.0852
0.6905
0.8333
0.1428
1976
1.3119
0.7288
0.5830
0.5563
−0.0267
1977
1.3834
0.6544
0.7290
0.5891
−0.1399
1978
0.1515
−0.9801
1.1316
0.6397
−0.4919
1979
0.2812
−0.6722
0.8534
0.8794
−0.0740
1980
0.8808
0.2776
0.6032
0.8631
0.2599
1981
1.5373
1.0680
0.4694
0.7983
0.3289
1982
2.5089
1.7979
0.7109
1.0479
0.3370
1983
1.0409
−1.6847
2.7256
1.3103
−1.4153
1984
3.0140
1.9915
1.0225
1.9498
0.9273
1974
6.0910
2.0761
4.0150
1.4191
−2.5959
1975
8.7938
7.9599
0.8339
1.6302
0.7963
1976
11.0573
10.3510
0.7063
2.1364
1.4301
1977
11.3162
10.6628
0.6534
2.6934
2.0400
1978
8.7605
7.5665
1.1940
2.9950
1.8010
1979
8.3175
5.7205
2.5970
3.0914
0.4944
1980
9.0590
4.2645
4.7945
2.9493
−1.8452
1981
8.3950
5.3863
3.0086
2.6245
−0.3841
1982
10.9948
8.6403
2.3545
2.8834
0.5289
1983
11.7115
9.474
2.2379
4.3775
2.1396
1984
7.6059
7.1778
0.4281
4.7293
4.3012
Singapore
Budget deficits and economic activity in Asia
Public sector borrowing requirement as percentage of nominal GDP
Public sector borrowing requirement as percentage of real GDP
Year
Monetary erosion of public sector liability as percentage of nominal GDP
22
Interest Inflation tax payments as as percentage of percentage GDP of GDP
Sri Lanka 1974
2.5746
−3.5929
6.1675
2.4399
−3.7276
1975
4.1803
2.1846
1.9957
2.6301
0.6344
1976
4.6287
2.1473
2.4814
2.7812
0.2998
1977
3.2631
−0.7771
4.0402
2.7962
−1.2440
1978
2.565
0.1450
2.4215
3.2064
0.7849
1979
5.2227
1.9325
3.9201
3.2355
−0.6846
1980
13.2743
9.3405
3.9338
3.4227
−0.5111
1981
5.8643
0.9013
4.9631
4.5362
−0.4269
1982
7.2210
4.2541
2.9669
5.1523
2.1854
1983
3.0805
−1.2309
4.3115
5.4473
1.1358
1984
1.0803
−3.5729
4.6533
4.8430
0.1897
1974
−0.5872
−3.0662
2.4839
1.1685
−1.3154
1975
0.7797
0.4094
0.3703
1.2446
0.8743
1976
3.0714
2.5814
0.4899
1.0550
0.565
1977
2.7453
1.6686
1.0768
1.0676
−0.0092
1978
2.7258
1.6407
1.0851
1.1322
0.0471
1979
2.3353
0.8977
1.4376
1.1330
−0.3046
1980
2.8631
1.0068
1.8562
1.4134
−0.4428
1981
2.2489
1.2145
1.0344
1.7620
0.7276
1982
3.8894
3.4035
0.4859
1.9630
1.4771
1983
2.7537
2.2261
0.5276
2.4142
1.8866
1984
3.3695
3.1119
0.2576
2.4993
2.2417
Thailand
A review of the record
23
Table 1.8 Government debt as percentage of GDP Year
India
Indonesia
South Korea
Malaysia
Pakistan
1974
34.2
23.19
13.83
40.17
67.80
1980
44.03
18.57
14.05
43.42
54.49
1981
43.66
16.92
15.62
54.05
48.20
1982
48.76
27.96
17.65
60.92
55.23
1983
45.98
30.11
17.56
74.29
54.86
1984
48.93
26.37
16.86
72.81
54.55
1985
—
55.79
16.80
—
59.41
1986
—
—
15.17
—
—
Year
The Philippines
Singapore
Sri Lanka
Thailand
1974
—
39.23
53.62
20.04
1980
15.47
64.70
78.39
22.96
1981
17.45
61.53
77.50
24.05
1982
20.78
66.18
81.57
28.45
1983
21.98
71.20
81.48
30.58
1984
29.44
75.79
69.22
32.05
1985
30.67
87.63
80.92
36.95
many reasons which motivate countries to engage in issuing public debt, the aim of this brief section is not to investigate the underlying reasons, merely to report the recent trends. This is done in Table 1.8, in which the data relate to total outstanding public debt as reported in Government Finances, a publication of the International Monetary Fund. The data are expressed here as a percentage of GDP. We can note a number of points from this table. Compared to 1974, the ratio in the 1980s was higher for all of the countries except for Pakistan. The ratio, however, remained relatively stable in India, South Korea, Pakistan, and Sri Lanka. It shows a consistently rising trend in Indonesia, Malaysia (except for 1984), the Philippines, Singapore, and Thailand. In terms of the values of the ratio, there are substantial intercountry differences. Thus the approximate mean values for the 1980s for these nine countries are 46.00 (India), 30.00 (Indonesia), 16.00 (South Korea), 61.00 (Malaysia), 54.00 (Pakistan), 26.00 (the Philippines), 71.00 (Singapore), 78.00 (Sri Lanka), and 29.00 (Thailand). The range is thus quite wide, from a low of 16.0 per cent for South Korea to a high of 78.0 per cent for Sri Lanka. It would be an interesting exercise to attempt to explain the reasons for these wide differences.
Budget deficits and economic activity in Asia
24
DEFICITS AND ECONOMIC ACTIVITY: A DESCRIPTIVE REVIEW While a formal quantitative analysis of the effects of deficits on various variables will be undertaken in the subsequent chapters, a review of movements in such variables in conjunction with the trends in deficits and debt observed above may reveal some patterns or lack thereof which would be instructive for the later analysis. This review is not meant to establish causality, but simply to see whether any patterns emerge which may shed some light on the hypotheses to be examined later on. In Tables 1.9 and 1.10 we examine the movements in the rates of growth of real income, reserve money, money supply, and the rate of inflation, in relation to the nominal and the real deficits and the rate of growth of nominal public debt. Money supply is represented by M1. Table 1.9 provides the mean values and variances of the variables for the period under consideration. Table 1.10 provides mean values for different time periods. Consider, first, column (1) of Table 1.9. Looking at the mean values of the real deficits as percentage of GDP and the rate of growth of real GDP, we find that countries with high rates of growth are associated both with low real deficits as well as with high. The only unambiguous case is that of India where low real deficits are also accompanied by low growth rate. But since low real deficits are also associated with high real growth for other countries, we cannot conclude that real deficits are either a necessary or a sufficient condition for rapid growth, assuming that causality runs from real deficits to growth. A similar lack of any consistent relationship between real growth and the rate of growth of nominal public debt can also be detected. The next important relationship is that between nominal deficits and reserve money, and between reserve money and money growth. These data are also given in Figure 1.3. It is interesting to note that the countries which experienced the highest mean nominal deficits, Malaysia and Sri Lanka, did not experience the highest rates of growth of reserve money. That distinction belonged to Indonesia and South Korea where the mean deficit ratio was one of the lowest in the sample. This would seem to suggest a lack of definite relationship between nominal deficits and the growth of reserve money supply. As is to be expected, there seems to be a close correspondence between the rates of growth of reserve money supply and M1. But given the lack of systematic relationships
Table 1.9 Variability of the major variables Variable
Mean
Variance
Minimum
Maximum
India nd
4.89
1.71
2.78
7.16
rd
1.34
4.83
−2.87
6.62
y
3.89
15.12
−5.05
9.85
rm
11.58
34.75
3.51
22.46
m
11.31
44.11
−11.69
24.87
π
7.79
30.23
−3.10
18.78
A review of the record
25
Indonesia nd
1.50
1.32
−1.00
3.20
rd
—
—
—
—
6.80
9.00
1.93
12.58
rm
27.00
188.91
4.77
56.83
m
29.14
171.21
6.40
51.12
π
17.57
140.46
3.16
47.15
nd
1.35
1.38
−0.37
3.86
rd
0.90
4.83
−2.77
7.88
y
8.23
40.40
−10.19
22.21
rm
27.69
448.49
−13.63
66.67
m
27.11
204.16
0.56
45.63
π
17.43
79.11
3.90
33.91
nd
9.92
15.70
5.55
19.12
rd
4.16
12.45
−0.92
11.58
y
23.89
10.02
18.20
28.19
rm
13.79
88.12
2.49
35.68
m
13.31
90.41
−0.56
37.57
π
5.02
48.71
−8.38
18.00
nd
6.22
5.60
−0.06
10.32
rd
3.29
83.75
−10.22
29.60
y
4.45
11.10
−4.66
8.78
rm
12.06
51.23
−3.13
26.74
m
12.89
71.15
−5.15
32.88
π
8.35
34.80
0.00
23.17
nd
1.03
2.52
−2.91
4.23
rd
0.39
1.51
−3.02
2.70
y
4.75
86.72
−24.39
28.14
y
South Korea
Malaysia
Pakistan
The Philippines
Budget deficits and economic activity in Asia
26
rm
15.70
105.47
−7.69
47.47
m
13.40
92.45
−1.65
38.14
π
11.86
102.60
1.82
49.14
Variable
Mean
Variance
Minimum
Maximum
Singapore nd
−1.03
4.28
−5.21
4.49
rd
6.43
15.03
−1.50
13.94
y
8.43
35.81
−3.45
24.24
rm
13.22
88.42
−10.26
37.27
m
11.58
71.53
−2.09
35.51
π
3.58
30.02
−9.30
15.47
nd
9.01
22.42
3.17
22.21
rd
1.86
12.19
−5.08
8.55
y
5.25
8.71
0.06
15.76
rm
13.58
174.73
−6.58
45.47
m
12.51
93.25
−3.52
34.89
π
8.63
73.67
−1.78
24.17
nd
2.63
3.34
−0.80
6.35
rd
1.46
2.72
−3.54
3.98
y
6.92
4.28
2.67
12.20
rm
10.65
16.44
3.79
17.88
m
9.39
53.22
−8.30
20.62
π
5.50
35.98
−1.75
20.09
Sri Lanka
Thailand
Notes nd and rd: nominal and real deficits as percentage of GDP, respectively. y, rm and m: rates of growth of real GDP, reserve money and M1, respectively. π: rate of inflation measured by the rate of change of GDP implicit price deflator.
between nominal deficits and growth of reserve money supply, we cannot be too certain about the strength of the relationship between nominal deficits and growth of money supply as measured by M1. Finally, we consider the mean rates of inflation. As we shall see later, the relationship between nominal deficits and inflation is not so simple, but still a comparison of these
A review of the record
27
two variables is instructive. The countries with the highest mean inflation rate, Indonesia and South Korea, are not the ones with the highest proportions of nominal debts. In fact they have two of the lowest. On the other hand, we also observe countries such as Pakistan and Sri Lanka where both the nominal deficits as well as the rate of inflation was high. Finally, we have the case of Singapore which experienced the lowest mean rate of inflation, but then it was also the only country in the sample to have experienced a nominal budget surplus. In short, no definite pattern is discernible in this case either.
Table 1.10 Mean values of the major variables for different periods Period
nd
rd
dg
y
rm
m
π
India 1961–65
4.93
0.60
8.3
3.0
7.5
9.5
6.9
1966–70
4.03
0.71
8.0
4.9
8.2
8.2
5.8
1971–75
3.87
0.35
11.1
3.0
10.3
12.6
10.0
1976–80
5.67
2.55
16.6
3.5
18.9
11.6
7.9
1981–85
6.65
3.20
20.0
5.5
16.4
15.1
7.9
1971–75
2.22
—
—
8.1
39.1
38.7
21.8
1976–80
0.80
—
—
7.9
25.6
31.9
20.0
1981–85
1.18
—
—
4.5
15.7
15.4
11.27
1961–65
0.56
−1.31
8.3
3.5
24.7
22.3
23.3
1966–70
0.72
1.26
80.9
12.8
44.6
36.4
13.5
1971–75
1.76
1.99
40.6
9.0
30.7
31.3
20.0
1976–80
1.67
0.81
29.2
7.7
26.0
26.6
20.8
1981–85
2.20
1.94
19.58
7.7
7.1
15.7
6.7
1971–75
7.40
3.41
15.4
7.5
18.1
16.7
5.5
1976–80
8.88
1.91
15.9
8.5
16.7
17.6
9.7
1981–85
13.36
5.69
17.6
5.2
8.5
7.8
2.6
1961–65
3.00
2.37
13.2
4.5
7.5
8.2
3.6
1966–70
7.32
3.81
16.4
2.2
8.4
9.5
5.4
1971–75
6.78
3.62
24.4
3.5
10.1
13.0
14.7
Indonesia
South Korea
Malaysia
Pakistan
Budget deficits and economic activity in Asia
28
1976–80
8.04
3.53
16.8
6.1
20.7
21.2
9.4
1981–85
5.91
3.15
18.1
6.6
12.9
13.2
8.2
1961–65
0.36
0.15
7.0
5.3
9.1
9.4
5.3
1966–70
0.98
0.44
15.5
2.8
13.7
10.9
9.8
1971–75
−0.47
0.63
24.2
7.3
17.2
19.2
14.6
1976–80
1.31
−0.10
10.4
6.0
19.1
17.1
11.7
1981–85
2.76
0.85
30.7
2.2
19.5
10.5
17.8
1963–65
3.09
1.34
9.4
2.1
10.8
2.4
0.98
1966–70
−0.97
6.57
32.9
13.1
10.5
13.4
1.7
1971–75
−1.52
4.89
23.4
9.6
20.4
16.8
8.0
1976–80
−0.40
7.55
20.9
8.8
14.6
12.1
4.3
1981–85
−2.99
8.47
17.1
6.1
10.0
7.7
2.8
1961–65
5.49
4.12
12.7
3.8
10.1
6.7
0.10
1966–70
6.64
3.20
11.4
7.9
3.0
5.4
3.1
1971–75
6.69
0.07
9.8
4.0
10.2
14.7
10.2
1976–79
13.26
2.08
22.6
5.5
25.1
19.6
14.0
1980–84
13.87
−0.68
14.7
5.0
21.1
17.1
17.5
Period
nd
rd
The Philippines
Singapore
Sri Lanka
dg
y
rm
m
π
Thailand 1961–65
0.52
0.79
9.6
7.3
7.3
5.5
2.0
1966–70
2.14
2.10
19.1
8.5
9.1
8.4
1.4
1971–75
2.69
0.56
13.5
6.3
14.2
12.5
10.4
1976–80
3.35
1.42
20.5
7.6
14.0
15.6
9.8
1980–85
4.49
2.42
17.9
4.9
8.6
4.9
3.7
Note dg: rate of growth of nominal public debt.
It is also instructive to consider the different measures of variability in Table 1.9. Consider once again the variances of real deficits and growth of real GDP. We can see that no systematic pattern is found. Thus, high variances in real deficits are accompanied by a high variance of real growth rates for South Korea, Singapore, and India, but by low
A review of the record
29
variances of y for Malaysia, Pakistan, Sri Lanka, and Thailand. On the other hand, low variance of real deficits are accompanied by low variance of y for Indonesia, but high variance for the Philippines. On the whole, therefore, an inter-country comparison of the mean values of the different variables does not reveal any consistent pattern about the relationship between nominal deficits and real deficits, and the other variables. This, of course, does not mean that intra-country relationships may not exist when viewed across time. To such an examination we now turn to in terms of Table 1.10. We proceed in the same order as above. Therefore, consider the relationship between real deficits and the rate of growth of real GDP. With the exception of Singapore and Thailand, there is no systematic direction in the movement of the two variables. Thus declines in real deficits are accompanied by increases in the real GDP growth as well as decreases. In Singapore, there is a positive relationship, as is the case for Thailand, except for the period 1981–85 when the relationship is negative. The intra-country comparison would thus seem to support the inter-country comparisons from Table 1.9. Turning to the behaviour of nominal deficits and the growth of reserve money supply, there are mixed results. Thus, for Pakistan, the Philippines, and Singapore the relationship is broadly positive. It is relatively negative for Malaysia and not at all clearcut for the other countries. However, the relationship between the growth of reserve money supply and the growth of M1 is closer and positive
Table 1.11 Correlation coefficients between nd, rd, and variables of main interest India
Indonesia
South Korea
Correlation coefficient
1961– 1961– 1974– 1970– 1970– 1974– 1961– 1961– 1974– 83 73 83 85 78 85 84 73 84
rrd,y
0.29
−0.16
0.52
—
—
—
0.09
0.20
−0.46
rnd,rm
0.24
−0.41
0.38
0.25
−0.64
0.11
−0.11
0.16
−0.24
rnd,m
0.08
−0.14
0.09
0.37
−0.38
0.23
−0.04
0.08
0.13
rnd,π
0.20
0.29
0.12
−0.09
−0.57
−0.02
−0.10
−0.26
0.09
Malaysia
Pakistan
The Philippines
Correlation coefficient
1971– 86
1971– 73
1974– 86
1961– 84
1961– 73
1974– 84
1961– 85
1961– 73
1974– 85
rrd,y
−0.62
−0.83
−0.62
−0.25
−0.23
0.03
0.67
−0.05
0.80
rnd,rm
−0.20
−0.30
0.02
0.25
−0.12
0.35
0.01
0.23
−0.35
rnd,m
−0.19
−0.50
0.02
0.25
−0.04
0.44
−0.06
0.38
−0.44
rnd,π
−0.37
−0.91
−0.34
0.42
0.33
0.18
−0.03
−0.37
0.06
Singapore Correlation coefficient
1964– 86
1964– 73
Sri Lanka 1974– 86
1961– 84
1961– 73
Thailand 1974– 84
1961– 85
1961– 73
1974– 85
Budget deficits and economic activity in Asia
30
rrd,y
−0.05
0.26
−0.48
0.25
0.28
0.43
−0.20
−0.42
−0.05
rnd,rm
−0.20
−0.47
0.21
0.24
−0.46
0.08
0.21
0.46
−0.24
rnd,m
−0.33
−0.67
0.22
0.42
0.24
0.03
0.05
0.25
−0.23
rnd,π
−0.10
−0.29
0.25
0.47
0.00
−0.03
−0.11
0.11
−0.67
Figure 1.3 Rates of growth of reserve money, M1, and nominal deficits for six of the countries, but not quite clear for India, Malaysia, and the Philippines. These results thus correspond broadly with those from inter-country comparisons. As for the relationship between the nominal deficits and the growth of M1, it appears to be negative for the Philippines, somewhat positive for Pakistan, Sri Lanka and Thailand, and not at all clear for the rest. Finally, we consider the rate of inflation and the nominal deficits. The relationship is not any clearer in this case either. For example, in the cases of India, Indonesia, South Korea, Malaysia, and the Philippines, there are no systematic patterns, while for Pakistan and Thailand it is mostly negative, and for Singapore and Sri Lanka it is mostly positive. We thus have a whole array of movements between these two variables. The above casual empiricism about the relationships between deficits, both nominal and real, and a number of variables can be further examined by looking at their correlation coefficients in Table 1.11. The coefficients are given for the entire sample period for each country and for two sub-periods, before 1974 and since 1974, the dividing
A review of the record
31
line being in terms of the first oil shock. This table would seem to confirm the conclusions suggested above. For example, the correlation coefficient between real deficit and the growth rate of real GDP ranges from positive and negative and sizeable to virtually non-existent. A similar lack of consistent patterns is evident from the correlation coefficient between nominal deficit and inflation, and nominal deficits and reserve money growth. It is clear from the discussion in this section that, contrary to popular perceptions, we cannot draw any general conclusions about the effects of budget deficits on real growth, money growth, and inflation. While the data are suggestive in some cases, the relationships are not clear-cut enough to warrant strong inferences. For this, we must turn to a more rigorous treatment of the issues. This is the subject matter of the chapters to follow.
2 Sustainability of perpetual deficits The issue to be explored in this chapter is: can the government follow a policy of perpetual primary deficits (that is, deficits excluding interest payments on federal debt) even if it wanted to? In other words, is such a policy feasible or sustainable? There is no consensus in the literature. It is generally agreed, however, that if the rate of interest at which government borrows is in excess of an economy’s growth rate, then the ratio of the debt to GNP will rise without bounds, so that a policy of perpetual primary deficits will be impossible. The same issue could also be looked at this way: if governments, like private individuals, are subject to a present-value borrowing constraint, then a policy of following a permanent primary deficit is impossible because the constraint will be violated. Thus, we have two ways of testing for the feasibility of fiscal policies practised by countries in our sample. This chapter presents evidence using both approaches.
INTEREST RATE, GROWTH RATE, AND SUSTAINABILITY In order to examine the feasibility of perpetual primary deficits according to the traditional approach, we start with the government budget constraint, given by ∆B=G−T+rB−1 (1) where B, G, T and r represent, in real terms, market value of government bonds, expenditure, tax revenue, and interest rate. It is assumed in (1) that all deficits are bond financed. Assuming that income grows at the rate g, the ratio of government debt to GNP, given by b, can be written as (2) where x is the ratio of the primary deficit G−T to GNP. Using the approximation (1+r)/(1+g)≈1+r−g, we can write (2) as b=x+(1+r−g)b−1 (3) In steady state, with constant x, g and r, we have (4)
Sustainability of perpetual deficits
33
In order to examine the stability of b*, we can solve the first-order non-homogeneous difference equation (3). The solution is given by b(t)=b*+(b0−b*)(1+r−g)t (5) It is clear that the stationary value b* is stable if g>r, and unstable if g
Table 2.1 Data for analysing the sustainability of public debt Average rate of interest on Year public debt
Money market rate
Inflation Growth rate rate of real GDP
Primary deficit/GDP
Debt/GDP
India 1974
0.0384
0.1352
0.1797
0.9845
0.0292
0.3420
1975
0.0424
0.1040
−0.0310
0.0984
−0.0031
0.3697
1976
0.0478
0.1128
0.0684
0.0132
0.0389
0.4169
1977
0.0429
0.1018
0.0340
0.0834
−0.0135
0.4129
1978
0.0490
0.0805
0.0210
0.0654
−0.0131
0.4088
1979
0.0469
0.0847
0.1587
−0.5054
−0.0036
0.4346
Budget deficits and economic activity in Asia
34
1980
0.0465
0.0724
0.1136
0.0692
−0.0125
0.4403
1981
0.0494
0.0861
0.0920
0.0611
−0.0103
0.4336
1982
0.0494
0.0727
0.0787
0.0365
−0.0096
0.4876
1983
0.0557
0.0830
0.0933
0.0747
−0.0064
0.4598
1984
0.0538
0.0995
0.0720
0.0331
−0.0004
0.4893
1985
0.0560
0.1000
0.0752
0.0605
−0.0005
0.4834
1974
—
0.1142
0.4715
0.0775
−0.0039
0.2319
1975
—
0.1341
0.1243
0.0501
0.0207
0.2598
1976
—
0.1417
0.1449
0.0680
0.0257
0.2459
1977
—
0.0723
0.1309
0.0872
0.0333
0.2001
1978
0.0320
0.0729
0.1082
0.0796
0.0299
0.2935
1979
0.0548
0.1323
0.3253
0.0623
0.0384
0.2161
1980
0.0478
0.1287
0.1287
0.2920
0.0004
0.1857
1981
0.0476
0.1626
0.1850
0.0793
0.0003
0.1692
1982
0.0392
0.1724
0.0514
0.0222
0.0012
0.2796
1983
0.0542
0.1317
0.1316
0.0924
−0.0047
0.3011
1984
0.0650
0.1863
0.1191
0.0613
−0.0390
0.2637
1985
0.0519
0.1033
0.0766
0.0192
−0.0125
0.5579
Primary deficit/GDP
Debt/GDP
Indonesia
Average rate of Interest on Year public debt
Money market rate
Inflation Growth rate rate of real GDP
South Korea 1974
0.0242
—
0.3391
0.0436
0.0011
0.1383
1975
0.0237
—
0.2597
0.0723
0.0018
0.1261
1976
0.0341
—
0.2010
0.1397
−0.0102
0.1064
1977
0.0563
0.181
0.1631
0.1048
0.0005
0.1029
1978
0.0638
0.193
0.2196
0.1053
−0.0165
0.1229
1979
0.0808
0.189
0.2118
0.0628
−0.0213
0.1131
1980
0.0813
0.229
0.2484
−0.0273
−0.0185
0.1405
1981
0.0772
0.181
0.1550
0.0738
−0.0223
0.1562
1982
0.0698
0.142
0.0649
0.0566
−0.0097
0.1765
1983
0.0641
0.130
0.0390
0.1096
−0.0512
0.1756
Sustainability of perpetual deficits
35
1984
0.0737
0.114
0.0391
0.0864
−0.0230
0.1686
1985
0.0778
0.094
0.0399
0.0544
−0.0208
0.1688
1974
0.0658
0.0337
0.1265
0.0837
0.0043
0.4017
1975
0.0680
0.0404
−0.0307
0.0080
0.0140
0.5089
1976
0.0780
0.0388
0.1268
0.1159
0.0149
0.4773
1977
0.0777
0.0309
0.0690
0.0771
0.0075
0.4877
1978
0.0964
0.0419
0.1001
0.0648
0.0082
0.4481
1979
0.0788
0.0350
0.1209
0.0931
−0.0307
0.4326
1980
0.0846
0.0488
0.0683
0.0748
−0.0018
0.4342
1981
0.0914
0.0565
0.0110
0.0689
0.0841
0.5405
1982
0.0944
0.0785
0.0247
0.0599
0.0460
0.6092
1983
0.1015
0.0777
0.0463
0.0624
−0.0027
0.7429
1984
0.1302
0.0950
0.0608
0.0779
−0.0396
0.7281
1985
—
0.0757
−0.0147
−0.0105
—
—
1986
—
0.0803
−0.0835
0.0096
—
—
Money market rate
Inflation Growth rate rate of real GDP
Primary deficit/GDP
Debt/GDP
Malaysia
Average rate of interest on Year public debt Pakistan 1974
0.0272
0.0577
0.2304
0.0555
0.0130
0.6780
1975
0.0284
0.0577
0.2317
0.0393
0.0320
0.5571
1976
0.0293
0.0904
0.1144
0.0521
0.0188
0.5901
1977
0.0315
0.0927
0.1069
0.0378
0.0094
0.6036
1978
0.0325
0.0948
0.0902
0.0806
−0.0254
0.5503
1979
0.0326
0.975
0.0547
0.0485
0.0149
0.6015
1980
0.0382
0.1120
0.1050
0.0878
−0.0081
0.5449
1981
0.0420
0.0940
0.1080
0.0696
0.0086
0.4820
1982
0.0408
0.0936
0.0902
0.0620
−0.0070
0.5523
1983
0.0535
0.0931
0.0571
0.0645
0.0060
0.5486
1984
0.0596
0.0925
0.0963
0.0532
−0.0048
0.5455
1985
0.0326
—
0.0585
0.0796
0.0030
0.5941
The Philippines
Budget deficits and economic activity in Asia
36
1974
0.0369
—
0.0831
0.0281
−0.0123
0.0941
1975
0.0836
—
0.0921
0.0531
−0.0029
0.1001
1976
0.0565
0.1018
0.0734
0.0887
0.0005
0.0984
1977
0.0592
0.1088
0.0931
0.0516
−0.0010
0.1550
1978
0.0732
0.1088
0.1518
0.0048
−0.0082
0.1564
1979
0.1198
0.1225
0.1561
0.0701
−0.0254
0.1679
1980
0.1239
0.1213
0.1100
0.0869
−0.0146
0.1547
1981
0.1049
0.1247
0.0837
0.0588
0.0032
0.1745
1982
0.1123
0.1378
0.1164
−0.0034
−0.0013
0.2078
1983
0.1407
0.1423
0.4914
−0.2439
−0.0140
0.2198
1984
0.2024
0.2852
0.1827
0.1897
−0.0264
0.2944
1985
—
0.2672
0.0181
0.1075
−0.0315
0.3067
Primary deficit/GDP
Debt/GDP
Average rate of interest on Year public debt
Money market rate
Inflation Growth rate rate of real GDP
Singapore 1974
0.0393
0.0890
0.1547
0.0643
−0.0949
0.3923
1975
0.0382
0.0439
0.0252
0.0398
−0.1175
0.4451
1976
0.0427
0.0415
0.0185
0.0756
−0.0587
0.5497
1977
0.0473
0.0476
0.0145
0.0790
−0.0665
0.6049
1978
0.0499
0.0493
0.0238
0.0857
−0.0667
0.6049
1979
0.0512
0.0776
0.0524
0.0936
−0.0681
0.6348
1980
0.0504
0.1098
0.1074
0.1040
−0.0934
0.6470
1981
0.0449
0.1154
0.0640
0.0989
−0.0897
0.6153
1982
0.0454
0.0792
0.0469
0.0635
−0.1244
0.6618
1983
0.0642
0.0711
0.0412
0.0797
−0.1376
0.7120
1984
0.0674
0.0767
0.0689
0.0077
−0.1691
0.7579
1985
0.0357
0.0538
−0.0196
−0.0188
−0.1390
0.8763
1986
—
0.0427
−0.0375
0.0188
—
—
1974
0.0712
0.0650
0.2417
0.0401
0.0403
0.5362
1975
0.0755
0.0650
0.0696
0.0452
0.0570
0.5699
1976
0.0788
0.0650
0.0881
0.0444
0.0636
0.5992
Sri Lanka
Sustainability of perpetual deficits
37
1977
0.0859
0.1000
0.1602
0.0389
0.0268
0.7047
1978
0.1058
0.1000
0.0956
0.0696
0.0854
0.7358
1979
0.1082
0.1000
0.1537
0.0642
0.1104
0.6880
1980
0.0929
0.1200
0.2005
0.0578
0.1773
0.7839
1981
0.1308
0.1400
0.2080
0.0577
0.1068
0.7750
1982
0.1395
0.1400
0.1109
0.0508
0.1184
0.8157
1983
0.1640
0.1300
0.1669
0.0500
0.1016
0.8148
1984
0.1770
0.1300
0.2152
0.0404
0.0641
0.6922
Primary deficit/GDP
Debt/GDP
Average rate of Interest on Year public debt
Money market rate
Inflation Growth rate rate of real GDP
Thailand 1974
0.0775
—
0.1887
0.0542
−0.0245
0.2004
1975
0.0860
—
0.0278
0.0582
0.0074
0.2003
1976
0.0664
—
0.0398
0.0266
0.0275
0.2320
1977
0.0651
0.0632
0.0857
0.0721
0.0221
0.2384
1978
0.0689
0.0704
0.0860
0.1009
0.0281
0.2466
1979
0.0820
0.0740
0.1156
0.0609
0.0220
0.2476
1980
0.0881
0.0916
0.1641
0.0577
0.0293
0.2296
1981
0.1087
0.1157
0.0800
0.0627
0.0219
0.2405
1982
0.1036
0.1164
0.0333
0.0415
0.0447
0.2845
1983
0.1201
0.0935
0.0313
0.0590
0.0163
0.3058
1984
0.1128
0.1000
0.0139
0.0552
0.0135
0.3205
1985
0.0799
0.1102
0.2571
0.0266
0.0227
0.3695
rate, that is the rate on financial instrument which could be found. To this extent, the use of the latter measure is not strictly comparable across the countries in our sample. The data on primary deficits and debt as ratios of GDP, two nominal interest rates, the rate of inflation and real rate of growth of GDP are given in Table 2.1. These data can be used to examine the question of stability according to equation (6). Before doing that, we can get a quick idea about this issue by looking at the values of (r−g). As we have seen, a necessary condition for potential instability is that (r−g) be positive. These values, derived from Table 2.1, are given in Table 2.2. A cursory glance at this table would seem to suggest that in the vast majority of cases, instability was not a problem. But there are exceptions in the latter period for a number of countries where r did exceed g. But then, it is also clear from Table 2.1 that in many cases there was also a primary budget surplus. Consequently, we must turn to an analysis based on equation (6).
Budget deficits and economic activity in Asia
38
These results are given in Table 2.3. This table gives estimates of (6) for three time periods including the latest year for which the data are available. One of the most remarkable aspects of this table is that on both definitions of the interest rate used, there is no evidence of instability for any of the time periods for India, South Korea, Pakistan, and Singapore. For Indonesia, instability is indicated only for the period 1981–85, and that too using the definition of r2. The same situation exists for Malaysia. For the Philippines, a similar situation exists but for 1985. The most unfavourable results for stability are those for Sri Lanka and Thailand. For these two countries, most of the evidence suggests that the deficit experienced during those time periods were not sustainable. In all these cases, however, an interesting aspect is that, of the cases of instability identified, more of them arise when we use r2 as a measure of interest rate. This would seem to support the position adopted by Barth and Russek (1986) about the appropriate interest rate to be used in this type of analysis. A more detailed analysis of the recent data was carried out using Tobin’s approach outlined above. In Table 2.4, row (1) gives primary deficit (surplus with a negative sign) as a ratio of GDP from Table 2.1. Rows (2) and (3) give values of the difference between the rate of growth of real GDP and the real rate of interest calculated from the data in Table 2.1. Rows (3a) and (3b) give the hypothetical equilibrium debt/GDP ratio, corresponding to the two interest rate measures. A negative of this statistic means that the government is a net lender to the private sector. Row (4)
Table 2.2 Values of real interest rate minus real growth rate India
Indonesia
South Korea
Year
r1−g
r2−g
r1−g
r2−g
r1−g
r 2−g
1974
−0.1423
−0.0455
—
−0.4389
−0.3585
—
1975
−0.0249
0.0366
—
−0.0404
−0.3084
—
1976
−0.0339
0.0311
—
−0.0713
−0.3066
—
1977
−0.0745
−0.0157
—
−0.1419
−0.2116
−0.0889
1978
−0.0375
−0.0060
−0.1559
−0.1150
−0.2011
−0.1319
1979
−0.0612
−0.0235
−0.3328
−0.2553
−0.1939
−0.0856
1980
−0.1313
−0.1055
−0.3426
−0.2617
−0.1397
−0.0080
1981
−0.1037
−0.0670
−0.2168
−0.1018
−0.1516
−0.0478
1982
−0.0659
−0.0426
−0.0345
−0.0987
−0.0518
0.0204
1983
−0.1124
−0.0851
−0.1198
−0.0423
−0.0845
−0.1186
1984
−0.0513
−0.0056
−0.1154
0.0059
−0.0518
0.1613
1985
−0.0989
−0.0357
−0.0441
0.0073
−0.0165
−0.0003
Malaysia Year
r1−g
Pakistan r2−g
r1−g
The Philippines r2−g
r1−g
r2−g
Sustainability of perpetual deficits
39
1974
−0.1444
−0.1765
−0.2588
−0.2280
−0.0709
—
1975
0.0909
0.0632
−0.2429
−0.2134
−0.0616
—
1976
−0.1647
−0.2039
−0.1372
−0.0761
−0.1056
−0.0603
1977
−0.0859
−0.0115
−0.1132
−0.0520
−0.0855
−0.0360
1978
−0.0686
−0.1231
−0.1383
−0.0760
−0.0835
−0.0479
1979
−0.1352
−0.1790
−0.0264
−0.0058
−0.1065
−0.1037
1980
−0.0586
−0.0944
−0.1547
−0.0809
−0.0731
−0.0756
1981
−0.0115
−0.0235
−0.1356
−0.0837
−0.0378
−0.0180
1982
−0.5301
−0.0062
−0.1114
−0.0587
−0.0035
0.0220
1983
−0.0072
−0.0373
−0.0681
−0.0285
−0.1068
−0.1052
1984
−0.0086
−0.0434
−0.0899
−0.0571
−0.1700
−0.0871
1985
—
−0.1010
−0.1055
—
—
0.1416
1986
—
0.1545
—
—
—
—
Singapore
Sri Lanka
Thailand
Year
r1−g
r 2−g
1974
−0.1797
−0.1301
−0.2107
−0.2168
−0.1654
—
1975
−0.0209
−0.0213
−0.0394
−0.0499
−0.0131
—
1976
−0.0514
−0.0527
−0.0537
−0.0675
−0.0600
—
1977
−0.0463
−0.0460
−0.1132
−0.0992
−0.0927
−0.0947
1978
−0.0597
−0.0503
−0.0594
−0.0652
−0.1180
−0.0853
1979
−0.0955
−0.0685
−0.1098
−0.1179
−0.0945
−0.1026
1980
−0.1609
−0.1016
−0.1654
−0.1383
−0.1337
−0.1302
1981
−0.1180
−0.0476
−0.1349
−0.1257
−0.0341
−0.0271
1982
−0.0651
−0.0134
−0.0223
−0.0218
−0.0287
0.0415
1983
−0.0568
−0.0499
−0.0530
−0.0870
0.0297
0.0031
1984
−0.0843
−0.0751
−0.0787
−0.1257
0.0436
0.0308
1985
0.0742
0.0923
—
—
0.0275
0.0578
r1−g
r2−g
r1−g
r2−g
Source: r1 and r2 refer to real interest rates based on Table 2.1 columns (1) and (2) respectively, and g is the rate of growth of real GDP from Table 2.1
Budget deficits and economic activity in Asia
40
Table 2.3 Values of the condition for instability of the debt/GDP ratio (equation (6)) Values of (br−g)+x when Year
r=r1
r=r2
India 1974–80
−0.0193
−0.0004
1981–85
−0.0225
−0.0060
1985
−0.0483
−0.0023
1974–80
−0.0413
−0.023
1981–5
−0.0251
−0.0067
1985
−0.0370
−0.0084
1974–80
−0.0324
−0.0115
1981–85
−0.0284
−0.0159
1985
−0.0236
−0.0208
1974–80
−0.0464
−0.0599
1981–84
−0.0679
−0.0128
1984
−0.0458
−0.0715
1974–80
−0.0716
−0.0566
1981–84
−0.0496
−0.0261
1984
−0.0538
−0.0795
1974–80
−0.0356
−0.0423
1981–85
−0.0339
−0.0205
1985
n.a.
−0.0119
1974–80
−0.1298
−0.1181
1981–85
−0.1555
−0.1749
1985
−0.0470
−0.0582
Indonesia
South Korea
Malaysia
Pakistan
The Philippines
Singapore
Sustainability of perpetual deficits
41
Sri Lanka 1974–80
−0.0083
−0.0081
1981–85
−0.0463
−0.0279
1985
−0.0097
−0.0229
1974–80
−0.0082
−0.0387
1981–85
−0.0261
−0.0279
1985
−0.0328
−0.0441
Thailand
reproduces the actual debt/GDP ratio from Table 2.1. The three sub-rows of row (5) and row (6) give the values of the debt/GDP ratio at intervals of five, ten, and fifteen years. Before we discuss this table, a clarification is in order about the use of the traditional approach towards the sustainability issue being discussed in this section. We have noticed that, ceteris paribus, the higher the real rate of interest and the lower the real rate of growth, the greater the possibility of debt instability. In all of the countries in our sample, the nominal interest rates have been generally administered which in general means that they probably have been lower than those dictated by market forces. This means that for a given rate of inflation the real rates of interest have been lower than they would be in a non-regulated environment. This, of course, means that the stability condition may in fact be violated when it is shown to be not, or, at the very least, that the range of values for stability is narrower than suggested by the actual values.3 With this caveat, we can turn to a discussion of Table 2.4. It is best to analyse it country by country and then summarize the general conclusions. India:
From 1980 to 1984, the real growth rate exceeded the real interest rate for both definitions of the latter, thus implying stable b. This process was helped by the fact that during these years there was a primary budget surplus. Given these facts, it follows that the hypothetical equilibrium b is negative during these years. In none of the years, therefore, does the initial b exceed the hypothetical equilibrium b, meaning that b does not increase over time. Looking at the time profile of b, we can look at the least favourable case, which is for 1974, using the real interest rate r2, and 1984, again for r2. Assuming 1974 values of the primary deficit and r2 and g configuration, debt would be only about 27 per cent of GDP after five years and only about 18 per cent after fifteen years. For 1984, these percentages would be about 48 and 45 respectively, which are not high by the historical standards.
Indonesia: In this case, towards the end of the period some possibility of instability begins to emerge as the real interest rate exceeds real growth rate. During
Budget deficits and economic activity in Asia
42
Table 2.4 Federal debt dynamics 1974
1980
1981
1982
1983
1984
1985
India (1)
Primary deficit
2.92
−1.25
−1.03
−0.96
−0.64
−0.04
—
(2)
Real GDP growth−real i (A)
14.23
13.13
10.37
6.594
11.24
5.13
—
(3)
Real GDP growth−real i (B)
4.548
10.55
6.701
4.264
8.514
0.56
—
(3a) (1)/(2)
0.2052 −0.0952 −0.0993
−0.1456 −0.0569 −0.0077
—
(3b) (1)/(3)
0.6420 −0.1185 −0.1537
−0.2251 −0.0751 −0.0714
—
(4)
Debt/GDP
(5)
After 5 years (A)
(6)
34.2
44.03
48.93
—
17.684 23.7169 26.6194
35.3925 26.9703 38.0998
—
After 10 years (A)
9.1921 12.7550 16.2146
25.6788 15.8101 29.6663
—
After 15 years (A)
4.8259
18.6202
9.2583 23.0992
—
6.8394
43.66
9.8617
48.76
45.98
After 5 years (B)
27.5089 −6.6191 31.5250
39.5299 30.5339 47.5807
—
After 10 years (B)
22.1519 16.0745 22.7510
32.0390 20.2681 46.2686
—
After 15 years (B)
17.8631
25.9596 13.4453 44.9926
—
9.6885 16.4071
Indonesia (1)
Primary deficit
−0.39
0.04
0.03
0.12
−0.47
−3.7
−1.25
(2)
Real GDP growth−real i (A)
—
34.26
21.68
3.448
11.98
11.54
4.407
(3)
Real GDP growth−real i (B)
43.89
26.17
10.18
−9.872
4.232
−0.589
−0.733
(3a) (1)/(2)
—
0.0011
0.0013
0.0348 −0.0292 −0.3380
0.2836
(3b) (1)/(3)
−0.0088
0.0015
0.0029
23.19
18.57
16.92
−0.0121 −0.1111
(4)
Debt/GDP
(5)
After 5 years (A)
—
4.2577
6.3440
After 10 years (A)
—
0.9769
2.3792
19.931
9.6854
8.6226 36.1462
After 15 years (A)
—
0.2248
0.8928
16.8292
5.4837
4.8522 27.0803
(6)
27.96
30.11
6.6214 −1.7053 26.37
55.79
23.6062 17.0835 15.1319 44.9135
After 5 years (B)
3.7522
5.8091 10.4216
47.0235 24.4532 26.9620 57.9443
After 10 years (B)
0.6008
1.8179
6.4194
79.0791 19.8553 27.5718 60.1794
After 15 years (B)
0.0899
0.5696
3.9546 132.9812 16.1180 28.1998 62.4982
Sustainability of perpetual deficits
1974
1980
1981
43
1982
1983
1984
1985
South Korea (1)
Primary deficit
0.11
−1.85
−2.23
−0.97
−5.12
−2.3
−2.08
(2)
Real GDP growth−real i (A)
35.85
13.97
15.16
5.176
8.452
5.175
1.65
(3)
Real GDP growth−real i (B)
—
−0.795
4.781
−2.044
11.86
−16.13
0.03
−0.1470 −0.1874 −0.6057
−0.4444
−1.2606
−0.1425 −69.3333
(3a) (1)/(2) (3b) (1)/(3)
— −0.1324 0.0030
2.3270
−0.4664
—
14.05
15.62
(4)
Debt/GDP
(5)
After 5 years (A)
13.83
After 10 years (A)
2.9914
17.56
16.86
16.88
7.2432
7.6376 13.6721 11.5020
13.0016
15.4547
3.7033
3.6965 10.5813
7.4643
10.0035
14.1414
1.8623
1.7506
8.1798
4.7731
7.6739
12.9313
0.1427 14.5273
12.2700 19.5183
9.8413
40.4269
16.7508
After 10 years (B)
— 15.0241
9.6176 21.5899
5.4341
97.2165
16.6218
After 15 years (B)
— 15.5410
7.5176 23.8868
2.9176 234.0636
16.493
After 15 years (A) (6)
0.4745 −0.4317
After 5 years (B)
0.6489
17.65
Malaysia (1)
Primary deficit
0.43
−0.18
8.41
4.60
−0.27
−3.96
—
(2)
Real GDP growth−real i (A)
14.44
−5.858
−11.45
53.01
0.724
0.860
—
(3)
Real GDP growth−real i (B)
17.65
9.438
2.349
0.621
3.725
4.38
—
(3a) (1)/(2)
0.0297 −0.0307
−0.7344
0.0867 −0.3729
−4.6046
—
(3b) (1)/(3)
0.0243 −0.0190
3.5802
7.4074 −0.0724
0.9041
—
74.29
72.81
—
(4)
Debt/GDP
(5)
(6)
40.17
43.42
54.05
60.92
After 5 years (A)
20.4796 32.6565
99.8926
7.3402 71.6449
69.5654
—
After 10 years (A)
10.4482 24.5594 184.0953
0.9516 69.0936
66.4567
—
After 15 years (A)
5.3375 18.4680 338.7572
0.1899 66.6326
63.4784
—
After 5 years (B)
17.8350 27.6529
48.5182 59.2889 61.8626
58.5887
—
After 10 years (B)
7.9260 17.6088
43.5927 57.7076 51.5120
47.1110
—
After 15 years (B)
3.5299 11.2104
39.2071 56.1765 42.8913
37.8476
—
Budget deficits and economic activity in Asia
1974
1980
1981
1982
44
1983
1984
1985
Pakistan (1)
Primary deficit
1.30
−0.81
0.86
−0.70
0.60
−0.48
0.30
(2)
Real GDP growth−real i (A)
25.88
15.47
13.56
11.14
6.813
2.993
10.554
(3)
Real GDP growth−real i (B)
22.80
8.085
8.366
5.866
2.853
5.708
—
(3a) (1)/(2)
0.0502 −0.0523
0.0634 −0.0628
0.0880 −0.0534 −0.0284
(3b) (1)/(3)
0.0570 −0.1002
0.1028 −0.1193
0.2103 −0.0841
(4)
Debt/GDP
(5)
After 5 years (A)
21.4853 26.5174 25.5522 32.5446 39.4827 35.4465 35.9851
After 10 years (A)
6.8320 12.8909 13.5599 19.1665 28.4227 23.0265 21.8009
After 15 years (A)
2.1959
(6)
67.80
54.49
6.2528
48.20
55.23
54.86
54.55
— 59.41
7.2099 11.2772 20.4677 14.9518 13.212
After 5 years (B)
24.3160 36.9071 32.2879 41.5033 47.6895 41.3087
—
After 10 years (B)
8.7442 24.9875 21.6400 31.1809 41.4597 31.2766
—
After 15 years (B)
3.1680 16.9070 14.5148 23.4184 36.0474 23.6759
—
The Philippines (1)
Primary deficit
−1.87
−1.46
0.32
−0.13
−1.40
−2.64
−3.45
(2)
Real GDP growth−real i (A)
7.094
7.305
3.775
0.3495
10.68
17.00
—
(3)
Real GDP growth−real i (B)
—
7.558
1.795
−2.2
10.52
8.711
14.16
0.0847 −0.3719 −0.1311 −0.1553
—
(3a) (1)/(2)
−0.2636 −0.1988
(3b) (1)/(3)
— −0.1932
(4)
Debt/GDP
(5)
(6)
9.41
30.67
After 5 years (A)
6.6033 10.8147 14.5131 20.4142 13.1816 13.3434
—
After 10 years (A)
4.6110
7.5424 12.0729 20.0548
7.8842
6.0016
—
After 15 years (A)
3.1967
5.2422 10.0455 19.7015
4.6947
2.6529
—
— 10.6878 15.9799 23.2178 13.2774 19.2862
15.709
After 10 years (B) After 15 years (B)
— —
17.45
0.0591 −0.1331 −0.3031 −0.2225 29.44
After 5 years (B)
15.47
0.1783
20.78
21.98
7.3656 14.6350 25.9424
7.9997 12.5987
7.994
5.0578 13.4046 28.9876
4.7991
4.015
8.1943
Sustainability of perpetual deficits
1974
1980
1981
45
1982
1983
1984
1985
Singapore (1)
Primary deficit
−9.49
−9.34
−8.97
−12.44
−13.76
−16.91
—
(2)
Real GDP growth−real i (A)
17.97
16.09
11.80
6.511
5.683
8.43
—
(3)
Real GDP growth−real i (B)
13.01
10.16
4.757
3.135
4.993
7.505
—
(3a) (1)/(2)
−0.5281 −0.7602 −0.7602 −1.9106 −2.4213 −2.0059
—
(3b) (1)/(3)
−0.7294 −0.9193 −1.8856 −3.9681 −2.7559 −2.2532
—
(4)
Debt/GDP
(5)
(6)
39.23
64.70
61.53
66.18
71.20
75.79
—
After 5 years (A)
16.8726 30.3801 34.9022 47.7618 53.4229 49.8991
—
After 10 years (A)
7.0876 14.1032 19.6573 34.3256 39.9383 32.6249
—
After 15 years (A)
2.8050
6.3835 10.9292 24.5284 29.7099 21.0996
—
After 5 years (B)
20.9494 37.5301 48.3812 56.1473 55.2098 52.0958
—
After 10 years (B)
11.0318 24.0147 37.9586 47.5494 42.6769 35.5953
—
After 15 years (B)
5.6513 14.4507 26.6972 40.1813 32.8537 24.1044
—
Sri Lanka (1)
Primary deficit
4.03
17.73
10.68
11.84
10.16
6.41
—
(2)
Real GDP growth−real i (A)
21.07
16.54
13.49
2.227
5.297
7.865
—
(3)
Real GDP growth−real i (B)
21.68
13.83
12.57
2.177
8.697
12.57
—
(3a) (1)/(2)
0.1913
1.0719
0.7917
5.3166
1.9181
0.8150
—
(3b) (1)/(3)
0.1859
1.2820
0.8496
5.4387
1.1682
0.5099
—
53.62
78.39
77.50
81.57
81.48
69.22
—
(4)
Debt/GDP
(5)
After 5 years (A)
20.7309 37.0391 41.5348 73.6182 63.3827 47.6624
—
After 10 years (A)
8.0873 17.8033 22.4322 66.4957 49.4019 32.8985
—
After 15 years (A)
3.2268
8.8551 12.2859 60.1159 38.6011 22.7875
—
After 5 years (B)
20.2178 41.6294 43.2530 73.7980 54.0970 38.5207
—
After 10 years (B)
7.6957 22.3941 34.3074 66.8194 36.0504 21.5377
—
After 15 years (B)
3.0012 12.3291 13.8266 60.5532 24.1570 12.1426
—
(6)
Budget deficits and economic activity in Asia
1974
1980
1981
1982
46
1983
1984
1985
Thailand (1)
Primary deficit
−2.45
−2.93
2.19
4.47
1.63
1.35
2.27
(2)
Real GDP growth−real i (A)
16.54
13.365
3.409
2.871
2.969
4.362
2.751
(3)
Real GDP growth−real i (B)
—
13.02
2.709
−4.151
−0.309
−3.082
−5.781
(3a) (1)/(2)
−0.1481
0.2192
0.6424
1.5569
0.5490
0.3095
0.8252
(3b) (1)/(3)
—
0.2250
0.8084 −1.0768 −5.2751 −0.4380
—
20.04
22.96
(4)
Debt/GDP
(5)
After 5 years (A)
9.2431 12.3646 20.4378 24.9009 26.4930 25.9485 32.3662
After 10 years (A)
4.2205
6.7058 17.3831 21.8203 22.9623 21.0199 28.3640
After 15 years (A)
1.8841
3.6835 14.7997 19.1461 19.9120 17.0388 24.8696
(6)
28.05
28.45
30.83
32.05
36.95
After 5 years (B)
— 12.5537 21.1424 35.4220 31.1391 37.5548 49.9010
After 10 years (B)
—
6.9106 18.5986 44.0403 31.7170 43.9922 67.3435
After 15 years (B)
—
3.8505 16.3730 54.6935 32.2837 51.5205 90.8353
this period, there also emerged a primary budget surplus. This means that the question of instability is not as serious as it could have been. But we do notice in this case that if the configuration of x, r, and g was as in 1985, then with r2 as the interest rate, b would increase over time, rising to about 58 per cent after five years, and to about 62 per cent after fifteen years. However, even this value is not inordinately high. South Korea:
Here the most interesting outcome is if we take 1984 and use r2 as the interest rate. In this case, debt would be more than twice of GDP at the end of fifteen years. But if we use r1, the value is not much different from that observed during the period under study.
Malaysia:
The only case of instability is in 1982 using r1 as the measure of interest rate. Here the debt to GDP ratio rises quite rapidly, becoming over three after fifteen years. Otherwise, the deficits in the recent years appear to be sustainable.
Pakistan:
There is no problem of instability in any of the years. Therefore the value of b declines over time. Once again, we find no problem of instability and of existing deficits.
The Once again, there is no problem of instability and the debt to GDP ratio is quite low. Philippines: Singapore:
Every year from 1980 to 1984 there are primary budget surpluses and the real GDP growth rate exceeds the real interest rate on both definitions of the latter, thus guaranteeing stability. While the value of b has been rising over time, the issue of instability does not arise in this case.
Sri Lanka:
In this case, although primary deficit exists in each of the years under consideration,
Sustainability of perpetual deficits
47
g also exceeds in every case, thus satisfying the condition for stability. It is interesting to note that in no case does the ratio b take any unreasonable value. Thailand: For this case, instability is indicated from 1982 to 1985 using r2 as the measure of r. But even in this scenario, the increase in b is very gradual. For example, in terms of the configuration of r2, g, and x values, b would increase from about 50 per cent to about 91 per cent after fifteen years. The increase is much more modest for the other three years.
In sum, it would appear that with few exceptions, the fiscal policies followed by the countries in our sample passed the test of Sustainability. But as pointed out above, the facts discussed may have over-estimated the existence of stable situations because of the possible under-estimates of nominal interest rates. One way to deal with this issue would be to use an approach which does not depend on the rates of interest. Thus we turn to direct tests of the present-value borrowing constraint of the government.
TESTING THE PRESENT-VALUE BORROWING CONSTRAINT In an innovative paper, Hamilton and Flavin (1986) have proposed a direct test of the present-value borrowing constraint. If this constraint is violated, it would imply that perpetual primary deficits are not feasible. Briefly, their approach can be described by using the budgeted constraint in equation (1). Equation (1) can be written as Bt=(1+r)Bt−1−St (7) where St is the budget surplus. By forward recessive substitution, (7) can be written as (8) Hamilton and Flavin note that equation (8) should cause little controversy since it is derived from an accounting identity. ‘What is of economic interest (and subject in principle to empirical refutation) is what creditors expect to happen to the second term in (8) as N gets large’ (1986:811). It is clear from (8) that the present-value borrowing constraint is satisfied provided that the expected value of the first term is equal to the lefthand side, which, of course, implies that in the limit the expected value of the second term is zero. This allows them to write the general solution to equation (7) as:4 (9) The present value borrowing constraint can now be verified by testing whether A0=0 in equation (9). If the constraint is violated, it would follow that the observed deficits are not feasible. They propose three tests. However, we use only the first, because of the lack of data for other tests. Their test proceeds as follows. For any stationary process for
Budget deficits and economic activity in Asia
48
when A0=0, Bt will be stationary, whereas for A0>0, Bt will not be stationary. What we need to do is to test the above. This can be done by using Dickey and Fuller’s (1981) tests for unit roots for nonstationarity for both B and S. If both are stationary, we can conclude that A0=0 and the present-value borrowing constraint holds. If the maintained hypothesis is satisfied, namely that S is stationary, but B is not stationary, we conclude that A0≠0 and the presentvalue borrowing constraint is violated. The inference, as explained by Wilcox (1989), is a little more complicated if the maintained hypothesis is violated. He shows that under special assumptions, the present-value borrowing constraint may hold even in the presence of the non-stationarity of S. For our purposes, if S turns out to be non-stationary, we would assume that the constraint is violated. Dickey and Fuller (1981) provide likelihood ratio tests for testing for the existence of unit roots. We examine the stationarity of Bt and St as follows. We assume that the time series is adequately represented by the following second-order autoregressive process: Zt=β0+β1t+β2Zt−1+β3(Zt−1−Zt−2)+et (10) where et are independent identically distributed (i.i.d.) (0, σ2) random variables. Using we first test the null hypothesis that β0=β1=0 and this likelihood ratio rest statistic β2=1 against the general alternative (10). Next we test the null hypothesis that the secondorder autoregressive process has a unit root with possible drift, namely that β1=0 and β2=1. This is done by using their test statistic For testing the stationarity of B and S, we use real values of government debt and primary deficits. As Wilcox (1989) points out, the appropriate variables should be the discounted values. However, lack of data on discount rates forced me to follow Hamilton and
Table 2.5 Unit root tests for budget deficits excluding interest payments Country India
23.76
26.55
Indonesia
2.27
3.03
South Korea
6.99
4.41
Malaysia
5.03
6.27
Pakistan
7.41
7.53
23.06
16.03
Singapore
7.56
9.74
Sri Lanka
6.33
6.64
Thailand
42.28
42.71
The Philippines
Sustainability of perpetual deficits
49
Table 2.6 Unit root tests for government debt Country India
3.61
3.24
Indonesia
3.67
3.03
South Korea
3.43
3.31
Malaysia
2.33
2.55
Pakistan
4.71
4.45
The Philippines
4.64
5.51
Singapore
3.86
3.92
Sri Lanka
3.19
3.97
Thailand
2.28
1.77
Flavin (1986) instead, and use only the undiscounted values. Without reporting the and for the regressions, since they are not of interest in themselves, the values of primary deficit and the government debt are given in Tables 2.5 and 2.6 respectively. The 5 per cent Dickey-Fuller critical values for and are 5.68 and 7.24 respectively. We first consider the null hypothesis of non-stationarity for primary deficits in terms of Table 2.5. In terms of
the null is rejected for all but Indonesia and
the null is rejected for all but Indonesia, South Korea, and Malaysia. In terms of Malaysia. Thus we can conclude that for the other six countries the maintained hypothesis of the stationarity of primary surpluses is accepted. In Indonesia, South Korea, and Malaysia, we must conclude that the deficits are not sustainable, although we must note the possible exception recognized by Wilcox (1989) above. Turning to the second part of the test, we look at the stationarity of government debt in Table 2.6. In terms of both and the null of non-stationarity with unit root is accepted for all countries. Thus, we must conclude that even for the six or seven countries, depending on whether or as the decision criterion, the present-value borrowing constraint is we use violated, thus implying the instability or the unsustainability of government deficits. These results are in sharp contrast to the more favourable results reported in the previous section. How can we explain this? The most plausible explanation would appear to be the one already alluded to, namely the administered nature of the interest rates in the countries in our sample, which thus understates the level of real interest rates, thereby overstating the possibility of stability. The Hamilton-Flavin test, on the other hand, is not plagued by this problem since it does not require the use of interest rate data. On the other hand, we cannot rule out measurement problems with the national accounts estimates of budget deficits as explained by Eisner and Piper (1984), among others. Further, as explained by Wilcox (1989), the appropriate variables to use for the Hamilton-Flavin test are their discounted values, which we have not been able to do. These measurement problems may affect the tests for the stationarity of the two variables and thus our
Budget deficits and economic activity in Asia
50
inferences about the validity of the present-value borrowing constraint. At this stage, it is difficult to conjecture on the seriousness of the bias which these measurement errors may introduce.5 The main conclusion of this chapter is that the problem of instability may well be serious in many of the countries in our sample and, that, at the very least, the subject needs further investigation with more refined data.
3 Deficits and seigniorage The last two chapters have presented some evidence on the dimensions of the problem under consideration and its proximate causes. This chapter and the subsequent ones are devoted to some of the implications of the budget deficits and the public debt. The issue to be dealt with in this chapter relates to that of the monetization of budget deficits. Monetization of budget deficits provides the governments with a source of revenue. This process of creating high-powered money, called seigniorage, is one of the most important and controversial aspects of deficit financing. This chapter considers four aspects of the issue. First, it provides alternate estimates of seigniorage and examines their importance as a source of revenue. Second, it considers the extent to which these revenues are accounted for by the treasury. Third, it provides estimates of the monetization of government debt. And fourth, it provides some quantification of the effects of budget deficits on seigniorage for the countries in our sample.
ALTERNATE ESTIMATES OF SEIGNIORAGE A variety of measures have been proposed in the literature to measure revenue from money creation.1 But, unfortunately, lack of data precludes the use of most of them and obliges us to concentrate on only two of them. These measures are defined as follows: S1: Change in reserve money as a percentage of GDP, calculated from the data in row 14 of the International Financial Statistics, various issues. S2: Change in the net claims on government, again expressed as a percentage of GDP. Net claims are measured by the difference between rows 12a and 16d, the former giving claims on
Table 3.1 Table for S1 and S2—alternate estimates of seigniorage India Year
Indonesia
S1
S2
S1
1961–69 average
0.8705
0.8842
—
1970–79 average
1.445
0.8446
1980–85
1.8983
1980 1981
S2
South Korea S1
S2
—
1.9477
0.4054
1.9105
0.1092
2.5892
0.4888
2.6945
1.0524
−0.809
9.193
6.2910
1.8297
3.2742
1.9738
0.7217
−0.5908
0.3007
1.2554
2.9043
0.9376
−0.0756
−0.9399
1.6353
Budget deficits and economic activity in Asia
52
1982
2.1297
3.0779
0.2993
−0.0112
1.9334
0.2268
1983
1.5423
0.9033
1.3989
−0.7599
0.4426
0.2557
1984
1.9038
1.9705
0.6432
−1.408
0.2222
−0.5852
1985
2.7288
4.0370
1.0618
−1.8799
0.0940
−0.0874
Malaysia Year
S1
Pakistan
S2
The Philippines
S1
S2
S1
S2
1961–69 average
—
—
1.252
0.9565
0.793
0.3935
1970–79 average
2.3787
−0.6578
2.2017
2.1761
1.0218
0.2149
1980–85
0.8140
1.2379
1.8620
—
0.9838
0.9812
1980
1.8665
3.5623
2.385
1.2382
0.7029
0.3195
1981
1.1647
0.7898
1.156
−0.6301
0.5634
1.4446
1982
1.9112
1.5405
2.327
3.2846
0.3083
1.9936
1983
0.5146
1.4993
1.894
−1.5007
2.4369
2.1609
1984
0.4023
1.0371
2.2197
4.9680
1.0824
−0.2516
1985
0.8910
−1.0019
1.1904
1.3773
0.8089
0.2199
Singapore Year
S1
Sri Lanka
S2
S1
Thailand
S2
S1
S2
1964–69 average
1.3649
—
—
—
—
—
1961–69 average
—
—
0.8884
1.6894
0.8623
−0.4762
1970–79 average
2.4868
—
1.3213
1.4317
1.3404
1.397
1980–85
1.5959
—
1.7733
2.1746
0.6415
1.0407
1980
2.0007
—
1.6399
1.1557
0.4808
1.5506
1981
1.5986
—
1.4034
3.5763
0.8651
1.8697
1982
2.6967
—
1.8713
3.9723
0.7758
1.4704
1983
1.4428
—
2.1061
1.3865
0.4268
−0.4561
1984
1.0887
—
1.2859
−1.5200
0.6501
0.8249
1985
0.7476
—
2.3329
4.4767
0.6501
0.8249
Deficits and seigniorage
53
Figure 3.1 Estimates of seigniorage government while the latter gives government deposits. These data are again derived from the International Financial Statistics.
The estimated values of these two measures are given in Table 3.1. The data generally cover the period from 1960 to 1985. For the convenience of visual comparison betw een the two measures, they are plotted in Figure 3.1. From the figure, a somewhat close parallel between the two measures can be seen in many, though not in all, cases. Concentrating on S1, since it is the traditional measure,2 we notice a number of similarities between the various countries. The first one is that this source of revenue was more important in the 1970s than in the 1960s for the countries for which we have the data. For India and Sri Lanka, the percentage was also higher in the 1980s compared with the 1970s. But, for the other countries in the sample, the trend seems to have been reversed in the 1980s. The inter-country differences can be highlighted more sharply by considering each country separately. In the case of India, there was a sharp increase in the value of S1 from the 1960s to the 1970s, almost doubling in the 1980s, touching almost 1.9 per cent. When viewed against the figure of almost 4.6 per cent from direct taxes during the same time period, seigniorage constituted a very significant source of revenue. In the case of Indonesia, the magnitude went down to almost half of what it was in the 1970s and, compared to the almost 14 per cent of revenues from the direct taxes, the value of 1 per cent from seigniorage stands in marked contrast with what is the case for India. The decline in the case of South Korea is even more remarkable. The value of S1 went down
Budget deficits and economic activity in Asia
54
from about 2.59 per cent in the 1970s to about 0.2 per cent in the 1980s, which is even lower than the value reported by King and Plosser for the USA.3 A comparison of India and South Korea is quite interesting here, because the percentage of revenue from direct taxes is virtually the same in the two countries and yet the value of S1 varies from 1.90 for India to only 0.2 per cent for South Korea. It would be instructive to see in the last section of this chapter whether the effect of deficits on monetization explains this striking difference. Malaysia seemed to have had the same kind of experience as Indonesia, both in terms of the decline in the value of S1 and in terms of its importance relative to the role of revenue from direct taxes, namely, relatively small. In the case of Pakistan, even though the value of S1 declined in the 1980s compared to the 1970s, its importance as a source of revenue continued to be even greater than in the case of India. The same is true of the Philippines; in fact, even more so. Sri Lanka showed a marked trend in the value of S1 and like Pakistan, seigniorage constitutes a very significant part of the revenues. In the case of Thailand, while the absolute value of S1 almost halved compared to what it was in the 1970s, when viewed against the share of direct taxes it still accounts for a substantial source. These brief comparisons clearly show that, with the exception of South Korea, revenue from money creation constitutes an important source of total revenues for the countries in our sample. This is a characteristic of developing countries in general and emphasizes the importance of the examination of the issue about the monetization of government deficits and its possible effects on other main variables.
SEIGNIORAGE AND THE SHARE OF GOVERNMENT The last section provided estimates of the total revenue from money creation. In this section, we examine the role of the government (the treasury) in the creation of reserve money. It was seen from the definition of the government budget constraint in the last two chapters that the government’s total borrowing requirements must be met by the sale of its debt either to the private sector or to the central bank. The latter, as we know, leads to the creation of reserve money. To these estimates we now turn. For this purpose, we report three estimates: the rate of growth of reserve money; the rate of growth of reserve money due to goverment; and the share of the government in the growth of the total. The relevant estimates are given in Table 3.2. Column (1) is calculated from row 14 of the International Financial Statistics. Column (2) is based on the rate of growth of net claims on the government as defined in the previous section. The final column is then calculated from the data in rows 14, 12a and 16d. It can be seen from this table that the growth rate of total reserve money almost doubled from the 1960s to the 1970s for India, Pakistan, the Philippines, Sri Lanka and Thailand. These increases were accompanied by increases in the growth rate due to government. However, the treasury’s share showed a decline in India, Pakistan, the Philippines, and Sri Lanka, suggesting that there were other factors operating which lead to the growth of the overall base. Only in the case of Thailand did the share of the government go up. The experience during the 1980s for these countries is somewhat different. Thus in the case of India the share of the treasury almost doubled as compared
Deficits and seigniorage
55
to the 1970s. This was due to the sharp increase in the growth rate due to government relative to the overall growth rate. In the case of Pakistan, although both rates declined, the treasury’s share still went up. Similar differences can be noticed for the other three countries. The other three countries, Indonesia, South Korea, and Malaysia stand out from the rest. Thus in Indonesia the government’s share continued to decline. In the case of South Korea, the same thing happened from the 1970s to the 1980s. Finally, Malaysia showed a decline in the growth rate due to the government. The main conclusion of this section is that there are sufficient differences in the experiences of the countries in our sample, as far as the behaviour of the growth of reserve money is concerned, to
Table 3.2 Monetary base growth and the share of government Year
Monetary base growth
Rate of growth of reserve money due to government
Share due to treasury
India 1961–69 average
7.5904
7.5492
106.460
1970–79 average
14.1230
28.7161
72.6760
1980–85
16.1739
26.4147
144.1332
1980
15.2189
40.9881
178.9451
1981
10.5013
29.8802
231.3376
1982
18.0276
27.2649
144.5266
1983
12.9984
7.3885
58.5700
1984
16.7954
17.7479
103.4783
1985
23.5019
35.2188
147.9418
1961–69 average
283.055*
29.9157*
38.01*
1970–79 average
32.0954
−76.8216
−18.5036
1980–85
18.5612
54.5942
−161.90
1980
36.1986
153.6070
−137.6812
1981
16.1482
88.5728
−331.3761
1982
4.7704
−5.2796
108.5561
1983
25.1035
21.4992
−75.9457
Indonesia
Budget deficits and economic activity in Asia
56
1984
10.9576
59.8192
−470.1599
1985
17.8916
9.3467
−64.8039
1961–69 average
34.2376
11.0870
13.9709
1970–79 average
23.9028
120.3766
19.177
1980–85
4.8153
40.4163
−85.2872
1980
−6.4591
45.0593
−50.8929
1981
−13.6252
209.5368
−173.9819
1982
36.5096
10.5634
11.7302
1983
7.0588
12.4204
57.7778
1984
3.7363
−28.5411
−263.3987
1985
1.6714
−6.5411
−92.9578
1961–69 average
176.072
36.6302
2.0911
1970–79 average
16.4403
14.6975
−36.1436
1980–85
10.4739
−48.8467
147.9230
1980
18.0975
−71.8230
190.8543
1981
10.3342
−61.0738
67.8092
1982
16.6946
−332.4138
80.6020
1983
4.2823
154.7478
291.3408
1984
3.6706
48.0489
257.8125
1985
9.7649
−30.5665
−0.8804
South Korea
Malaysia
Year
Monetary base growth
Rate of growth of reserve money due to government
Share due to treasury
Pakistan 1961–69 average
7.5299
8.3100
101.1223
1970–79 average
13.8905
19.5056
41.610
1980–85
13.5480
14.1295
59.7941
1980
16.6637
9.0374
51.9221
Deficits and seigniorage
57
1981
8.2029
−5.0461
−55.0436
1982
17.6780
31.7743
141.1348
1983
13.7588
−12.3974
−79.2274
1984
16.3667
54.0971
223.8070
1985
8.6179
7.3115
76.0724
1961–69 average
11.5150
14.756
56.8003
1970–79 average
18.3842
16.1139
23.7307
1980–85
17.6379
36.3963
208.3704
1980
12.8107
18.4382
42.2886
1981
6.1017
80.7692
408.3333
1982
5.1118
68.7943
707.2917
1983
47.5684
49.8199
88.3919
1984
20.1167
−5.4487
−23.2082
1985
14.1183
5.6780
27.1255
1961–69 average
7.6112
12.6355
109.6497
1970–79 average
15.3306
14.7686
34.10
1980–85
21.8652
29.7011
197.8062
1980
19.7003
109.4052
704.7663
1981
17.9967
20.6564
254.8198
1982
23.7407
22.1997
212.2779
1983
26.4593
7.7699
65.8337
1984
16.7974
−21.3427
−242.7529
1985
26.4969
39.5183
191.8955
1961–69 average
7.6917
−94.622
−58.7764
1970–79 average
13.6809
61.6011
110.549
1980–85
8.5879
17.9983
160.4005
The Philippines
Sri Lanka
Thailand
Budget deficits and economic activity in Asia
58
1980
13.9674
39.4027
214.2450
1981
6.5992
22.9092
322.4868
1982
11.9882
24.1896
216.1202
1983
10.4855
16.7323
189.5397
1984
5.5857
−4.7569
−106.8720
1985
8.4869
9.5127
126.8833
Note * 1966–69 average
warrant an answer to the question: does monetization of the deficit alone explain these differences?
MONETIZATION OF GOVERNMENT DEBT We can get a better idea about the movements in budget deficits and the revenue from the creation of reserve money if we can look at the direct estimates of the monetization of government debt. These estimates are given in Table 3.3. The first column for each country gives budget deficits as a percentage of GDP and the second the degree of monetization. The degree of monetization was calculated as the ratio of net credit to the government by the central bank, as already defined, to the government’s net borrowing requirements as given in row 84 of the International Financial Statistics. Once again, there are important inter-country differences. For India, Malaysia, Pakistan, and Thailand, the degree of monetization increased in the 1980s compared to the previous decade, although the rate of increase was not identical for all. In the case of Indonesia, South Korea and the Philippines, the situation was just the opposite. For Sri Lanka, the situation remained relatively unchanged after the 1960s. It is also interesting to consider the trends in the budget deficits and in the degree of monetization for each country across time. In the case of India, the deficits and the degree of monetization seem to move positively. The deficit went up from 4.38 in the 1970s to 7.28 per cent in the 1980s and the degree of monetization increased from −22.47 to 43.11 per cent, although within the 1980s we can see various other possibilities. The same is the case with Indonesia. In the case of South Korea, however, the opposite is the case. While the deficit increased from 1.57 to 2.05 per cent, the degree of monetization went down from 28.71 to 13.03 per cent. For Malaysia, again the relationship is positive. But for Pakistan, the three periods show a certain lack of consistency. Thus the deficit went up from the 1960s to the 1970s and so did the degree of monetization. But, although the deficit went down from the 1970s to the 1980s, the degree of monetization went up. A similar negative relationship is evident in the case of the Philippines. Sri Lanka shows the same type of lack of consistency between the three periods as displayed by Pakistan. This kind of relationship, we shall see below, can be explained by considering not the one period budget constraint, as we have done so far, but the intertemporal budget constraint. Finally, Thailand shows a positive relationship
Deficits and seigniorage
59
Table 3.3 Budget deficits and the degree of monetization India Year
Deficit
Indonesia
Monetization
Deficit
Monetization
1961–69
4.60
33.59
—
—
1970–79
4.38
−22.47
1.85
44.93
1980–85
7.28
43.11
1.39
−228.08
1980
71.6
53.79
2.42
−116.18
1981
6.17
58.80
1.36
−210.86
1982
6.74
38.39
2.26
−59.31
1983
7.06
20.61
1.41
17.72
1984
8.71
28.12
0.56
−577.88
1985
7.81
58.94
0.29
−481.98
South Korea Year
Deficit
Malaysia
Monetization
Deficit
Monetization
1961–69
—
—
4.26
−13.43
1970–79
1.57
28.71
8.52
−4.16
1980–85
2.05
13.03
13.32
22.60
1980
2.24
21.80
13.17
82.17
1981
3.37
73.52
19.11
9.86
1982
3.13
12.31
17.85
15.85
1983
1.09
60.00
13.20
23.51
1984
1.22
−76.33
8.89
25.86
1985
1.25
−13.15
7.73
−21.84
Pakistan Year 1961–69
Deficit 4.82
The Philippines
Monetization
Deficit
Monetization
14.84
—
—
a
83.31a
1970–79
7.67
54.42
1.26
1980–85
6.06
62.72
2.56
36.33
1980
5.69
81.41
1.52
25.11
1981
5.81
−23.70
3.98
36.28
1982
4.77
182.20
4.23
47.11
Budget deficits and economic activity in Asia
60
1983
6.84
−37.50
1.94
111.14
1984
6.20
145.91
1.84
−13.66
1985
7.07
32.99
1.83
12.01
Sri Lanka Year
Deficit
Thailand
Monetization
Deficit
Monetization
1961–69
6.01
42.51
—
—
1970–79
8.41
20.48
2.68
8.93b
1980–85
14.57
21.82
4.36
28.23
1980
22.20
63.25
3.76
60.99
1981
15.60
28.57
3.26
42.72
1982
17.61
27.44
6.35
29.83
1983
13.61
12.87
3.96
33.03
1984
8.87
−48.69
3.44
−12.81
9.52
46.99
5.36
15.61
1985 a
b
Notes: 1976–79: 1972–79.
between deficits and the rate of monetization between the two periods.
DETERMINANTS OF SEIGNIORAGE The question in this section is: what determines the rate of growth of reserve money? Since this essentially represents the stance of monetary policy, in the literature this question has been dealt with under the rubric of the reaction function of the monetary authority. In other words, what does the monetary authority react to in determining the rate of growth of reserve money? Is it the deficit? Or are there other goals? In the ultimate analysis, only empirical results can resolve the issue for any particular case. While there is considerable literature on this issue, there is no consensus on the objectives of monetary authority. The matter is even more problematic in our case because there is little empirical literature on this issue for the developing countries in general and for our sample of countries in particular. The most promising approach to consider for our case is the one proposed by Sargent and Wallace (1981). Very simply, their approach is that in a regime of fiscal dominance, where the fiscal authorities dictate the time path of government finances (given the constraint imposed by the absorption of government debt by the private sector), in order to satisfy the inter-temporal budget constraint, the monetary authority must monetize the government debt. Of course, the monetary authority may choose a different time path of monetization than implied by the traditional approach. It may, for example, monetize in the future and not in the current period. This implies that the hypothesis of fiscal dominance is compatible with lack of contemporaneous relationship between deficits and growth in base money. In any event,
Deficits and seigniorage
61
the point is that, according to this hypothesis, budget deficits become a determinant of the growth of reserve money. The attraction of this hypothesis for our study lies in the fact that in the countries of our sample, the capital markets being very shallow and the interest rates being generally administered, the scope for the absorption of government debt by the private sector is rather limited. Therefore, a recourse to the monetary authority is the only option if development plans are to be pursued. At least, this is alleged to be the case in the literature.4 The question for us, then, is to test whether the hypothesis of fiscal dominance holds for the countries under study. How does one test this hypothesis? Two approaches have been implemented in the empirical literature. One is the so-called ‘structuralist’ approach, and the other is the non-structure approach. In the first approach, the basic idea is to specify a structural model which could lead to the specification of a reaction function for the monetary authority. To the extent that the reaction function is model-specific, the arbitrariness of this approach is obvious. Besides this, if we are ultimately to estimate a reduced form equation and not the full structural model, the interpretation of the estimates may not be always unambiguous. These shortcomings of the structuralist approach have led to the use of the alternate approach, as proposed by King and Plosser (1985). This approach is based on the notion that, since the true structure is not known, it is better not to impose any a priori restrictions. It simply specifies that the behaviour of the monetary authority is determined by a variety of fiscal and other variables. The basic motivation of both approaches, however, is the same in our context, namely, to test for the significance of budget deficit in determining growth of base money and hence for the validity of the fiscal dominance hypothesis. Consequently, we would present results using both approaches. The structural approach As pointed out above, this approach consists of specifying and estimating a model which examines the relationship between the objectives of the monetary authority and its financing of budget deficits. The empirical literature in this, with minor exceptions, relates to the developed countries. In almost all of these studies, single equations have been estimated in which the monetary base, or its rate of growth, is the dependent variable, and budget deficit—nominal or real—is one of the explanatory variables.5 One of the few exceptions is Levy (1981) who specifies a complete IS-LM-AS type of model. However, even he does not derive the monetary base equation from the underlying structural model. Instead, the model is used to identify instruments used in the two-stage least squares estimation of the monetary base equation which is specified on an ad hoc basis. The findings for the USA and a number of other developed countries are not conclusive.6 In order to implement this approach for this study, I follow others in the field and specify a reaction function for the monetary authority without articulating the underlying structural model, even though it can be easily rationalized within the framework of the standard IS-LM-AS type of a model. Furthermore, since data limitations preclude experimentation with some richer specifications, I confine myself to the relatively simple reaction function. I then assume that the desired level of reserve money or the monetary base is given by:7
Budget deficits and economic activity in Asia
62
(1) where RM* stands for the desired level of reserve money, πe for expected inflation, GD for government debt, u for the unemployment rate and T for time. I further assume that the actual level of RM is adjusted to the desired level according to the simple Koyck scheme, namely (2) Substituting (1) in (2) gives (3) where b4=1−λ and bi=ai/1−λ, bi=1, 2, 3, and 5. Since we do not have the data on unemployment for most countries in our sample, following Ferguson and Gupta (1979) I assume that (4) represent the actual and the trend rate of growth of real GDP, where and respectively. Substituting for ut in (3) we get (5) In order to avoid spurious regression results, first differences of the data were used, assuming that first differencing was sufficient to induce stationarity. This reduces equation (5) to (6) Since, from the definition of the government budget constraint, we know that ∆GDt=DEFt, where DEF is the budget deficit, (6) can also be written as (7) Since the variables GD, DEF and in equations (6) and (7) are endogenous, these equations were estimated by the 2SLS estimator, with the instruments in the first stage being πe and the lagged values of the endogenous variables. πe was approximated by a three-year moving average of the actual rate of inflation based on the GDP implicit price deflator and the trend rate of growth of real GDP was estimated by a simple regression of the log of real GDP on time. Relatively simple though it is, the model does capture the more familiar goals of monetary policy in developing countries. The inclusion of expected inflation is easy to
Deficits and seigniorage
63
understand on the grounds that monetary authority may try to be accommodative with respect to inflationary expectations, whereas the inclusion of allows us to consider the role of monetary authority in matters relative to cyclical movements in the economy. In terms of the most important aspect of the model, namely, the inclusion of government debt/ deficit, while a number of explanations have been offered in the literature,8 we consider the one by Sargent and Wallace (1981), as already discussed, as the most pertinent for our purpose. The one variable not included, but included in many studies,9 is the rate of interest. It has often been argued that the monetary authority, in its desire to maintain interest rate stability, reacts to it by changing the monetary base. But, as already pointed out, interest rates in our sample of countries are generally administered and are not generally determined by the supply of and demand for credit. Consequently, its omission is not as serious as it may appear at first sight. Besides, even for the developed countries, where it has been included, it has not been found to be significant.10 Now, it may be argued that in terms of the effects on the real side of the economy, what matters is the change in the real value of the monetary base and not a change in its nominal value. To examine this possibility, I estimated equations (6) and (7) in their nominal as well as their real versions. In the real version, the dependent variable is the change in the real value of RM and real government debt and deficits are measured as described in Chapter 1. The estimates of equations (6) and (7), both their nominal as well as the real versions, are given in Table 3.4. Instead of discussing the results for each country separately, we examine them together in terms of their general characteristics and in terms of the effect of each explanatory variable separately. We can make the following observations: (i) The model does not fit equally well for all of the countries in the sample. With the exception of South Korea and Malaysia, the nominal version fits better than the real. This may merely indicate that the monetary authority’s target is the nominal monetary base. In virtually every case, the results, using change in government debt as a measure of deficits, performs as well or better as the NIPA definition of budget deficits. This presumably reflects the fact that
Table 3.4 Estimates of equations (6) and (7) India Nominal (1) ∆RM−1 ∆π
e
∆GD
DEF
Real (2)
(1)
(2)
0.47659
0.23455
0.4423
0.49039
(3.0773)
(0.72482)
(3.2172)
(2.7525)
−0.23949
−0.30913
−0.0071512
−0.013955
(−0.66534)
(−0.69596)
(−1.4977)
(−3.0058)
0.15463
—
0.13261
—
(3.96)
—
(2.648)
—
—
0.2046
—
0.040612
Budget deficits and economic activity in Asia
64
—
(2.4321)
—
(0.65920)
−0.15652
−0.29329
0.0047459
0.0053828
(−0.6567)
(−1.0670)
(1.5578)
(1.4928)
1.2933
0.8285
0.0090319
0.0079907
Constant
(1.0512)
(0.55565)
(0.63089)
(0.25515)
2
R
0.8744
0.8270
0.6777
0.5740
DW
2.1857
2.0532
2.7384
2.5918
S.E.
3.7486
4.3985
0.048509
0.055772
Indonesia Nominal (1)
Real (2)
∆RM−1
(1)
(2)
—
0.15573
—
−0.15509
—
(0.52565)
—
(−0.49613)
—
10.915
—
−0.0396508
—
(0.91997)
—
(−0.54078)
∆GD
—
—
—
—
DEF
—
0.14828
—
−0.018004
—
(0.77269)
—
(−0.16326)
—
−35.005
—
−0.055668
—
(−1.0853)
—
(−1.24775)
—
333.21
—
1.7216
∆π
e
Constant
—
(2.3669)
—
(1.8331)
2
R
—
0.2254
—
0.0850
DW
—
2.2325
—
1.9571
S.E.
—
270.33
—
1.5803
South Korea Nominal (1) ∆RM−1 ∆πe
Real (2)
(1)
(2)
0.17276
0.20581
0.0717
0.084283
(0.82114)
(0.98792)
(0.36683)
(0.38951)
−11.403
−7.2238
−0.18326
−0.1799
(−0.55296)
(−0.35016)
(−0.80931)
(−0.79714)
Deficits and seigniorage ∆GD
65
0.050788
—
0.027807
—
(0.41097)
—
(−0.14539)
—
—
0.14932
—
0.0080289
—
(0.92334)
—
(0.032003)
6.6387
10.098
0.42275
0.43013
(0.40663)
(0.64214)
(2.4687)
(2.4396)
127.27
99.187
1.1127
0.95916
(1.2634)
(1.0060)
(0.95897)
(0.63286)
R2
0.0693
0.0984
0.3290
0.3283
DW
2.0271
2.0500
2.3276
2.3115
S.E.
314.81
309.84
3.5902
3.5919
DEF
Constant
Malaysia Nominal (1) ∆RM−1 ∆π
e
∆GD
DEF
Constant
Real (2)
(1)
(2)
−0.14471
−0.26002
−0.33157
0.077664
(−0.49837)
(−0.86727)
(−0.99867)
(0.30457)
0.73874
−2.0960
−0.006762
−0.75150
(0.020408)
(−0.066164)
(0.012665)
(−2.2146)
0.096385
—
0.16166
—
(1.5215)
—
(1.7889)
—
—
0.055558
—
0.016107
—
(1.8985)
—
(0.49361)
30.693
27.232
0.15070
−0.36631
(1.4914)
(1.3994)
(0.42797)
(−1.6972)
410.48
449.48
1.3884
2.0838
(2.3188)
(2.8266)
(1.0999)
(1.1890)
2
R
0.2773
0.3334
0.6491
0.5707
DW
1.8485
2.0835
1.6235
2.0938
S.E.
242.51
232.90
2.2705
2.5111
Budget deficits and economic activity in Asia
66
Pakistan Nominal (1) ∆RM−1 ∆π
e
∆GD
(2)
(1)
(2)
0.5662
0.18665
0.14158
0.1237
(5.7513)
(0.90634)
(1.2314)
(0.7715)
−0.10676
−0.10388
−0.004884
−0.0066786
(−1.5637)
(−1.1072)
(−4.2628)
(−4.1729)
0.13288
—
0.086011
—
(7.2387)
—
(4.3236)
—
—
0.26174
—
−0.04032
—
(4.2354)
—
(−0.48258)
0.057844
0.03528
0.0019146
0.0010122
(0.97022)
(0.4306)
(2.1943)
(0.86607)
0.11365
0.021291
0.0056196
0.016958
DEF
Constant
Real
(0.39412)
(0.051608)
(1.6308)
(1.4593)
2
R
0.9172
0.8439
0.7614
0.5540
DW
2.6655
2.7806
2.7746
2.3157
S.E.
0.78999
1.0846
0.0127
0.017364
The Philippines Nominal (1) ∆RM−1 ∆π
e
∆GD
DEF
Constant
Real (2)
(1)
(2)
0.38161
0.53082
−0.082652
−0.053935
(1.3832)
(2.6505)
(−0.37098)
(−0.24325)
0.21625
0.24157
−3
−0.40717 ×10−3
(2.5501)
(2.6144)
(−1.10128)
(−0.6874)
0.17082
—
−0.05998
—
(1.4615)
—
(−0.66233)
—
—
0.11577
—
0.0072813
—
(1.3662)
—
(0.13336)
−0.067018
−0.045548
−3
0.26884×10−3
(−1.8816)
(−1.0714)
(1.3887)
(1.1506)
0.4698
0.39543
0.004643
0.0037352
−0.5809×10
0.32656 ×10
Deficits and seigniorage
67
(1.4577)
(1.1973)
(2.1796)
(1.5552)
2
R
0.7073
0.7040
0.2061
0.1909
DW
2.2858
2.3127
1.9031
1.9843
S.E.
1.2066
1.2134
0.0075987
0.007671
Singapore Nominal (1) ∆RM−1 ∆π
e
∆GD
(2)
(1)
(2)
0.14856
0.51341
−0.041069
0.0016861
(0.5669)
(2.6699)
(−0.21301)
(0.008444)
3.7371
−6.255
−0.21872
−0.40617
(0.2631)
(−0.43741)
(−1.0394)
(−2.4212)
0.11281
—
0.10379
−0.10292
(2.1462)
—
(1.6590)
(−0.98456)
—
−0.08192
—
—
—
(−0.95217)
—
—
12.748
6.1748
0.15793
0.061102
(8.3163)
(0.74412)
(1.4050)
(0.62986)
111.23
145.45
1.3868
2.3342
DEF
Constant
Real
(1.8134)
(2.2750)
(1.3643)
(3.0887)
2
R
0.5174
0.432
0.3715
0.3181
DW
2.1786
2.1717
2.2666
2.1111
S.E.
146.48
158.92
1.7755
1.8494
Sri Lanka Nominal (1) ∆RM−1 ∆π
e
∆GD
DEF
Real (2)
(1)
(2)
0.76355
0.40841
0.16584
0.15024
(6.7469)
(2.5791)
(1.0851)
(1.0379)
−0.057372
−0.047722
−0.001241
−0.0012363
(−2.1793)
(−2.0207)
(−3.9725)
(−4.2461)
0.090646
—
0.005992
—
(2.6944)
—
(0.1334)
—
—
0.07069
—
0.014688
Budget deficits and economic activity in Asia —
Constant
68
(3.8395)
—
(0.7871)
−0.026543
−0.02448
3
−0.1428×10−3
(1.1308)
(−1.1741)
(−0.55242)
(−0.55418)
0.081437
0.0800502
0.0020253
0.13005×10−3
−0.15509×10−
(0.86042)
(0.98757)
(2.1308)
(1.0075)
2
R
0.8309
0.8663
0.4748
0.4894
DW
2.6081
2.5433
1.6921
1.7671
S.E.
0.31075
0.27632
0.0034583
0.0034099
Thailand Nominal (1) ∆RM−1 ∆π
e
∆GD
DEF
Constant
Real (2)
(1)
(2)
0.40085
0.44443
−0.88071
0.038374
(2.4390)
(3.3764)
(−0.54723)
(0.21908)
0.16027
0.18081
−0.0010653
−0.0027811
(1.6985)
(2.1121)
(−0.82983)
(−2.2273)
0.12688
—
0.14092
—
(3.0667)
—
(3.050)
—
—
0.091112
—
0.048859
—
(4.0079)
—
(1.4493)
0.32451
0.12061
0.0012596
0.93972×10−3
(0.22611)
(0.88402)
(0.80686)
(0.4764)
0.84718
0.77496
0.01186
0.012981
(1.8525)
(1.8597)
(2.31179)
(1.8510)
2
R
0.7412
0.7865
0.5152
0.3703
DW
1.9740
1.6295
2.0073
1.8868
S.E.
1.2023
1.0921
0.013846
0.01578
changes in government debt, rather than the measured deficit, better reflect government’s total borrowing requirements. On the whole, with the exception of Indonesia and South Korea, some or the other version of the model fits well the rest of the countries. There is no evidence of serial correlation in any of the estimated equations. In view of these results, the rest of the discussion is based on the estimates of equation (6) in nominal form, except where expressly stated otherwise.
Deficits and seigniorage
69
(ii) Turning to the individual coefficients, we start with the coefficient of the lagged dependent variable RM(t−1). Its coefficient is given by 1−λ where λ, as already seen, is the speed of adjustment of the actual to the desired value of the monetary base. A significant and positive coefficient of RM(t−1) has been interpreted by Levy (1981) to indicate the continuity of monetary policy. However, in our context, a better interpretation would be in terms of λ which can be used to examine the speed with which the monetary authority adjusts the monetary base in response to changes in its determinants, including the deficit. For example, we might say that it indicates how quickly the monetary authority monetizes the budget deficit. Keeping in mind this interpretation and the fact that we are using annual data, we can easily verify that the monetary authority adjusts 52, 100, 82, 100, 43, 62, 85, 24 and 60 per cent of the actual monetary base to its desired value in one year for India, Indonesia, South Korea, Malaysia, Pakistan, the Philippines, Singapore, Sri Lanka, and Thailand, respectively. The relatively low values for Pakistani and Sri Lanka are particularly noteworthy. They suggest, for example, that the monetary authority in these two countries does not monetize the budget deficit as rapidly as in, say, Indonesia and Malaysia. An examination of the underlying causes for these differences would be interesting. (iii) The results with respect to the coefficient of expected inflation are rather mixed. Thus for India, Indonesia, South Korea, Malaysia, and Singapore, the coefficient is not statistically significantly different from zero, whereas it is significant for the other four countries. However, the signs are not identical. Thus, the coefficient is positive in the case of the Philippines and Thailand, indicating an accommodative policy but it is negative for Pakistan and Sri Lanka, where a counter-cyclical stance is suggested. (iv) In terms of the variable reflecting cyclical fluctuations, the coefficient of is not at all significant for India, South Korea, Pakistan, and Thailand. However, it is marginally significant for Indonesia, Malaysia, and Sri Lanka, and significant for the Philippines and Singapore. But the signs are not the same for all five. The sign suggests that the monetary authority pursued a pro-cyclical policy in the cases of Indonesia, the Philippines, and Sri Lanka. (v) Finally, consider the variable of the major interest, namely government debt/deficit. Changes in government debt have a significant effect on the monetary base in the cases of India, Pakistan, Singapore, Sri Lanka, and Thailand and a marginally significant effect in Malaysia and the Philippines. Only in the case of South Korea is the coefficient not statistically significantly different from zero. We cannot say anything definitive about Indonesia, because of the lack of data on government debt. But in terms of the sign of the deficit, the NIPA concept, the coefficient is not significant. The coefficient has a positive sign in all cases, thus suggesting that there is monetization of government debt. The magnitude of the coefficient is also quite interesting. It varies between about 90 to 170 million for each one billion worth of change in government debt, all measurements being in the units of local currency. This compares well with the figure of 60 to 65 million reported by Levy for the USA. We can also use the estimates of Table 3.4 to examine the extent to which the observed changes in monetary base can be attributed to the current deficits, as well as to the current and past deficits. Once again, concentrating on the nominal estimates with ∆GD as the measure of deficit in column (1) for each country, this can be done by using
Budget deficits and economic activity in Asia
70
the coefficient of ∆GD and its steady-state value. The pertinent calculations are shown in Table 3.5 for different time periods. The explanation of the various columns of this table is as follows. The first column gives the change in reserve money or the monetary base in billions of local currency; the second column gives the change in government debt, again in the same units; the third column gives the produce of column (2) and the coefficient of ∆GD in column (1) of Table 3.4 for each country; column (4) gives the product of the steady-state value of the coefficient of ∆GD and ∆GD; column (5) expresses column (3) as a percentage of column (1), which gives the proportion of the given increase in the monetary base which can be attributed to the current change in government debt or the current borrowing requirements; and finally, column (6) gives the percentage of the increase in the monetary base which can be attributed to the current and the past changes in government debt. In other words, columns (5) and (6) represent the monetization of current and past deficits. Turning to the estimates themselves, and recalling that the data used are annual, we can draw a number of conclusions. First, there are wide variations in the effects of both current and cumulated deficits on the expansion of the monetary base. Thus, for South Korea only about 7 per cent of the increase in the monetary base can be attributed to the current change in government debt for the entire period, whereas the corresponding figure for India is about 45 per cent and for Singapore even higher at 62 per cent. With the exception of South Korea, these percentages are quite substantial. Turning to the effects of both current and past changes in the debt, the results are again quite variable. Thus, South Korea still stands in a class by itself, with only 8 per cent of the increase in the monetary base being accounted for by the monetization of the current and past borrowing requirements, whereas the corresponding figures for India and Sri Lanka are as high as almost 85 and 98 per cent respectively. To the extent that inflation is ultimately a monetary
Table 3.5 Increases in reserve money attributable to current and past government borrowing requirements India Period
1974– 75
∆RM
3.1
∆GD 0.15463×∆GD 0.2954×∆GD
Column (3) as percentage of column (1)
Column (4) as percentage of column (1)
7.31
1.13
2.159
36.45
69.65
1976– 80
20.46 50.86
7.86
15.02
38.42
73.41
1981– 83
27.88 97.82
15.13
28.896
54.27
103.64
1974– 83
19.2142 56.23
8.695
16.61
45.25
86.45
Deficits and seigniorage
71
South Korea Period
∆RM
∆GD
0.050788× ∆GD
0.0614× ∆GD
Column (3) as percentage of column (1)
Column (4) as percentage of column (1)
1974– 75
226.5
−10.8
−0.5485
−0.6631
−0.242
−0.293
1976– 80
433.4
272.62
13.8458
16.7389
3.195
3.862
1981– 84
251.0
861.0
43.7285
52.8654
17.42
21.06
329.451 435.036
22.0946
26.7112
6.707
8.1078
1974– 84
Malaysia Period
∆RM
∆GD
1974– 75
211.0 −1685.0
1976– 80
697.8
1981– 85 1974– 85
0.096385× ∆GD
0.0842× ∆GD
Column (3) as percentage of column (1)
Column (4) as percentage of column (1)
−162.41
−141.94
−76.97
−67.27
1621.6
156.30
136.60
22.40
19.58
647.2
6066.0
584.67
510.999
90.34
78.96
581.387
1350.5
130.12
113.77
22.38
19.59
Pakistan Period
∆RM
∆GD
0.13288× ∆GD
0.3063× ∆GD
Column (3) as percentage of column (1)
Column (4) as percentage of column (1)
1974– 75
0.7325
1.552
0.2062
0.4754
28.15
64.90
1976– 80
4.7662
3.6904
0.4904
1.1304
10.289
23.717
1981– 84
6.7112
9.8678
1.3112
3.0225
19.537
45.04
1974– 84
4.7401
5.5479
0.7372
1.6993
15.55
35.85
Budget deficits and economic activity in Asia
72
The Philippines Period
∆RM
∆GD
0.17082× ∆GD
0.2762× ∆GD
Column (3) as percentage of column (1)
Column (4) as percentage of column (1)
1974– 75
1.045 1.6905
0.2888
0.4669
27.64
44.68
1976– 80
1.966 0.4928
0.0841
0.1361
4.282
6.92
1981– 85
4.582 8.1778
1.3969
2.2587
30.49
49.295
1974– 85
2.9025 3.8945
0.6653
1.0757
22.92
37.06
Singapore Period
∆RM
∆GD
0.11281× ∆GD
0.1325× ∆GD
Column (3) as Column (4) as percentage of column percentage of column (1) (1)
1974– 75
209.0 929.50
104.857
123.16
50.17
58.93
1976– 80
428.60 1933.4
218.1069
256.18
50.89
59.77
1981– 85
520.80 3175.8
358.262
420.79
68.79
80.797
1974– 85
426.149 2352.4
265.37
311.69
62.272
73.14
Sri Lanka Period ∆RM ∆GD 0.090646×∆GD 0.4391×∆GD Column (3) as percentage of column (1)
Column (4) as percentage of column (1)
1976– 80
0.8854 2.0268
0.1837
0.8899
20.63
100.51
1981– 84
1.8970 4.1865
0.3795
1.8383
20.00
96.9
1976– 84
1.335 2.9867
0.2707
1.311
20.28
98.2
Thailand Period
∆RM
∆GD
1974–
3.035
3.578
0.12688× ∆GD
0.4540
0.2115× ∆GD
0.7567
Column (3) as percentage of column (1) 14.96
Column (4) as percentage of column (1) 24.93
Deficits and seigniorage
73
75 1976– 80
5.508
7.620
0.9668
1.6116
17.55
29.26
1980– 85
5.8520 23.659
3.0019
5.0040
51.297
85.5
1974– 85
5.2392 13.629
1.7292
2.8825
33.005
55.02
phenomenon, these calculations might suggest that budget deficits have been the major cause of inflation in these countries. We consider this issue in Chapter 5. Within-country over-time variations in these percentages are also revealing. Thus, in the case of India, there has been a continuous rise in the proportion of the increase in the monetary base attributable to current and past budget deficits, so much so that the figure exceeded 100 per cent in the early 1980s. Similar patterns can also be seen for South Korea, Malaysia, Singapore, and Thailand. An interesting aspect of the results for these countries is the sharp jump in the figures for the early 1980s. Pakistan and the Philippines show a different pattern in that there was no continuous rise in the percentage, the 1980s figure being almost at the same levels in the 1970s though still higher than the one for the mid-1970s. Finally, Sri Lanka showed a decline in the 1980s compared to the 1970s, though in absolute value the percentage was still very high, almost 98 per cent. The above results would seem to suggest a relatively strong support for the fiscal dominance hypothesis in the countries of our sample, with the possible exception of South Korea, and perhaps Indonesia for which we do not have complete results. But since, as King and Plosser (1985) have pointed out, these kind of results may be model specific, it would be useful to look at the sensitivity of these results to an alternate approach. To this aspect we now turn. The atheoretical approach King and Plosser (1985) have suggested an alternative strategy to test the fiscal dominance hypothesis. This approach starts by specifying the intertemporal consolidated government budget constraint. Using this constraint, they show that the implication of the fiscal dominance hypothesis is that if seigniorage and deficits are not contemporaneously correlated, then there must be a dynamic relationship between them. Their test strategy is twofold. Assuming that seigniorage is a function of certain policy variables, we first estimate regression equations between a measure of seigniorage and the contemporaneous values of the other policy variables. For our purpose, in the absence of data on tax rates, it was assumed that seigniorage depended on a measure of deficit and the rate of growth of real GDP. The two sets of equations were then specified as follows: Si=αi+βijDj+γig (8) where Si is the ith measure of seigniorage, Dj is the jth measure of the deficit and g is the rate of growth of real GDP. For the dynamic equation, the specification used was
Budget deficits and economic activity in Asia
74
(9) On the basis of the estimates of equation (9), three specific restrictions on the coefficient of the debt/deficit variable are tested: (i) H0: β1j=β2j=…=βnj=0 (ii) H0: Σβij=0 (iii) H0: Σβij|(1−Σθi)=0 The first restriction is straightforward and says that the coefficients of the lagged deficits are equal to zero. The next restriction is less severe than the first in that it merely requires their sum to be equal to zero. These two restrictions relate to the short term. The last restriction, on the other hand, relates to the long term and says that the sum of the longterm effects of all past deficits be equal to zero. For purposes of estimation of equations (8) and (9), it was decided to use both S1 and S2 as the dependent variable to ascertain the sensitivity of the results to alternate measures of seigniorage. In the spirit of the last section, deficit was represented by both its nominal as well as its real value. For the dynamic equation (9) it was assumed that n=3 in all cases. Considering first the contemporaneous correlations between seigniorage and the alternate measures of budget deficits, we can draw a number of conclusions from Table 3.6. (i) With the exception of South Korea, the results for the other countries are similar to those obtained by the structuralist approach, particularly when we look at the estimates using S1. This is not really surprising when we consider the fact that the dependent variable in both cases is similar. In other words, there is significant support for the fiscal dominance hypothesis. For South Korea, however, this hypothesis did not receive much support in the structuralist approach, but it does receive it here. (ii) The support for the hypothesis seems to be sensitive to the measure of the seigniorage measure used in some of the cases. For example, when the measure of budget deficit is nominal, deficits
Table 3.6 Contemporaneous correlations between seigniorage and deficit India Independent variables nd
rd
g
S1
S2
0.1817
—
0.3681
—
(0.28)
—
(0.46)
—
—
7.7801
—
7.3376
—
(21.07)
—
(4.82)
−0.0046
0.0043
−0.0045
0.0062
Deficits and seigniorage
Constant
75
(0.42)
(2.51)
(0.32)
(0.88)
1.3084
0.7308
0.2368
0.4606
—
—
(0.07)
(0.57)
−2
−0.08
0.95
−0.08
0.48
See
2.86
0.61
3.62
2.51
DW
1.16
1.91
1.88
2.67
R
Indonesia Independent variables nd
rd
g
Constant
S1
S2
0.1746
—
0.0830
—
(0.68)
—
(0.77)
—
—
—
—
—
—
—
—
—
−0.001
—
−0.001
—
(2.11)
—
(4.00)
—
2.1896
—
0.1866
—
(3.43)
—
(0.70)
—
−2
R
0.23
—
0.56
—
See
1.10
—
0.46
—
DW
1.82
—
2.02
—
South Korea Independent variables nd
S1
S2
−0.1062
—
0.3834
—
(0.36)
—
(1.87)
—
—
1.0385
—
0.5605
—
(3.78)
—
(2.38)
−0.0002
−0.0002
−0.0001
−0.0001
(1.95)
(2.47)
(1.64)
(0.69)
2.7073
2.3424
0.3642
0.5597
(5.63)
(7.23)
(1.09)
(2.02)
R−2
0.13
0.47
0.09
0.16
See
1.63
1.27
1.13
1.09
DW
1.38
1.62
0.75
1.38
rd
g
Constant
Budget deficits and economic activity in Asia
76
Malaysia Independent variables nd
rd
g
Constant
S1
S2
0.0952
—
0.2538
—
(0.48)
—
(1.83)
—
—
12.085
—
−4.4848
—
(6.64)
—
(1.72)
−0.0001
−0.0001
−0.0001
0.0001
(1.06)
(1.88)
(0.11)
(1.32)
4.2542
1.8044
−2.6838
0.6120
(2.88)
(3.23)
(2.57)
(0.72)
−2
R
0.11
0.78
0.24
0.22
See
2.36
1.17
1.65
1.67
DW
1.49
2.14
2.33
2.28
Pakistan Independent variables nd
S1
S2
−0.1740
—
−36.3210
—
(0.50)
—
(0.13)
—
—
9.0620
—
6.1775
—
(5.56)
—
(4.13)
−0.0122
0.0181
2.6166
19.7870
(0.34)
(0.77)
(0.09)
(0.91)
3.8670
0.9149
2.3070
0.7633
—
(1.16)
(1.29)
(1.05)
R−2
−0.08
0.55
−0.08
0.39
See
3.91
2.54
3.10
2.32
DW
1.38
1.90
1.57
2.06
rd
g
Constant
The Philippines Independent variables nd
rd
S1
S2
−0.0891
—
0.3018
—
(0.39)
—
(2.22)
—
—
11.085
—
5.1729
—
(10.46)
—
(4.12)
Deficits and seigniorage g
Constant
77
−0.0054
0.0062
−0.0076
—
(0.27)
(0.83)
(0.64)
—
1.3929
0.7412
0.4218
0.0125
(3.21)
(3.92)
(1.64)
(1.42)
−2
−0.06
0.81
0.11
0.38
See
1.57
0.66
0.93
0.78
DW
1.24
2.39
1.78
1.74
R
Singapore Independent variables nd
rd
g
Constant
S1
S2
−0.1209
—
−2.444
—
(0.32)
—
(1.78)
—
—
20.449
—
15.834
—
(2.52)
—
(0.49)
−0.0003
−0.0002
−0.0015
−0.0010
(1.19)
(0.80)
(2.56)
(1.70)
3.3514
1.0857
14.283
15.111
(3.47)
(0.88)
(4.71)
(3.47)
−2
−0.02
0.21
0.22
0.09
See
3.21
2.82
7.54
8.16
DW
1.36
1.42
2.02
2.20
R
Sri Lanka Independent variables nd
S1
S2
−0.0130
—
0.0620
—
(0.08)
—
(4.03)
—
—
9.2933
—
8.9770
—
(5.49)
(3.22)
—
0.0005
0.0384
−0.0141
−0.0041
(0.01)
(1.56)
(2.83)
(1.01)
1.7602
0.2884
−1.5637
0.9342
(1.50)
(0.57)
(1.35)
(1.14)
R−2
−0.09
0.54
0.37
0.26
See
2.52
1.64
2.48
2.70
rd
g
Constant
Budget deficits and economic activity in Asia DW
1.35
78
2.02
1.48
1.23
Thailand Independent variables
S1
nd
rd
g
Constant
S2
−0.3072
—
0.5610
—
(0.93)
—
(2.44)
—
—
28.603
—
15.789
—
(8.65)
—
(3.92)
−0.0024
−0.0005
−0.0048
0.0095
(0.25)
(0.14)
(0.74)
(2.13)
2.3623
0.3152
−0.3488
−0.4717
(2.97)
(0.85)
(0.63)
(1.04)
−2
0.0003
0.76
0.17
0.37
See
2.38
1.17
1.65
1.43
DW
1.18
0.88
1.09
1.00
R
Table 3.7 Dynamic estimates for S1 [(1) nd, (2) rd Independent variables nd−1, rd−1
India (1) 0.24257 (1.1445)
nd−2, rd−2
−0.0942 (−0.32041)
nd−3, rd−3
−0.10209
(2)
Independent Indonesia Independent variables variables (1)
11.995 rd−1 (2.4793) 4.8462 nd−2 (1.1019) −1.6549 nd−3
(−0.53783) (−0.41751) F(3, 10) g−1 g−2
0.7304
(1)
(2)
0.47476 nd−1, rd−1
−0.024126
−5.7848
(1.1266)
(−0.06916)
(−1.6718)
−0.33859 nd−2, rd−2
−0.64925
2.2169
(−0.8949)
(−2.1104)
(0.84668)
0.2884
(0.57475)
(0.92842)
(0.26989)
1.8542
0.97096
0.39073 nd−3, rd−3 — 0.4868 F(3, 11)
0.05773
−0.00981 g−1
−0.029699 g−1
0.064541
0.0797
(1.5102)
(−0.2443)
(−0.14906)
(0.86017)
(0.98676)
−0.09399
0.0873
(−1.2167)
(0.87695)
0.02692
0.11079
(0.51108)
(1.5599)
1.03697
1.00134
0.36444
0.3436
0.09146 (1.9974)
g−3
2.61046 F(3, 3)
South Korea
0.022396
0.04025 g−2
0.08916 g−2
(0.99844)
(0.71425)
−0.02134 g−3
−0.22764 g−3
(0.40029) (−0.45132) F(3, 10)
1.88067
1.13716 F(3, 3)
S1
0.19344
0.09919 S1
(−0.90629) 0.5238 F(3, 11) 0.752 S1
Deficits and seigniorage (0.55676) S−2
0.11953 (0.37029)
S−3
0.11596 (0.41287)
F(3, 10)
0.4729867
Constant
−0.11469 (−0.15285)
(0.28829)
79
(1.0274)
0.23036 S−2 (0.2824)
−0.59981 S−2 (−0.83496)
0.13584 S−3 (0.23195)
0.78307 S−3 (0.98281)
(1.2250)
(1.1227)
−0.010986
−0.4265
(−0.0332)
(−1.2127)
−0.01996
0.04808
(0.07224)
(0.17524
1.04408 F(3, 3)
(0.5358) F(3, 11)
0.5142
0.78877
0.45739 Constant
0.92663 Constant
1.8320
−0.36633
(0.91064)
(0.4754)
R
0.5398
0.6854 R
0.5864 R
0.4973
0.4015
DW
1.9958
2.1078 DW
2.5955 DW
1.6977
2.1367
1.209
0.820
F(9, 10)
Independent variables nd−1, rd−1
1.303
2.420 F(9, 3)
Malaysia (1) −0.09755 (−0.97565)
nd−2, rd−2
2
(1.3887) (−0.29817)
2
0.17535
Independent variables
(2)
8.1314 nd−1, rd−1 (0.50533) −4.3502 nd−2, rd−2
(1.1012) (−0.39002) nd−3, rd−3
−0.27402 (−2.1552)
F(3, 3) g−1
3.1936 0.12105 (1.1626)
g−2
0.1348 (0.71501)
g−3
0.05647 (0.3987)
F(3, 3) S1
S−2
0.5443
−3.7355 nd−3, rd−3 (−0.2796) — F(3, 11) 0.20373 g−1 (0.86001) 0.19074 g−2 (0.53419) 0.20945 g−3 (0.61900) — F(3, 11)
−0.82891
−0.04557 S1
(−1.7005)
(−0.0742)
−0.25417
−0.4385 S−2
(−0.575) (−0.45482) S−3
−0.41814
−0.46587 S−3
2
0.473 F(9, 11)
Pakistan (1)
(2)
The Philippines (1)
(2)
−0.11249
−0.88182
−0.2246
10.930
(−1.2117)
(−0.6667)
(−1.1883)
(1.0638)
0.15227
−2.5995
0.46157
1.0209
(1.4330)
(−2.0675)
(1.7932)
(0.09881)
0.13945
−0.86782
0.05446
−2.7432
(1.1361) (−0.70636)
(0.2454)
(−0.4479)
2.4503113
1.431057
1.3986
0.7466
−0.06788
−0.0636
0.03283
−0.01779
(−1.0553) (−0.87878)
(1.2460)
(−0.8948)
0.057114
−0.07136
0.07125
−0.00708
(0.96310) (−0.93653)
(1.3527)
(−0.2984)
0.000532
0.00429
(1.0123) (−0.30371) (0.013411)
(0.17815)
0.064109
−0.0236
0.89596
0.490676
0.8257
0.3602
0.09367
0.34601
−0.14716
−0.278
(0.31658)
(0.38470)
−0.23873
0.13935
−0.39131
0.01488
(−1.0138)
(0.33694)
(−1.1680)
(0.0424)
−0.17076
−0.09311
−0.1125
−0.4733
(−0.4576) (−0.92174)
Budget deficits and economic activity in Asia (−1.1411)
(−0.6476)
80
(−0.76270)
(0.24909)
(−0.2933)
(−0.3262)
F(3, 3)
0.9998
— F(3, 11)
1.098746
0.4889
0.5616
1.0326
Constant
3.2667
−1.5656 Constant
1.2130
2.3456
0.78235
1.6292
(1.2199)
(2.8348)
(1.0485)
(2.6315
(1.5736) (−0.33119) 2
2
R
0.8060
0.4187 R
0.5175
0.4211
0.4513
0.3702
DW
2.8526
2.0665 DW
2.3658
2.2175
1.9719
1.8049
1.311
0.889
1.005
0.718
F(9, 3)
1.385
Independent variables
Singapore (1)
nd−1, rd−1
0.4681
3.006
0.18321
(2.1131)
(0.3724)
(2.3167)
0.4391
12.938
0.0220
(1.853)
(1.9552)
(0.25319)
−0.8583
−4.3822
−0.0645
(−3.748)
(−0.7844)
(−0.7263)
4.7129
1.7035
3.25129
−0.0174 −0.003337
−0.00621
nd−2, rd−2 nd−3, rd−3 F(3, 10)/F(3, 11) g−1 g−2 g−3
0.240 F(9, 11)
Sri Lanka
(2)
(−0.2592)
(−0.0426)
0.1220
0.00632
(2.1597)
(0.07599)
0.0883
0.1191
(1)
(2)
Independent variables
0.37671 nd−1, rd−1 (0.11276) 3.4552 nd−2, rd−2 (0.89953) −11.597 nd−3, rd−3 (−3.8397) 6.14254 F(3, 12) −0.02389 g−1
(−0.0756) (−0.30593) −0.00335
−0.0366 g−2
3.7194
(−0.30578)
(1.1446)
0.001882
−0.342
(0.02257) (−0.10243) 0.00243
2.135
(0.03824)
(0.75511)
0.400
0.74276
−0.05388 −0.0293 (−1.0598) (−0.70276)
(0.5723)
(0.74049
−0.03298
0.00628
−0.01049
0.0971 g−3
F(3, 10)/F(3, 11)
3.08169
0.9985
0.06195
0.84008 F(3, 12)
S1
0.09116
0.08991
−0.11022
F(3, 10)/F(3, 11)
−0.02207
(−0.0405) (−0.46220)
(1.4661)
S−3
(2)
0.03087
(1.629) (−0.42135)
S−2
(1)
0.03025
(1.6674)
(0.40556)
Thailand
(0.2648) (−0.54567)
−0.547
0.31244
−0.21569
(−2.0332)
(0.8791)
(−1.1116)
−0.3173
0.37810
−0.40290
(−1.1358)
(1.1172)
(−2.090)
2.2934
0.68388
1.77669
−0.0379 S1 (−0.2256) 0.0760 S−2 (0.46218) −0.2855 S−3 (−1.8273) 1.27278 F(3, 12)
(0.1343) (−6.24225) 0.5702
0.30735
0.67521
0.498
(2.3010)
(1.7080)
−0.16635
−0.0645
(−0.4587) (−0.19977 −0.18299
0.00316
(−0.60024)
(0.01087)
2.3503
1.14996
Deficits and seigniorage Constant
2.2152
81
−1.8333
0.84747
1.7769 Constant
0.85189
0.4732
(1.8025) (−0.94806)
(0.7828)
(1.899)
(0.9730)
(0.85779
2
R
0.7125
0.5551
0.5196
0.6612 R
0.3906
0.4809
DW
1.9075
1.9294
1.3385
2.3772 DW
1.9192
2.0051
2.878
1.386
1.322
0.855
1.23
F(9, 10)/F(9, 11)
2
2.385 F(9, 11)
have a positive and significant effect on seigniorage in the cases of South Korea, Malaysia, the Philippines, Sri Lanka, and Thailand when S2 rather than S1 is used as the dependent variable. A different result is obtained for Singapore where the coefficient turns out to be negative. (iii) On the whole, the results with the real deficit and S1 are more clear cut. In fact, with the exception of Indonesia where the data on the stock of government debt are not available, the fiscal dominance hypothesis is supported in all cases. In the case of Indonesia, just as in the structuralist case, there is no support for the hypothesis at all. (iv) What is also remarkable about these results is that in all cases, if we take the nominal deficit and S1 as the independent and the dependent variables respectively, there is no support for the fiscal dominance hypothesis whatever. This is in sharp contrast with the results of the structuralist estimates. But, as already pointed out, lack of contemporaneous correlations is not necessarily inconsistent with the validity of the fiscal dominance hypothesis, because it may simply mean that the timing of monetization has been shifted to future years. In order to find out whether such was indeed the case, we turn to the dynamic estimates. The dynamic estimates are given in Table 3.7 for S1 and in Table 3.8 for S2. As for the contemporaneous correlations, these estimates also use both the nominal and the real versions of the budget deficit. Instead of commenting on these tables in detail, we concentrate rather on their two aspects. First, we consider the three null hypotheses described above. Second, we consider the quantitative magnitudes of the coefficients and their statistical significance implied by the second and the third null hypotheses. The first null hypothesis is the relevant one for testing whether the fiscal dominance hypothesis holds. Results for the F statistic for testing the null that the hypothesis does not hold in the dynamic sense, namely that the coefficients of the three lagged deficit terms are jointly equal to zero, as given in columns (1) and (4) of Table 3.9 for S1 and in Table 3.10 for S2. Column (1) in both tables refers to the case where the deficit measure is nominal and column (4) when it is in real terms. The results of Table 3.9 for S1 reject the null only for Singapore in terms of the nominal deficit and only for Sri Lanka when the real deficit measure is used. In terms of Table 3.10 for S2. the null is not
Budget deficits and economic activity in Asia
82
Table 3.8 Dynamic estimates for S2 [(1) nd, (2) rd] India Independent variables nd−1, rd−1 nd−2, rd−2 nd−3, rd−3 F(3, 10)/F(3, 3) g−1 g−2 g−3 F(3, 10)/F(3, 3) S2−1 S2−2 S2−3 F(3, 10)/F(3, 3) Constant
(1)
Indonesia (2)
(1)
(2)
0.35318
32.720
−0.06856
—
(0.3676)
(1.4965)
(−0.21447)
—
1.1140
26.722
−0.13302
—
(0.87675)
(1.6073)
(−0.40313)
—
−0.50643
38.503
−0.03598
—
(−0.63865)
(2.5046)
(−0.17865)
—
1.3168
3.4676
0.08634
—
−0.30867
−0.4843
0.08606
—
(−1.6116)
(−2.5929)
(0.41233)
—
0.078553
−0.24618
0.092346
—
(0.40439)
(−1.3326)
(1.0363)
—
−0.12433
−0.48159
0.03271
—
(−0.52732)
(−2.3021)
(0.19485)
—
1.4219
2.5950
0.45677
—
−0.37295
−0.40091
0.06275
—
(−1.2585)
(−1.7681)
(0.10198)
—
−0.35851
−3.5155
0.71714
—
(−1.0517)
(−1.3953)
(0.4833)
—
−0.30331
−0.13342
0.41988
—
(−0.97704)
(−0.56412)
(0.3119)
—
0.7728
1.3791
0.80094
—
−1.0637
4.8164
−1.5008
—
(−0.31125)
(2.6019)
(−0.4476)
—
2
R
0.4808
2.019
0.7209
—
DW
1.9243
1.6527
2.1150
—
1.029
2.019
0.861
—
F(9, 10)/F(9, 3)
Deficits and seigniorage
83
South Korea Independent variables nd−1, rd−1 nd−2, rd−2 nd−3, rd−3
(1)
g−2 g−3
(2)
(1)
−1.6419
0.40845
−12.323
(−0.2465)
(−0.747)
(1.2368)
(−0.3328)
0.071882
3.2230
−0.17602
−8.8582
(0.2419)
(2.121)
(−0.4282)
(−0.3634)
0.25584
0.7594
0.20655
−36.404
(0.7886)
(0.5216)
(0.54369)
(−0.8584)
0.3194
1.7492
0.7260
0.6918
−0.02315
−0.04309
0.30552
0.37132
(−0.4581)
(−0.987)
(1.3758)
(1.4933)
0.03992
0.05173
0.28959
0.4312
(0.7939)
(0.9147)
(0.76802)
(0.6342)
0.03903
0.01541
−0.10872
−0.6678
(0.7914)
(0.3359)
(−0.2437)
(−1.763)
South Korea Independent variables F(3, 11)/F(3, 3)
(2)
−0.076065
F(3, 11)/F(3, 3) g−1
Malaysia
(1)
Malaysia
(2)
(1)
(2)
0.4999
0.6283
1.3168
1.9577
0.09312
−0.04326
−0.60263
−0.1371
(0.2332)
(−0.14172)
(−0.88238)
(−0.26079)
0.13197
−0.03335
0.10664
0.9405
(0.3322)
(−0.1246)
(0.1741)
(1.4398)
−0.16751
−0.000825
0.10199
0.1835
(−0.4312)
(−0.003226)
(0.25057)
(0.39898)
F(3, 11)/F(3, 3)
0.10637
0.1448
0.5609
0.80099
Constant
−0.5375
0.08418
−7.8279
1.6655
S2−1 S2−2 S2−3
(−0.5433)
(0.09612)
(−0.75638)
(0.3121)
2
R
0.2596
0.4551
0.6972
0.6911
DW
1.9445
1.7686
1.6859
2.0975
0.429
1.021
0.768
0.746
F(9, 11)/F(9, 3)
Budget deficits and economic activity in Asia Pakistan Independent variables nd−1, rd−1
(1)
84 The Philippines
(2)
(1)
(2)
−177.87
−2758.1
0.13684
13.842
(−0.88724)
(−1.1509)
(0.5458)
(1.2094)
191.79
−2820.5
0.4265
−3.4203
(0.87337)
(−1.2027)
(1.7281)
(−0.3223)
445.38
292.33
−0.06272
−9.3305
(1.9421)
(0.1326)
(−0.2864)
(−1.389)
2.585
0.7814
2.1605
1.3524
−153.98
−43.371
0.06799
0.0554
(−1.0691)
(−0.30117)
(2.3455)
(2.229)
188.94
30.820
0.01724
0.0322
(1.5548)
(0.20337)
(0.3209)
(1.0287)
82.338
94.947
−0.0511
−0.01163
(0.63565)
(−0.584)
(−1.2824)
(−0.40168)
1.179
0.1828
3.3404
2.1851
−0.48286
−0.16724
0.2587
0.4484
(−1.8040)
(−0.5377)
(0.60237)
(1.5113)
0.10774
0.30281
−0.28813
0.6496
(0.4241)
(0.93159)
(−0.7461)
(1.5956)
−0.71468
−0.37354
−0.1462
−0.005025
(−2.6370)
(−1.1698)
(−0.3891)
(−0.01395)
F(3, 11)
2.9325
0.8228
0.3522
2.3982
Constant
−74.923
2989.9
0.00196
−0.3249
nd−2, rd−2 nd−3, rd−3 F(3, 11) g−1 g–2 g−3 F(3, 11) S2−1 S2−2 S2−3
(−0.04249)
(1.7001)
(0.003009)
(−0.68751)
2
R
0.5376
0.3501
0.5974
0.5325
DW
2.3184
2.0711
2.187
1.8076
1.421
0.658
1.813
1.392
F(9, 11)
Deficits and seigniorage
85
Sri Lanka Independent variables nd−1, rd−1
(1)
Thailand (2)
(1)
(2)
557.74
−2013.3
0.2538
20.703
(3.0056)
(0.15498)
(0.88529)
(2.4644)
228.92
−2086.1
0.415
−35.366
(1.274)
(−0.1399)
(1.6862)
(−3.8963)
−903.59
−9114.3
0.475
10.67
(−4.235)
(−0.7687)
(2.0497)
(1.2284)
7.9107
0.3929
2.0327
6.1735
−48.034
−4.3714
−0.08511
0.12685
(−0.3017)
(−0.0139)
(−0.59596)
(1.0718)
−134.85
−76.926
0.51101
0.52977
(−0.8671)
(−0.2514)
(3.3257)
(5.2428)
−226.90
−121.87
−0.04779
−0.24979
(−1.5144)
(−0.4656)
(−0.2460)
(−1.412)
0.8733
0.0987
4.3668
9.318
0.11163
0.3287
0.60661
0.86421
(0.3981)
(−1.0121)
(1.8326)
(3.2159)
−0.1988
−0.2376
0.21461
−0.2474
(−0.7922)
(−0.6567)
(0.7783)
(−0.99378)
0.5922
0.1365
−0.32379
0.3786
(1.9926)
(−0.40664)
(−1.1303)
(2.0473)
F(3, 11)/F(3, 12)
1.8853
0.3805
2.3429
10.9158
Constant
3428.6
3192.8
−3.0807
−2.6632
nd−2, rd−2 nd−3, rd−3 F(3, 11) g−1 g−2 g−3 F(3, 11)/F(3, 12) S2−1 S2−2 S2−3
(1.764)
(0.99032)
(−0.99674)
(−1.8676)
2
R
0.7407
0.2597
0.7515
0.8526
DW
1.5299
1.6781
2.2661
1.7073
3.486
0.429
4.031
7.714
F(9, 11)/F(9, 12)
rejected for any of the countries reported. In other words, at the very most, the dynamic version of the test supports the fiscal dominance hypothesis only for two countries. Combined with the static version of the test in Table 3.6, what can we conclude? Concentrating on the results for S1, first we notice that while there was no support for the hypothesis for Singapore in the static test, when nominal deficit is used there is support
Budget deficits and economic activity in Asia
86
when we consider the dynamic version. This is an interesting result in that it points out the dangers of rejecting the fiscal dominance hypothesis just on the basis of the lack of contemporaneous relationship. Second, the results for Sri Lanka are interesting in that they show that in this case both static and the dynamic links between seigniorage and deficits are important. Finally, the strong contemporaneous linkages
Table 3.9 Test statistics for the null hypotheses for S1 nd Country
(1)
(2)
(3)
India
F(3, 10)=0.73
(t/10)=0.29
(t/10)=0.319
Indonesia
F(3, 3)=0.487
(t/3)=0.520
(t/3)=0.0655
South Korea
F(3, 11)=1.854
(t/11)=0.942
(t/11)=0.818
Malaysia
F(3, 3)=3.194
(t/3)=2.554*
(t/3)=2.470*
Pakistan
F(3, 11)=2.45
(t/11)=1.617
(t/11)=1.63
The Philippines
F(3, 11)=1.398
(t/11)=1.562
(t/11)=1.35
Singapore
F(3, 10)=4.713**
(t/10)=0.262
(t/10)=0.2614
Sri Lanka
F(3, 11)=3.251
(t/11)=2.34
(t/11)=2.658**
Thailand
F(3, 12)=0.040
(t/12)=0.201
(t/12)=0.197
(1) Ho: nd−1=nd−2=nd−3=0 (2) Ho: nd−1+nd−2+nd−3=0 (3) Ho: (nd−1+nd−2+nd−3)/[1−(S1−1+S1−2+S1−3)]=0 rd Country
(4)
(5)
(6)
India
F(3, 10)=2.61
(t/10)=1.77
(t/10)=1.79
Indonesia
—
—
—
South Korea
F(3, 11)=0.971
(t/11)=−0.905
(t/11)=−0.867
Malaysia
F(3, 3)=0.399
(t/3)=0.0017
(t/3)=0.00175
Pakistan
F(3, 11)=1.431
(t/10)=1.699
(t/10)=0.868
The Philippines
F(3, 11)=0.747
(t/11)=0.720
(t/11)=0.726
Singapore
F(3, 10)=1.7035
(t/10)=0.952
(t/10)=0.2907
Sri Lanka
F(3, 11)=6.143**
(t/11)=−2.499**
(t/11)=−2.534**
Thailand
F(3, 12)=0.743
(t/12)=1.45
(t/12)=1.008
(4) Ho: rd−1=rd−2=rd−3=0
Deficits and seigniorage
87
(5) Ho: rd−1+rd−2+rd−3=0 (6) Ho: (rd−1+rd−2+rd−3)/[1−(S1−1+S1−2+S1−3)]=0
between S1 and real deficits and the absence of dynamic linkages imply that there is a strong tendency to monetize budget deficits in the very short-run in the countries of our sample. These findings are consistent with those of the structuralist approach when we recall the high percentage of the deficit monetized in the short run implied by the estimates. These results are also consistent with the descriptive data reported in Chapter 1. In order to shed some further light on the issue, we now consider the second aspect of the estimates in Tables 3.7 and 3.8 referred to above, the null hypotheses two and three. The test-statistics for the second hypothesis are given in columns (2) and (5) and for the third
Table 3.10 Test statistics for the null hypotheses for S2 nd Country
(1)
(2)
(3)
India
F(3, 10)=1.3168
(t/10)=1.3615
(t/10)=1.6046
Indonesia
F(3, 3)=0.08634
(t/3)=0.4639
(t/3)=0.07923
South Korea
F(3, 11)=0.3194
(t/11)=0.7002
(t/11)=0.7804
Malaysia
F(3, 3)=0.7260
(t/3)=0.9535
(t/3)=1.9387
Pakistan
F(3, 11)=2.585
(t/11)=1.961
(t/11)=2.0796
The Philippines
F(3, 11)=2.1605
(t/11)=1.1719
(t/11)=2.242
Singapore
—
—
—
Sri Lanka
F(3, 11)=7.9107
(t/11)=−0.4835
(t/11)=0.314
Thailand
F(3, 12)=2.0327
(t/12)=0.6751
(t/12)=1.09548
(1) Ho: nd−1=nd−2=nd−3=0 (2) Ho: nd−1+nd−2+nd−3=0 (3) Ho: (nd−1+nd−2+nd−3)/[1−(S2−1+S2−2+S2−3)]=0 rd Country
(4)
(5)
(6)
India
F(3, 10)=3.4676
(t/10)=2.813
(t/10)=2.3601
Indonesia
—
—
—
South Korea
F(3, 11)=1.7492
(t/11)=−0.9684
(t/11)=−0.9872
Malaysia
F(3, 3)=0.6918
(t/3)=−0.8649
(t/3)=−0.0171
Budget deficits and economic activity in Asia
88
Pakistan
F(3, 11)=0.7814
(t/11)=−1.068
(t/10)=−0.9019
The Philippines
F(3, 11)=1.3524
(t/11)=0.07261
(t/11)=−0.06865
Singapore
—
—
—
Sri Lanka
F(3, 11)=0.3929
(t/11)=−0.7832
(t/11)=−0.7698
Thailand
F(3, 12)=6.1735
(t/12)=−0.4267
(t/12)=−0.02584
(4) Ho: rd−1,=rd−2=rd−3=0 (5) Ho: rd−1+rd−2+rd−3=0 (6) Ho: (rd−1+rd−2+rd−3)/[1−(S2−1+S2−2+S2−3)]=0
hypothesis in columns (3) and (6) of Tables 3.9 and 3.10. Once again, we concentrate on the estimates for S1. From column (2), the second null is rejected from Malaysia and Sri Lanka for the nominal deficit case and for Sri Lanka only for the real deficit case. It is the same story for the third null hypothesis. These findings say that the short-run and the long-run total effect of past deficits for the creation of money is significant only in the case of Sri Lanka and Malaysia. The total effects, both the short-term and the long-term, of the past deficits, both nominal and real, on S1 and S2 are summarized in Tables 3.11 and 3.12 respectively. The negative dynamic effects of
Table 3.11 Total short-term and long-term effects of past deficits on seigniorage S1 Sum of deficit coefficients Country India
Indonesia
South Korea
Malaysia
Pakistan
The Philippines
Singapore
nd
Long-term effect
rd
nd
rd
0.0462
15.186
0.0808
28.407
(0.29)
(1.77)
(0.32)
(1.79)
—
—
—
—
—
—
—
—
−0.3850
−2.9932
−0.6144
−2.8923
(0.94)
(0.91)
(0.82)
(0.87)
−0.1962
0.0456
−0.0785
−0.0234
(2.55)
(0.01)
(2.47)
(0.01)
0.1792
−4.3492
0.1362
−7.1561
(1.62)
(1.70)
(1.63)
(0.86)
2.2915
9.2080
0.1766
5.3029
(1.56)
(0.72)
(1.35)
(0.73)
—
—
—
—
Deficits and seigniorage
Sri Lanka
Thailand
89
—
—
—
—
0.1407
−7.7677
0.0139
−6.2268
(2.34)
(2.50)
(2.66)
(2.53)
−0.0178
5.1523
−0.0263
9.7970
(0.20)
(1.45)
(0.20)
(1.01)
Note t values are given in parentheses
past deficits in the cases of Malaysia and Sri Lanka are curious when judged against the very strong and positive contemporaneous effects. In fact, it should be noted that the combined negative effect is outweighed by the positive contemporaneous effect.
SOME FURTHER DISCUSSION One of the most important ways in which budged deficits are alleged to affect the economy is through their effect on money creation, or, more properly, on the creation of reserve or high-powered money. There are alternate rationale for this linkage and alternate ways of testing for the existence of such a linkage. In this chapter, we have paid particular attention to the rationale offered by Sargent and Wallace (1981), using testing strategies used by the structural approach and the one suggested by King and Plosser (1985). Our results, from both approaches, provide strong support for the fiscal dominance hypothesis. In particular, our findings suggest that contemporary deficits are rapidly monetized, thus leading to
Table 3.12 Total short-term and long-term effects of past deficits on seigniorage S2 Sum of deficit coefficients Country India
Indonesia
South Korea
Malaysia
Pakistan
nd
Long-term effect
rd
nd
rd
0.9608
97.946
0.2670
−4390.1722
(1.36)
(2.82)
(1.60)
(2.36)
—
—
—
—
—
—
—
—
0.2517
2.3405
0.2670
2.1722
(0.70)
(0.97)
(0.78)
(0.99)
0.4390
−57.586
0.3149
−4390.8
(0.95)
(0.86)
(1.94)
(0.01)
459.29
−5286.3
219.78
−4270.1
Budget deficits and economic activity in Asia
The Philippines
Singapore
Sri Lanka
Thailand
90
(1.96)
(1.07)
(2.08)
(0.90)
0.5006
1.0911
0.4258
−11.740
(1.72)
(0.07)
(2.24)
(0.07)
—
—
—
—
—
—
—
—
−116.94
−9.1872
−236.26
−11.896
(0.48)
(0.78)
(0.31)
(0.76)
0.3135
−3.9930
0.6237
−869.81
(0.68)
(0.43)
(1.09)
(0.02)
Note t values are given in parentheses
increases in money creation. The dynamic effects of past deficits on money creation, on the other hand, are not that pervasive. This finding, given the sizeable magnitudes of the budget deficits in the countries of our sample, as discussed in Chapter 1, is intriguing. But it may well be due to the fact that since current deficits are so substantial, the amount of monetization implied by them alone will be more than what the economy can bear, so that the question of the monetization of past deficits simply does not arise. In any event, the findings from both approaches do suggest that there is considerable support for the fiscal dominance hypothesis; although at the same time, monetary authority does respond to other factors too. These findings have a bearing on the ability of the monetary authority in these countries to pursue an independent monetary policy. Given the strong support for the fiscal dominance hypothesis, it would be difficult not to conclude that this independence is relatively weak in most of the countries of our sample. Our findings do not suggest that future deficits will also need to be monetized to the same extent. This will clearly depend on the future fiscal policies pursued. What the findings do suggest is that there has been a strong linkage between deficits and seigniorage in the period covered and to that extent deficits may have had an effect on money growth, inflation, and interest rates. To these effects we turn in the next three chapters.
4 Deficits and money supply The objective of this chapter is to examine the effects of deficits on money supply. This involves two steps. The first, as seen in the previous chapter, is the relationship between the monetary base or high-powered money and the budget deficit, and the second is the relationship between money supply and the monetary base. A rigorous approach to the question under discussion would be to specify the determinants of the monetary base, as in Chapter 3, and then the determinants of the relationship between money supply and the monetary base. This latter, as we know, requires the speficiation of the determinants of money multipliers. Since the money multiplies, depending on the definition of money supply used, consist of the ratio of currency to demand deposits and that of bank reserves to demand deposits in the case of M1, for example, it follows that the determinants of the multiplier implies specifying the determinants of these ratios. Since these ratios are the outcome of the portfolio behaviour of the non-bank public and the banks, we need to specify models determining such behaviour. Once we have done that, we can combine the analysis of the previous chapter and these models, and examine the relationship between budget deficits and money supply. This is a huge task beyond the scope of this study. Our approach, instead, is to estimate alternate reduced-form equations detailing the relationship between money supply and budget deficits. This is also the approach which has been commonly used in the literature.1
THE MONEY MULTIPLIERS Assuming a stable relationship between the monetary base and the budget deficit, if the money multipliers are stable, we can derive a stable relationship between money supply and the budget deficit. Before proceeding to the estimation of such a relationship, therefore, we first consider the behaviour of money multipliers in the countries of our sample. Two types of multipliers are considered for this study: those relative to the narrow definition of money, M1, as given in row 34 of the International Financial Statistics, and the broader definition of money, M3, as given in row 35i of the same publication. Assuming that the multipliers can be approximated by the ratios of M1 and M3 to the monetary base, the two multipliers, denoted by k1 and k3, are plotted in Figure 4.1. Their average values for three different time periods are given in Table 4.1. As before, all data are annual.
Budget deficits and economic activity in Asia
92
Figure 4.1 Money multipliers for M1 and M3
Deficits and money supply
93
Consider the average values in Table 4.1 first. It is immediately clear that the values of k3 are larger than those of k1 for every country for each time period. This is not surprising when we note that the numerator for k3 is larger than that for k1 and the denominator is the same for both. Further, it can also be seen that neither of the two multipliers are constant over the time period covered, although the direction or the magnitude of the change are not necessarily the same for all countries. Thus, concentrating on k1 first, its average value increased for India, Indonesia, and Pakistan, whereas it declined for South Korea, the Philippines, Singapore, Sri Lanka, and Thailand. For Malaysia, it declined in the 1970s, but rose slightly in the 1980s, though still remaining lower than in the 1960s. The behaviour of k3 is somewhat more consistent across the countries, though once again k3 did not remain constant. It consistently increased in value for all of the countries in the sample.
Table 4.1 Money multiplier values k1 Country
1961–70
1971–80
k3 1981–86
1961–70
1971–80
1981–86
India
1.389
1.478
1.610
1.969
2.700
3.065
Indonesia
1.178*
1.269
1.548
1.362
1.897
3.000
South Korea
1.260
1.115
1.168
2.260
3.175
5.929
Malaysia
1.518
1.473
1.489
2.586
3.450
4.866
Pakistan
1.420
1.687
1.707
1.948
2.460
2.640
The Philippines
1.436
1.404
1.038
2.754
2.941
3.273
Singapore
1.771*
1.565
1.377
3.435
3.644
4.094
Sri Lanka
1.390
1.348
1.164
1.962
2.360
3.084
Thailand
1.333
1.256
1.125
2.213
3.630
6.141
Note * Indonesia 1966–70; Singapore 1963–70
But it is interesting to note that the extent of the variability is more marked for k3 than for k1 both over time for the same country and also across the countries. Just consider a few extreme cases. For India, k1 increased from 1.389 in 1961–70 to an average of 1.610 over 1981–86, whereas the value of k3 nearly tripled over the same period. This lack of constancy of the two multipliers and the greater variability of k3 can be appreciated better by considering Figure 4.1. This figure makes it quite clear that there has been considerable variability about the trend for both multipliers, but much more so for k3. Further, it is also clear that this variability is more marked for some countries than for others; for example, for South Korea, the Philippines, Singapore, and Sri Lanka. What this evidence suggests is that neither of the two multipliers has been constant over the period. Given this, we may speculate that even if the relationship between the
Budget deficits and economic activity in Asia
94
monetary base and the budget deficit was a stable one over the period covered for the countries in our sample, this may not turn out to be the case for the effect of the budget deficit on money supply. To the quantification of this surmise we now turn.
MONEY SUPPLY AND BUDGET DEFICITS For purposes of the quantification of the effects of budget deficits on the growth of money supply, we proceed as follows. We start with the conventional view that the basic relationship is between money supply and the monetary base via the money multiplier. Thus, we can write M=kMB (1) where M is some definition of money stock, k the corresponding money multiplier, and MB the monetary base. With k constant, the relationship between M and MB is straightforward. But, as we have already seen, k is far from constant for both definitions of the money supply used in this study. Consequently, we may write M=k(X)MB (2) where X is a vector of the determinants of k. In the last chapter, we examined the determinants of the monetary base. Following from that we can write (2) as M=k(X)MB(DEF…) (3) where DEF is some measure of budget deficit. As pointed out in the last section, no attempt is made here to estimate k. Instead, it is assumed that the budget deficits affect growth of money supply through their effect on the monetary base and other channels, and we simply estimate reduced form equations. We estimate three reduced form equations. The first one is a conventional equation which can easily be rationalized from standard macro models (see, for example, Dornbusch and Fischer (1987)). Its main advantage is that it can be used to compare our results with some other studies for the developing countries. The two other versions are in the spirit of the previous chapter. Of these two, the first one measures the nature of the contemporaneous relationship between budget deficits and money supply growth and the second one estimates the dynamic relationship between the same variables. The motivation for these two versions comes from Sargent and Wallace (1981), as was the case in the previous chapter. However, we must be careful about a mechanical application of their argument in this case. Recall that their argument, based on an intertemporal government budget constraint, applies to changes in the monetary base and not to changes in money supply, unless, of course, the two were to coincide. Since that is not necessarily the case, their results may not hold for money supply. Part of the problem relates to how the marginal reserve requirements are defined in different countries. Sometimes they are defined in terms of last period’s deposits, sometimes contemporaneous, or a mix of the two.2 This can affect the timing of the effect of changes in the monetary base as a result of budget deficit and the subsequent
Deficits and money supply
95
changes in money supply. However, we can say that the time path of the monetary base in response to budget deficits or changes in government borrowing requirements may indirectly have something to do with the time path of budget deficits and money supply. Since the relationship between budget deficits and money supply essentially involves two types of time profiles, there is no way to tell a priori how the results will be affected between budget deficit and money supply. Only empirical estimates can shed some light on this issue. The data used are, again, annual, which may well affect our estimates. To judge the sensitivity of our estimates to alternate definitions of money supply, both M1 and M3 are used. In order to conserve space, we report detailed results of the conventional model, but only summarize the results for the static and the dynamic versions. We turn to the estimates of the conventional version first. The general form of the equation estimated was ∆logMt=a+Σbi ∆logMt−i+clogYt+dNGRt+eDEFt (4) where Y refers to real GDP, NGR to government expenditure as a proportion of GDP, and DEF to budget deficit as a proportion of money supply only. Different versions of this model were estimated, particularly in terms of the number of lags on ∆logMt−i. However, the intention was not to seek a ‘best’ fit in any particular sense. The results are reported in Table 4.2. Deficit is nominal and given by the NIPA version.
Table 4.2 Money growth estimates (M1): the conventional model Variables Constant
∆logM−1 ∆logM−2 LogY
NGR
DEF
India
Indonesia
South Korea
Malaysia
Pakistan
0.0080
0.0105
0.1174
0.0505
0.1181
(0.08)
(0.20)
(0.88)
(0.80)
(1.18)
−0.0461
0.4318
0.3393
0.0754
0.1935
(0.18)
(2.28)
(1.10)
(0.25)
(0.9)
−0.5799
—
−0.2994
—
−0.2497
(2.38)
—
(1.12)
—
(1.21)
−0.4738
2.3628
0.3227
1.3160
1.0219
(1.13)
(3.48)
(0.50)
(1.98)
(2.08)
2.6327
—
0.3700
—
−0.3767
(1.57)
—
(0.31)
—
(0.58)
2.4708
2.1949
−1.3689
0.2164
−0.5541
(1.19)
(1.32)
(0.44)
(0.50)
(0.77)
−2
R
0.057
0.664
0.012
0.264
0.018
See
0.061
0.063
1.949
0.069
0.065
Budget deficits and economic activity in Asia DW Variables Constant
1.96 The Philippines
1.72
0.111
Singapore
96 2.22
Sri Lanka
Taiwan
1.88 Thailand
0.1259
0.2650
−0.0779
−0.2674
−0.0138
(1.02)
(1.64)
(0.67)
(1.03)
(0.06)
−0.0739
−0.1929
0.3060
−0.0600
0.2452
(0.33)
(0.98)
(1.29)
(0.30)
(1.06)
∆logM−2
—
—
—
—
—
LogY
−0.4101
0.5908
0.0297
1.3383
1.2288
(2.07)
(1.92)
(0.04)
(2.06)
(1.42)
0.2796
−1.0398
0.5596
1.4814
−0.1723
(0.29)
(1.39)
(1.28)
(1.40)
(0.12)
1.4588
1.3279
0.1717
2.2854
−1.0106
∆logM−1
NGR
DEF
(1.21)
(1.39)
(0.30)
(1.07)
(0.58)
−2
R
0.672
0.243
0.047
0.088
0.166
See
0.083
0.063
0.083
0.088
0.068
DW
2.00
1.69
1.74
1.90
1.87
Note t values are given in parentheses
Judged by the conventional criteria, these results are not very favourable to the hypothesis that budget deficits are an important determinant of growth in money supply as measured by M1. Of the ten countries reported, the coefficient of the budget deficit is not significant at the 5 per cent level for any of the countries. Further, it has the wrong sign in four cases. Taking a more liberal interpretation, deficits exercise a marginally positive effect only in the cases of India, Indonesia, the Philippines, Singapore, and Taiwan. One reason for these relatively unfavourable results may be that the conventional approach does not adequately capture the effect as postulated by Sargent and Wallace. It may well be that the King and Plosser methodology as used in the last chapter may provide a better test of the hypothesis. Therefore, we now consider the estimates based on our alternate models. These estimates are provided for both M1 and M3 and the models estimated are similar to those in the last chapter. More specifically, the dynamic version is given by GM=a+ΣbiGMt−i+Σcigt−i+ΣdiDEFt−i (5) where GM is the rate of growth of money stock, M1 and M3 alternately, g is the rate of growth of real GDP and DEF, as before, is the ratio of nominal budget deficit to the GDP. Because of the shortness of the time series, three lags were used in each case. Instead of
Deficits and money supply
97
reporting the regression results of the static and dynamic versions in detail, we report two sets of summary results. The estimates for GM1 are given in Table 4.3 and those for GM3 in Table 4.4. Each table gives the sum of the lagged deficit coefficients and the final column gives the long-run effect which is equal to the sum in column (2) divided by one minus the sum of the coefficients of the lagged dependent variables from the dynamic version. Consider Table 4.3 for growth of M1. In terms of the contemporaneous effect, only South Korea and Malaysia have a positive and marginally significant coefficient for the budget deficit. For the others, either the coefficient has a negative sign, as in the cases of India, Indonesia, the Philippines, and Singapore, or is not at all significant. But, as
Table 4.3 Effects of deficits on money growth, M1: the static and the dynamic models Country India
Indonesia
South Korea
Malaysia
Pakistan
The Philippines
Singapore
Sri Lanka
Thailand
Effect of contemporaneous deficit
Sum of lagged deficit coefficients
Long-term effect
−0.8878
−0.6157
−0.4779
(0.51)
(0.46)
(0.46)
−3.1688
4.2100
6.4551
(1.19)
(0.71)
(0.62)
2.9978
−5.0417
−5.5193
(1.18)
(1.29)
(1.06)
0.8483
−1.7894
−1.3096
(1.77)
(2.85)
(2.49)
0.2423
2.3258
1.4889
(0.30)
(2.24)
(2.53)
−1.1228
−1.0775
−1.0945
(0.65)
(0.29)
(0.29)
−2.444
2.1174
2.4657
(1.78)
(1.56)
(2.27)
1.4511
−0.7336
−0.5938
(0.41)
(0.21)
(0.21)
0.1963
−0.9652
−1.1255
(0.20)
(0.41)
(0.36)
Note t values are given in parentheses
Budget deficits and economic activity in Asia
98
pointed out in the previous chapter, such a lack of contemporaneous relationship is not inconsistent with the Sargent-Wallace story. Consequently, we then consider the dynamic effects. The results improve greatly for two countries, Pakistan and Singapore, but deteriorate for South Korea and Malaysia. But, of course, this may simply indicate different time paths for the effects of deficits on money supply in these countries. The long-term effects do not add much to the information in column (2). On the whole, therefore, we can conclude from this table that deficits affect the growth of money stock M1 in South Korea, Malaysia, Pakistan, and Singapore. The results for the growth of M3 in Table 4.4 are no more favourable to the hypothesis. In column (1), the only significant coefficient is for Sri Lanka, whereas in column (2), this is the case for South Korea and Singapore, but for the former, the coefficient has the wrong sign. The long-term effects provide some additional information in the case of Pakistan only where the coefficient is now marginally significant and of the correct sign. Given the much greater volatility of k3, these results are not that surprising. Before turning to a further discussion of these results, it may be useful to consider them in conjunction with the results of Chapter 3. Recall that the effect of budget deficits on the growth of money supply is a two-step sequential process, namely, from deficit to the monetary base and from the monetary base to the money supply.
Table 4.4 Effects of deficits on money growth, M3: the static and the dynamic models Country India
Indonesia
South Korea
Malaysia
Pakistan
The Philippines
Singapore
Effect of contemporaneous deficit
Sum of lagged deficit coefficients
Long-term effect
−1.0456
−0.8589
−2.3856
(1.18)
(1.10)
(0.89)
−0.5440
2.9660
7.0889
(0.15)
(0.52)
(0.50)
−0.9051
−10.662
−17.1510
(0.28)
(2.82)
(3.03)
−7.4416
−2.5733
−1.9254
(0.28)
(0.04)
(0.04)
0.5600
0.8800
0.7200
(0.73)
(0.88)
(1.20)
0.0319
−1.9390
−5.4242
(0.03)
(0.72)
(0.42)
−0.6570
2.4636
2.9743
(0.61)
(2.27)
(1.48)
Deficits and money supply Sri Lanka
Thailand
99
1.4608
0.7178
1.1136
(2.08)
(0.67)
(0.98)
0.4728
−0.6346
−0.8146
(0.68)
(0.53)
(0.45)
Note t values are given in parentheses
The estimates which bear upon these two steps are brought together in Table 4.5. Table 4.5 sheds interesting light on the money supply process in the countries of our sample in so far as the effect of deficits is concerned. We consider the result for each country separately in order to examine this aspect. Thus for India, while budget deficits have a significant effect on the monetary base, the deficits do not appear to have any effect on the growth of M3, though in one type of the model, column (2), there is some evidence of a positive effect on the growth of M1. For Indonesia, the results are rather curious. There is no effect on the monetary base and yet some marginal effect, in column (2), is indicated on the growth of M1, but again none on the growth of M3. In the case of South Korea, the results are somewhat similar to those for Indonesia, except for the negative and significant effect on the growth of M3. Malaysia shows somewhat consistent results in that both the MB and the growth of M1 are positively affected, but again the growth of M3 is not affected. The results for Pakistan are similar to those for Malaysia,
Table 4.5 Comparisons of the effects of budget deficits on monetary base and money growth Country
Effect on MB(1)
Effect on M1(2)
Effect on M1(3)
Effect on M1(4)
Effect on M3(5)
Effect on M3(6)
India
+(s)
+(ms)
−(ns)
−(ns)
−(ns)
−(ns)
Indonesia
+(ns)
+(ms)
−(ms)
+(ns)
−(ns)
+(ns)
South Korea
+(ns)
−(ns)
+(ms)
−(ms)
−(ns)
−(s)
Malaysia
+(ms)
+(ns)
+(ms)
−(s)
−(ns)
−(ns)
Pakistan
+(s)
−(ns)
+(ns)
+(s)
+(ns)
+(ns)
The Philippines
+(ms)
+(ms)
−(ns)
−(ns)
+(ns)
−(ns)
Singapore
+(s)
+(ms)
−(ms)
+(ms)
−(ns)
+(s)
Sri Lanka
+(s)
+(ns)
+(ns)
−(ns)
+(s)
+(ns)
Thailand
+(s)
−(ns)
+(ns)
−(ns)
+(ns)
−(ns)
Notes (1) Based on the estimates of the structural equation in Chapter 3
Budget deficits and economic activity in Asia
100
(2) Based on Table 4.2 (3) Based on column (1) Table 4.3 (4) Based on column (2) Table 4.3 (5) Based on column (1) Table 4.4 (6) Based on column (2) Table 4.4 ‘s’, ‘ns’ and ‘ms’ stand for significant at 5 per cent level, not significantly different from zero, and that the coefficient exceeds its own standard error, respectively
except that they are stronger, the effect on both the monetary base and M1 growth being sigificant. The Philippines displays results virtually identical with those for Malaysia. Singapore has the most unambiguous results of all countries, all three variables, MB, M1, and M3 being positively affected. The results for Sri Lanka are rather interesting in that they are the reverse of those for Malaysia, Pakistan, and the Philippines in that now deficits have no effect on the growth of M1 but do affect the growth of M3. Finally, the results for Thailand are different from those for the other eight countries. In this case, while deficits have a positive and significant effect on the monetary base, they do not have any effect on the growth of either M1 or M3. Before speculating on these results further, it may be interesting to compare them with other studies for the developing countries. Unfortunately, there are a few studies which provide such estimates. I have been able to find only two. Their results are summarized in Table 4.6. Given the fact that only one country, Sri Lanka, is common to this table and our study, it is interesting to note that for Sri Lanka our results are similar to those reported by Dornbusch and Fischer, in spite of the differences in the sample period and the
Table 4.6 Effects of budget deficit on money growth in other studies for LDCs Country
Author (s)
Monetary aggregate
Effect
South Africa
Dornbusch-Fischer
M1
−(ns)
Brazil
Edwards
M2
+(s)
Chile
Edwards
M2
+(ns)
Colombia
Edwards
M2
+(s)
Guatemala
Dornbusch-Fischer
M1
+(s)
Israel
Dornbusch-Fischer
M1
+(s)
Mexico
Edwards
M1
+(s)
Peru
Edwards
M1
+(s)
Sri Lanka
Dornbusch-Fischer
M1
+(ns)
Source: Dornbusch-Fischer (1987) and Edwards (1983)
Deficits and money supply
101
Note ‘s’ and ‘ns’ stand for significant and not significant at 5 per cent, respectively
models estimated. On a broader point, it is clear from this table that for the developing countries also, the effects of deficits on money growth are sensitive to the definition of money supply used and, further, that budget deficits do not appear to affect money growth in all developing countries.
SUMMING UP The evidence on the relationship between budget deficits and money growth in the developed countries is very mixed.3 The evidence for the Asian countries in our sample does not turn out to be any more unequivocal. Certainly, on the basis of the results in this chapter, it is difficult not to conclude that budget deficits do not exercise a strong influence on the growth of money supply, particularly that of M3. This lack of a firm relationship is contrary to the widespread belief in these countries.4 It may be that quarterly data will reveal more significant effects. But given the preponderance of the unfavourable evidence from the developed countries and the evidence from this study, we must conclude for now that perhaps we need to have a fresh look at the role of deficits in affecting money growth in developing countries as well, rather than go on believing that deficits always have an expansionary impact on money supply. Our results also point to the need for carefully distinguishing between the two-steps in the money supply process as far as the role of deficits is concerned. Our results also warn against the mechanical application of the money multiplier analysis to the determination of money supply in developing countries.
5 Deficits and inflation Are budget deficits always inflationary? Within the context of the countries in our sample, this is the question we examine in Chapter 5. Using the Sargent and Wallace (1981) framework, it was pointed out in the previous chapter that in a regime of fiscal dominance, given that the inter-temporal budget constraint of the government must be satisfied, the central bank will be obliged to monetize the deficit either now or in later periods. To the extent that such monetization takes place, it will lead to increases in money supply and, ceteris paribus, in the rate of inflation, at least in the long run. This quantity theory result is widely accepted and need not be explored in great detail here. However, it is important to keep in mind that this result cannot be guaranteed, at least in the short run, in view of the results of the previous chapter. There we found that only a relatively weak relationship existed between budget deficits and money growth, whether in the static sense or in the dynamic sense. Apart from this indirect effect of budget deficits on inflation, this chapter also deals with the more conventional view that budget deficits exercise a direct effect on inflation, even if there is no change in money supply. The channels through which this might happen are numerous and controversial, for example, the mechanism could be via interest rates, wealth effects, the underemployment of resources and their utilization, and so on. Unfortunately, disentangling the separate effects of the various channels is not an easy matter, although some attempts at identifying a few of these channels are made in the next two chapters. In this chapter, we adopt a less ambitious and more conventional approach, namely, we estimate single equation, reduced form models, to ascertain the indirect effect and the direct effect via money growth. Before proceeding to the quantification of these effects, we present a brief review of the literature in so far as it relates to the developing countries.
A BRIEF SURVEY OF THE LITERATURE While there is substantial literature on this topic for the developed countries, particularly for the USA, the evidence is quite sparse for the developing countries. It is much more substantial about the direct effect of budget deficits. This section provides a brief review of this literature. The works surveyed can be broadly classified into four groups, as follows: (1) works which only conduct Granger-Sims type causality tests to determine the direction of causality between inflation and money growth; (2) works which only specify and estimate structural models of inflation; (3) works which combine (1) and (2); and
Deficits and inflation
103
(4) works where no such formal testing or estimation is done, but diagrammatic and/or tabular data representations are used to draw inferences about causality. The specific studies included in each of the groups are as follows: (1) Nachane and Nadkarni (1985), Ramachandran (1983) and Siddiqui (1989); (2) Bhalla (1981a), Dornbusch and Fischer (1981) and Saini (1982); (3) Aghveli and Khan (1978), Bhalla (1981b), Darrat (1986), and Onis and Ozmucur (1990); and (4) Minhas (1987). Instead of reviewing each one of these works separately, an attempt is made to offer an integrated view of their main findings. We first concentrate on the findings with regard to the direction of causality. The seven studies in groups (1) and (2) have used different prewhitening filters to achieve stationarity of the time series used, different measures of money supply and inflation, and different tests to identify the direction of causality. The results have been found to be sensitive to these differences. For example, Siddiqui (1989) has reported that for Pakistan the results are sensitive to alternate prewhitening filters used as well as the tests used. For example, on the basis of the Sims (1972) test WPI causes both M1 and M2, but on the basis of the Pierce and Haugh test (1977), there is two-way causality between CPI and both measures of money, but independence between WPI and money. The two-way causality is also reported by Aghevli and Khan (1978) for four countries using only the Pierce and Haugh cross-correlation method. Ramachandran (1983) reported the same result for India using the Sims test. Finally, Onis and Ozmucur (1990), using a four-variable VAR model, also reported a similar finding for Turkey. However, this two-way causality is not supported uniformly. Thus, Ramachandran’s findings for India are contradicted by those of Bhalla (1981b) and Nachane and Nadkarni (1985). Darrat (1986) also reported unidirectional causality from money to inflation for four African countries. The main findings of these studies are summarized in Table 5.1. From these results, it is difficult to conclude unequivocally whether money is exogenous to inflation or not. But from our point of view, what is more important is that these studies do not give any idea as to the direct effect of budget deficits on inflation, nor indeed of the indirect effect, even if we assume that money was causing inflation. Given the ambiguities pointed out by Siddiqui, and the unidirectional results by Bhalla, Darrat, and Nachane and Nadkarni, the most that we can say is that there is some support for the monetarists’ proposition that budget deficits, in so far as the lead to increases in money supply, do cause inflation. Of the studies surveyed above, Aghveli and Khan (1978), Bhalla (1981b), Darrat (1986), and Onis and Ozmucur (1990) carry the results of their causality tests a step further. Of particular interest
Budget deficits and economic activity in Asia
104
Table 5.1 Results of causality tests for the works surveyed Author Aghveli-Khan (1978)
Country
Period
Direction of causality
Brazil
1964–74
MP
Colombia
1961–74
MP
Dominican Republic
1964–74
MP
Thailand
1964–74
MP
Bhalla (1981b)
India
1956–76
M→P
Darrat (1986)
Libya
1960–80
M→P
Morocco
1960–80
M→P
Tunisia
1960–80
M→P
Nachane-Nadkarne (1985)
India
Onis-Ozmucur (1990)
Turkey
Ramachandran (1983)
India
Siddiqui (1989)
Pakistan
M→P 1979–87
MP MP
1971–82
Ambiguous
are the studies by Aghveli and Khan, and Oniz and Ozmucur. Before discussing these studies, a brief word about the other two. Bhalla and Darrat, having established the direction of causality from money to inflation for India and for those African countries in Table 5.1, estimated reduced form equations which may be regarded as modified forms of the monetarist hypothesis. For the period 1956–76 and the sub-period 1956–72, Bhalla reports that for India the monetarist model performed well in some equations but not in others. On the whole, though, it would appear that changes in M1 were a significant determinant of inflation in the period under consideration. But these findings are contradicted by Saini (1982) for the period 1953–80. Using a similar kind of model, he reports that the coefficient of Mt is both insignificant and negative in sign, although that of the lagged one period is significant and positive. He concludes that ‘since the sum of the coefficients varies from equation to equation, the impact of money growth on inflation is not clear’ (Saini 1982:878). Darrat, using quarterly data for the period 1960– 80, estimates a monetarist model for Libya, Morocco, and Tunisia and concludes that the monetarist model provides an adequate explanation of inflation in these countries. Turning to the other two studies in the group, namely those of Aghveli and Khan, and Onis and Ozmucur, the interesting thing about them is that having established the twoway (and in the case of Onis and Ozmucur four-way) causality between money and prices, they specify and estimate models which provide a rationale for this phenomenon. In both cases, the models are estimated and simulated with some success. The structure of the Aghveli and Khan model is as follows. It consists of five equations, four structural and one definition. The structural equations relate to the demand for real cash balances, government revenue and expenditure, and the money supply. The definition relates to the formation of expectations about inflation. The potentially destabilizing element of the
Deficits and inflation
105
model came from the lags in the response of revenues to inflation but of instantaneous adjustment of expenditures to inflation. The self-perpetuating nature of inflation or the two-way causality between money and inflation occurs as follows. Starting from a position of equlibriums, suppose a monetary shock occurs. This leads to increases in price via the quantity theory mechanism, the increase in inflation leads to an increase in government expenditure but not to a corresponding increase in revenues, thus creating a budget deficit, which is financed by money creation, which then leads to a further increase in price and so on. The model is estimated for Brazil, Colombia, Dominican Republic, and Thailand for the period 1964–74, and for the others the period 1961–74. The data used are quarterly. The estimated parameters and the dynamic simulation results show a considerable success of the model. The Onis and Ozmucur model, on the other hand, rationalizes the four-way causality between high-powered money, prices, exchange rate, and exports, by specifying a model which interrelates these four variables. The model consists of four equations, namely, a price equation, and equations for high powered money, exchange rate, and exports. According to their ‘vicious circle’ hypothesis, ‘under a floating exchange rate regime, an initial disturbance (either domestic or foreign) can set into motion a cumulative process of inflation and exchange rate devaluation, through which the exchange rate effect is rapidly translated into domestic prices and costs and back to the exchange rate (Onis and Ozmucur 1990:135–6). This model is estimated and simulated for the period 1981–87, using montly data. This model also provides a reasonable explanation of the observed causality patterns. From our point of view, an important point about both these studies is that the change in money supply is specified in terms of budget deficits and thus there is a mechanism to determine the effect of budget deficits on inflation. However, like the other studies so far, these also concentrate only on the indirect or the monetary effects of budget deficits on inflation. In the same spirit as the Bhalla and Darrat studies, but without any tests of causality, are the studies in group (2). All of these studies estimate reduced form equations. The Bhalla and the Saini studies estimate equations which are essentially monetarist model and its augmented versions, increased by the inclusion of additional variables. The Dornbusch and Fischer estimating equation, on the other hand, is derived from a standard IS-LM-AS model which is general enough to encompass a variety of hypotheses. A basic difference between the Dornbusch-Fischer equation estimated and the other studies surveyed is that theirs is the only study which provides estimates for the indirect as well as the direct effects of budget deficits on inflation. In this sense, this is really the only study which is strictly comparable to the results to be reported on later in this chapter. Continuing with the survey, Dornbusch and Fischer estimate an equation for inflation which includes both deficit and money growth as arguments. Only three countries in this sample can be classified as developing—Guatemala, Israel, and Sri Lanka. Current money growth is not significant in any of the three and, in fact, has the wrong sign in two. As for the budget deficit, it has a positive and significant effect only in Israel and even in this case the quantitative impact is very small. For Guatemala, the coefficient even has the wrong sign. They also point out that the use of current and lagged monetary growth as the sole determinant of inflation also performed poorly. Their results for Sri Lanka contrast sharply with those reported by Bhalla (1980a) and Saini (1982). In both cases, the monetarist model provides a reasonable explanation of inflation, except that in both cases the sum of the coefficients of the current and lagged money growth varies
Budget deficits and economic activity in Asia
106
from 0.18 to 0.38, thus implying an estimate of much less than unity. These differences may reflect the different time periods covered by the three studies. Bhalla and Saini also present results for other countries in addition to India and Sri Lanka. In all, Bhalla reports results for twenty-nine countries and Saini for six. Of the twenty-nine, twenty-five countries have a positive and significant sum of current and lagged money growth in Bhalla’s sample, thus providing strong support for the indirect effect of budget deficits on inflation. But in the countries in our sample, Bhalla’s results are not uniform. He includes eight of the countries in our sample, India, Korea, Malaysia, Pakistan, the Philippines, Sri Lanka, Taiwan and Thailand. The sum of money growth is significant and positive in six cases, but not in that of Korea and Pakistan. In the other cases, the individual coefficients are not always significant. This is so for India and Sri Lanka. In the cases of the Philippines and Taiwan, the current and the lagged growth rates, respectively, are not significant. Saini’s results for India and Sri Lanka have already been considered. His results for the Philippines and Korea are similar to those of Bhalla—not very supportive of the indirect effect. His results for Taiwan and Thailand are much weaker than those of Bhalla. On the basis of the above survey, we cannot draw any unequivocal conclusions about either the direct or the indirect effects of budget deficits on inflation in developing countries. The evidence on the direct effect is too sparse to be categorical at all, but even for the indirect effect, the evidence is mixed. At the very least, the survey points to the need for further work before we can draw any definitive conclusions about the issues under consideration. This task is undertaken in the following sections.
Figure 5.1 Nominal deficits, growth of M1, and rate of inflation
Deficits and inflation
107
A DETOUR Before we proceed to the specification and estimation of our models, it would be useful to have a closer look at the data on money growth, budget deficits, and inflation presented in Chapter 1. For visual examination, the data are presented in Figure 5.1. Instead of drawing any general conclusions first, we consider the evidence on a countryby-country basis. The figures are used to consider both the direct and indirect effects of the budget deficits on inflation. India: Consider the relationship between the deficit and inflation first. It is clear that after the early 1970s, the relationship between the two variables is not close, except towards the end of the period. The relationship between money growth and inflation seems to be somewhat less pronounced, particularly towards the end of the period. Indonesia:
In this case, there seems to be little relationship between the nominal deficit and inflation. On the other hand, with the exception of the period from 1983 to 1985, the growth of money and inflation move together nicely. The reversal towards the end of the period is intriguing. In this case, then, the direct effects seems to be minimal, but the indirect effect considerable.
South Korea:
There is virtually no relationship either between the budget deficit and inflation or between inflation and money growth. In other words, neither the direct nor the indirect effect seems to exist in the case of South Korea. The nature of indirect effect is consistent with the regression results reported by Bhalla (1981a) and Saini (1982) for somewhat shorter periods.
Malaysia:
In terms of the direct effect, the nominal deficit and inflation move together in the 1970s but the relationship diverges in the later period. The indirect effect, on the other hand, seems more consistent over the period.
Pakistan:
There is far greater variability in the rate of money growth than in inflation and the direction of movement between the two series is not at all clear. On the other hand, nominal deficit and inflation show a closer correspondence, except for the period 1973–76. It would thus seem that in this case the indirect effect is rather weak, if it is apparent at all, but the direct effect is more pronounced.
The Except for the first two years of the period, the relationship between money growth Philippines: and inflation is rather close, thus suggesting a strong indirect effect. However, there is virtually no direct effect. Singapore: Sri Lanka:
Once again the direct effect is not in evidence, but the indirect effect, while more pronounced, is not as clear cut as, say, in the case of the Philippines. In this case both direct and the indirect effects seem to exist, with the direct effect being a little more prominent.
Thailand: The situation in this case seems to be somewhat like that of Sri Lanka, with some evidence of both types of effects.
These observations, of course, do not necessarily establish either the direction of causality or the magnitude of the two effects. But they do suggest some general observations. First, the experience of the nine countries during the period covered was not uniform. And second, that even for the same country, the experience over the period was not the same. In other words, the response of inflation to budget deficits, either
Budget deficits and economic activity in Asia
108
directly or indirectly through money growth, in the countries of our sample has not been uniform or as strong as often alleged to be the case. But for more specific inferences, we must turn to quantitative estimates.
THE MODELS In the spirit of the treatment in Chapter 3, in this case we also follow two approaches to examining the impact of budget deficits on inflation, the structural approach and the nonstructural approach, along the lines of King and Plosser (1985). The structural approach adopts the by now familiar route in which we derived a reduced form equation for inflation from a standard IS-LM-AS model. Since such versions are rather well known, the treatment here is rather brief and is largely meant to highlight the features of interest for this study. We start by specifying the Lucas (1973) aggregate supply function (1) with (2) where
is the full employment level of output and
desired price level
is the rate of change of the
The aggregate demand function is given by (3)
where
is the rate of growth of money supply and DEF is the budget deficit.
Deducting
from both sides of (3) and using (2), we get (4)
In view of the difficulties in the measurement of GAP and in a suitable proxy and the insignificance of this variable in Bhalla (1981a) and Dornbusch and Fischer (1981), this equation is rewritten as (5) Solving for
we get (6)
Deficits and inflation
109
Given the relatively underdeveloped nature of the financial markets in the developing countries, it may be the case that not only current but past changes in money supply will also affect inflation (Saini 1982), so that (6) can be written in a more generalized form as (7) Now p* not, of course, observable. But more importanly, for a variety of reasons, prices may not adjust instantaneously to the changes in the right-hand side variables of (7). In order to deal with this issue, one’s first approach is to use a Koyck-distributed lag adjustment function (Niskanen 1978), such that (8) where λ is the speed of adjustment of the actual to the desired rate of inflation. Substituting (7) in (8), we get (9) In order to highlight more sharply the distinction between the direct and the indirect effects of budget deficits on inflation, we estimate two other versions of (9). In the initial version, it is assumed that inflation is entirely caused by current and past changes in money supply and the adjustment of price changes to these changes is instantaneous, i.e. that λ=1. With these assumptions, (9) is reduced to (10) Next, it is assumed that although budget deficits also affect inflation directly, λ is still unity, so that (9) now becomes (11) An alternative approach to modelling the adjustment issue is to use the ‘error-correction’ modelling approach as proposed by Hendry et al. (1984). Using this approach, we can write the ‘error-correction’ model of inflation as ∆logPt=a+b∆logPtA1+c∆logMt−1+d∆logMt−2 +cLPMt−1+fDEFt (12) where LPMt−1 is the ‘error correction’ term defined as LPMt−1=(P−M)t−1 (13) The existence of two-lagged terms in money supply is justified by the fact that higher lagged terms turned out to be insignificant in the majority of cases.
Budget deficits and economic activity in Asia
110
Therefore, as far as the structural approach is concerned, we report estimates of equations (9), (10), (11) and (12). Turning to the unstructural or the atheoretical approach, a la King and Plosser, we again proceed in two steps. First we examine the contemporaneous relationships and then the dynamic relationships. The equation for the static version is (14) and the dynamic version is (15) where g refers to the rate of growth of real GDP. The rationale for this approach has already been explained in the previous chapter. However, a few other points are in order. It was pointed out in Chapter 3 that lack of contemporaneous relationship between changes in money supply and inflation does not imply violation of the fiscal dominance hypothesis as proposed by Sargent and Wallace (1981). It only implies the existence of a dynamic relationship as specified in equation (15). King and Plosser have further explained that this indirect effect, that is via changes in money supply (or more appropriately, high powered money), can be further illuminated if we make two simplifying assumptions, namely, that the demand for real cash balances is completely insensitive to changes in expected inflation and that the income elasticity of demand for real cash balances is unity. According to the results reported by Bhalla (1981a), these assumptions are not implausible for the developing countries. Thus, assuming that these assumptions are satisfied, King and Plosser show that a smaller increase in the money supply in the current period, meaning less current inflation, will necessitate a larger increase in money supply later and hence a higher rate of inflation. This points to the importance of the dynamic effects as specified in equation (15). The above discussion has concentrated on the indirect effects only and does not imply that the direct effects must follow an identical course. For example, there is nothing in the Sargent and Wallace model to suggest that a lack of contemporaneous relationship between money growth, caused by budget deficits, and inflation will be incompatible with a direct effect of such deficits on inflation. Similar comments can be made about the dynamic effects. However, since this discussion closely relates to the mechanisms through which direct effects take place, the topic is not pursued here, but will be taken up in the next two chapters. Here, we simply allow the empirical results to tell the story.
THE ESTIMATES In Tables 5.2 and 5.3, DEF represents nominal deficits as a proportion of the GDP and the money supply variable refers to M1. We start with a discussion of the estimates of equations (9), (10), and (11) in Table 5.2. A striking feature of these results is that the maximum that any of the models explains for any of the countries is 59 per cent of the variation in the rate of inflation. For a number of countries, that is, India, Indonesia, South Korea, and Taiwan, the fit is extremely poor. The most generalized version of the
Deficits and inflation
111
model in equation (9) performs the best for four countries. In the rest of the cases, the coefficient of the lagged dependent variable is not statistically significant from zero. As for the importance of the direct effect of budget deficits, the coefficient has the correct sign and is marginally significant in only two cases. The indirect effects are discussed below. However, before we draw any general conclusions, a country-by-country discussion of the estimates would be useful. India: Neither of the three models provides any explanation of inflation. Considering the results of equation (9), the lagged dependent variable is marginally significant, suggesting a lag in the
Table 5.2 Estimates of equations (9), (10), and (11) India Variables Constant
DEFt
(9)
Indonesia
(10)
(11)
(9)
(10)
(11)
4.5693
8.1064
6.5954
1.2141
1.9791
1.5412
(0.80)
(1.83)
(1.20)
(0.12)
(0.22)
(0.16)
0.1986
0.1869
0.1679
0.4650
0.4501
0.4622
(1.00)
(0.98)
(0.84)
(1.50)
(1.63)
(1.58)
−0.2166
−0.1506
−0.1614
0.0836
0.1463
0.1265
(1.17)
(0.86)
(0.89)
(0.19)
(0.43)
(0.35)
−0.0133
−0.0439
−0.0544
−0.0476
−0.0145
−0.0371
(0.07)
(0.23)
(0.28)
(0.14)
(0.05)
(0.11)
−33.777
—
−39.109
−124.11
—
−92.653
(0.80)
—
(0.40)
(0.34)
—
(0.30)
0.2667
—
—
0.0738
—
—
(1.16)
—
—
(0.21)
—
—
−2
R
0.06
0.04
0.08
0.03
0.22
0.14
See
5.50
5.44
5.55
11.80
10.64
11.16
DW
1.79
1.43
1.4
2.09
2.16
2.05
South Korea Variables Constant
(9)
Malaysia
(10)
(11)
(9)
(10)
(11)
2.4836
11.4200
11.8140
−4.5079
−4.2995
−4.9527
(0.33)
(1.83)
(1.63)
(0.71)
(1.26)
(0.91)
−0.0836
−0.0157
−0.0195
0.3929
0.4107
0.4137
(0.58)
(0.10)
(0.12)
(1.78)
(2.49)
(2.37)
0.0422
−0.0380
−0.0353
0.3544
0.3217
0.3256
Budget deficits and economic activity in Asia
DEFt
112
(0.30)
(0.25)
(0.22)
(0.41)
(1.93)
(1.84)
0.2300
0.2168
0.2142
0.0203
−0.0057
−0.0028
(1.66)
(1.44)
(1.37)
(0.09)
(0.04)
(0.01)
39.247
—
20.449
−1.6799
—
−5.2595
(0.25)
—
(0.11)
(0.04)
—
(0.17)
0.5259
—
—
−0.0847
—
—
(2.44)
—
—
(0.17)
—
—
−2
R
0.15
0.00
0.09
0.45
0.55
0.51
See
8.18
8.99
9.23
5.35
4.80
5.05
DW
2.14
1.08
1.06
1.91
1.98
1.98
Pakistan Variables Constant
DEFt
(9)
(10)
The Philippines (11)
(9)
(10)
(11)
5.9077
7.6480
6.5422
−1.8407
5.2274
−1.7213
(1.92)
(2.00)
(1.35)
(0.40)
(0.97)
(0.37)
−0.3000
−0.1424
−0.1512
0.6583
0.6229
0.6637
(2.68)
(0.86)
(0.88)
(4.54)
(3.35)
(4.62)
0.1631
0.0236
0.0192
0.1840
0.1753
0.2718
(1.51)
(0.15)
(0.11)
(1.01)
(0.91)
(1.81)
0.0697
0.2111
0.2031
−0.1293
−0.2286
−0.1011
(0.69)
(1.42)
(1.32)
(0.82)
(1.19)
(0.66)
43.867
—
−22.295
−317.60
—
−323.50
(1.13)
—
(0.38)
(3.63)
—
(3.74)
0.7870
—
—
0.1444
—
—
(5.78)
—
—
(0.85)
—
—
−2
R
0.59
0.02
0.02
0.59
0.32
0.59
See
3.67
5.67
5.80
6.61
8.52
6.57
DW
1.83
0.60
0.59
1.87
1.60
1.64
Singapore
Sri Lanka
Variables
(9)
(10)
(11)
(9)
(10)
(11)
Constant
−1.4015
−2.7046
−2.0176
−0.1171
1.4554
−2.1424
(0.46)
(0.95)
(0.88)
(0.04)
(0.47)
(0.66)
Deficits and inflation
DEFt
113
−0.1782
−0.0778
−0.1133
−0.3573
−0.0029
−0.0322
(1.16)
(0.63)
(0.87)
(2.00)
(0.02)
(0.22)
0.3787
0.3318
0.3623
0.2869
0.1745
0.0771
(2.97)
(2.75)
(2.91)
(1.75)
(1.00)
(0.46)
0.2507
0.2883
0.3060
0.1673
0.4633
0.3589
(1.76)
(2.35)
(2.46)
(1.09)
(2.83)
(2.28)
−76.436
—
−63.524
−30.804
—
−70.907
(1.15)
—
(0.99)
(0.95)
—
(2.15)
0.2059
—
—
0.6435
—
—
(0.81)
—
—
(2.63)
—
—
−2
R
0.33
0.35
0.34
0.59
0.33
0.44
See
4.65
4.60
4.60
5.51
7.01
6.40
DW
1.81
1.58
1.57
1.88
1.10
1.19
Taiwan
Thailand
Variables
(9)
(10)
(11)
(9)
(10)
(11)
Constant
−1.9234
−0.2767
−2.0581
−2.9309
−2.3921
−3.6377
(0.41)
(0.06)
—
(1.25)
(1.07)
(1.38)
−0.0608
−0.0193
−0.0529
(0.2361)
(0.2739)
(0.2621)
(0.47)
(0.14)
(0.43)
(1.93)
(2.00)
(1.89)
0.2661
0.2154
0.2618
0.3036
(2.62)
(2.04)
(2.67)
(2.07)
(3.18)
(2.88)
0.0331
0.1128
0.0268
−0.1451
0.0792
0.1054
(0.31)
(1.09)
(0.26)
(0.88)
(0.54)
(0.70)
338.27
—
289.85
−71.683
—
−52.857
(1.66)
—
(2.19)
(1.83)
—
(0.88)
−0.1061
—
—
0.5068
—
—
DEFt
0.4737 0.4431
(0.31)
—
—
(2.50)
—
—
−2
R
0.20
0.08
0.25
0.57
0.46
0.45
See
6.49
6.97
6.30
4.02
4.54
4.57
DW
1.58
1.29
1.67
1.60
1.01
0.98
Note
Budget deficits and economic activity in Asia
114
t values are given in parentheses adjustment of actual to the desired rate of inflation. The deficit variable is not only insignificant, but even has the wrong sign. As for the indirect effect of the deficit, the sum of the three coefficients on money growth add up to a negative magnitude, with the first two being positive and negative in sign respectively, and marginally significant. It is thus clear that on the basis of the structural models of equations (9) to (11), budget deficits do not seem to have much of an impact on inflation in India, either directly or indirectly. Indonesia: The situation is only marginally better for Indonesia. The only difference now is that the strictly monetarist model explains about 22 per cent of the variation in the observed inflation, and the current rate of money growth is marginally significant and positive in all three equations, suggesting some indirect effect of the budget deficit. The three coefficients on the money
Table 5.3 Estimates of equation (12) Variables Constant
∆logP−1 ∆logM−1 ∆logM−2 LPM−1 DEF
India
Indonesia
South Korea
Malaysia
Pakistan
0.0241
−0.4244
0.0871
−0.8164
0.1216
(0.55)
(0.96)
(0.78)
(1.05)
(2.46)
0.3343
—
0.5221
0.7696
0.8472
(1.44)
—
(2.28)
(2.81)
(5.76)
0.2639
0.7926
−0.1424
0.4120
−0.2393
(1.47)
(2.54)
(0.88)
(2.28)
(2.17)
−0.2614
—
0.0378
0.2756
0.1416
(1.41)
—
(0.21)
(1.27)
(1.31)
−0.0978
−0.1017
−0.0074
−0.1731
−0.0744
(1.06)
(0.99)
(0.21)
(1.66)
(1.75)
0.6200
−1.7868
2.0078
0.8800
0.8355
(0.90)
(0.83)
(0.92)
(2.25)
(1.94)
−2
R
0.222
0.307
0.077
0.631
0.629
See
0.049
0.080
0.073
0.041
0.031
DW
1.83
1.87
2.12
2.06
2.00
Variables Constant
∆logP−1
The Philippines
Singapore
Sri Lanka
Taiwan
Thailand
0.6694
−0.0093
0.9663
0.0243
0.0940
(2.48)
(0.08)
(4.60)
(0.85)
(1.44)
0.1705
0.3944
0.4514
0.1397
0.1596
(0.84)
(1.52)
(2.66)
(0.92)
(0.83)
Deficits and inflation ∆logM−1 ∆logM−2 LPM−1 DEF
115
0.4289
−0.2089
−0.2466
0.0136
0.2764
(2.44)
(1.10)
(1.97)
(0.14)
(2.30)
−0.2089
0.3772
−0.0189
0.1002
0.2092
(0.91)
(2.49)
(0.12)
(1.17)
(1.31)
−0.2809
−0.0052
−0.3199
−0.0192
−0.0851
(2.36)
(0.18)
(4.54)
(1.69)
(1.71)
1.6397
−0.5136
0.6708
3.9675
1.3278
(1.38)
(0.56)
(2.61)
(4.50)
(1.73)
−2
R
0.408
0.168
0.796
0.623
0.569
See
0.064
0.050
0.035
0.039
0.037
DW
1.68
2.06
2.10
1.66
1.46
Note t values are given in parentheses growth also sum up to a positive magnitude though well below unity. The direct effect of the budget deficits, once again, is negative and not at all significant. South Korea:
The overall explanatory power of the three models is not much of an improvement over Indonesia, but the estimates are different. For one thing, the lagged dependent variable is significant. For another, the budget deficit has a positive, albeit insignificant, coefficient. And further, only the two-year lagged money growth has any significance, suggesting delayed indirect effects of budget deficits. On the whole, however, the results must be considered unsatisfactory for the ‘debt-inflation’ hypothesis.
Malaysia:
The results are a vast improvement over the last three countries. Estimates of equation (10) explain about 55 per cent of the variation in observed inflation. The first two terms of the money growth are positive and significant and the sum of the three coefficients is also positive. Thus the indirect effect of the budget deficits in this case is relatively clear and strong, and is consistently so in all three equations. However, the direct effect is insignificant and even has the wrong sign. In this respect, the results are similar to those of the last three countries. The insignificance of the lagged dependent variable is again noteworthy.
Pakistan:
Equation (9) fits best in this case, with the lagged dependent variable being quite significant. The direct effect of budget deficit in this case is both positive and marginally significant. However, the indirect effect is ambiguous. Thus, while the first lagged term has a positive and marginally significant coefficient, the current money growth has a negative and significant coefficient and the sum of the coefficients is negative although very small.
The The lagged dependent variable does not add anything to the explanatory power of the Philippines: model, so that equation (11) performs as well as equation (9). A marginally aspect of these results is that, in all three equations, current money growth is highly significant, and in (11) the first lagged term is also significant. But the direct effect
Budget deficits and economic activity in Asia
116
of budget deficits is curious in that the coefficient has a negative sign and is also highly significant. Thus, in this case, it would appear that while the indirect effect is unambiguously significant and positive, the direct effect is not reliable. Singapore: The results for Singapore are somewhat similar to those for the Philippines. Once again, equation (11) performs best, but the direct effect has the wrong sign, though in this case it is not significant. The indirect effect, however, is significant and the sum of the coefficients is positive, although the current money growth does not exercise any explanatory power. Sri Lanka: The general model of equation (9) performs best in this case, with the coefficient of the lagged dependent variable suggesting a significant lag in the adjustment between the actual and the desired rate of inflation. And the indirect effect of budget deficits adds up to be positive, even though the negative and significant effect of current money growth is strange. The direct effect, once again, is insignificant and of the wrong sign. Taiwan:
The best result is given by the estimates of equation (11), with the lagged dependent variable not being significant in equation (9). This is the first, and only, case with a positive and significant direct effect of budget deficits on inflation. The indirect effect is also clearly positive and significant. Thus, even though the overall explanatory power of the model is low, one can conclude in this case that the budget deficits exercise, both directly and indirectly, a significant effect on inflation.
Thailand:
With the lagged dependent variable being significant, the best fit is given by equation (9), but the direct effect is, once again, of perverse sign although insignificant. However, the indirect effect is clearly positive and significant.
Before drawing general conclusions from the structural estimates, we should consider estimates of equation (12). It may well be the case that the estimates of the models in equations (9) to (11) are sensitive to the particular assumption made about the dynamics. This sensitivity can be judged by looking at the estimates of the ‘error-correction’ model in Table 5.3. The first thing to note about these results is that the ‘error-correction’ term LPM(−1) has the correct sign in all ten cases and is significant or marginally significant in seven cases. In contrast, the partial adjustment coefficient of the previous model had the correct sign in eight cases and was significant or marginally significant in only five cases. Thus, at least by this criterion, the ‘error-correction’ model provides a better specification of the dynamics. But to go further, we must look once again at the direct and indirect effects of the budget deficits. As far as the direct effects are concerned, they are positive in eight cases and significant or marginally significant in six cases. As far as the indirect effects are concerned, they are positive in seven cases, and of these seven they are significant or marginally significant in six. On the whole, then, the ‘errorcorrelation’ model seems to provide a much stronger support for both the direct and the indirect effects of budget deficits on inflation. However, since a priori no particular adjustment hypothesis is pre-eminent, in order to draw general conclusions on the basis of the estimates of structural, we summarize the results of Tables 5.2 and 5.3 in Table 5.4. From the summary results in Table 5.4, we can draw the following conclusions:
Deficits and inflation
117
(i) In terms of the structural models, budget deficits seem to exercise direct effects on inflation in Malaysia, Pakistan, the Philippines, Sri Lanka, Taiwan, and Thailand. But there is no evidence of such direct effects in India, Indonesia, South Korea, and Singapore. (ii) The indirect effects of budget deficits on inflation exist in
Table 5.4 Summary of structural models’ results Partial adjustment model Country
Direct effect
Indirect effect
Error-correction model Direct effect
Indirect effect
India
−(ns)
−(ns)
+(ns)
+(ns)
Indonesia
−(ns)
+(ns)
−(ns)
+(s)
South Korea
+(ns)
+(ns)
+(ns)
−(ns)
Malaysia
−(ns)
+(ns)
+(s)
+(s)
Pakistan
+(ns)
−(ns)
+(s)
−(s)
The Philippines
−(s)
+(s)
+(ns)
+(s)
Singapore
−(ns)
+(s)
−(ns)
+(s)
Sri Lanka
−(ns)
+(s)
+(s)
−(s)
Taiwan
+(s)
+(ns)
+(s)
+(ns)
Thailand
−(ns)
+(s)
+(ns)
+(s)
India, Indonesia, Malaysia, the Philippines, Singapore, Sri Lanka, Taiwan, and Thailand. (iii) On the basis of these results, the indirect effects of budget deficits on inflation are more pervasive than the direct effects, and when the two are put together, the strongest effects are exhibited in the cases of Malaysia, the Philippines, Taiwan, and Thailand. We now turn to the estimates based on the non-structural approach according to equations (14) and (15). In this case, results are reported both with M1 and M3 and with the nominal and real deficits as defined in Chapter 1. This should provide some evidence about the sensitivity of results to alternate definitions of money supply and budget deficits. The estimates of equation (14) are given in Table 5.5. Since the numerical estimates are not of critical importance, we only provide the sign and the statistical significance of the coefficients of current money growth and budget deficits. For each country, there are eight estimates. Columns (1) and (2) provide the indirect effects due to growth of M1 and M3, respectively, when the measure of budget deficits is nominal. Columns (3) and (4) do the same thing when budget deficit is measured in real terms. Columns (5) and (6) give the direct effects of nominal and real budget deficits respectively when the money growth is based on M1, and columns (7) and (8) do the same thing for M3. Considering the indirect effects first, we can see from columns (1) and (2) that the results are sensitive to the measure of money supply used. Thus in the cases of India and Malaysia, for
Budget deficits and economic activity in Asia
118
example, is significant or marginally significant while is not, the reverse being the case for Sri Lanka. Similar sensitivity can be seen when the deficit is in real terms in columns (3) and (4). But the more important finding is that the current indirect effect exists for only five countries: India, Malaysia, Singapore, Sri Lanka, and Thailand. The contemporaneous direct effects are even less pervasive regardless of the measure of money supply used. Thus, of the nine cases only in the case of Pakistan does the budget deficit have a positive and significant effect. In all other cases, either the sign is insignificant or the sign is the wrong one, or both. But as already explained, the absence of contemporaneous effect, whether direct or indirect, does not necessarily imply the absence of such effects. It may only mean that static effects are unimportant and that dynamic effects exist. To see whether such is the case here, we turn to the estimates of equation (15).
Table 5.5 Contemporaneous relationships of inflation with money growth and budget deficits: equation (14) DEF
RDEF
(1)
(2)
(3)
(4)
(5) DEF
(6) RDEF
(7) DEF
(8) RDEF
India
+(ns)
+(ns)
+(ns)
−(ns)
+(ns)
−(s)
+(ns)
−(s)
Indonesia
−(ns)
−(ns)
—
—
−(ns)
—
−(ns)
—
South Korea
−(ns)
−(s)
−(ns)
−(s)
+(ns)
−(s)
+(ns)
−(s)
Malaysia
+(s)
+(ns)
+(s)
−(ns)
−(ns)
−(s)
−(ns)
−(s)
Pakistan
−(ns)
−(ns)
+(ns)
+(ns)
+(s)
−(s)
+(s)
−(s)
The Philippines
−(ns)
+(ns)
+(ns)
+(ns)
−(ns)
−(ns)
−(ns)
−(ns)
Singapore
−(ns)
+(ns)
−(ns)
+(ns)
−(ns)
−(ns)
−(ns)
−(s)
Sri Lanka
−(ns)
+(ns)
−(ns)
+(ns)
+(ns)
−(s)
−(ns)
−(s)
Thailand
+(s)
+(s)
+(s)
+(s)
−(ns)
−(s)
−(s)
−(s)
Note DEF, nominal deficit; RDEF, real deficits; respectively
and
are the real rates of growth of M1 and M3,
In order to avoid cumbersome presentation, the detailed regression results are not presented. Instead, summary statistics of the coefficients of major importance are reported. We start with Table 5.6. This table reports the F statistic for the hypothesis that the coefficient of each of the money growth terms be zero. Table 5.7 then reports the sum of the coefficients of the money growth terms along with their t values. These two tables thus summarize the indirect effects of budget deficits according to the unstructured
Deficits and inflation
119
approach. As with Table 5.5, here also we have estimates based on two definitions of money growth and deficits. From Table 5.6 it can be seen that the individual coefficients of money growth are significantly different from zero only for Malaysia using M1 as the definition of money, regardless of the definition of deficits. A less stringent hypothesis, that the sum of the money growth coefficients be zero, is tested in Table 5.7. Using M1 as the measure of money supply, from column (1) we can see that the sum
Table 5.6 F values for the effects of each money growth term to be zero on inflation (Inf) Country
(1) DEF
(2) RDEF
(3) DEF
(4) RDEF
India
0.82
0.60
0.60
0.48
(2, 12)
(2, 12)
(2, 12)
(2, 12)
0.05
—
0.69
—
Indonesia
(2, 5) South Korea
Malaysia
Pakistan
The Philippines
Singapore
Sri Lanka
Taiwan
(2, 5)
1.40
1.71
0.83
1.05
(2, 13)
(2, 13)
(2, 13)
(2, 13)
5.51*
1.19
3.05
2.81
(2, 5)
(2, 5)
(2, 5)
(2, 5)
3.05
0.21
1.71
0.59
(2, 13)
(2, 13)
(2, 13)
(2, 13)
0.14
0.05
0.58
0.42
(2, 14)
(2, 14)
(2, 14)
(2, 14)
1.39
2.25
1.54
0.54
(2, 12)
(2, 12)
(2, 12)
(2, 12)
1.40
1.47
0.15
1.22
(2, 13)
(2, 13)
(2, 13)
(2, 13)
1.99
—
—
—
0.79
1.32
1.43
0.26
(2, 14)
(2, 14)
(2, 14)
(2, 14)
(2, 12) Thailand
Note * Significant at the 5 per cent level. Degrees of freedom are given in parentheses
Budget deficits and economic activity in Asia
120
Table 5.7 Total short-run effect of money growth on inflation Country India
Indonesia
(1) DEF
(2) RDEF
Malaysia
Pakistan
The Philippines
Singapore
Sri Lanka
Taiwan
(4) RDEF
−0.1459
0.0313
1.1011
0.3156
(0.46)
(0.09)
(1.09)
(0.89)
0.0361
—
0.2884
—
(0.26) South Korea
(3) DEF
(1.15)
0.3090
0.3281
0.2185
0.1279
(1.26)
(1.35)
(0.57)
(0.81)
1.4365
0.2729
−0.0175
−0.0039
(2.53)
(0.36)
(1.04)
(0.32)
0.3757
0.0051
0.0267
−0.0811
(2.41)
(0.32)
(0.08)
(0.41)
−0.0942
−0.1534
−0.8326
0.4163
(0.24)
(0.32)
(0.91)
(0.91)
0.3033
0.4893
0.5820
−0.0484
(1.51)
(2.12)
(1.41)
(0.15)
0.0777
0.0949
0.1791
0.2918
(1.58)
(1.71)
(0.55)
(1.56)
0.0604
—
—
—
0.1869
0.1760
−0.0175
−0.0039
(0.67)
(0.63)
(1.04)
(0.32)
(0.31) Thailand
Note t values are in parentheses
is positive and significant or marginally significant for South Korea, Malaysia, Pakistan, Singapore, and Sri Lanka when the deficit measure is in nominal terms, and for South Korea, Singapore, and Sri Lanka when the deficit is in real terms. When M3 is used to measure money supply, the sum of the coefficients is positive and significant or marginally significant for India, Indonesia, and Singapore when the deficit is in nominal terms, and for Sri Lanka alone when the deficit is in real terms. We can conclude, therefore, that budget deficits exercise indirect influence on inflation in the cases of India,
Deficits and inflation
121
Indonesia, South Korea, Malaysia, Pakistan, Singapore, and Sri Lanka. The sensitivity to alternate measures of money supply in the case of India and Indonesia should be noted. We now consider the direct effect of budget deficits. Since this is the more important part of the story, we examine this issue not only in terms of the two hypotheses above for the indirect effect, but also in terms of the long-term effect which is give by the sum of the coefficients of the budget deficit terms divided by one minus the sum of the coefficients of the lagged dependent variable. Consider the F values in Table 5.8 first. It can be seen that only in the case of Indonesia can we reject the null hypothesis that each of the coefficients of budget deficit is zero, regardless of whether money supply is measured by M1 or M3. The sum of the deficit coefficients and their t values are given in Table 5.9. With M1 as the measure of money stock, nominal budget deficit has a positive and significant or marginally significant effect in the Philippines and Sri Lanka, and with real deficits in India, Malaysia, and Pakistan. With M3 as the definition of money, the nominal deficit shows significance in the Philippines and Sri Lanka and real deficits in Malaysia, Pakistan, and Thailand. In other words, as measured by the sum of the coefficients of the deficit terms, it seems to exercise some direct influence on inflation in India, Malaysia, Pakistan, the Philippines, Sri Lanka, and Thailand, but in the remaining four countries there
Table 5.8 F values for the effects of each deficit term to be zero on inflation* Country
(1) DEF
(2) RDEF
(3) DEF
(4) RDEF
India
0.33
1.31
0.19
1.26
(2, 12)
(2, 12)
(2, 12)
(2, 12)
7.91*
—
0.19
—
Indonesia
(2, 5) South Korea
Malaysia
Pakistan
The Philippines
Singapore
(2, 5)
0.43
0.73
0.30
0.53
(2, 13)
(2, 13)
(2, 13)
(2, 13)
3.23
1.77
0.27
1.78
(2, 5)
(2, 5)
(2, 5)
1.55
2.25
1.16
3.72
(2, 13)
(2, 13)
(2, 13)
(2, 13)
2.96
0.02
3.05
0.02
(2, 14)
(2, 14)
(2, 14)
(2, 14)
0.76
0.86
1.61
0.003
(2, 12)
(2, 12)
(2, 12)
(2, 12)
(2, 5)
Budget deficits and economic activity in Asia Sri Lanka
Taiwan
122
2.56
0.74
1.33
0.70
(2, 13)
(2, 13)
(2, 13)
(2, 13)
0.80
—
—
—
0.45
0.90
2.13
0.80
(2, 14)
(2, 14)
(2, 14)
(2, 14)
(2, 12) Thailand
Note * See Table 5.7
Table 5.9 Total short-run effect of deficits on inflation Country India
Indonesia
(1) DEF
(2) RDEF
(3) DEF
1.0309
79.377
0.7102
66.419
(0.80)
(1.10)
(0.55)
(0.90)
1.4249
—
0.4305
—
(0.57) South Korea
Malaysia
Pakistan
The Philippines
Singapore
Sri Lanka
Taiwan
(4) RDEF
(0.19)
0.9155
−0.0452
0.4907
4.0452
(0.39)
(0.03)
(0.17)
(0.22)
0.4534
308.32
−0.3988
298.06
(0.65)
(1.78)
(0.74)
(1.86)
−0.6487
17.042
−0.5125
21.266
(1.09)
(2.02)
(0.81)
(2.58)
5.9365
5.5059
8.3486
5.9801
(2.43)
(0.03)
(2.46)
(0.03)
0.2667
−12.556
−0.4136
−1.2255
(0.37)
(0.57)
(0.54)
(0.05)
0.9094
6.0044
0.8076
26.896
(2.05)
(0.13)
(1.40)
(0.50)
−0.5696
—
—
—
56.448
0.8552
71.179
(0.22) Thailand
0.7174
Deficits and inflation (0.76)
123 (0.97)
(0.62)
(1.27)
is no evidence of such an influence. To ensure that no information is lost from the estimates, Table 5.10 reports the longterm direct effects of the budget deficits. With M1 as the definition of money, the longterm effect is positive and significant or marginally significant for the Philippines and Sri Lanka only, and for the real deficit for none of the countries. For M3 measure of money, the nominal deficit affects inflation only in Sri Lanka and the real deficits in none. In comparison with the results in Table 5.9, this table does not shed any additional light on the issue under discussion. A comparison of the results for the structural models in Table 5.4 and of the unstructured models in Tables 5.7 and 5.9 shows that the direct effects are the strongest for Malaysia, Pakistan, the Philippines, Sri Lanka, and Taiwan, with suggestive evidence for India and Thailand. The indirect evidence is significant for Indonesia, Malaysia, Pakistan, the Philippines, Singapore, Sri Lanka, and Thailand. Taking the two effects together, the effects of budget deficits on inflation are the strongest for Malaysia, Pakistan,
Table 5.10 Long-term effects of deficits on inflation Country India
Indonesia
(1) DEF
(2) RDEF
(3) DEF
0.8056
72.288
0.6939
61.199
(0.84)
(0.78)
(0.56)
(0.66)
0.8606
—
0.2801
—
(0.59) South Korea
Malaysia
Pakistan
The Philippines
Singapore
Sri Lanka
(4) RDEF
(0.19)
3.1333
−0.1472
1.5044
14.235
(0.32)
(0.03)
(0.16)
(0.19)
0.7823
−233.26
−0.2157
−277.45
(0.29)
(1.15)
(0.79)
(1.06)
−0.7514
74.569
−1.4240
131.27
(0.50)
(0.86)
(0.60)
(0.73)
9.4843
6.3938
27.152
5.371
(1.11)
(0.03)
(0.51)
(0.03)
0.3593
−13.453
−0.6753
−2.0631
(0.36)
(0.55)
(0.54)
(0.05)
1.1572
18.111
1.1109
−31.954
(2.75)
(0.11)
(1.55)
(0.79)
Budget deficits and economic activity in Asia Taiwan
−2.6112
124
—
—
1.1168
105.82
1.7249
190.48
(0.67)
(0.59)
(0.51)
(0.63)
(0.35) Thailand
the Philippines, and Sri Lanka. For India there is little, and for South Korea none whatsoever.
CONCLUDING REMARKS This chapter carried out an extensive investigation into the effects of budget deficits on inflation in the countries of our sample. This was done by using alternate structural models and the unstructured approach. Since a priori both approaches have something to recommend themselves, our general conclusions are based on both, although it should be noted that the results of the two approaches in a number of cases are similar. Our results suggest that the effects of deficits on inflation have not been uniform in the countries of the sample, so that one should be careful in drawing general conclusions about this topic for the Asian economies. Further, there seems to be somewhat greater evidence of the indirect effect of deficit on inflation than the direct effect, although in four countries, Malaysia, Pakistan, the Philippines, and Sri Lanka, both effects exist strongly although it must be reiterated that since all of the growth in money supply cannot be attributed to budget deficits, not all of the effects of money growth can be assigned to budget deficits. The results for a number of countries, particularly those for India and South Korea, are interesting because of the widespread belief that the deficits are the major cause of inflation, at least in India (Minhas 1987). It is also interesting to see that a purely monetarist explanation, that is, the indirect effects of deficits alone, is not adequate to explain inflationary behaviour for the period covered.
6 Deficits and aggregate demand Quite apart from the effects of budget deficits on the growth of money supply and inflation, there are two other contentious aspects, although these are related to the direct effects of deficits on inflation discussed in the last chapter. One is the possible crowdingout effect of government expenditures on private expenditures. To the extent that such crowding out occurs, the effectiveness of fiscal policy is reduced correspondingly. The other aspect deals with the neutrality of the mode of financing such excess expenditures, i.e. budget deficits. The issue here is whether the Ricardian equivalence proposition as restated by Barro (1974) holds. This proposition states that it is irrelevant whether a given budget deficit is financed by tax increases or by debt issue. But the outcome, as discussed below, is based on a number of assumptions. A priori, it is neither possible to decide whether crowding out occurs, and even if it does to what extent, nor whether the Ricardian equivalence proposition holds. The only way to deal with these issues for any given country is via empirical evidence. Consequently, the aim of this chapter is to provide estimates on both aspects for each of the ten countries. Since the policy implications of both crowding out effects and the Ricardian equivalence proposition are serious, the evidence presented in this chapter should allow us to discuss such implications in a concrete way.
CROWDING OUT AND RICARDIAN EQUIVALENCE The issue of crowding out has been well known in the literature for quite some time and does not need any extensive elaboration. However, the mechanism of mechanisms through which such crowding out occurs, if it does, are still being debated and there is no unanimity. This issue is dealt with in the next chapter. The Ricardian equivalence proposition, though known for a long time, has really become germane to the discussion of fiscal policy since Barro revived it in 1974. Essentially it claims that, under certain conditions, the effect of government expenditures on aggregate demand is insensitive to whether such expenditures are financed by taxes or by debt (see Ricardo 1951, Buchanan 1958, Barro 1974). The basic idea is that debt and tax financing are equivalent, because financing by debt implies future tax liabilities, which are perfectly foreseen by economic agents and hence debt is not viewed as private wealth. This equivalence proposition is based on a number of assumptions. More specifically that: (a) capital markets are perfect with no constraints on borrowing by consumers; (b) taxes are non-distortionary; (c) economic agents are fully aware of the path of future fiscal policies; and (d) both public and private sectors have equal planning horizons.1 Clearly, violations of one or more of these assumptions could lead to deviations from the equivalency proposition. Superficially, at least, we would tend to
Budget deficits and economic activity in Asia
126
think that these assumptions are more likely to be violated than upheld in the developing countries. For example, there is plenty of evidence that in most of the countries in our sample capital markets are far from perfect, with severe liquidity constraints, and certainly taxes are, more often than not, distortionary. Whether the other two assumptions are also violated is less certain. Be that as it may, the approach in this chapter is not to verify these assumptions, but to test the proposition directly. As in other areas of macroeconomics, in this case also two approaches have been used, first an ad hoc approach in which private consumption is specified as a function of government expenditure, taxes, and private wealth including government debt and second the proposition is tested in terms of certain restrictions on the parameters of these and other variables. This approach is used, for example, by Boskin (1988), Feldstein (1982), Buiter and Tobin (1979), Kormendi (1983) and Seater and Mariano (1985), among others. These aproaches do not explicitly incorporate the third assumption listed above, namely, perfect information about future government fiscal policies or, more appropriately, rational expectations. The other approach is the one in which the estimating equation is explicitly derived from an optimizing framework which also incorporates the assumption of rational expectations. Such an approach has been used by Aschauer (1985). Since there is no unanimity on the applicability of rational expectations as a maintained hypothesis (see, for example, Lovell 1986), it may be useful to test the proposition both without and with the assumption. In other words, use both the ad hoc and the Aschauer approaches. Both approaches allow us to test the ‘crowding-out’ hypothesis as well as the Ricardian equivalence proposition. The ad hoc approach essentially exploits what is considered to be the appropriate definition of disposable income assuming that the Ricardian proposition holds. It is well known by now that this definition is given by total income minus government expenditure, y−g, where y is income and g is government expenditure, both in per capita real terms. Following Buiter and Tobin (1979), this may be redefined as y−t−d where t is taxes and d is budget deficit g−t, again in real per capita terms. Given this, and defining private real per capita consumption as a function of thus-defined disposable income, the ad hoc approach tests the two hypotheses by testing certain restrictions on the signs and the magnitudes of the components of disposable income. More concretely, we can show this as follows. Let ct=a(yt−tt−dt) (1) In its unrestricted form, (1) can be written as ct=a1yt−a2tt−a3dt (2) Buiter and Tobin (1979) test the proposition by testing whether |a1|=|a2|=|a3| and whether all three coefficients are statistically significant. If the two conditions are satisfied, it is taken to support the Ricardian equivalence hypothesis. Since yt−tt−dt=yt−gt, they also estimate an alternative version of (2), namely, ct=a1yt−a2gt (3)
Deficits and aggregate demand
127
and the proposition is examined by testing whether |a1|=|a2| and whether a2 is statistically significant. This version also allows a test of the ‘crowding-out’ hypothesis. Using the definition d=g−t, equation (2) can be rewritten as ct=a1yt−(a2−a3)tt−a3gt (4) Equation (4) is in accordance with Kormendi’s (1983) ‘augmented consolidated approach’. If the coefficient of t is not significantly different from zero, it is taken to imply a support for the Ricardian equivalence proposition. Once again, the coefficient of g is used to test the ‘crowding-out’ hypothesis. Boskin (1988) estimates ct=b1(yt−gt)+b2(gt−tt) (5) and tests the Ricardian equivalence proposition by examining the sign and the significance of b2. If it is found to be positive and significant, it is interpreted to refute the Ricardian equivalence proposition and support the traditional proposition in which, given g, a tax cut stimulates private consumption. It is easy to see that (5) is a special case of (2). Thus (5) can be written as ct=b1y−b2t−(b1−b2)gt (6) But (6) is precisely equivalent to (4) which is equivalent to (2). A shortcoming of Boskin’s approach is that it imposes the restriction that the effect of g and t are equal (although different in sign), a restriction, according to Kormendi (1983), not justified on any a priori grounds. In any event, it is preferable if such restrictions are treated as testable hypotheses rather than imposed. Finally, we consider the approach by Seater and Mariano (1985). Their model, which is an extended version of Barro (1983), in our notation, is (7) where superscripts P and T represent permanent and transitory magnitudes respectively. However, because of severe multicollinearity between gP and gt, they replace gt by its transitory component (gt−gP), so that their estimating equation is (8) In this equation, the crowding-out effect is measured by the sign and the significance of and the Ricardian equivalence proposition is tested by the significance of a5. The coefficient of gP gives the effect of the cost of government. It is easy to see that the appropriate form of their model is in fact (8) rather than (7), so that (8) is not the consequence of estimation problems as stated by them. Using the definitions d=g−t, y=yP+yT and g=gP+gT, we can write equation (2) as
Budget deficits and economic activity in Asia
128
(9) In its unrestricted form, it can be written as (10) But (10) is equivalent to (8). However, it is not based on any considerations of difficulties in estimation. For our purpose, as a representative of the ad hoc approaches, we use equation (10). Unlike the approaches by Buiter and Tobin, Boskin, and Kormendi, it is more in the spirit of Barro (1983) in so far as it clearly distinguishes between the effects of permanent and transitory government expenditures as well as being a more faithful representation of the modern theories of consumer behaviour. It will be estimated in two steps. First, without including t, so that we would be able to test only the ‘crowding-out’ hypothesis. In the second stage, both hypotheses will be tested by including t as well. A brief description of the alternate approach follows. It is based on an explicit model of utility maximization subject to appropriate budget constraints (Aschauer 1985). Assuming the utility function to be quadratic, and that a representative household maximizes present discounted value of utility of consumption in current and future periods, Aschauer concentrates on the Euler equation, (11) where E is the expectations operator and c* is the effective private consumption, c* is given by (12) where c gives actual private consumption and g government consumption. In this specification, government spending is allowed to affect household utility. Each unit of g is assumed to yield the same utility as θ units of private spending. The parameters α and β are explicitly derived and are shown to be non-linear functions of the discount rate and the real rate of interest. Lagging equation (12) by one period and substituting in (11), we get (13) Taking expectations of (12) gives (14) Substituting (14) in (13) gives ct=α+βct−1+βθgt−1−θEt−1gt+ut (15)
Deficits and aggregate demand
129
which is Aschauer’s equation (13). He assumes that Et−1gt is given by Et−1gt=γ+ε(L)gt−1+ω(L)dt−1 (16) where L is the lag operator and d is per capita real government deficit. Substituting (16) in (15) gives ct=(α−θγ)+βct−1+θ[β−ε(L)]gt−1−θω(L)dt−1+ut (17) The estimating model is given by equations (16) and (17) with cross-equation restrictions on the parameters. To bring out the implications of his model explicitly, we take a special case of these two equations in which only two lags of g and d are involved. In that case (16) can be written as gt=γ+ε1gt−1+ε2gt−2+ω1dt−1+ω2dt−2+vt (18) and (17) can be written as Ct=δ+βct−1+η1gt−1+η2gt−2+µ1dt−1+µ2dt−2+ut (19) where δ=α−θγ η1=θ(β−ε1) η2=−θε1 µ1=−θω1 µ2=−θω2
(20)
The cross-equation restrictions in (20) are the essence of the rational expectations approach. If these restrictions are not violated, we can conclude that the Ricardian proposition under rational expectations is supported by the data. His approach then consists of estimating equations (18) and (19) under the restrictions in (20), then estimating the unrestricted form of (19) and testing whether the restrictions in (20) are violated. The most restrictive feature of this approach, apart from the assumptions of the model, relates to equation (16) or (18) which describes the process generating the expected values of g. This equation creates two problems. The first is the arguments included in the information set. It is a moot point whether only past values of government expenditures and budget deficits are sufficient to determine Et−1g. Like the rest of the empirical literature on rational expectations, there is really no a priori satisfactory way of dealing with this issue. Consequently we follow Aschauer, but check the sensitivity of the results by following the alternate Buiter-Tobin approach, which does not embody the assumption of rational expectations. The second problem relates to the number of lag lengths used. Our problem on this score is even more acute because of the short time series and hence the ‘fewness’ of the degrees of freedom. We deal with this problem in two ways. First, we start with two lags. Given that the data used are annual, this lag is
Budget deficits and economic activity in Asia
130
deemed adequate. In the second stage, we retain θ and β, since these parameters are of crucial importance, but drop those variables whose coefficients did not exceed their own standard error in the first stage. This allows us to follow a non-uniform lag structure for each country, with respect to both g and d.
THE EVIDENCE In this section we present evidence based on both models. The variables are in real per capita terms and the time period covered is generally from 1960 to 1985. We start with the estimates of equation (10). The results are given in Table 6.1. For each country, two sets of estimates are presented: in column (1), without taxes to test for the ‘crowding-out’ hypothesis first and in column (2), with taxes to test for the ‘crowding-out’ as well as the Ricardian equivalence proposition. In all cases, permanent variables were defined as a simple three-year moving average and the corresponding transitory magnitudes were defined as the difference between the actual and the permanent estimate. In each column, the estimates were corrected for first order serial correlation. Concentrating on the estimates in column (1) for each country, we can see that the model fits reasonably well. Instead of interpreting the complete model, in this case we only consider the coefficient of transitory government expenditure, gT, which represents the crowding-out effect. We can see that the coefficient is significant for India and Indonesia, marginally significant for Malaysia, and not significantly different from zero for the other countries. Of the two countries where it is significant, it exceeds unity only for India. Thus except for India, government expenditures are a very poor substitute for private expenditure, thus suggesting little crowding-out. In terms of the results of the full model in column (2), the evidence about crowdingout changes somewhat. The coefficient of gT is no longer significant for India or Indonesia and is positive and almost significant for the Philippines and Sri Lanka, suggesting complementarity between the two types of expenditures for these two countries. Turning to the Ricardian equivalence proposition, the coefficient has the correct (negative) sign and is statistically significant only for Sri Lanka. For India, Indonesia, and the Philippines, some borderline significance is indicated. For Malaysia, although significant, the sign is wrong. Thus on the basis of strict interpretation, the Ricardian equivalence proposition is rejected
Table 6.1 Estimates of equation (10) Independent variables Constant
y
P
y
T
India
Indonesia
South Korea
0.0057
0.0058
1.2058
1.0977
1.0470
1.0398
(3.61)
(4.40)
(16.72)
(9.39)
(5.65)
(4.62)
0.4401
0.4836
−0.0661
0.0107
0.6120
0.6144
(2.86)
(3.67)
(1.08)
(0.12)
(7.26)
(6.54)
0.9418
0.9695
−0.0801
0.0642
0.2108
0.2145
(6.65)
(6.86)
(0.48)
(0.31)
(2.23)
(1.88)
Deficits and aggregate demand gP T
g
t
131
−0.4647
0.0157
1.4475
1.6642
−0.3902
−0.3741
(0.83)
(0.03)
(8.09)
(6.59)
(0.80)
(0.71)
−1.1287
−0.6345
−0.4191
−0.1018
−0.0626
−0.0533
(2.20)
(0.93)
(2.52)
(1.15)
(0.26)
(0.18)
—
−1.0902
—
−0.4534
—
−0.0252
(1.48) −2
R
(1.15)
(0.06)
0.828
0.832
0.982
0.981
0.995
0.995
DW
1.78
1.81
2.13
2.28
1.06
1.06
Independent variables
Malaysia
Constant
y
P
y
T
P
g
Independent variables T
g
t
Pakistan
The Philippines
10.3380
12.9710
0.0009
−0.0011
0.0067
0.0080
(8.32)
(8.72)
(0.026)
(0.09)
(1.83)
(2.22)
0.1083
−0.1368
0.5243
0.5620
0.7168
0.7605
(1.71)
(1.28)
(5.34)
(5.12)
(5.00)
(5.45)
−0.2638
−0.2651
1.5653
1.5266
0.7500
0.8001
(2.11)
(2.61)
(6.96)
(6.63)
(13.11)
(12.22)
0.5440
0.3624
1.3947
0.9640
−1.2882
−1.0575
(4.84)
(3.28)
(2.76)
(1.24)
(1.72)
(1.43)
Malaysia
Pakistan
The Philippines
0.0729
−0.1230
−0.2250
−0.2714
0.3122
0.7032
(0.84)
(1.12)
(0.53)
(0.64)
(1.19)
(1.90)
—
0.8761
—
0.2834
—
−0.8798
(2.82)
(0.72)
(1.44)
R−2
0.974
0.981
0.996
0.996
0.950
0.951
DW
1.83
1.89
1.91
1.90
1.11
1.16
Independent variables
Singapore
Constant
y
P
y
T
P
g
Sri Lanka
Thailand
17.0370
17.1440
0.0004
−0.0004
0.0090
0.0092
(5.68)
(5.85)
(0.19)
(0.34)
(3.31)
(3.56)
0.2421
0.2516
0.6248
0.8649
0.5726
0.5044
(2.87)
(3.06)
(5.52)
(8.64)
(5.82)
(4.65)
0.4289
0.4421
0.7207
0.6362
0.4227
0.3729
(3.39)
(3.59)
(2.13)
(2.07)
(1.37)
(1.24)
0.4273
0.2089
0.3915
0.3171
0.1220
−0.0168
Budget deficits and economic activity in Asia
T
g
t
(1.42)
(0.60)
(1.65)
(1.83)
(0.31)
(0.04)
−0.1370
−0.1088
−0.0462
0.4101
0.2438
0.3080
(0.67)
(0.38)
(0.28)
(1.96)
(0.61)
(0.79)
—
0.0887
—
−0.8571
—
−0.6136
(1.19) −2
R
DW
132
(3.19)
(1.27)
0.990
0.990
0.979
0.984
0.991
0.991
1.59
1.60
1.20
1.62
1.52
1.46
Note t values are given in parentheses
only for Sri Lanka and marginally rejected for India, Indonesia, and the Philippines. The rest of the variables generally behave as expected, although there are some exceptions as, for example, income in the case of Indonesia and permanent government expenditure in Malaysia. Before discussing any implications of these findings, we turn to the estimates of the Aschauer model based on equations (18), (19), and (20). The estimation consists of two steps. For the first step, equations (18) and (19) are estimated subject to the restrictions in (20). This provides estimates of the free parameters (α, β, θ, γ, ε1, …, εn, ω1…, ωm). These estimates help us to examine the importance of the crowding-out effect of government expenditures as well as the importance of α, β, and γ. In the second step we test the validity of the null hypothesis of Ricardian equivalence with rational expectations. This is done by calculating the appropriate log-likelihood ratio test statistic. The calculation is carried out as follows. The statistic is −2 loge(Lr/Lu) where Lr is the value of the loglikelihood function under constrained maximization and Lu is its value under unconstrained maximization. This statistic is distributed as a χ2(k) random variable with k=[2(n+m)+3]−(n+m+4) =n+m−1 degrees of freedom. If the estimated value of the test statistic is less than the critical value of the χ2 distribution at a predetermined level, then we conclude that the null hypothesis of Ricardian equivalence with rational expectations cannot be rejected. The estimates of equations (18) and (19) subject to the constraints of equation (20) as well as without the constraints are given in Table 6.2 when the number of lags for g and d is equal to two. This table also provides the hypothesized values of the unconstrained parameters which are calculated by using the estimated constrained parameters and the restrictions in (20). We first consider the estimated values of β and θ. β is statistically significant and positive in all cases and is not significantly different from unity for Indonesia, Malaysia, South Korea, Pakistan, the Philippines, Singapore, Taiwan, and Thailand. This means that, with the level of government expenditure being given, private consumption expenditure follows a random walk for these countries. This result is similar to the one reported by Aschauer for the USA. The only two exceptions to this result are India and Sri Lanka. The estimated value of θ is of greater interest since it indicates the extent of the ex ante crowding out of private consumption expenditure by government expenditure.
Deficits and aggregate demand
133
The results for the ten countries are not the same. For five countries, South Korea, Malaysia, the Philippines,
Table 6.2 Estimates of the Aschauer model: Version 1 India (1960–83) Constrained
Unconstrained
Hypothesized
α=124.64 (1.95)
δ=119.98 (1.97)
δ=119.49
β=0.65 (4.11)
β=0.65 (3.86)
β=0.69
θ=−3.12 (6.46)
η1=0.15 (0.14)
η1=1.11
—
η2=1.60 (1.67)
η2=0.31
ε1=1.00 (3.47)
µ1=−0.08 (0.06)
µ1=0.63
ε2=0.10 (0.33)
µ2=0.10 (0.09)
µ2=−0.511
ω1=0.20 (0.60)
γ=−1.45 (0.23)
γ=−1.65
ω2=−0.16 (0.52)
ε1=1.29 (5.17)
ε1=1.00
—
ε2=−0.19 (0.69)
ε2=0.10
—
ω1=0.37 (1.10)
ω1=0.20
—
ω2=−0.30 (0.91)
ω2=−0.16
Indonesia (1960–84) Constrained
Unconstrained
Hypothesized
α=841.61 (0.11)
δ=2519.6 (0.67)
δ=−1398.89
β=0.76 (1.94)
β=0.78 (2.80)
β=0.76
θ=−2.44 (1.08)
η1=1.45 (7.98)
η1=1.342
γ=918.27 (1.86)
η2=−1.03 (1.79)
η2=0.854
ε1=1.31 (8.03)
µ1=1.06 (3.65)
µ1=−1.147
ε2=−0.35 (3.10)
µ2=−1.10 (2.64)
µ2=0.488
ω1=0.47 (1.27)
γ=1894.8 (5.25)
γ=918.27
ω2=−0.20 (1.01)
ε1=0.97 (5.72)
ε1=1.31
—
ε2=0.04 (0.23)
ε2=−0.35
—
ω1= 0.78 (2.36)
ω1=0.47
—
ω2=1.62 (4.32)
ω2=−0.20
Budget deficits and economic activity in Asia
134
South Korea (1960–84) Constrained
Unconstrained
Hypothesized
α=636.10 (1.42)
δ=477.57 (1.19)
δ=394.77
β=1.04 (76.83)
β=1.01 (11.71)
β=1.04
θ=1.45 (0.60)
η1= 0.02 (0.08)
η1=0.05
γ=166.73 (1.06)
η2= 1.13 (0.41)
η2=−0.04
ε1=1.01 (5.34)
µ1=−0.22 (0.50)
µ1=−0.22
ε2=0.03 (0.13)
µ2=0.56 (1.08)
µ2=0.37
ω1=0.15 (0.50)
γ=110.35 (0.55)
γ=166.73
ω2=−0.25 (0.83)
ε1=0.92 (3.83)
ε1=1.01
—
ε2=0.17 (0.66)
ε2=0.03
—
ω1=0.18 (0.46)
ω1=0.15
—
ω2=−0.04 (0.08)
ω2=−0.25
Malaysia (1970–86) Constrained
Unconstrained
Hypothesized
α=2137.5 (1.29)
δ=6568.0 (3.62)
δ=1865.73
β=0.98 (21.77)
β=0.43 (2.43)
β=0.98
θ=0.24 (0.60)
η1=0.68 (2.15)
η1=0.05
γ=1153.0 (0.87)
η2=0.10 (0.29)
η2=1.14
ε1=0.76 (4.30)
µ1=0.51 (1.41)
µ1=−0.02
ε2=0.59 (1.58)
µ2=−0.04 (0.08)
µ2=−0.20
ω1=0.09 (0.48)
γ=−10.83 (0.01)
γ=1153.00
ω2=0.87 (1.58)
ε1=0.87 (1.60)
ε1=0.76
—
ε2=0.65 (0.90)
ε2=0.59
—
ω1=−0.09 (0.14)
ω1=0.09
—
ω2=1.36 (1.40)
ω2=0.87
Pakistan (1960–84) Constrained
Unconstrained
Hypothesized
α=−2.36 (0.69)
δ=−0.09 (0.02)
δ=−3.20
β=1.07 (37.88)
β=0.80 (4.02)
β=1.07
θ=−0.68 (1.64)
η1=0.70 (1.03)
η1=−0.18
γ=−2.13 (1.37)
η2=0.45 (9.00)
η2=0.31
Deficits and aggregate demand
135
ε1=0.80 (4.95)
µ1=0.56 (1.47)
µ1=0.39
ε2=0.45 (2.72)
µ2=−0.74 (1.88)
µ2=−0.24
ω1=0.57 (3.47)
γ=−2.13 (1.38)
γ=−2.13
ω2=−0.35 (1.98)
ε1=0.80 (5.03)
ε1=0.80
—
ε2=0.44 (2.24)
ε2=0.45
—
ω1= 0.57 (3.51)
ω1=0.57
—
ω2=−0.36 (2.17)
ω2=−0.35
The Philippines (1960–85) Constrained
Unconstrained
Hypothesized
α=8.85 (0.72)
δ=−2.89 (0.24)
δ=8.74
β=0.96 (8.03)
β=0.81 (2.79)
β=0.96
θ=−0.97 (0.74)
η1=1.28 (0.81)
η1=−0.762
γ=−0.11 (0.06)
η2=0.37 (0.20)
η2=0.97
ε1=0.18 (0.75)
µ1=3.37 (2.25)
µ1=0.01
ε2=0.99 (0.41)
µ2=−1.57 (0.82)
µ2=1.24
ω1=0.01 (0.04)
γ=−0.84 (0.50)
γ=−0.11
ω2=1.27 (4.25)
ε1=0.27 (1.16)
ε1=0.18
—
ε2=0.94 (3.96)
ε2=0.95
—
ω1=0.19 (0.81)
ω1=0.01
—
ω2=1.13 (4.00)
ω2=1.27
Singapore (1963–88) Constrained
Unconstrained
Hypothesized
α=394.24 (2.60)
δ=433.88 (1.65)
δ=374.22
β=1.03 (72.74)
β=1.01 (17.38)
β=1.03
θ=0.39 (2.35)
η1=0.05 (0.17)
η1=0.14
γ=51.37 (0.52)
η2=−0.05 (0.13)
η2=−0.17
ε1=0.68 (3.02)
µ1=−0.35 (1.53)
µ1=51.37
ε2=0.42 (1.68)
µ2=0.25 (0.82)
µ2=0.60
ω1=0.80 (4.55)
γ=51.73 (0.52)
γ=51.37
ω2=−0.43 (1.94)
ε1=0.67 (2.92)
ε1=0.68
—
ε2=0.43 (1.68)
ε2=0.42
—
ω1=0.79 (4.45)
ω1=0.80
Budget deficits and economic activity in Asia —
ω2=−0.42 (1.87)
136 ω2=−0.43
Sri Lanka (1960–84) Constrained
Unconstrained
Hypothesized
α=1.71 (0.64)
δ=−1.28 (0.97)
δ=−1.15
β=0.72 (9.22)
β=0.76 (10.25)
β=0.72
θ=−2.00 (4.09)
η1=0.42 (1.75)
η1=0.47
γ=−1.43 (0.77)
η2=0.85 (3.15)
η2=0.91
ε1=0.96 (7.51)
µ1=0.49 (1.46)
µ1=0.52
ε2=0.45 (2.41)
µ2=0.92 (2.74)
µ2=0.95
ω1=0.26 (1.27)
γ=−3.43 (1.62)
γ=−1.42
ω2=0.47 (2.22)
ε1=1.22 (1.22)
ε1=0.96
—
ε2=0.55
ε2=0.45
—
ω1=1.26
ω1=0.26
—
ω2=0.11
ω2=0.47
Taiwan (1961–84) Constrained
Unconstrained
Hypothesized
α=−6.69 (1.28)
δ=−16.21 (1.07)
δ=−6.32
β=0.96 (43.31)
β=1.00 (17.12)
β=0.96
θ=−0.08 (0.75)
η1=−0.12 (1.04)
η1=−0.02
γ=4.86 (0.81)
η2=0.16 (1.36)
η2=0.03
ε1=0.63 (2.32)
µ1=−0.04 (0.21)
µ1=0.03
ε2=0.40 (1.39)
µ2=0.19 (1.04)
µ2=0.08
ω1=0.38 (1.05)
γ=5.04 (0.85)
γ=4.86
ω2=1.09 (2.64)
ε1=0.55 (2.02)
ε1=0.63
—
ε2=0.49 (1.69)
ε2=0.40
—
ω1=0.30 (0.79)
ω1= 0.38
—
ω2=1.15 (2.52)
ω2=1.09
Thailand (1960–85) Constrained
Unconstrained
Hypothesized
α=10.13 (1.87)
δ=−3.98 (0.43)
δ=11.08
β=1.03 (59.80)
β=1.29 (9.29)
β=1.03
θ=0.21 (0.37)
η1=−0.82 (1.05)
η1=−0.12
Deficits and aggregate demand
137
γ=−4.62 (2.66)
η2=−0.18 (0.27)
η2=0.06
ε1=1.60 (6.75)
µ1=−0.83 (1.61)
µ1=−0.16
ε2=−0.31 (1.19)
µ2=0.17 (0.35)
µ2=0.04
ω1=0.76 (4.55)
γ=−4.63 (2.57)
γ=−4.62
ω2=0.21 (1.09)
ε1=1.61 (6.82)
ε1=1.60
—
ε2=−0.31 (1.22)
ε2=−0.31
—
ω1=0.75 (4.62)
ω1=0.76
—
ω2=0.22 (1.20)
ω2=0.21
Note t values in parentheses
Taiwan, and Thailand, θ is not significantly different from zero, implying that in these countries government expenditure has no crowding-out effect at all. In the case of India, Indonesia, Pakistan, and Sri Lanka, the coefficient is statistically significant or at least marginally so, but the sign is negative, thus suggesting a certain degree of complementarity between the government and private expenditures. Such possibility has often been suggested in the development literature. Only in the case of Singapore is θ positive and significant, its value being 0.39, which is not too far apart from that reported by Aschauer (1985) and Kormendi (1983) for the USA. But still, it suggests that government spending is a poor substitute for private spending even in the case of Singapore. Thus it is safe to conclude from these estimates that in all of these countries, government spending will exercise expansionary effects on aggregate demand even if the Ricardian equivalence proposition were to hold. To check whether this remarkable result is sensitive to the particular lag structure adopted for g and d, we report the results of two modifications to the estimates of Table 6.2. First, the estimates are done by assuming that there is only one lag for both g and d. For this version of the model, called Version 2, we only report the estimates of θ and the log-likelihood ratio test statistic. This is done in Table 6.4. The second variation is that the model of Table 6.2 is re-estimated by retaining θ and β, but by dropping all those coefficients which are smaller than their own standard errors. We
Table 6.3 Values of the log-likelihood ratio test statistic: Version 1 (n=m=2, k=3) Country India Indonesia South Korea
−2 loge(Lr/Lu) 6.64 18.18 0.82
Malaysia
10.08
Pakistan
2.68
Budget deficits and economic activity in Asia
138
The Philippines
7.46
Singapore
0.20
Sri Lanka
6.72
Taiwan
2.71
Thailand
5.48
Note k=degrees of freedom for the likelihood ratio test; n and m represent lags for g and d
Table 6.4 Estimates of θ and log-likelihood ratio test statistic: Version 2 (n=m=1, k=1) Country India
θ
−2 loge(Lr/Lu) −3.09
1.19
(7.06) Indonesia
−1.77
0.94
(17.10) South Korea
1.31
0.07
(0.54) Malaysia
0.41
2.12
(0.89) Pakistan
−0.08
2.08
(0.07) The Philippines
−3.45
0.18
(3.30) Singapore
0.04
0.00
(1.96) Sri Lanka
−2.22
2.32
(2.72) Taiwan
−0.08
3.27
(0.77) Thailand
0.84 (1.13)
Note n and m are the number of lags for g and d
3.00
Deficits and aggregate demand
139
call this Version 3 and its results are reported in Table 6.5. It can be see from these two tables that there is virtually no difference in the outcome of the three estimates as far as the estimated value of θ is concerned. The only slight change is in the case of Thailand where in Version 2, the coefficient is just equal to its own standard error, thus indicating some marginal significance, but still poor substitutionability of the government spending for private spending remains intact. Thus it is clear that the estimates of θ and, therefore, our inferences from it are not sensitive to the lag structure adopted in equation (19). Next, we consider the joint hypothesis of the Ricardian equiva-\ lence proposition and rational expectations. Heuristically, we could examine it by comparing the unconstrained and the hypothesized estimates of the parameters in columns (2) and (3) of Table 6.2. If these two estimates are reasonably close, we may hazard the guess that the joint hypothesis holds. However, given that we have ten countries, such casual observational inference would be a bit messy. Consequently, we turn to the statistical test discussed above. The log-likelihood ratio test statistics for Version 1 of the estimated model are given in Table 6.3. In this case, we consider the value of χ2 for k=3 at 5, 10, and 20 per cent levels of significance. The critical values at these levels are 7.81, 6.25, and 4.64 respectively. At the 5 per cent level, the joint null hypothesis is rejected for Indonesia and Malaysia only. But if we use 10 per cent critical value as the decision criterion, as done by Aschauer, then the null hypothesis is also rejected for India, the Philippines, and Sri Lanka. At the 20 per cent level, we can also add Thailand to this list. But before pursuing the implications of these results, we once again check their sensitivity to alternate lag structures on g and d. The log-likelihood ratio test statistic for one period lag for both g and d are given in Table 6.4 and those for Version 3 are given in Table 6.6. Concentrating on the 5 per cent critical value of χ2 as the decision criterion, we can see from Table 6.4 that the joint null hypothesis is not rejected for any of the countries. At the 10 per cent level, it is rejected only for Taiwan and Thailand. For Version 3, it is not rejected for any of the countries, even at the 10 per cent level. As already pointed out, a comparison of the ad hoc approach of equation (10) and the Aschauer model is not strictly valid in view of the fact that the former does not embody the hypothesis of rational expectations. However, since both try to answer the same questions raised in this chapter, a comparision may nevertheless be useful. As far as the crowding-out effect is concerned, both models yield
Table 6.5 Estimates of the modified Aschauer model: Version 3 India Constrained
Unconstrained
Hypothesized
α=114.14 (1.93)
δ=114.14 (1.92)
δ=114.14
β=0.68 (4.39)
β=0.68 (4.40)
β=0.68
θ=−3.11 (7.55)
η1=1.19 (2.43)
η1=1.19
ε1=1.06 (63.95)
η2=1.06(63.29)
η2=1.06
Indonesia
Budget deficits and economic activity in Asia Constrained
140
Unconstrained
Hypothesized
β=0.80 (6.63)
δ=4240.9 (1.02)
δ=2833.72
θ=−2.71 (9.28)
β=0.68 (2.16)
β=0.80
γ=1047.5 (2.32)
η1=1.21 (7.18)
η1=1.26
ε1=1.26 (9.04)
η2=−0.61 (1.06)
η2=−0.83
ε2=−0.31 (3.29)
µ1=0.83 (2.48)
µ1=1.20
ω1=0.44 (3.86)
γ=1183.9 (2.48)
γ=1047.48
—
ε1=1.31 (5.82)
ε1=1.26
—
ε2=−0.33 (1.51)
ε2=−0.31
—
ω1=1.10 (2.32)
ω1=0.44
South Korea Constrained
Unconstrained
Hypothesized
α=335.90 (306.97)
δ=336.44 (0.85)
δ=336.44
β=1.03 (21.24)
β=1.03 (10.72)
β=1.03
θ=−1.09 (0.40)
η1=0.05 (0.18)
η1=0.05
ε=1.07 (58.2)
η2=1.08(58.19)
η2=1.08
Malaysia Constrained
Unconstrained
Hypothesized
β=1.05 (101.0)
β=1.08 (27.0)
β=1.05
θ=0.05 (0.19)
η1=0.19 (1.50)
η1=0.02
ε1=0.66 (4.04)
η2=−0.34 (1.46)
η2=−0.04
ε2=0.86 (2.69)
µ2=−0.16 (0.58)
µ2=−0.06
ω1=1.20 (2.51)
ε1=0.94 (3.98)
ε1=0.66
—
ε2=0.57 (1.58)
ε2=0.86
—
ω1=1.25 (2.93)
ω1=1.20
Pakistan Constrained
Unconstrained
Hypothesized
β=1.05 (120.99)
β=0.87 (6.40)
β=1.05
θ=−0.65 (1.48)
η1=0.50 (0.96)
η1=−0.16
ε1=0.80 (4.80)
η2=0.31 (0.65)
η2=0.25
ε2=0.38 (1.84)
µ1=0.50 (1.36)
µ1=0.37
ω1=0.56 (3.34)
µ2=−0.69 (1.85)
µ2=−0.23
Deficits and aggregate demand
141
ω2=−0.35 (1.90)
ε1=0.81 (4.91)
ε1=0.80
—
ε2=0.36 (1.87)
ε2=0.38
—
ω1=0.56 (3.34)
ω1=0.56
—
ω2=−0.37 (2.18)
ω2=−0.35
The Philippines Constrained
Unconstrained
Hypothesized
β=0.80 (4.90)
β=0.57 (2.61)
β=0.80
θ=−1.33 (1.51)
η2=2.84 (2.15)
η2=1.58
ε2=1.19 (44.27)
µ2=1.57 (1.18)
µ2=1.9
ω2=1.42 (9.53)
ε2=1.18 (44.95)
ε2=1.18
—
ω2=1.37 (9.37)
ω2=1.42
Singapore Constrained
Unconstrained
Hypothesized
α=385.95 (2.49)
δ=445.52 (1.69)
δ=385.95
β=1.03 (71.70)
β=1.01 (17.28)
β=1.03
θ=0.39 (2.39)
η1=0.05 (0.17)
η1=0.14
ε1=0.68 (2.93)
η2=−0.05 (0.14)
η2=−0.18
ε2=0.45 (1.76)
µ1=−0.34 (1.51)
µ1=−0.30
ω1=0.77 (4.54)
µ2=0.25 (0.86)
µ2=0.18
ω2=−0.47 (2.14)
ε1=0.67 (2.88)
ε1=0.68
—
ε2=0.46 (1.79)
ε2=0.45
—
ω1=0.76 (4.51)
ω1=0.77
—
ω2=−0.46 (2.11)
ω2=−0.47
Sri Lanka Constrained
Unconstrained
Hypothesized
β=0.73 (10.97)
β=0.76 (10.22)
β=0.73
θ=−2.31 (10.82)
η1=0.36 (1.51)
η1=0.40
ε1=0.90 (11.21)
η2=0.74 (2.95)
η2=0.80
ε2=0.35 (3.28)
µ1=0.34 (1.10)
µ1=0.34
ω1=0.15 (1.26)
µ2=0.77 (2.52)
µ2=0.84
ω2=0.37 (2.94)
ε1=1.04 (3.08)
ε1=0.90
—
ε2=0.22 (0.60)
ε2=0.35
Budget deficits and economic activity in Asia
142
—
ω1=0.84 (1.78)
ω1=0.15
—
ω2=−0.30 (0.60)
ω2=0.37
Taiwan Constrained
Unconstrained
Hypothesized
β=0.93 (93.2)
β=0.94 (102.27)
β=0.93
θ=0.02 (0.47)
η1=−0.16 (1.52)
η1=0.003
ε1=0.79 (3.32)
η2=0 17 (1.51)
η2=−0.005
ε2=0.24 (0.94)
µ1=−0.14 (1.00)
µ1=−0.01
ω1=0.54 (1.75)
µ2=0.16 (0.89)
µ2=−0.02
ω2=0.91 (2.34)
ε1=0.62 (2.47)
ε1=0.79
—
ε2=0.43 (1.61)
ε2=0.24
—
ω1=0.40 (1.18)
ω1=0.54
—
ω2=1.09 (2.61)
ω2=0.91
Note t values in parentheses
Table 6.6 Values of the log-likelihood ratio test statistic for the Aschauer model: Version 3 Country
−2 loge(Lr/Lu)
India
0.00 (k=0)
Indonesia
1.95 (k=2)
South Korea
0.00 (k=1)
Malaysia
3.78 (k=2)
Pakistan
2.78 (k=3)
The Philippines
2.06 (k=1)
Singapore
0.18 (k=3)
Sri Lanka
5.06 (k=3)
Taiwan
2.39 (k=3)
virtually identical results in that they both suggest government expenditures to be a poor substitute for private expenditures in all countries, thus indicating that government expenditures would exercise an expansionary effect even if the mode of financing these expenditures was irrelevant. Regarding the Ricardian equivalence proposition, if we adopt the 5 per cent level of significance as the decision criterion, the only difference between the two sets of results is about Sri Lanka. In the ad hoc model the proposition is
Deficits and aggregate demand
143
rejected, whereas in Version 3 of the Aschauer, the proposition is upheld. But it is interesting to note that the proposition for Sri Lanka is also rejected by Version 1 of the Aschauer model at the 10 per cent level. Another interesting comparison of our results can be made with the only other study of this topic for the developing countries which I have been able to find. This one is by Haque (1987). He examines the joint null hypothesis of Ricardian equivalence with rational expectations using a variant of Blanchard’s (1985) model. His approach is to test one of the underlying assumptions of the proposition, that the government and the private sector have the same planning horizons. A refutation of this assumption is interpreted as a rejection of the joint hypothesis. His sample includes five countries also included in our sample, Indonesia, South Korea, Malaysia, Pakistan, and the Philippines. The time period covered is roughly the same as in this study. He reports that the joint hypothesis is supported in all five cases. Since the joint hypothesis is also supported for these countries by our Versions 2 and 3 of the Aschauer’s model and the Ricardian equivalence proposition without rational expectations for these countries is also supported by our ad hoc model, it may seem reasonable to conclude that the results for these countries are relatively robust and the conclusion that the Ricardian equivalence holds in these countries is warranted. As far as the other countries are concerned, we have seen that in the case of Sri Lanka the two approaches give a different answer, although for the remaining three, the Ricardian hypothesis is upheld. It is important to emphasize, however, that if we use a lower level of significance than 5 per cent, the proposition tends to become invalid in a few other countries in terms of both approaches, but particularly in terms of our Version 1 of the Aschauer model.
CONCLUDING REMARKS In this chapter we have examined the role of deficits in affecting aggregate demand. More specifically, two questions were asked. First, does government expenditure crowd out private expenditure, thus reducing or even nullifying the effect of government expenditures? Second, is the effect of government expenditure sensitive to the mode of financing this expenditure, that is, by taxes or by government debt? The evidence on the first question is unambiguous. It suggests that government expenditure is a poor substitute for private expenditure so that even if the mode of financing is irrelevant, government expenditures can be expected to exercise significant expansionary effects on aggregate demand in the countries in our sample. This evidence thus refutes the alleged fears about massive crowding-out effects in these countries. The evidence on the Ricardian equivalence proposition is less clear. Thus the evidence is reasonably unambiguous in the case of South Korea, Pakistan, Singapore, and Thailand in the sense that the proposition is supported. But for the other countries, the outcome is not so clear as already discussed. The problem arises because of the sensitivity of the estimates to the lag structure used for the g and d. But if we assume that excluding one whole year, since we have only annual data, may amount to losing more information than justified, then we need to take the results of our Version 1 of Aschauer model seriously. Once we do that we find that the results of our ad hoc approach and Version 1 lead, at least marginally, to rejection of the proposition for India, Indonesia, the Philippines, and
Budget deficits and economic activity in Asia
144
Sri Lanka. As we shall see in the next chapter, these results are also consistent with the findings using an altogether different approach. This would seem to suggest that economic agents in some of the economies in our sample may be liquidity constrained and/or myopic. The results in this chapter suggest that we cannot treat the countries in our sample as homogeneous in terms of the effects of budget deficits on aggregate demand. In other words, the timing of taxes does matter for some, but not for the others. For those countries where the Ricardian equivalence proposition holds, the problem of raising national saving for accelerated growth is even more acute, because contrary to the widespread belief that this could be achieved by reducing the budget deficits, it is clear from our findings for these countries that such a policy will achieve nothing of the sort.
7 Deficits and interest rates We saw in the previous chapter that some crowding-out effects did exist in a number of countries. Given this, the question arises as to the possible mechanism(s) which might lead to this result. The most common route suggested is via increases in interest rates caused by high and persistent budget deficits. Here two alternate schools of thought exist. The traditional, or the Keynesian, one states that budget deficits, by affecting aggregate demand, do exercise a positive effect on interest rates. But the Ricardian view, as explained in Chapter 6, postulates no such effect. Unfortunately, not only is there ambiguity in the theoretical literature, the empirical findings are inconclusive also. The situation for the developing countries is even more complicated because all of the evidence on this issue is confined to the developed countries.1 Consequently, the aim of this chapter is to make a beginning in filling this gap.
INTEREST RATE VARIATIONS It has often been claimed that since interest rates in developing countries are essentially administered, they do not require any explanation.2 But this type of claim raises two questions. First, what do the data suggest—do they show that rates have varied over time? Second, even if the rates are administered does it imply that they are not changed in response to economic factors? Clearly, if the answer to the first question is in the affirmative, then we must provide an explanation for the observed variations, even if these variations were administered. What the administered nature of interest rate implies is that they are not allowed to reach the levels of market clearing rates or react fully to economic factors or, in the language of Shaw and McKinnon,3 there is financial repression in the economy. The issue of financial repression in the countries of our sample was examined at length in Gupta (1984). That study reported that there was serious doubt whether there was significant financial repression in the countries in our sample. Or at the very least, the question whether the observed interest rates significantly deviated from market cleaning rates was an open question. Given this state of affairs, we first look at the observed behaviour of interest rates in these countries. The ideal treatment of this part should include an analysis of both short-term and long-term rates, in nominal as well as in real terms. Unfortunately, we are not in a position to undertake such an analysis due to lack of data. The data on interest rates for the developing countries are very sparse. The only consistent time series data I have for all of the countries is on the rate of interest on twelve-month time deposits.4 For a few countries, we also have data on money market rates and the government bond rates. But in order to provide comparable evidence on the
Budget deficits and economic activity in Asia
146
movement of interest rates in these countries, we only use the twelve-month interest rate in this section.
Figure 7.1 Nominal and real interest rates
Deficits and interest rates
147
The information on the nominal and the real rates is given in Figure 7.1. The real rate refers to the ex post rate measured by the difference between the nominal rate and the rate of inflation based on the GDP implicit price deflator. Using these graphs, a country by country description of the two sets now follows. India:
The nominal rate shows only marginal variations. But note that the rates are generally higher after 1975. The real rate, on the other hand, shows wide fluctuations reflecting wide variations in the rate of inflation. It is thus clear that for India neither the nominal rate nor the actual rate on the twelve-month deposits remained constant over the period covered.
Indonesia:
The behaviour of the nominal rate is quite different in this case. The rate declined steadily throughout the period with some signs of stabilization towards the end. The real rate was negative for a good part of the period, with the average rate being negative over the period.
South Korea:
The nominal rate shows considerable variations over the period, unlike both India and Indonesia. The real rate, on the average, was positive for the 1960s and negative for the 1970s.
Malaysia:
From 1970 to 1974, the nominal rate rose steadily, but remained relatively constant thereafter. The real rate, on the other hand, showed considerable variation, once again implying larger variations in the rate of inflation.
Pakistan:
The nominal rate showed a steady rise throughout the period, thus showing an altogether different behaviour than in the case of the above four countries. But as above, the real rate showed much greater variations.
The Philippines:
In this case, the nominal rate behaved just as in the case of Pakistan, namely, it rose steadily throughout the period, though at a higher rate. And once again the variations in the actual real rate were much wider.
Singapore:
The nominal rate in this case seemed to have remained relatively constant until 1973, then changed its course, rising steadily from 1980 onwards. The real rate again was more volatile, but alternated between positive and negative for well defined periods of time.
Sri Lanka:
The nominal rate of interest behaved very much like that in the cases of Pakistan and the Philippines, rising throughout the period, except that the rise was quite rapid after 1975. But interestingly enough, the actual real rate after 1975 was negative for a good part of the period.
Taiwan:
The nominal rate remained relatively unchanged until 1972, but showed some variation thereafter. The real rate was positive for most of the period, while the nominal rate was steadily rising; the real rate was negative.
Thailand:
The nominal rate showed little variation except towards the end. The real rate, on the other hand, showed much greater variation.
Keeping in mind the fact that we are dealing with only one interest rate, albeit perhaps a more representative one in terms of the consumers, a number of points are clear from the above discussion. First, in virtually no case did either the nominal or the real rate remain constant. Second, all of the countries did not display the same pattern of the nominal rate. There were sufficient differences in the experiences of the countries to make us want to ask the obvious question: what accounts for these differing experiences? In short, then,
Budget deficits and economic activity in Asia
148
we can conclude that the countries in our sample have experienced enough variations in the interest rates as to warrant an investigation of the causes of these variations. Before turning to formal analysis, we consider the descriptive evidence about the relationship between budget deficits and the nominal interest rates in the ten countries. The relevant data are shown in Figure 7.2. In these diagrams, the nominal rate again refers to the rate of interest on the twelve-month time deposits and the budget deficit (DEF) is expressed as a percentage of the GDP. Looking at these figures, it is difficult to draw any clear-cut conclusions about most of the countries. There seems to be some comovement of the two variables in India, Indonesia, the Philippines, Sri Lanka, and Taiwan. But for the other five countries, no such pattern is easily detectable. Consequently, casual empiricism does not suggest any strong relationship between the two variables. Therefore, for a more reliable indicator we turn to the quantitative analysis.
THE MODELS The basic model is an extension of the loanable funds model of Sargent (1969). Sargent’s starting point is that the nominal rate of interest can be expressed as the sum of three components, these being: an ex ante real rate of interest which equilibrates the level of desired investment and saving; the difference between the nominal rate and the market rate of interest; and the difference between the market and the real rate of interest. The determinants of these three components are then specified and a reduced form equation for the nominal rate is then derived. More specifically, he starts with the identity. rn(t)=re(t)+[rm(t)−re(t)]+[rn(t)−rm(t)] (1) when rn(t) is the nominal rate of interest; rm(t) is the market rate of interest which is also a nominal rate; and finally, re(t) is the real rate of interest which equilibrates the level of desired investment and saving. We first derive an equation for the equilibrium real rate of interest re(t). For this purpose the model is given by I(t)=a0+a1re(t)+a2∆Y(t) (2) S(t)=b0+b1re(t)+b2Y(t) (3) ∆B(t)=G(t)−T(t) (4) I(t)+∆B(t)=S(t) (5)
Deficits and interest rates
149
Figure 7.2 Nominal interest rates and budget deficits
Budget deficits and economic activity in Asia
150
where equation (2) specifies the demand for desired investment and is based on the wellknown accelerator principle along with the cost of capital, which is here proxied by the real rate of interest. Equation (3) is a simple saving function. More complicated saving functions could have been specified but lack of data precluded such an exercise. Equation (4) is the government budget constraint in the spirit of Echols and Elliott (1976). Equation (5) describes the equilibrium condition in terms of the equality of ex ante investment and ex ante saving. In these equations, all variables are in real terms where I stands for desired investment, S for desired saving, Y for real income, ∆Y for change in real income, B for government bonds, G for government expenditure, and T for government taxes. Using equations (2) to (5), we can solve for re(t) so that (6) Following Sargent, we assume that the spread rm(t)−re(t) between the market rate and the equilibrium rate is determined by the rate of growth of real money supply. Assuming linearity again, we then have rm(t)−re(t)=c0+c1m(t) c1>0 (7) where m is the rate of growth of real money stock. The last term in equation (1) is assumed to depend on the expected rate of inflation. Again assuming a linear relationship, this gives rn(t)−rm(t)=d0−d1πe(t) d1>0 (8) Combining equations (1), (6), (7) and (8), we can now solve for rn(t). This gives (9)
Equation (9) can also be specified for an open economy by redefining the equilibrium condition (5) to include the balance on current account. In the next section, wherever relevant, the results for the open economy version will also be presented. It has been argued in a number of studies that the relevant deficit variable in the above model should be its expected or future value and not just the current value.5 The incorporation of this view presents the practical difficulty in that we do not have any data on future projections of deficits for the countries in our sample. An alternative would be to assume that current deficits accurately reflect future deficits. However, such an assumption of perfect foresight is not likely to be valid. The approach we take is that the expected deficits are determined by current and past deficits and therefore we respecify equation (9) as
Deficits and interest rates
151
(10)
where L is the lag operator. Needless to say, the models specified above are relatively simple. But given the problems of data availability and the shortness of the time series used, these models adequately capture the essence of the issue being considered here.
THE ESTIMATES The models in equations (9) and (10) were estimated for nine countries. Two estimates of budget deficit were used as in the previous chapter. The first one was nominal deficit as a proportion of the trend level of output (DEF), the second one was based on the change in the real value of government debt, again expressed as a proportion of the trend level of output (RGD). As far as the dependent variable is concerned, we had even more problems. As already pointed out, data on this variable are quite sparse. Our strategy was therefore to use whatever data we could. Consequently we have done the following. We were able to find data on three rates: twelve-month time deposit rate, money market rate, and the government bond rate, but not for all countries. The source for the money market and the government bond rates was the International Financial Statistics, IMF, various issues. The sources for the twelve-month rate have already been explained. The real rate of growth of money was measured by the rate of growth of real M1. The expected rate of inflation was measured by three-year moving average of the rate of change of the GDP implicit price deflator. Y and ∆Y were measured by real GDP. The net exports are denoted by NE, again in real terms. For equation (10), two lags of the deficit were used. For each country, two sets of estimates are reported, two for each of equation (9) and equation (10). The two sets for each are accounted for by the two definitions of the budget deficit used. Except in one case, net exports either had the wrong sign or were statistically insignificant, or both. Consequently, results with NE included are not reported. Instead of discussing the results together initially, we examine them for each country first and then offer a comparative analysis. India: For India, estimates are presented for all three rates. Tables 7.1, 7.2, and 7.3 give the results for the twelve-month time deposit rate, for the money market rate, and for the government bond rate. From Table 7.1, judged by the standard criteria, the models fit reasonably well. The size of the Fisher effect is very small, about 0.1, which is far from the theoretical value of unity. The effect of real money growth, while of the correct sign, is again very small in magnitude. While the effect of the budget deficits will be discussed later on in detail, it is nevertheless interesting to see that current deficit has a positive and significant or marginally significant effect in all cases. The effects of the lagged deficits are negative, but differ in significance for the two definitions of the deficit. For the money market in Table 7.2, the effect of money supply is even less marked, but the Fisher effect, in three of the four equations, is much bigger, ranging from 0.3 to 0.4. But
Budget deficits and economic activity in Asia
152
there does not appear to be any effect of budget deficits. In fact, even the sign is wrong. The fit of the model in this case is somewhat poorer than
Table 7.1 India: twelve-month deposit rate* DEF Coefficients Constant
Y
∆Y
m
π
e
DEF, RGD**
DEF−1, RGD−1
(1)
RGD (2)
(3)
(4)
1.61
1.70
2.24
2.69
(1.00)
(1.40)
(1.90)
(2.46)
0.0020
0.0035
0.0025
0.0025
(1.41)
(2.93)
(2.17)
(1.88)
−0.0012
−0.0018
−0.0023
−0.0048
(0.73)
(1.29)
(3.17)
(3.05)
0.0474
0.0565
0.0408
0.0359
(4.25)
(5.08)
(3.12)
(3.55)
0.1019
0.1225
0.1213
0.0922
(3.21)
(4.06)
(3.29)
(1.59)
22.101
19.236
7.503
9.165
(2.22)
(2.08)
(1.57)
(1.40)
—
−26.291
—
−1.882
(2.54) DEF−2, RGD−2
—
−1.575
(0.25) —
(0.15)
−12.739 (2.52)
−2
R
0.801
0.818
0.780
0.853
See
0.366
0.322
0.385
0.290
DW
1.03
1.05
1.45
1.27
Notes * All estimates are based on first order correction for autocorrelation ** DEF refers to the ratio of nominal deficits to the trend level of nominal GDP and RGD to the ratio of the change in the real value of government debt trend level of real GDP for the twelve-month deposit rate. The results for the government bond rate are given in Table 7.3. The results in terms of the effect of expected inflation and real money growth are not too different from those for the twelve-month deposit rate. The resuls for the effect of the budget deficit are different from those for the other two rates. The current deficits have a positive and significant or marginally significant effect in all cases. But the lagged effects, unlike in Table 7.1, are positive in the case of the real deficit and negative and positive in the nominal deficit case.
Deficits and interest rates
153
What is most interesting about these results is that for the interest rate which is the least regulated, namely the money market rate, budget
Table 7.2 India: money market rate* DEF Coefficients Constant
Y
∆Y
m
π
e
DEF, RGD*
DEF−1, RGD−1
(1)
RGD (2)
(3)
(4)
−0.12
1.36
−1.66
3.58
(0.04)
(0.43)
(0.50)
(1.05)
0.0075
0.0101
0.0064
0.0136
(2.38)
(3.21)
(2.07)
(2.71)
−0.0116
−0.0130
−0.0109
−0.0143
(1.63)
(1.90)
(1.47)
(1.73)
0.0281
0.0374
0.0407
0.0082
(0.55)
(0.76)
(0.72)
(0.15)
0.3122
0.3878
0.2972
0.1253
(2.30)
(2.93)
(2.00)
(0.46)
−54.794
−50.005
−7.171
−52.483
(1.25)
(1.20)
(0.41)
(1.85)
—
−62.846
—
−68.112
(1.44) DEF−2, RGD−2
—
−40.273
(1.88) —
(0.95)
−27.604 (1.08)
−2
0.544
0.544
0.514
0.524
See
1.55
1.44
1.59
1.47
DW
1.81
1.91
1.82
1.78
R
Note * See Table 7.1 deficits do not have any effect, a result in accordance with the Ricardian equivalence proposition. Indonesia: We have rather limited results for Indonesia, because of the lack of data on interest rates and government debt. The only results, given in Table 7.4, are those for the rate on twelve-month time deposits and the nominal deficit. The Fisher effect and the liquidity effects are again quite weak. The effects of budget deficits are positive for the current period, but negative for the second lagged deficit, which is also significant. But we postpone a more detailed analysis of this for later treatment.
Budget deficits and economic activity in Asia South Korea:
154
The estimates for South Korea, given in Table 7.5, are again available only for twelvemonth time deposit rate, but this time for both measures
Table 7.3 India: bond rate* DEF Coefficients Constant
Y
∆Y
m
π
e
DEF, RGD*
DEF−1, RGD−1
(1)
RGD (2)
(3)
(4)
1.50
1.500
1.75
2.05
(4.53)
(4.47)
(5.29)
(6.68)
0.0034
0.0034
0.0036
0.0028
(10.89)
(9.70)
(11.44)
(5.25)
−0.0014
−0.0013
−0.0016
−0.0019
(1.50)
(1.42)
(1.68)
(1.68)
0.0101
0.0107
0.0059
0.0077
(1.58)
(1.63)
(0.82)
(1.06)
0.0577
0.0580
0.0635
0.1054
(3.68)
(3.56)
(3.62)
(3.61)
9.558
10.116
2.588
8.119
(1.84)
(1.86)
(1.11)
(2.36)
—
−1.832
—
7.602
(0.32) DEF−2, RGD−2
—
1.766
(1.83) —
(0.32)
0.131 (0.04)
−2
R
0.940
0.932
0.953
0.938
See
0.187
0.186
0.196
0.177
DW
1.54
1.56
1.50
1.55
Note * See Table 7.1 of the budget deficit. The most curious aspect of these results is the persistently negative and significant Fisher effect. The models were estimated with first differences but without any difference in the results in this respect. A possibility worth exploring may be to compute expected inflation differently. But is is interesting to note that Gupta (1984) reported a similar result for South Korea when expected inflation was estimated in an altogether diffferent way. In any event, the model does not on the whole seem to fit well in this case. As for the effect of budget deficits, it is always negative, while being insignificant in some cases and significant in others. More on this later.
Deficits and interest rates
155
Malaysia: For Malaysia also, we present the same set of results as for South Korea, and they are given in
Table 7.4 India: bond rate* DEF Coefficients Constant
(1)
(2) −2.06
−0.578
(0.56)
(0.26)
0.0003
0.0003
(4.35)
(7.25)
−0.0005
−0.0008
(1.07)
(2.15)
−0.0700
−0.0497
(0.94)
(1.10)
0.1913
0.1664
(1.52)
(2.14)
106.91
150.104
(1.48)
(1.03)
—
111.81
—
(4.62)
—
−59.306
—
(2.09)
R−2
0.615
0.802
See
1.55
0.91
DW
2.60
2.99
Y
∆Y
m
π
e
DEF, RGD*
DEF−1, RGD−1 DEF−2, RGD−2
Note * See Table 7.1 Table 7.6. There is virtually no liquidity or Fisher effect. Fisher effect even has the wrong sign. This simply suggests that the money market rates were not responding to expected inflation at all. Current deficit has no effect, but lagged deficits do, therefore suggesting that anticipated deficits may have an effect. Pakistan: In this case, we are able to present the results for all three rates and both measures of the deficit. Consider Table 7.7 which gives estimates for the twelve-month deposit rate. The liquidity and the Fisher effects are non-existent or weak in most cases. In a few cases, the Fisher effect is significant or marginally significant. In terms of the individual effects of the budget deficits, current deficits have a positive and significant
Budget deficits and economic activity in Asia
156
Table 7.5 South Korea: twelve-month deposit rate* DEF Coefficients Constant
Y
∆Y
m
π
e
DEF, RGD**
DEF−1, RGD−1
(1)
RGD (2)
(3)
(4)
27.709
24.95
28.72
38.307
(6.23)
(8.94)
(6.61)
(13.57)
−0.0001
0.0002
−0.0001
−0.0004
(1.62)
(2.15)
(1.63)
(1.28)
−0.0004
0.0004
−0.0006
0.0003
(1.05)
(0.95)
(1.35)
(0.90)
−0.0612
0.0169
0.0639
0.1183
(1.42)
(0.40)
(1.52)
(2.72)
−0.3412
−0.3144
0.3824
0.9530
(1.70)
(2.57)
(1.95)
(7–40)
−5.012
−106.25
−20.395
−33.325
(0.14)
(3.46)
(0.97)
(1.85)
—
−202.46
—
−96.816
(4.82) DEF−2, RGD−2
—
−73.357
(5.47) —
(1.93)
−33.257 (2.12)
−2
0.500
0.629
0.522
0.710
See
2.53
2.02
2.47
1.78
DW
1.90
1.92
1.88
2.20
R
Notes * All estimates are based on second order correction for autocorrelation ** See Table 7.1 effect when measured by the change in the real value of government debt. Some of the lagged terms are also marginally significant. The results for the money market rate are given in Table 7.8. The Fisher effect is stronger though far from unity. The effect of current deficit is either not significant or has the wrong sign. However, some of the lagged terms have significant and positive effect according to both measures of the deficit. Finally, the estimates for the government bond rate are given in Table 7.9. The liquidity effects are significant and positive as predicted, but the Fisher effect is nonexistent. The effect of the budget deficit is similar to that displayed by the money market rate.
Deficits and interest rates
157
Table 7.6 Malaysia: money market rate* DEF Coefficients
(1)
Constant
Y
∆Y
m
π
e
DEF, RGD**
DEF−1, RGD−1
RGD (2)
(3)
(4)
2.33
2.28
1.349
−0.4145
(0.86)
(0.81)
(0.54)
(0.11)
0.0001
0.0001
−0.0001
−0.0001
(2.31)
(2.50)
(3.13)
(3.39)
−0.0001
0.0003
−0.0001
0.0002
(0.45)
(1.24)
(0.09)
(0.98)
−0.0998
−0.1355
−0.0763
−0.0913
(1.57)
(1.52)
(1.35)
(1.43)
−0.2300
−0.1825
0.1769
0.0265
(1.78)
(1.30)
(1.38)
(0.10)
7.671
−8.369
4.785
9.393
(0.85)
(0.62)
(0.43)
(0.63)
—
31.085
—
24.290
(1.45) DEF−2, RGD−2
—
−20.130
(1.50) —
(0.83)
2.511 (0.14)
−2
R
0.831
0.847
0.820
0.836
See
0.916
0.870
0.946
0.901
DW
2.53
2.57
2.27
2.17
Notes * All estimates are based on OLS ** See Table 7.1 The Philippines:
We have results only for the twelve-month deposit rate in this case, as given in Table 7.10. The liquidity and the Fisher effects are strong, though very small in absolute magnitude. The effect of the budget deficit displays the same pattern as in Pakistan, namely, little effect of the current deficit, even has the wrong sign, but significant and positive effect of some of the lagged terms.
Singapore:
Here again, only the results for the twelve-month deposit rate are available, as given in Table 7.11. Little liquidity effect, but strong Fisher effect though small in magnitude, exist. Current deficit has no effect, but some of the lagged terms do. Whether that effect matters is examined below.
Budget deficits and economic activity in Asia
158
Table 7.7 Pakistan: twelve-month deposit rate* DEF Coefficients Constant
Y
∆Y
m
π
e
DEF, RGD**
DEF−1, RGD−1
(1)
RGD (2)
(3)
(4)
−1.00
−1.72
−1.49
−1.47
(0.61)
(1.42)
(0.99)
(1.26)
0.0408
0.0391
0.0432
0.0450
(4.44)
(6.10)
(4.89)
(6.46)
−0.0158
−0.0114
−0.0858
−0.0147
(1.34)
(0.79)
(0.07)
(2.20)
−0.0007
−0.0017
−0.0108
−0.4610
(0.10)
(0.21)
(1.47)
(0.01)
0.0583
0.0752
0.0508
0.0317
(1.34)
(2.04)
(1.20)
(0.90)
1.970
2.901
1.027
−0.3636
(0.73)
(0.77)
(2.28)
(0.98)
—
4.894
—
1.188
(1.35) DEF−2, RGD−2
—
5.061
(2.42) —
(1.48)
0.4371 (1.08)
−2
R
0.971
0.968
0.978
0.991
See
0.345
0.345
0.313
0.184
DW
1.81
1.85
1.79
1.30
Notes * All estimates are based on second order correction for autocorrelation ** See Table 7.1 Sri Lanka:
The estimates for the twelve-month deposit rate, corrected for second-order autocorrelation, are given in Table 7.12. Very little liquidity or Fisher effect is found in this case. But all terms in budget deficit have a positive sign and are either significant or marginally significant. At least, in terms of the individual coefficients, budget deficit seems to have the strongest effect in this case.
Thailand: For Thailand, we have estimates for the twelve-month deposit rate in Table 7.13 and for the government bond rate in Table 7.14. In both cases, there is virtually no liquidity effect. However, the Fisher effect is different in the two cases. In the case of the twelvemonth deposit rate, there is virtually none, but in the case of the government bond rate, it is both significant and
Deficits and interest rates
159
Table 7.8 Pakistan: money market rate* DEF Coefficients Constant
Y
∆Y
m
π
e
DEF, RGD*
DEF−1, RGD−1
(1)
RGD (2)
(3)
(4)
3.28
0.753
1.70
1.341
(1.81)
(0.95)
(1.43)
(1.02)
0.0198
0.0194
0.0211
0.0232
(2.14)
(4.19)
(2.99)
(3.31)
−0.0442
−0.0298
−0.0189
−0.0592
(1.29)
(0.81)
(0.41)
(1.33)
−0.0491
−0.0498
−0.0485
−0.0136
(2.75)
(2.82)
(1.79)
(0.46)
0.2509
0.2261
0.2548
0.2622
(3.64)
(4.41)
(4.37)
(3.83)
−19.477
8.027
0.4331
−1.909
(2.41)
(0.99)
(0.20)
(0.24)
—
20.106
—
−0.5136
(2.89) DEF−2, RGD−2
—
13.821
(0.24) —
(2.06)
3.966 (1.86)
−2
R
0.842
0.876
0.810
0.830
See
0.787
0.651
0.863
0.763
DW
1.58
1.69
1.77
1.69
Note * See Table 7.1 sizeable, being as high as 0.51, though, of course, still well below unity. Similar difference can be noticed in terms of the effect of budget deficit. While there does not seem to be any for the twelve-month rate, there appears to be some for the government bond rate. Just like Sri Lanka, all of the coefficients have a positive sign, though unlike Sri Lanka, not all are significant or even marginally significant. For a proper inference, we must turn to the next section.
Budget deficits and economic activity in Asia
160
Table 7.9 Pakistan: bond rate* DEF Coefficients
(1)
Constant
Y
∆Y
m
π
e
DEF, RGD*
DEF−1, RGD−1
RGD (2)
(3)
(4)
1.15
−1.12
0.728
−0.267
(0.60)
(1.00)
(0.41)
(0.10)
0.0303
0.0306
0.0312
0.0324
(3.16)
(5.50)
(3.39)
(3.93)
−0.0132
−0.0152
−0.0092
−0.0366
(0.42)
(0.56)
(0.26)
(0.97)
0.0345
0.0306
0.0352
0.0572
(2.09)
(2.24)
(1.66)
(2.28)
0.0173
−0.0212
0.0103
0.0995
(0.27)
(0.41)
(0.16)
(1.36)
−4.987
5.817
−0.2609
0.3455
(0.67)
(0.97)
(0.16)
(0.18)
—
24.625
—
3.218
(4.29) DEF−2, RGD−2
—
2.832
(1.66) —
(0.50)
3.292 (1.75)
−2
R
0.868
0.913
0.864
0.866
See
0.742
0.563
0.751
0.698
DW
1.71
1.62
1.84
1.88
Note * See Table 7.1
FORMAL TESTS FOR THE SIGNIFICANCE OF BUDGET DEFICITS ON INTEREST RATES In the previous section, we briefly commented on the individual coefficients of budget deficits. In this section, a more formal analysis of their effects is undertaken. For this purpose we concentrate on the estimates in columns (2) and (4) of the above tables which suggest that it is the future or the expected deficits that matter. For these columns, we test two null hypotheses, first that each one of the three deficit coefficients is zero, and second that their sum is zero. The first null hypothesis is tested by using the F test and the second by the t test. The relevant F and t statistics are given in Table 7.15.
Deficits and interest rates
161
Consider columns (1) and (3) of Table 7.15 for testing the first null hypothesis. Using the national accounts definition of deficits, column (1) suggests that the first null hypothesis is rejected for India, Indonesia, Pakistan, the Philippines, South Korea, Sri Lanka, and Thailand, for at least one of the interest rates. When deficit is
Table 7.10 The Philippines: twelve-month deposit rate* DEF Coefficients Constant
Y
∆Y
m
π
e
DEF, RGD**
DEF−1, RGD−1
(1)
RGD (2)
(3)
(4)
−1.21
−1.12
−1.07
−1.19
(2.58)
(3.26)
(2.46)
(3.22)
0.0596
0.0565
0.0555
0.0545
(11.50)
(13.02)
(21.93)
(25.57)
−0.0357
−0.0194
−0.0195
−0.0113
(1.88)
(1.26)
(1.37)
(0.93)
0.0342
0.0459
0.0293
0.0205
(1.93)
(3.24)
(1.76)
(1.46)
0.0472
0.0743
0.0794
0.0931
(1.14)
(2.22)
(2.71)
(3.73)
−12.739
−13.586
−10.99
−18.585
(0.98)
(1.42)
(0.89)
(1.59)
—
20.044
—
22.365
(2.28) DEF−2, RGD−2
—
−24.263
(1.89) —
(3.84)
15.346 (1.383)
−2
R
0.971
0.985
0.970
0.979
See
0.497
0.356
0.499
0.415
DW
2.03
2.13
1.97
2.34
Notes * All estimates are based on OLS ** See Table 7.1
measured by the change in real government debt, column (3) shows that the null hypothesis is rejected for India, South Korea, Pakistan, and the Philippines, although it must be noted that for South Korea the coefficients have the wrong sign in both cases. If we consider that the definition of the deficit in column (3) is the more appropriate one,
Budget deficits and economic activity in Asia
162
then the null gets a rejection in only two cases, India and the Philippines. And in case of India and Pakistan, the null is rejected for only one of the three interest rates. Unfortunately, these kinds of sensitivity results are not available for the Philippines. But it still follows that this null is not very supportive of the hypothesis that budget deficits, whether current or past, affect interest rates. But, of course, we must keep in mind the much wider rejection of the null hypothesis when the nominal definition of the deficit is used as in column (1). These results suggest that the outcome is sensitive to the interest rate and the measure of the deficit used. If we assume that future budget deficits, as approximated here by
Table 7.11 Singapore: twelve-month deposit rate* DEF Coefficients Constant
Y
∆Y
m
π
e
DEF, RGD* DEF−1, RGD−1
(1)
RGD (2)
(3)
(4)
3.49
3.41
3.76
3.23
(1.11)
(4.32)
(4.29)
(3.76)
0.0001
0.0001
0.0001
0.0001
(1.51)
(1.56)
(1.85)
(2.76)
0.0007
0.0008
0.0005
0.0004
(1.64)
(1.81)
(1.10)
(1.64)
−0.0180
−0.0194
−0.0068
−0.0059
(1.18)
(1.32)
(0.46)
(0.77)
0.1615
0.1890
0.1724
0.2226
(2.43)
(2.86)
(2.42)
(4.24)
−16.075
−10.603
−0.4963
2.834
—
13.978
—
−9.751
(1.11) DEF−2, RGD−2
—
−4.255
(5.03) —
(0.37)
4.848 (1.95)
−2
R
0.785
0.764
0.772
0.904
See
0.608
0.582
0.627
0.371
DW
1.95
1.94
1.96
1.52
Note * See Table 7.1
current and past deficits, are the appropriate variable to measure the effect of deficits on interest rates, then the second null hypothesis is the more relevant one. Consequently, we then consider the evidence in these columns which give the sum of the three coefficients
Deficits and interest rates
163
of the deficit along with the t value of the sum, so that whether the null is rejected can be tested in the standard way. Once again, we start with the nominal definition of the budget deficit, that is, column (2). The first thing to note is that the sign of the sum of the coefficients is negative in six out of the fourteen cases and contrary to our theoretical expectations. Of the other eight, the sum is positive and significant or marginally significant for India (bond rate), Indonesia (money market rate), Pakistan (all three), Sri Lanka (twelve-month rate), and Thailand (bond rate). Thus on the nominal definition of budget deficits, future budget deficits exercise some effect on interest rates in India, Indonesia, Pakistan, Sri Lanka, and Thailand. There are some changes in these results when the deficit is measured in real terms as in column (4). Note first that we do not have the results for Indonesia in this column due
Table 7.12 Sri Lanka: twelve-month deposit rate* DEF Coefficients Constant
Y
∆Y
m
π
e
DEF, RGD**
DEF−1, RGD−1
(1)
RGD (2)
(3)
(4)
−7.46
−2.62
−8.27
−6.81
(2.69)
(1.21)
(2.90)
(2.65)
0.3231
0.0823
−0.3617
−0.2676
(4.61)
(1.04)
(5.34)
(3.61)
−0.1217
−0.0550
−0.1711
−0.2450
(1.04)
(0.78)
(1.30)
(2.24)
−0.0043
0.0197
−0.0006
0.0244
(0.30)
(2.02)
(0.04)
(1.50)
−0.0571
0.1217
−0.0675
0.1628
(0.55)
(1.37)
(0.60)
(1.04)
8.833
6.996
3.654
11.946
(1.58)
(1.36)
(0.67)
(1.75)
—
16.745
—
12.676
(2.28) DEF−2, RGD−2
—
31.942
(1.54) —
(4.68)
19.346 (2.17)
−2
R
0.960
0.976
0.956
0.958
See
0.941
0.672
0.987
0.896
DW
1.86
2.12
1.90
2.10
Notes * See table 7.7
Budget deficits and economic activity in Asia
164
** See table 7.1
to lack of data on government debt. For South Korea, Malaysia, and Singapore the results are similar to those in column (2), in that deficits do not affect interest rates. For the other five countries, there are some changes. For India, deficit now affects two rates (bond rate and money market rate) rather than one, as in column (2); for Pakistan, the deficit affects only one rate (bond rate) now as against two in column (2) and at a lower level of significance; for the Philippines, unlike in column (2), the effect is now positive and marginally significant; for Sri Lanka, both definitions of the deficit suggest a positive and significant effect; and for Thailand also, the outcome is largely similar, though the level of significance has gone down. We thus see that there is some evidence of a positive effect of future budget deficits on interest rates in six of the nine countries. Only for South Korea, Malaysia, and Singapore can we say that no evidence of any such effect exists. Depending on one’s view about the appropriate level of statistical significance as the decision criterion, our results can be seen to be more favourable to the
Table 7.13 Thailand: twelve-month deposit rate* DEF Coefficients Constant
Y
∆Y
m
π
e
DEF, RGD**
DEF−1, RGD−1
(1)
RGD (2)
(3)
(4)
5.01
4.97
4.89
5.03
(5.96)
(9.53)
(6.64)
(7.89)
0.0057
0.0107
0.0062
0.0078
(2.10)
(4.57)
(2.73)
(3.38)
−0.0076
−0.0247
−0.0082
−0.0092
(0.41)
(0.99)
(0.43)
(0.46)
−0.0354
−0.0478
−0.0307
−0.0362
(1.36)
(1.38)
(1.23)
(1.42)
0.0576
−0.0308
0.0536
−0.0273
(0.88)
(0.51)
(0.70)
(0.29)
12.787
−3.831
8.570
−0.329
(1.06)
(0.28)
(0.75)
(0.03)
—
−11.728
—
−7.696
(0.89) DEF−2, RGD−2
—
−18.497
(0.73) —
(1.52)
−7.673 (0.84)
2
0.736
0.764
0.731
0.707
See
0.644
0.728
0.650
0.628
R−
Deficits and interest rates DW
1.74
165 2.02
1.80
1.87
Notes * All estimates except those of column (2) are corrected for first-order autocorrelation. Column (2) used OLS ** See Table 7.1
Ricardian equivalence proposition if we use 5 per cent as the critical level. In that case India, Pakistan, Sri Lanka, and Thailand are the only cases where, on one or the other measure of the deficits, expected deficits have a positive and significant effect on at least one of the interest rates. It is interesting to recall here the results of the last chapter. If we ignore the specific level of significance as the defining criterion, then we can see that there was some evidence to reject the Ricardian equivalence proposition for India, Indonesia, Malaysia, the Philippines, Sri Lanka, and Thailand. But with the exception of Malaysia these are also the countries for which we have suggestive evidence in this chapter to reject the Ricardian equivalence proposition. This outcome is quite striking when we note that the questions asked and the models estimated in the two chapters are quite different.
Table 7.14 Thailand: bond rate* DEF Coefficients Constant
Y
∆Y
m
π
e
DEF, RGD*
DEF−1, RGD−1
(1)
RGD (2)
(3)
(4)
−5.59
−5.51
−5.68
−5.98
(5.06)
(6.22)
(5.48)
(5.81)
0.0184
0.0171
0.0188
0.0184
(7.89)
(7.19)
(8.10)
(8.07)
−0.0016
−0.0158
−0.0087
−0.0180
(0.04)
(0.40)
(0.22)
(0.45)
−0.0317
−0.0226
−0.2402
−0.0204
(0.57)
(0.42)
(0.44)
(0.38)
0.3831
0.3716
0.4463
0.5144
(4.68)
(4.44)
(4.44)
(4.61)
26.108
24.361
30.716
31.678
(1.40)
(1.23)
(1.42)
(1.40)
—
13.466
—
20.554
(0.69) DEF−2, RGD−2
—
23.671 (1.27)
(1.02) —
3.480 (0.20)
Budget deficits and economic activity in Asia
166
R−2
0.912
0.914
0.912
0.906
See
1.378
1.285
1.380
1.338
DW
1.93
1.90
1.86
1.89
Note * See Table 7.1
CONCLUDING REMARKS This chapter has examined one possible mechanism through which budget deficits could crowd out private expenditure—via the effect of sustained budget deficits on interest rates. Using a simple loanable funds model, and alternate measures of interest rates and budget deficits, it was found that at 5 per cent or better levels of significance, future budget deficits had a positive effect on interest rates in India, Pakistan, Sri Lanka, and Thailand. At somewhat lower levels, we would also include Indonesia and the Philippines. But for the other three countries, South Korea, Malaysia, and Singapore, there is no evidence of such effect whatever. Thus it would appear that the Ricardian equivalence proposition receives substantial support for these three countries, but for the other six countries the evidence is not so strong. The findings in this chapter are broadly consistent with those of the previous chapter in so far as the support for the Ricardian equivalence proposition is concerned. Keeping in mind the various qualifications about the data and the
Table 7.15 Test statistics for restrictions on the coefficients of deficits1 DEF F values for e1=e2=e3=0 (1)
RGD t values for e1+e2+e3=0 (2)
F values for e1=e2=e3=0 (3)
t values for e1+e2+e3=0 (4)
India Money market rate
1.74 (3, 21)
−153.12 (2.12)*
1.32 (3, 21)
148.26 (1.83)
Bond rate
1.19 (3, 21)
10.05 (1.23)
2.29 (3, 21)
15.85 (1.75)*
3.89 (3, 19)*
−8.63 (0.48)
6.36 (3, 19)*
−5.40 (0.31)
8.74 (3, 12)*
102.61 (1.69)*
—
—
9.00 (3, 20)*
−382.07 (4.65)*
27.7 (3, 20)*
−163.40 (8.26)
12 month deposit rate Indonesia Money market rate South Korea 12 month
Deficits and interest rates
167
deposit rate Malaysia Money market rate
1.22 (3, 6)
2.58 (0.15)
1.00 (3, 6)
36.19 (0.89)
6.00 (3, 22)*
25.89 (2.31)*
2.08 (3, 22)
1.54 (0.33
Bond rate
6.21 (3, 22)
33.27 (2.83)*
1.19 (3, 22)
6.88 (1.61)*:
12 month deposit rate
0.98 (3, 20)
12.85 (1.57)
18.67 (3, 20)*
−1.11 (1.00)
Pakistan Money market rate
DEF F values for e1=e2=e3=0 (1)
RGD t values for e1+e2+e3=0 (2)
F values for e1=e2=e3=0 (3)
t values for e1+e2+e3=0 (4)
The Philippines 12 month deposit rate
5.71 (3, 12)
−17.81 (1.27)
3.13 (3, 12)*
19.13 (1.21)
0.95 (3, 18)
−0.88 (0.04)
12.43 (3, 18)
−1.89 (0.34)
10.27 (3, 20)*
3.48 (12.09)*
1.96 (3, 20)
43.97 (2.07)*
1.25 (3, 15)
56.16 (1.83)*
0.72 (3, 15)
44.45 (1.17)
6.31 (3, 20)*
−39.06 (3.58)*
0.75 (3, 20)
−15.69 (0.76)
Singapore 12 month deposit rate Sri Lanka 12 month deposit rate Thailand Bond rate 12 month deposit rate 1
Columns (1) and (3) give the null for the individual coefficients to be equal to zero; the F values and their degrees of freedom in parentheses are given; columns (2) and (4) give the sum of the individual coefficients and their t values in parentheses. One asterisk indicates significance at the 5 per cent level and two at the 10 per cent level
model used, a number of conclusions can be drawn from these results. First, that not all countries in the sample display the same outcome in terms of the effect of budget deficits on interest rates. It is not an accident that the three countries, South Korea, Malaysia, and Singapore, show complete support for the Ricardian equivalence proposition. At least two of these are also the most developed in the sample, with relatively well-developed capital markets. Second, the results are sensitive to the measures of interest rate and deficit used. Third, the fact that in six of the countries the evidence suggests a positive
Budget deficits and economic activity in Asia
168
effect of deficits on interest rates implies that perhaps one or more of the assumptions required for the validity of the Ricardian equivalence proposition are not quite satisfied by these countries, so that as interest rates become freer in these countries, the greater might become the effect of rising deficits. In this sense, the crowding out consequences of budget deficits in these countries might become more severe. But at the same time, the evidence for the other three countries indicates that such an outcome is not inevitable even for the developing countries.
8 Conclusions This study has been concerned with the implications of budget deficits for ten Asian economies. The ground covered, though selective, is still quite wide. This chapter is an attempt to bring together briefly the major findings and their policy implications.
MAJOR FINDINGS AND POLICY IMPLICATIONS It is best to summarize the findings in terms of the topics covered. (i) The two approaches to the sustainability of the budget deficits gave somewhat conflicting results. According to the traditional approach, with a few exceptions, the fiscal policies followed by the countries in our sample passed the test of sustainability. However, a number of these stable situations were reversed when the Hamilton– Flavin test was applied. Given the different data requirements of the two tests and, therefore, different potential measurement problems, it is difficult to tell at this stage which of the two sets of results are most reliable. However, it would be safer to err on the side of conservatism and conclude that problems of instability may well be serious in many of the countries covered and therefore, at the very least, the subject needs further investigation with more refined data. (ii) The issue of monetization of budget deficits in the developing countries is an important one. Our estimates clearly show that, with the exception of South Korea, revenue from money creation constitutes an important source of total revenue for the countries in the sample. This naturally calls for an examination of the role of budget deficits in this process. The data suggest important differences both on an inter-country basis and inter-temporal basis for a given country. These differences are sought to be explained by using different models. The basic approach is to use the Sargent-Wallace hypothesis of fiscal dominance. This hypothesis is sought to be tested by estimating a reaction function of the monetary authorities and the atheoretical approach due to King and Plosser. In both approaches, the basic motivation is the same, namely to examine the effect of budget deficits in the creation of high powered money. The results of the reaction function approach suggest a relatively strong support for the fiscal dominance hypothesis in nine of the countries, the only exception being South Korea, and possibly Indonesia for which we do not have complete results. In other words, in these countries the effects of budget deficits in the creation of high-powered money tend to be considerable. The alternate approach also supports the fiscal dominance hypothesis. In particular, our findings suggest that contemporary deficits are rapidly monetized, thus leading to increases in money supply. The dynamic effects of past deficits on money creation, on the other hand, are not that pervasive. This finding, while
Budget deficits and economic activity in Asia
170
surprising at first blush, may be explained by noting that the amount of monetizaton necessitated by current deficits alone will be more than what the economy can bear, so that the question of the monetization of the past deficits does not arise. These findings suggest that the ability of monetary authority to pursue independent monetary policy in these countries is relatively limited or, at least, severely circumscribed. (iii) The close relationship between budget deficits and reserve money noted above immediately raises the question of the nature and the extent of the relationship between money supply and budget deficits. It was mentioned that the evidence for the developed countries is a mixed one on this issue. Our findings for the Asian countries covered do not provide an unequivocal answer either. Certainly, on the basis of the results reported, it is difficult not to conclude that budget deficits do not exercise a strong influence on the growth of money supply, particularly that of M3. This lack of a firm relationship is contrary to the widespread belief in these countries. It may be that the results of quarterly data will show stronger results. But given the preponderance of the evidence presented, we must conclude that there is at least a need for a fresh look at the role of budget deficits on money supply in these countries rather than perpetuating the belief that deficits are always expansionary in their effects on money supply. Our results in (ii) and (iii) also point to the importance of the two-stage process in analysing the effect of budget deficits, namely, on reserve money growth and then on money supply. Our results also warn against the mechanical application of the money multiplier analysis to the determination of money supply in developing countries. (iv) Regarding the effects of budget deficits on inflation, no blanket statement can be made about all of the ten countries. However, there seems to be somewhat greater evidence of indirect effect than the direct effect. But it should be reiterated that to the extent that all of the increases in money supply cannot be attributed to budget deficits, all of the indirect effects via money growth cannot be assigned to budget deficits either. It is also interesting to note that a purely monetarist explanation alone, that is, the indirect effects of budget deficits, is not adequate to explain the behaviour of inflation in the sample of the countries over the period covered. A surprising finding is that both effects are the weakest in India and South Korea. The finding with regard to India is all the more surprising because of the widespread belief to the contrary. (v) The evidence on the effects of budget deficits on the crowding out of private expenditure is unambiguous. It suggests that the crowding out is very partial, so that even if the mode of financing these expenditures is irrelevant, government expenditures can be expected to exercise significant expansionary effects on aggregate demand in the countries of our sample. Thus evidence thus refutes any fears of massive crowding out and clearly establishes the role of an activist fiscal policy in these countries. The evidence on the neutrality of the mode of finance, that is, on the Ricardian equivalence proposition is, however, mixed. Thus, the evidence is reasonably unambiguous in the case of South Korea, Pakistan, Singapore, and Thailand in the sense that the proposition is supported. But for the other countries the outcome is not so clear. Some versions of the models lead to a marginal rejection of the proposition for India, Indonesia, the Philippines, and Sri Lanka. These latter results suggest that economic agents in these economies may be liquidity constrained and/or be myopic.
Conclusion
171
The results of Chapter 6 thus suggest that we cannot treat the ten countries as being homogeneous in terms of the effects of budget deficits on aggregate demand. In other words, the timing of taxes does matter in some, but not in other, countries. (vi) The evidence on the effects of budget deficits on interest rates and thus indirectly on the Ricardian equivalence, is broadly consistent with the findings in (v) above. It turns out that future budget deficits have a positive effect on interest rates in India, Pakistan, Sri Lanka, and Thailand at the 5 per cent level of significance. At some lower level, we could also include Indonesia and the Philippines. But for South Korea, Malaysia, and Singapore, there is no evidence of such effect whatever. In other words, the Ricardian equivalence proposition is supported for these three but not for the six, thus broadly supporting the findings in (v). These results suggest that if the financial markets in the six countries were to be completely deregulated and interest rates allowed to be determined freely by market forces, budget deficits may lead to even greater effects on the interest rates. Were this to be the case, it could have serious consequences for the sustainability results discussed earlier. The results reported in the previous chapters and summarized above are by no means definitive. But they do suggest that the two extreme views about the effects of budget deficits, that either they do not matter or that they have devastating effects, are not supported. At the same time, the results are sufficiently reliable to warrant the conclusion that budget deficits have had some deleterious economic effects in a number of countries in our sample, and further that in some of them it may be difficult to sustain current levels of budget deficits and therefore a restructuring of fiscal policies may be necessary. Before we could go any further in making more reliable judgements about the effects of budget deficits, better data must be made available, in particular on budget deficits and government debt.
Notes INTRODUCTION 1 See Yellen (1989). 2 See, for example, Minhas (1987).
CHAPTER 2 1 This treatment follows Tobin (1982). 2 See Dornbusch, Fischer, and Sparks (1989). 3 For evidence on the extent of financial repression in the countries being considered here, see Gupta (1984). 4 Hamilton and Flavin justify the error terms as picking up ‘expected changes in real short-term interest rates, the term structure of low rates, and measurement errors’ (1986:815). 5 In this context, it is interesting to note that Wilcox (1989) reported a complete reversal of the results reported by Hamilton and Flavin (1986) when the discounted values of the variables were used.
CHAPTER 3 1 For a discussion of these measures, see King and Plosser (1985), Fischer (1982), Auernheimer (1974), among others. 2 Fischer (1982). 3 King and Plosser (1985). 4 Minhas (1987). 5 Giannaros and Kolluri (1985), Demopoulos et al. (1987), Barro (1978), Niskannen (1978), among others. 6 Levy (1981), Demopoulos et al. (1987), and references cited therein. 7 This specification is based on Levy (1981), although he does not start by distinguishing between desired and actual levels or use an explicit adjustment function to relate the two. 8 Levy (1981), Demopoulos et al. (1987), among others. 9 Levy (1981). 10 Levy (1987).
CHAPTER 4 1 See King and Plosser (1985), Dornbusch and Fischer (1981), Barro (1978), Levy (1981), Niskanen (1978), Hamberger and Zwick (1981), Willet and Laney (1978), Akhtar and Wilford (1979), Demopoulos et al. (1987), among others. 2 See Bond, Shearer and Chant (1981) for a discussion of this point.
Notes
173
3 References in note (1) above. 4 For example, Minhas (1987).
CHAPTER 6 1 See Liederman and Blejer (1988) for an excellent review.
CHAPTER 7 1 See, for example, Cebula (1987, 1988, 1989), Echols and Elliot (1976), Evans (1985, 1987), Gupta (1989), among others. 2 The seminal work on administered interest rates and their implications is by McKinnon (1973) and Shaw (1973). But also see Gupta (1984). 3 McKinnon, op. cit.; Shaw, op. cit. 4 The data are from Fry (1981) and Gupta (1984). 5 Evans (1987), Feldstein (1986), among others.
Selected bibliography Aghveli, B.B. and Khan, M.S. (1978) ‘Government deficits and the inflationary process in developing countries’, IMF Staff Papers 25, 3:383–415. Akhtar, M.A. and Wilford, D.S. (1979) ‘The influence of the United Kingdom’s public sector deficit on its money stock. 1963–76’, Bulletin of Economic Research 31, 1:3–13. Aschauer, D.A. (1985) ‘Fiscal policy and aggregate demand’, The American Economic Review 75, 1:117–27. Auernheimer, L. (1974) ‘The honest goverment’s guide to the revenue from the creation of money’, Journal of Political Economy 82, 3:598–606. Barro, R.J. (1974) ‘Are government bonds net wealth?’ Journal of Political Economy 82, 6:1095– 117. ——(1978) ‘Comments from an unreconstructed Ricardian’, Journal of Monetary Economics 4, 4:569–81. ——(1978) ‘The impact of social security on private saving’, The American Enterprise Institute Studies, 199. ——(1983) Macroeconomics, New York: Wiley. Barth, J.G. and Russek, F. (1986) ‘The economic consequences of federal deficits: an examination of the net wealth and instability issues’, Southern Economic Journal 52, 1:27–50. Bhalla, S.S. (1981a) ‘The transmission of inflation into developing economies’, in W.R.Cline and Associates (eds) World Inflation and the Developing Countries, Washington, DC: Brookings Instutition. ——(1981b) ‘India’s closed economy and world inflation’, in W.R.Cline and Associates (eds) World Inflation and the Developing Countries, Washington, DC: The Brookings Institution. Blinder, A.S. (1983) ‘On the monetization of deficits’, in L.H.Meyer (ed.) The Economic Consequences of Government Deficits, Boston: Kluwer-Nijhoff Publishing Company. Blanchard, O.J. (1985) ‘Debt, deficits and aggregate demand’, Journal of Political Economy 93, 2:223–47. ——(1984) ‘Current and anticipated deficits, interest rates and economic activity’, European Economic Review 25, 1:7–27. Boskin, M.J. (1988) ‘Concepts and measures of federal deficits and debt and their impact on economic activity’, in K.J.Arrow and M.J.Boskin (eds) The Economics of Public Debt, London: Macmillan. Buchanan, J.M. (1958) ‘Barro on the Ricardian equivalence theorem’, Journal of Political Economy 84, 2:337–42. Buiter, W. and Tobin, J. (1979) ‘Debt neutrality: a brief review of doctrine and evidence’, in G.M.von Furstenberg (ed.) Social Security versus Private Saving, Cambridge, Massachusetts: Ballinger Publishing Company. Burdekin, R.C.K. and Laney, L.O. (1988) ‘Fiscal policy making and the Central Bank instutitional constraint’, Kyklos 41, 4:647–62. Carmichael, J. (1982) ‘On Barro’s theorem of debt neutrality: the irrrelevance of net wealth’, The American Economic Review 72, 1:202–13. Cebula. R.J. (1987) ‘Federal deficits and the real rate of interest in the United States’, Public Choice 58, 1:97–100. ——(1988) ‘Federal government budget deficits and interest rates: an empirical analysis for the United States 1955–84’, Public Finance 43, 3:337–48.
Selected bibliography
175
——(1989) ‘A brief empirical note on the federal budget deficits and the yield curves in the United States’, Public Finance 44, 2:316–19. Darby, M. (1984) ‘Some pleasant monetarist arithmetic’, Quarterly Review Federal Reserve, Bank of Minneapolis: 15–20. Darrat, A.F. (1986) ‘Money, inflation, and causality in the North African countries: an empirical investigation’, Journal of Macroeconomics 8, 1:87–103. Demopoulos, G.D., Katsimbris, G.M., and Miller, S.M. (1987) ‘Monetary policy and Central Bank financing of government budget deficits’, European Economic Review 31, 5:1023–50. Diamond, P. (1965) ‘National debt in a neoclassical growth model’, The American Economic Review 55, 5:1126–50. Dickey, D.A. and Fuller, W.A. (1981) ‘Likelihood ratio statistics for autoregressive time series with a unit root’, Econometrica 49, 4:1057–72. Dornbusch. R. and Fischer, S. (1981) ‘Budget deficits and inflation’, in M.J. Flanders and A.Razin (eds) Development in an Inflationary World, New York: Academic Press. Dornbusch, R., Fischer, S., and Sparks, G.S. (1989) Macroeconomics, Toronto: McGraw-Hill Ryerson Limited. Dwyer, G. (1982) ‘Inflation and deficits’. Economic Inquiry 20, 3:315–29. Echols, M.E. and Elliott, J.W. (1976) ‘Rational expectations in a disequilibrium model of the term structure’. The American Economic Review 66, 1:28–44. Edwards, S. (1983) ‘The short-run relation between growth and inflation in Latin America: comment’, The American Economic Review 73, 3:477–82. Eisener, R. and Pieper, P.J. (1984) ‘A new view of the federal debt and budget deficits’, The American Economic Review 74, 1:11–29. Evans, P. (1985) ‘Do large deficits produce high interest rates’, The American Economic Review 74, 1:68–87. ——(1987) ‘Do budget deficits raise nominal interest rate?: evidence from six countries’. Journal of Monetary Economies 20:281–300. Feldstein, M. (1982) ‘Government deficits and aggregate demand’, Journal of Monetary Economies 9 1:1–20. Feldstein, M.S. (1986) ‘Budget deficits, tax rules, and real interest rates’, Working Paper No. 1970, Cambridge, MA: National Bureau of Economic Research. Ferguson, B.L. and Gupta, K.L. (1979) ‘On the dynamics of inflation and unemployment in a quantity theory framework’, Economica 46:51–9. Fischer, S. (1982) ‘Seigniorage and the case for a national money’, Journal of Political Economy 90, 2:295–313. Fry, M.J. (1981) ‘Interest rates in Asia’, mimeo. Fuller, W.A. (1976) Introduction to Statistical Time Series, New York: John Wiley. Giannaros, D.S. and Kolluri, B.R. (1985) ‘Deficit spending, money and inflation: some international empirical evidence’, Journal of Macroeconomics 7, 3:401–18. Gupta, K.L. (1984) Finance and Economic Growth in Developing Countries, London: CroomHelm. (1989) ‘Budget deficits and interest rates in the United States’, Public Choice 60, 1:87–92. Hall, R.E. (1978) ‘Stochastic implications of the life-cycle permanent income hypothesis: theory and evidence’, Journal of Political Economy, 86, 6:971–88. Hamberger, M. and Zwick, B. (1981) ‘Deficits, money, and inflation’, Journal of Monetary Economies 7, 2:141–50. Hamilton, J.D. and Flavin, M.A. (1986) ‘On the limitations of government borrowing: a framework for empirical testing’, The American Economic Review 76, 4:808–19. Hanson, J.A. (1980) ‘The short-run relationship between growth and inflation in Latin America’, The American Economic Review 70, 5:972–89. Haque, N.U. (1987) ‘Fiscal policy and private saving behavior: tests of Ricardian equivalence in some developing economies’, IMF Working Paper.
Selected bibliography
176
Harberger, A.C. (1976) ‘A primer on inflation’, Journal of Money, Credit and Banking 10, 4:505– 21. Hendry, D.F., Pagan, A.R., and Sargan, J.D. (1984) ‘Dynamic specifications’, in Z.Griliches and M.D.Intrilligator (eds) Handbook of Econometrics, Amsterdam: North-Holland Publishing Company. Hubbard, G.R. and Judd, K.L. (1986) ‘Liquidity constraints, fiscal policy, and consumption’, Brookings Papers on Economic Activity 1:1–50. King, R.G. and Plosser, C.I. (1985) ‘Money, deficits, and inflation’, Carnegie-Rochester Conference Series on Public Policy 22:147–96. Kochin, L.A. (1974) ‘Are future taxes anticipated by consumers?’ Journal of Money, Credit and Banking 6, 4:385–94. Kolluri, B.R. and Giannaros, D.S. (1987) ‘Budget deficits and short-run real interest rate forecasting’, Journal of Macroeconomics 9, 1:109–25. Kormendi, R.C. (1983) ‘Government debt, government spending and private sector behavior’, The American Economic Review 73, 5:994–1010. Laney, L. and Willett, T. (1983) ‘Presidential politics, budget deficits, and the money supply in the United States: 1960–1976’, Public Choice 40, 1:53–69. Leiderman, L. and Blejer, M.I. (1988) ‘Modelling and testing Ricardian equivalence’, IMF Staff Papers 35, 1:1–35. Leiderman, L. and Razin, A. (1988) ‘Testing Ricardian equivalence in an intertemporal stochastic model’, Journal of Money, Credit and Banking 20, 1:1–22. Levy, M. (1981) ‘Factors affecting monetary policy in an era of inflation’, Journal of Monetary Economics 7, 3:351–73. Lovell, M.C. (1986) ‘Tests of the rational expectations hypothesis’, The American Economic Review 76, 1:110–24. Lucas, R.E. (1973) ‘Some international evidence on output inflation tradeoffs’, The American Economic Review 63, 3:326–34. Masera, R. (1987) ‘Four arguments for fiscal recovery in Italy’, in M.J. Boskin, J.S.Flemming, and S.Gorini, Private Saving and Public Debt, Oxford: Basil Blackwell. McKinnon, R.I. (1973) Money and Capital in Economic Development, Washington, DC: Brookings Institution. Minhas, B.S. (1987) ‘The planning process and the annual budgets: some reflections on recent Indian experience’, Indian Economic Review 22, 2:115–47. Mitchell, D.W. (1988) ‘The feasibility of perpetual deficits’, Journal of Macroeconomics 10, 3:407–19. Nachane, D.M. and Nadkarni, R.M. (1985) ‘Empirical testing of certain monetarist propositions via causality theory: the Indian case’, Indian Economic Journal 32, 1:13–41. Naqui, S.N.H., Khan, A.H., Khilji, N.M., and Ahmad, A.M. (1983) The P.I.D.E. Macroeconometric Model of Pakistan’s Economy, Islamabad: Pakistan Institute of Development Economics. Niskanen, W. (1978) ‘Deficits, government spending, and inflation’, Journal of Monetary Economics 4, 3:591–602. Onis, Z. and Ozmucur, S. (1990) ‘Exchanges, rates, inflation, and money supply in Turkey’, Journal of Development Economies 32, 1:133–54. Park, Y.C. (1973) ‘The ability of the monetary authorities to control the stock of money in LDCs’, IMF Staff Papers 20, 2:379–418. Pierce, D.A. and Haugh, J.D. (1977) ‘Causality in temporal systems—characterizations and a survey’, Journal of Econometrics 5, 3:265–94. Protopapadakis, A.A. and Siegel, J.J. (1987) ‘Are money growth and inflation related to government deficits? Evidence from ten industrialized economies’, Journal of International Money and Finance 6, 1:31–47.
Selected bibliography
177
Ramachandran, V.S. (1983) ‘Direction of causality between monetary and real variables in India: an empirical result’, Indian Economic Journal 30, 1:65–74. Ricardo, D. (1951) ‘Funding system’, in P.Sraffa (ed.) The Works and Correspondence of David Ricardo, Volume IV, Pamphlets and Papers 1815–1823, Cambridge: Cambridge University Press. Saini, K.G. (1982) ‘The monetarist explanation of inflation: the experience of six Asian countries’, World Development 10, 10:871–84. Sarantis, N. (1985) ‘Fiscal policies and consumer behaviour in Western Europe’, Kyklos 38, 2:233– 48. Sargent, T.J. (1969) ‘Commodity price expectations and the interest rate’, Quarterly Journal of Economics 83, 1:127–40. Sargent, T. and Wallace, N. (1981) ‘Some unpleaseant monetarist arithmetic’, Quarterly Review, Federal Reserve Bank of Minneapolis, 5, 3:1–18. Seater, J.J. and Mariano, R.S. (1985) ‘New tests of the life cycle and tax discounting hypothesis’, Journal of Monetary Economies 15, 2:195–215. Seater, J. (1982) ‘Are future taxes discounted?’ Journal of Money, Credit and Banking 14, 1:76–83. Shaw, E.S. (1973) Financial Deepening in Economic Development, New York: Oxford University Press. Shearer, R.A., Chant, J.F., and Bond, D.E. (1981) The Economics of the Canadian Financial System, Scarborough: Prentice-Hall Inc. Sheehey, E.T. (1980) ‘Money, income and prices in Latin America’, Journal of Development Economies 7, 3:345–57. Siddiqui, A. (1989) ‘The causal relation between money and inflation in a developing economy’, International Economic Journal 3, 2:79–96. Siegel, J. (1979) ‘Inflation induced distortions in government and private saving statistics’, Review of Economies and Statistics 61, 1:83–90. Sims, C.A. (1972) ‘Money, income, and causality’, The American Economic Review 62, 4:540–52. Spaventa, L. (1988) ‘Introduction: is there a public debt problem in Italy?’ in F.Giavazzi and L.Spaventa (eds) High Public Debt: The Italian Experience, Cambridge, UK: Cambridge University Press. Tanner, J.E. (1979) ‘An empirical investigation of tax discounting’, Journal of Money, Credit and Banking 11, 2:214–18. Tobin, J. (1982) ‘Discussion’, Saving and Government Policy, Proceedings of a Conference, New Hampshire: Federal Reserve Bank of Boston. Turnovsky, S.J. and Wohar, M.E. (1987) ‘Alternative modes of deficit financing and endogenous monetary and fiscal policy in the USA 1923–1982’, Journal of Applied Econometrics 2, 1:1–25. von Furstenberg, G.M. (1983) ‘The uncertain effects of inflationary finance on growth in developing countries’, Public Finance 38, 2:232–65. Wilcox, D.W. (1989) ‘The sustainability of government deficits: implications of the present-value borrowing constraint’, Journal of Money, Credit and Banking 21, 3:291–307. Willett, T.D. and Laney, L.O. (1978) ‘Monetarism, budget deficits, and wage push inflation: the case of Italy and the UK’, Banca Nationale del Lavoso Quarterly Review 31, 127:315–21. Yellen, J.L. (1989) ‘Symposium on the budget deficit’, The Journal of Economic Perspectives 3, 2:7–93.
Index Aghveli, B.B., 112–14, 191 Ahmad, A.M., 194 Akhtar, M.A., 190 Aschauer, D.A., 139–40, 142–43, 147, 151, 153, 156–8, 191 Asia, 1, 110, 185 Aurenheimer, L., 189, 191 Barro, R.J., 4, 39, 138–9, 141–2, 190–1 Barth, J.G., 39, 45, 191 Bhalla, A.S., 112–16, 120, 122, 191 Blanchard, O.J., 191 Blejer. M.I., 190, 193 Blinder. A.S., 191 Bond, D.E., 190–1 bond rate, 167–8, 170–2, 176, 178, 182–3 Boskin, M.J., 139–40, 142, 191 Brazil, 113–14 Buchanan, J.M., 139, 192 budget deficits, 1–6, 24 nominal, 2, 4, 7–8, 10, 14–15, 24, 31–2, 34, 84, 91, 105, 117, 185; primary, 8–9, 45; real, 2, 7, 14–15, 31, 34; and aggregate demand, 1–2, 138, 157–8; and inflation, 1–2, 30, 36–7, 83, 99, 111, 113, 117, 187; and interest rates, 2, 5, 99, 111, 159, 163, 164, 167–8, 170–82, 187; and monetization, 1–3, 59, 62, 66, 95, 97–9, 111, 185–6; and money growth, 1, 3–4, 36, 68, 99–100, 103–4, 106–8, 110, 113, 117–19, 122, 132, 136; and reserve money, 32, 34, 37, 63, 68, 77–8, 97, 186; and seigniorage, 3–4, 59, 71, 78, 83, 94, 96, 99; their sustainability 1–3, 24, 36, 45, 48, 55, 58, 185, 188 budget surplus, 32 Buiter, W., 139–40, 142–3, 192 Burdekin, R.C.K., 192 Carmichael, J., 192 causality, 112–15, 118 Cebula, R.J., 190, 192 Chant, J.F., 190–1 Colombia, 113, 115 crowding out effect, 45, 138, 144, 151, 153, 156–7, 159, 181, 184, 187; and fiscal policy, 4, 138
Index
179
Darby, M., 39, 192 Darrat, A.F., 112–15, 192 Demopolous, G.D., 189–90, 192 Diamond, P., 192 Dickey, D.A., 56–7, 192 Dominican Republic, 113, 115 Dornbusch, R., 104, 109, 112, 115, 120, 189–90, 192 Dwyer, G., 192 Echols, M.E., 165, 192 economic growth, 1, 30, 34, 37 Eisener, R., 58, 192 Elliot, J.W., 165, 190, 192 error-correction model, 128–9 Evans, P., 192 expectations formation, 2 expenditures, 2, 10–11 Feldstein, M., 139, 192 Ferguson, B.L., 70, 193 financial repression, 159–60 fiscal dominance hypothesis, 39, 68, 71, 83, 91, 94, 97–8, 111, 196 Fischer, S., 104, 109, 112, 115, 120, 189–90, 193 Fisher effect, 167, 170–4 Flavin, M.A., 55–6, 58–9, 193 Fry, M.J., 190 Fuller, W.A., 56–7, 192–3 Giannaros, D.S., 189, 193 government budget constraint, 3, 14–25, 24, 38–9, 70, 83, 111 Guatemala, 115–16 Gupta, K.L., 70, 159, 170, 193 Hall, R.E., 193 Hamberger, M., 190, 193 Hamilton, J.D., 55–6, 58–9, 193 Hanson, J.A., 193 Haque, N.U., 157, 193 Harberger, A.C., 193 Haugh, J.D., 113, 194 Hendry, D.F., 121, 193 high powered money, 4 Hubbard, G.R., 193 India, 1, 7, 9, 11, 29–30, 34, 36, 45, 48, 61, 63, 66–7, 77–8, 83, 102, 106, 108, 113–14, 116–17, 122–3, 126, 129–32, 136–7, 144–8, 151–2, 154–6, 158, 160, 163, 167–8, 170–1, 176–8, 180–2, 187 Indonesia, 1, 8–9, 29–30, 32, 36, 45, 48, 57, 62–3, 66, 76–7, 83, 102, 106, 108, 118, 122–3, 125–6, 129–30, 144–8, 151–4, 156–8, 160, 163, 169, 176, 178, 180–2, 188 inflation, 1, 24, 32, 78 expected, 167, 170
Index
180
inflation tax, 2 interest rates, 5, 159 ex ante, 163 ex post, 160 market clearing, 160 nominal, 5, 159–63 real, 5, 159–63 Israel, 115–16 Judd, K.L., 193 Katsimbris, G.M., 192 Khan, A.H., 194 Khan, M.S., 112–14, 191 Khilji, N.M., 194 King, R.G., 14, 62, 69, 83, 97, 119, 121–2, 186, 189–90, 193 Kockin, L.A., 193 Kolluri, B.R., 189, 193 Kormendi, R.C., 139–40, 142, 151, 193 Laney, L., 190, 193 Levy, M., 69, 76, 78, 189–90, 194 Libya, 113–14 Liederman, L., 190, 193 likelihood ratio test, 56, 147, 151–3, 156 Lovell, M.C., 139, 194 Lucas, R.E., 119, 194 Malaysia, 1, 7, 9, 11, 15, 29–30, 34, 36, 45, 54, 57, 63, 71, 77, 83, 91, 96–7, 106–9, 116, 118, 123, 126–7, 129–36, 144–5, 149, 152–4, 156–7, 162, 170, 173, 179, 180–2, 184, 188 Masera, R., 14–15, 194 Mariano, R.S., 139–40, 195 McKinnon, R.I., 159, 194 Miller, S.M., 192 Minhas, B.S., 112, 137, 189–90, 194 Mitchell, D.W., 194 models, 1; non-structural, 4, 69; single equation, 1, 11; structural, 4, 69, 91, 95, 112 money supply, 2, 30 money multipliers, 4, 100, 102–3, 110, 187 monetarist hypothesis, 114, 116 money market rate, 167, 171–3, 178, 182 monetary base, 101, 103–4, 107, 109 Morocco, 113–14 Nachane, D.M., 112–13, 194 Nadkarni, R.M., 112–13, 194 Naqui, S.N.H., 194
Index
181
Niskannen. W., 120, 190, 194 Onis, Z., 112–15, 194 Ozmucur, S., 112–15, 194 Pagan, A.R., 193 Pakistan, 1, 7, 9, 11, 29, 32, 34, 36, 45, 54, 62–3, 66, 77, 83, 102, 106–9, 112–13, 116, 118, 124, 126, 127, 129, 131–6, 145, 147, 149, 151–2, 155–8, 162, 171, 173, 178, 180–2, 187–8 Park, Y.C., 194 Philippines, 1, 7, 9, 15, 29, 34, 36, 45, 54, 62–3, 66, 77, 83, 91, 102, 106, 109, 116, 118, 124, 126– 7, 129–36, 144–5, 147, 149, 152–3, 155–8, 162–3, 173, 176–7, 180–3, 187–8 Pierce, D.A., 113, 194 Piper, P.J., 58, 192 Plosser, C.I., 14, 62, 69, 83, 97, 119, 121–2, 186, 189–90, 193 present value borrowing constraint, 55, 58, 111 Protopapadakis, A.A., 194 public debt, 1, 3, 8–9, 29 Ramachandran, V.S., 112–13, 194 rational expectations, 139, 147, 157 Razin, A., 194 reserve money, 30, 78; estimates, 63, 66 revenues, 2, 10, 11 Ricardian equivalence proposition, 4–5, 138–41, 143–4, 147, 151, 153, 156–8, 169, 180–1, 184, 187 Ricardo, D., 139, 194 Russek, F., 45, 191 Saini, K.G., 112, 114–16, 120, 194 Sarantis, N., 194 Sargan, J.D., 193 Sargent, T.J., 39, 68, 71, 97, 104, 106, 111, 122, 186, 194 Seater, J.J.. 139–40, 195 seigniorage, 2, 83–4; determinants, 68; estimates, 61–3 Shaw, E.S., 159, 195 Shearer, 190–1 Sheahey, E.T., 195 Siddiqui, A., 112–13, 195 Siegel, J.J., 194 Sims, C.A., 112–13, 195 Singapore, 1, 7–8, 29, 34, 36, 45, 54 77, 83, 91, 94, 102–3, 106–7, 109, 118, 124, 126, 128–33, 146–7, 150–2, 155–6, 158, 162, 173, 178–9, 181, 183–4, 187–8 South Korea, 1, 7–8, 11, 15, 29–30, 32, 34, 36, 45, 54, 57, 62, 66, 71, 76–8, 83–4, 91, 102–3, 106– 8, 116, 118, 122–3, 126, 129, 131–7, 145, 147–8, 152, 154, 156–8, 160, 169, 172, 176–7, 179, 181– 2, 184–8 Sparks, G.S., 190–1 Spaventa, L., 195
Index
182
Sri Lanka, 1, 7–9, 11, 14–15, 29–30, 32, 34, 36, 45, 54, 61–3, 66, 77–8, 91, 94, 96–7, 102, 107, 109, 115–16, 118, 124, 126, 128–36, 144, 146–7, 150–1, 155–8, 162–3, 175–6, 178–81, 183, 187–8 Taiwan, 1, 106, 116, 122, 125–6, 128–30, 132–3, 136, 147, 150–3, 156, 162–3 Tanner, J.E., 195 Thailand, 1, 7–9, 11, 15, 29, 34, 36, 45, 55, 62–3, 77, 83, 91, 102, 109, 113, 115–16, 118, 125–6, 128–36, 146–7, 151–3, 158, 162, 174, 176, 178–81, 183, 187–8 time deposit rate, 167–9, 173–4, 178, 182–3 Tobin, J., 39, 139–40, 142–3, 195 Tunisia, 113–14 Turkey, 113 Turnovsky, S.J., 195 unit roots, 56, 58 USA, 69, 78, 151 von Furstenberg, G.M., 195 Wallace, N., 39, 68, 71, 97, 104, 106, 111, 122, 186, 194 Wilcox, D.W., 56–8, 195 Wilford, D.S., 190–1 Willet, T.D., 190, 195 Wohar, M.E., 195 Yellen, J.L., 189, 195