Mechanisms of Democracy: Institutional Design Writ Small

Mechanisms of Democracy Institutional Design Writ Small

Adrian Vermeule Harvard Law School

Mechanisms of Democracy Institutional Design Writ Small

1

1

Oxford University Press, Inc., publishes works that further Oxford University’s objective of excellence in research, scholarship, and education.

Copyright © 2007 by Oxford University Press, Inc. Published by Oxford University Press, Inc. 198 Madison Avenue, New York, New York 10016

Oxford is a registered trademark of Oxford University Press Oceana is a registered trademark of Oxford University Press, Inc. All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without the prior permission of Oxford University Press, Inc.

Cataloging-in-Publication information is available from the Library of Congress. ISBN: 978-0-19-533346-6 (Cloth) Printed in the United States of America on acid-free paper.

Note to Readers: This publication is designed to provide accurate and authoritative information in regard to the subject matter covered. It is based upon sources believed to be accurate and reliable and is intended to be current as of the time it was written. It is sold with the understanding that the publisher is not engaged in rendering legal, accounting, or other professional services. If legal advice or other expert assistance is required, the services of a competent professional person should be sought. Also, to confirm that the information has not been affected or changed by recent developments, traditional legal research techniques should be used, including checking primary sources where appropriate. (Based on the Declaration of Principles jointly adopted by a Committee of the American Bar Association and a Committee of Publishers and Associations.)

You may order this or any other Oxford University Press publication by visiting the Oxford University Press website at www.oup.com

To Yun Soo, Emily and Spencer

This page intentionally left blank

Contents

Introduction: Democracy and Institutional Design . . . . . . . . . . . . . . . . .

1

Part I. Impartiality and Uncertainty . . . . . . . . . . . . . . . . . 25 1. The Veil of Uncertainty . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 2. The Limits of Uncertainty . . . . . . . . . . . . . . . . . . . . . . . . . . 55

Part II. Accountability and Voting . . . . . . . . . . . . . . . . . . . 73 3. Submajority Voting Rules: Forcing Accountability . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85 4. Absolute Majority Rules: Optimizing Accountability . . . . . . . . . . . . . . . . . . . . . . . . . 115 5. Delegation, Accountability and Judging . . . . . . . . . . . . . . 143

Part III. Deliberation and Transparency . . . . . . . . . . . . . . 177 6. Optimizing Transparency: The Budget Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183 7. Optimizing Deliberation: Constitutional Issues in Congress . . . . . . . . . . . . . . . . . . . 217

Conclusion Institutional Design as a Going Concern . . . . . . . . . . . . . . 245 Acknowledgments & Sources . . . . . . . . . . . . . . . . . . . . . . . . 249

This page intentionally left blank

Mechanisms of Democracy Institutional Design Writ Small

This page intentionally left blank

Introduction

Democracy and Institutional Design

W

hat institutional arrangements should a well-functioning constitutional democracy have? Most of the relevant literatures in law, political science, political theory, and economics address this question by discussing institutional design writ large.1 The debates center on mass elections, the separation of powers, federalism, and other large-scale institutional structures. Answers to the large-scale question differ, of course. In democratic theory, for example, followers of Joseph Schumpeter’s elitist vision of democracy argue that mass elections are both necessary and sufficient to constitute meaningful democracy,2 while critics of Schumpeter argue, from various perspectives, that a thicker set of democratic institutions is required.3 At the stage of constitutional design, standard controversies 1

2 3

See, e.g., Geoffrey Brennan & Alan Hamlin, Democratic Devices and Desires: Theories of Institutional Design (2000); Robert D. Cooter, The Strategic Constitution (2002); Dennis C. Mueller, Constitutional Democracy (2000). See, e.g., Richard A. Posner, Law, Pragmatism and Democracy (2003). See, e.g., Gerry Mackie, Schumpeter’s Leadership Democracy (University of Chicago Political Theory Workshop, 2005), available at http://ptw.uchicago.edu/mackie01.pdf.

Institutional Design Writ Small

1

Introduction

center on the relative merits of different electoral systems, of parliamentary versus presidential regimes, and of various types and degrees of judicial independence and judicial review of democratic processes and outcomes. In what follows, my ambition is to move beyond these debates, changing the subject to institutional design writ small. I assume the basic or large-scale institutions of constitutional democracy. Even within such a framework, there is much work for institutional designers to do. In established constitutional polities, I argue, law can and should—and to some extent already does—provide mechanisms of democracy: a repertoire of small-scale institutional devices and innovations that promote democratic values against the background of standard large-scale institutions. The bulk of the book is devoted to proposing such mechanisms and indicating the conditions under which they will, or will not, succeed. Examples include legal rules that promote impartiality by depriving officials of the information they need to act in self-interested ways (Part I); voting rules that create the right kind and amount of accountability for political officials and judges; (Part II); and legislative rules that structure deliberation, in part by adjusting the conditions under which deliberation occurs transparently or instead secretly (Part III). Why focus on institutional design writ small? The principal reason is that “[d]emocracy is inherently a device for regulating marginal political conflicts.”4 This reads as an essentialist claim about the very concept of democracy; in context, however, it is a claim about the insuperable costs of changing the large-scale structures of an ongoing democratic order on which the whole society has coordinated.5 The fact is that in most democratic polities, the basic constitutional arrangements are no longer up for grabs. New polities forming new constitutions may be interested in the basic structures of constitutionalism, as evident in the wave of Eastern European constitution-making in the 1990s. In places and times like that, there is a point to large-scale debates over parliamentary systems versus the separation of executive and legislative powers, or over proportional representation versus other representative forms. In the consolidated democracies of North America, the British Commonwealth, Western Europe and Scandinavia, and parts of Latin America and Eastern Europe, however, small-scale institutional design is all that is on offer. It is unlikely 4 5

2

Russell Hardin, Liberalism, Constitutionalism and Democracy 309 (1999). See id.

Mechanisms of Democracy

Democracy and Institutional Design

in the extreme that the United States will switch to a parliamentary system, or that the United Kingdom will switch to a system of separated powers. The point is not that these background institutions are adequately democratic; it is that they are not going anywhere. Even or especially if one believes that the background constitutional order in, say, the United States is pervasively undemocratic,6 it is all the more important to consider feasible smallscale reforms to promote democratic values within that order. In many of these polities, however, there are ongoing experiments with small-scale arrangements of the sort discussed here. In the United Kingdom, consider Parliament’s Joint Committee on Human Rights, charged with reviewing proposed legislation for conformity with the United Kingdom’s statutory and treaty-based obligations to respect rights.7 Chapter 7 generalizes this approach from rights to all constitutional questions and proposes analogous institutions in the U.S. Congress, especially an internal Office of Constitutional Issues, as one of several changes intended to promote legislative deliberation on constitutional questions. It is crucial not to assume that institutional design writ small necessarily has small effects, or that design at the margin has only “marginal” effects in the colloquial sense. On the contrary; one of my central claims in what follows is that small changes can have large democratizing effects. Theoretically, this is a consequence of the sensitivity of institutional equilibria. It is not systematically true that small institutional changes produce small effects. Where discontinuities in important variables occur, small changes can produce large effects, while large changes can produce small effects.8 Seemingly minor changes in voting rules, for example, can profoundly affect both processes and outcomes in multimember institutions. Empirically, I will offer examples throughout in which small-scale mechanisms have produced or would produce important democratizing effects. Chapters 3 and 4, for instance, suggest alternative voting rules that can force majorities to make clear decisions on important questions. Chapter 6 suggests that modest reductions in the transparency of the budget process would produce large improvements in legislative deliberation on budgeting, even in the face of powerful interest-group pressures. 6 7 8

See Robert Dahl, How Democratic Is the Constitution? (2002); Sanford R. Levinson, Our Undemocratic Constitution (2006). For a useful treatment of the Joint Committee, see Janet L. Hiebert, Parliament and the Human Rights Act: Can the JCHR Help Facilitate a Culture of Rights?, 4 Int’l. J. Con. Law 1 (2006). See Bruce Talbot Coram, Second Best Theories and Institutional Design, in Theory of Institutional Design 90, 91 (Robert E. Goodin ed., 1996).

Institutional Design Writ Small

3

Introduction

Mechanisms of Democracy Defined In this framework, the mechanisms of democracy are small-scale rules that structure the process by which laws are made, given the background institutions of the relevant constitutional democracy. So defined, mechanisms of democracy may be contrasted, on the one hand, with large-scale institutional arrangements, as above; they may also be contrasted with constitutional or legal rules that directly police the outcomes of democratic institutions, such as rules protecting individual rights and liberties. At several points, especially in Chapter 1, I will emphasize the latter comparison, suggesting that mechanisms of democracy can substitute for constitutional rules that protect rights directly. Such mechanisms can even do a better job, in the sense that second-order institutional arrangements can promote the very goals that underlie substantive protections, but with greater collateral benefits or fewer collateral costs.

Democratic Values What goals, exactly, are the mechanisms of democracy supposed to serve? How might we judge whether or not they are successful, and under what conditions? I will propose mechanisms that advance four core values of democratic constitutionalism: impartiality, accountability, transparency, and deliberation. These values, I suggest, form something like the common ground among democrats of differing stripes. Although each of them can be specified in different ways, and although each is controversial in some quarters, anyone who rejected all of them simultaneously as basic aims of institutional design would cease to be a recognizable democrat. I use thin conceptions of these values—least-common-denominator definitions that capture premises widely shared by different democratic views. As used here, impartiality means that government officials do not act in self-interested ways, giving preference to their own good at the expense of other members of the political community. Rather, officials act to promote the public good, somehow defined—even if that good is defined in extremely thin procedural terms as the choices made through voting by democratic majorities. Accountability means that officials must answer for their actions to other members of the political community. Elections are the most obvious mechanism of democratic accountability, but I shall argue that other mechanisms are necessary as well. Transparency means that governmental secrecy is presumptively bad in a well-functioning 4

Mechanisms of Democracy

Democracy and Institutional Design

democracy; citizens should in general be able to observe official decisionmaking in some way, although I will argue that there are good democratic reasons to overcome the presumption of transparency in a range of decisionmaking settings. Deliberation means that officials should make decisions on the basis of public-regarding reasons, rather than on the basis of “naked preferences.”9 In all of these cases, there are lurking questions about the boundaries of the political community—about how the demos is constituted.10 Officials are to serve impartially, and should be accountable to, voters and citizens; but a well-known set of issues in democratic theory involves the further extension of the democratic community. Does it include resident aliens, for example? Such questions, however, do not prove relevant to the mechanisms I will discuss, which are designed for the core case of governance affecting citizen-voters. In what follows, then, I will speak loosely of the political community of “citizens.” The values I have identified are not in general controversial among democrats, although there is appropriate disagreement about their meaning, nature and limits. A variety of high-level approaches to democratic theory converge to support some conception or other of each of these values. We may consider each in turn, as follows: Impartiality is underwritten by consequentialist theories of democracy, in welfarist or even utilitarian forms, that require government to weigh the interests of each citizen equally—in Bentham’s sense that “each person is to count for one, no person for more than one.” Nonconsequentialist democratic theories argue for a right of citizens, or an obligation of government, such that each person should be treated with equal concern and respect. A contractualist variant argues that impartial government would be chosen by reasonable persons deciding on arrangements for their common life. Whatever else such theories include or exclude, it is clear that self-interested behavior by government officials is ruled out. Accountability is also common ground among democratic theorists, although here too interpretations and conceptions differ.11 Schumpeterian 9 10 11

See Cass R. Sunstein, Naked Preferences and the Constitution, 84 Colum. L. Rev. 1689 (1984). See Robert Goodin, Enfranchising All Affected Interests, and Its Alternatives, 35 Phil & Pub. Aff. 40-68 (2007). For a taxonomy of various senses of accountability, see Jerry Mashaw, Accountability and Institutional Design: Some Thoughts on the Grammar of Governance, in Michael Dowdle ed., public accountability: designs, dilemmas and experiences 115–156 (2006). In

Institutional Design Writ Small

5

Introduction

theorists of elitist democracy suggest that accountability through elections is all there is to democratic institutions, but other varieties of democratic theory have a thicker conception of what accountability means. Political economists interpret accountability through the lens of political principal-agent models where voters lack full information about representatives’ behavior (the problem of moral hazard) or their type (the problem of adverse selection).12 In a very different interpretation, an essential component of accountability is that the reasons that ground official decisions be contestable by democratic citizen-subjects.13 Where possible, my aim will be to bracket such high-level controversies and find the common core of these differing conceptions. The accountability-inducing mechanisms I will discuss in Chapters 3, 4, 5 and 6 can appeal to all camps. Transparency is necessary, at least to some degree, to any conception of accountability, however thin. How can voters meaningfully participate in elections if they have no idea what decisions officials have made or (more controversially perhaps) how they have made them? Transparency is necessary to democratic governance, but here too the details are all-important. In Chapter 6, I suggest that transparency is one democratic good among others, and trades off against democratic goods such as well-functioning deliberation. Transparency should thus be optimized rather than maximized; I examine some institutional structures for securing the democratic benefits of transparency while reducing its costs. Deliberation is a controversial ideal in its thickest conceptions. Some theorists believe that the aggregation of predeliberative preferences is the only feasible or even desirable form of collective action in a democracy, whereas many deliberative democrats believe or hope that deliberation can itself launder preferences14 to filter out morally objectionable desires and motivations, or more generally can mold preferences in socially or morally desirable ways. Even here, however, there is a thinner version of deliberation that commands widespread assent, which is that deliberation can at least constitute useful talk by which democratic actors exchange

12 13 14

6

Mashaw’s terms, the different senses of accountability I examine here are all forms of political accountability. See Timothy Besley, Principled Agents? The Political Economy of Good Government (2006). Philip Petit, Democracy Electoral and Contestatory, in Designing Democratic Institutions 105, 119–20 (Ian Shapiro & Stephen Macedo eds., 2000). See Robert E. Goodin, Laundering Preferences, in Foundations of Social Choice Theory 75 (Jon Elster & Aanund Hylland eds., 1986).

Mechanisms of Democracy

Democracy and Institutional Design

information and opinions.15 In this minimal and widely endorsed sense, deliberative capacity is a property of institutions that should and indeed must be one of the goods that optimal democratic design will consider. Deliberation, information and epistemic democracy. Deliberation is distinct from the aggregation of preexisting information, which can occur through voting without talk. However, deliberation can precede information aggregation without necessarily undermining the independent epistemic value of voters’ information.16 More generally, both deliberative democrats and epistemic democrats hope that democratic decision procedures can be designed to produce right factual or moral answers, where such answers exist.17 Democrats who emphasize the epistemic advantages of democracy, as opposed to its deliberative credentials or procedural fairness, use the Condorcet Jury Theorem and other mathematical insights to justify majority voting in democratic institutions (or plurality voting, where there are three or more options in play).18 I take account of these insights in many contexts, particularly in Chapters 3, 4 and 5, which all concern voting mechanisms. In some of those contexts, however, I suggest that nonmajority voting rules are not epistemically objectionable, even where there are right answers. The claim, made explicit throughout Part II, is that epistemic democracy is compatible with nonmajority voting more often than is believed. A common mistake is to assume that where there are right answers, epistemic democracy requires majority voting in every multimember institution, taken one by one. This is false; nothing in the Jury Theorem tells us what group of voters should be governed by majority rule. To think that jury-theoretic reasoning requires majority voting in every institution, taken one by one, is to fall into a version of the fallacy of composition. It is always necessary to ask “a majority of what?” Thus in Chapter 5, I suggest a supermajority voting rule for multimember courts, weighted in favor of government agencies. Even where 15 16

17

18

See James D. Fearon, Deliberation as Discussion, in Deliberative Democracy 44 (Jon Elster ed., 1998). See Jeremy Waldron, Democratic Theory and the Public Interest: Condorcet and Rousseau Revisited, 38 Am. Pol. Sci. Rev. 1322 (1989); see also David Estlund, Opinion Leaders, Independence, and Condorcet’s Jury Theorem, 36 Theory & Decision 131 (1994). There is a large literature on epistemic democracy. For an overview by a leading contributor, see Christian List, Democracy and Epistemic Justification (Sept. 15–21, 2002), http:// personal.lse.ac.uk/LIST/PDF-files/KonstanzLectures.pdf. See Robert Goodin, Reflective Democracy (2003).

Institutional Design Writ Small

7

Introduction

there are right answers, this rule has the effect of counting the agency’s view in the judicial voting calculus, and thereby approximates the epistemic effect of majority rule over a group defined to include the agency officials as well as the judges. To the extent that epistemic democrats can be persuaded to support such aggregation of the voting calculus across institutions, then democrats of different stripes—including democrats with different commitments to the value or feasibility of obtaining “right” answers in political life—can approve of nonmajority voting in particular institutional settings, albeit on different grounds.

Information, Institutions and Mechanism Design How can these democratic values be advanced? Throughout, I suggest mechanisms that in one way or another act on the information held by institutional actors. In Part I, the basic strategy is to promote impartiality by depriving officials of information they might use in self-interested ways; the basic mechanism is that constitutional rules can create a veil of uncertainty that forces officials to act as though motivated by impartial considerations. It is not the case that more information always produces better decisions. Even apart from the costs of gathering information, it is often best if officials lack precise information about which policies will advance their personal interests—although Chapter 2 emphasizes a series of tradeoffs and costs that place limits on this principle in turn. In Part III, the corollary strategy is to promote deliberation and accountability by depriving some subset of constituents of information about what elected representatives are doing in some circumstances. It is not the case that all official decisionmaking should be fully transparent, even putting aside state secrets and executive confidences. Sometimes, transparency can make deliberation worse, by encouraging various forms of posturing and herding; and transparency can empower narrow interest groups, thereby enforcing a kind of bad accountability, at the expense of broader democratic values. The general point that transparency has costs is well understood. I go farther, however, by advocating selective transparency—disclosure of government decisionmaking to some audiences but not to others. Here the crucial mechanism is delayed disclosure, which takes advantage of the timing of election cycles: information of interest to multiple audiences can be released at a time when constituents may use it to inform their voting decisions, but when narrow interest groups can no 8

Mechanisms of Democracy

Democracy and Institutional Design

longer use the same information to fund captive candidates. In a related vein, Part II suggests novel voting rules that structure information so as to promote democratic values in collective choice. The voting mechanism discussed in Chapter 3, the submajority voting rule, allows legislative or judicial minorities to force the majority to reveal democratically desirable information. The voting mechanism discussed in Chapter 4, absolute majority rule, allows faithful representatives to disguise or ambiguate their resistance to political pressures emanating from third parties rather than constituents; it thus promotes deliberative values and the best form of democratic accountability. Information is a crucial fulcrum for mechanism design because institutions are, among other things, systems for managing information. Institutions aggregate information held by individuals and attempt to reduce uncertainty in order to promote the (collective) goals of the institution’s members.19 A special case involves information that determines mutual expectations and common knowledge—beliefs about what others think about what others think (and so on) are the proper roles and activities of the institution’s members, and what sanctions will be applied for deviating from those roles.20 The fragility of equilibria based on common knowledge creates opportunities for institutional design writ small. By structuring mutual expectations in subtle ways, institutional rules can profoundly affect the behavior of individuals within the institution. If institutions are informational systems, mechanism design can in general play one of two roles, either improving upon or deliberately frustrating the extant institutional arrangements for managing information. Parts II and III illustrate the former, meliorative role. In Part II, the basic idea is to propose improvements in the process by which individually held information is aggregated through voting, in order to promote democratic values; these improvements are compatible with the motivations and beliefs of the institutional actors themselves. Likewise in Part III, which suggests improvements in the structures of democratic transparency that both promote democratic values and that representatives might themselves favor. From the perspective of the wider democratic polity, on the other hand, it is sometimes best to devise institutional arrangements that thwart 19 20

See Barry C. Burden, Introduction: Everything but Death and Taxes: Uncertainty in American Politics, in Uncertainty in American Politics 3, 12 (Barry C. Burden ed., 2003). See generally H.L.A. Hart, The Concept of Law (1961).

Institutional Design Writ Small

9

Introduction

the attempts of particular institutions to manage information in the ways their members would otherwise choose. In Part I, the basic idea is to hamstring representatives in their attempts to reduce uncertainty about which policies will promote their partial interests. Because self-interested representatives would not lower the veil of uncertainty over their own eyes, however, uncertainty must be created by constitutional rules. I return to this point below.

Theories, Conflicts and Tradeoffs: A Challenge to Institutional Design The most obvious, if not the most serious, problem with this approach arises from persistent disagreement about democratic values. Even if there is an abstract consensus across democratic theories that attach value to impartiality, accountability, transparency and deliberation, perhaps intractable theoretical conflicts remain. After all, different theories favor different conceptions of these concepts; it seems that, inevitably, high-level theoretical conflicts will arise and must be ironed out before we can speak coherently of designing democratic mechanisms. On this view, absent a full specification of theoretical goals, the criteria for sorting good institutional designs from bad ones are not well defined. Moreover, as I will emphasize throughout the following chapters, democratic goods trade off against one another, and against other goods as well. What is worse, there are democratically desirable and undesirable forms of these goods, so the tradeoffs are internal, within values, as well as external, across values. Detailing these claims is a central project of the book, but a brief sketch of the central tradeoffs might run as follows: Impartiality vs. Information, Motivation and Commitment. In Chapter 1, I suggest a range of mechanisms that generate a veil of uncertainty and thereby promote impartial decisionmaking by officials. Three tradeoffs are important, however, and are discussed in Chapter 2. First, impartiality trades off against information. Decisionmaking behind the veil of uncertainty is less informed decisionmaking. Second, and less obviously, impartiality trades off against motivation. Where decisionmakers have agenda control, a veil of uncertainty reduces the decisionmakers’ incentive to take any action at all, perhaps resulting in excessively low activity levels. Third, impartiality trades off against commitment. Impartial enactments are more likely to be undone in future periods, when it becomes clear whose 10

Mechanisms of Democracy

Democracy and Institutional Design

ox has in fact been gored. Commitment problems are more serious for impartial enactments than for enactments that represent victories by selfinterested coalitions in the first instance. Transparency vs. Deliberation. Chapter 6 suggests that transparency deters self-dealing or corrupt bargains by officials, and thereby promotes a kind of impartiality. But transparency can also harm deliberation, in part because the glare of public attention causes officials to posture for the cameras, for interest groups and outside audiences, or for posterity; moreover, transparency can silence potential dissenters, who hesitate to take public risks and thus end up following the view of apparent experts, or of the crowd. If any of this is so, then transparency trades off against deliberative virtues. Transparency vs. Accountability. Transparency helps promote oversight by voters or constituents, and thus promotes the accountability of representatives to voters or constituents. Yet it also promotes oversight or monitoring by third parties—narrow interest groups, self-interested officials elsewhere in government, and others. In some cases, as Chapters 4 and 6 emphasize, the latter sort of monitoring can be undesirable, if it allows third parties to divert representatives from faithfully promoting the interests of or executing the wishes of constituents. Accountability (good) vs. Accountability (bad). The last point can also be put in terms of an internal tradeoff between two different types of accountability. The same structures and mechanisms that promote the accountability of officials to one group may have the unintended effect of promoting the accountability of the same officials to a different group. Under some range of circumstances, the former type of accountability might be democratically desirable, whereas the latter type would be democratically harmful. If so, there is a kind of tradeoff within accountability itself. Deliberation (good) vs. Deliberation (bad). Even in the minimal sense of deliberation as talking together, the very conditions necessary for deliberation to improve decisionmaking can also worsen decisionmaking. An exchange of views among officials, including information and instrumental arguments, can make decisions better. But deliberation in common can also produce decisional pathologies.21 Decisionmakers may herd in

21

See Cass R. Sunstein, Deliberative Trouble? Why Groups Go to Extremes, 110 Yale L.J. 71 (2000).

Institutional Design Writ Small

11

Introduction

the same direction because some have private information that others rationally follow, because of concerns for their reputations in the group or elsewhere, or because of the phenomenon of group polarization, which under particular circumstances causes opinions to become more extreme within deliberating groups.22 The conditions that are necessary for good deliberation can also create a risk of deliberative pathologies. It does not follow, however, that it is feasible or desirable to stamp out deliberative pathologies. For one thing, some such pathologies may be the unavoidable price of deliberative structures that are optimal overall. For another, just as private vices sometimes confer public benefits, so too deliberative vices may contribute to improving group deliberation in the aggregate. Hot rhetoric and extreme views can illuminate the concrete implications of abstract principles, introduce new perspectives, and challenge complacent premises that are widely shared.23 All this poses two basic challenges to the project of democratic institutional design. First, if institutional design is thoroughly theorydependent, we cannot say anything very useful about democratic institutions without a fully worked-out conception of constitutional democracy, one that takes sides on all relevant theoretical questions. The prospect is a depressing one, because of the interminable quality of high-level debates over democratic theory. It seems we will be doomed to theoretical discord into eternity, never legitimately reaching the stage of institutional design— although during our theoretical purgatory we will necessarily have institutions of some sort or another, just unexamined ones. To make things worse, if it is really the case that theory comes first, institutional design writ large is also ruled out. The literatures in law, comparative political science and economics that address large-scale issues of constitutional design would have to be stamped as philosophically confused, because these literatures often fail to resolve all relevant and logically prior theoretical questions before addressing desirable institutions. Second, how can tradeoffs between or within democratic goods, or between democratic goods and other goods, be resolved without a fully specified theory of democracy? We would need a theory to specify, for example, how much transparency to give up in return for how much deliberative benefit. It is incoherent to make democratic tradeoffs 22 23

12

See id. at 71–72. Don Herzog, Romantic Anarchism and Pedestrian Liberalism (unpublished draft).

Mechanisms of Democracy

Democracy and Institutional Design

without a full theory of what is democratically desirable—or so the argument runs.

The Place of Institutional Design The burden of the following chapters is to argue, and to illustrate, that the foregoing challenges fail. They fail, I will suggest, on three basic counts. First, there is the important possibility of operating-level convergence. Rival conceptions of democracy can and do converge on particulars at the operating level, in a type of overlapping consensus.24 Suppose that some democratic mechanism can help to dampen or eliminate selfinterested behavior on the part of legislators. We need not take sides on any (contested) questions of high-level democratic theory or the theory of representation to approve of such a mechanism. On some theories of representation, legislators are to represent specific constituencies or groups, geographically defined or otherwise; on other theories of representation, each legislator is charged with representing the general interests or welfare or good of the whole body politic. All such theories, however, unite in rejecting self-interested representation by legislators. To the extent that the mechanisms of democracy address the problem of self-interested representation, no ambitious theorizing is necessary. Another way to put this point is that the mechanisms discussed here aim to secure the minimal conditions of democratic governance. In many settings, the realistic task of institutional design is not the ambitious one of promoting perfected or rarefied democratic aspirations; it is the preliminary and mundane one of enforcing the basic ground rules and preconditions that any recognizably democratic set of institutions needs to function tolerably, let alone well. As to these minimum conditions, there is rarely much consequential disagreement among competing highlevel accounts of democracy; they are the common ground held by competing views. Securing the minimal conditions of democratic governance is as pressing a task in advanced constitutional democracies as in new or transitional democracies. Although basic institutions like elections and the separation of powers weed out the crudest forms of democratic dysfunction, there is plenty of dysfunction left to go around, as any observer of American government will attest. 24

See John Rawls, A Theory of Justice 340 (rev. ed. 1999).

Institutional Design Writ Small

13

Introduction

The second basis for rejecting the challenge is that the mechanisms discussed here can be defended as optimizing devices. In easy cases, some institutional arrangement will dominate another in the decision-theoretic sense. One arrangement will, under any state of the world, produce desirable accountability at no harm or cost to other democratic values, while another arrangement will produce lesser accountability with no offsetting benefit. In such cases no tradeoff arises, and the institutional choice is obvious. The harder cases arise when there is genuine conflict between or among democratic values. Even in such cases, however, it is sensible to respect a principle of marginalism. If a design innovation that would produce a small loss of transparency would also produce major gains in the quality of deliberation, then most democrats of most theoretical varieties will favor the change, because the new arrangement represents a move towards the democratic optimum. In what follows, I try as much as possible to consider the mechanisms of democracy in a non-sectarian way, appealing to all varieties of democratic theory. Where that hope fails, however, my premises are openly consequentialist: I assume that democratic values are comparable and commensurable goods subject to aggregation according to the principle of marginalism. Democracy, I assume, can be optimized. To be sure, even when marginal tradeoffs produce large gains in some democratic values at small costs to others, some band of theorists will protest. In Chapter 6, for example, those who attach intrinsic importance and lexical priority to transparency—theorists who favor maximal transparency at all costs—will jump ship, complaining that their philosophical arguments have been assumed away by the new institutional arrangements I suggest, which diminish transparency for larger gains in impartiality and the quality of government deliberation. Yet few such theorists exist; and one wonders why those who do exist would be entitled to block the large body of those who disagree from adopting optimizing innovations, at least until all the philosophical issues have been settled by persuading the obdurate outliers. Because we must have institutions of some kind or another while the theoretical conflicts rage, no particular theoretical view can be given veto power over the search for institutional innovations and improvements. In other words, the theory-dependence of institutional design cannot be converted into a kind of unanimity requirement, such that nothing can be done about institutions until all theoretical dissent has been squelched. 14

Mechanisms of Democracy

Democracy and Institutional Design

There will always be some theoretical resistance to innovation, but I will be content if the mechanisms discussed here appeal to the broad middle range of democratic commitments. Throughout, I implicitly imagine a sort of second-order assembly of democrats, of varying theoretical views, who vote upon proposals according to whether the suggested innovations promote their preferred, particular version of democratic theory. If the proposals suggested here are carried in the assembly of democrats by large supermajorities (quite possibly by different supermajorities on different proposals), that is good enough, even if some vote nay. It should be mentioned that an important subcamp of democratic theorists will think that the marginalist or optimizing approach to tradeoffs among democratic values is itself dictated by their theory. This camp is comprised of consequentialist theorists of democracy, who believe that democracy is justified only if and because it produces better consequences than other political arrangements.25 Consequentialists need a value theory specifying what counts as “better,” and most theorists of this sort are welfarists, who see (human) well-being as the test of value, or utilitarians, who go further to favor a particular conception of well-being, usually subjective happiness or pleasure. Whatever the details, these theorists necessarily unite in rejecting the idea that some democratic values might have lexical priority over others, such that no tradeoffs are permitted. If large gains in, say, accountability require small sacrifices in, say, transparency, and if the shift produces net improvements in welfare or utility overall, then such theorists will think the shift mandatory, not merely permissible. On the welfarist view, the democratic values I have identified are but rules of thumb, considerations that enter into the optimal design of democratic institutions, but that are of no intrinsic value in themselves, apart from their tendency to promote the well-being of the political community. The third response is the most modest of all, but perhaps the most important: institutional design is perfectly sensible, and indeed unavoidable, conditional on adopting some high-level theory or other.26 Democratic theorists just cannot move directly from some abstract principle—say, 25 26

See Richard J. Arneson, Democracy Is not Intrinsically Just, in Justice and Democracy 40 (Keith Dowding et al. eds., 2004). See Archon Fung, Practical Reasoning About Democratic Institutions: Governance Innovations in Democratic Theory (Working Paper, Feb. 2006), http://www.archonfung.net/papers/Fung PracticalReasonFeb2006.pdf.

Institutional Design Writ Small

15

Introduction

equal concern and respect—to concrete arrangements on the ground— say, a simple majority voting rule. Empirical questions and tricky problems of design always and necessarily intervene between the high-level principles and the operating-level outcomes. In reality, it is not the case that there is a prior stage of theory and a subsequent stage of institutional design. We will be living in the midst of some set of institutions or other while theorizing proceeds. It is not sensible—perhaps it is not even coherent—to say that we have no basis for examining, evaluating and improving those institutions unless and until we have a fully specified theory about what counts as an improvement, democratically speaking. Overall, the mid-level questions about institutional design are unavoidable, for two basic reasons. One is the ever-present possibility of unanticipated consequences and surprising second-best effects. A given institutional arrangement may, in certain circumstances, prove futile or even counterproductive relative to the high-level aims it is meant to serve. In Chapter 6, for example, one theme is that legal rules mandating the transparency of legislative deliberations backfire if they simply cause legislators to move real deliberation to other, less formal and indeed less transparent venues. Another reason is that institutions are frequently underdetermined by high-level commitments. Thus in Chapter 4 I argue that a commitment to equality within a voting group, for example, can be consistent with both a simple majority voting rule that looks only to a majority of the votes cast, and also with an “absolute” voting rule that looks to a majority of all the votes that could have been cast. An invocation of “equality” will not help us, even in principle, to figure out which of these two voting rules is best, under what conditions. The choice between them must be made on other grounds. At a minimum, institutional design can go forward without theoretical strain relative to some high-level value or set of values. The most modest ambition of the following, then, is to provide a repertoire of tools for democratic design that people of varying commitments can pick up and put down at will. But I have less modest ambitions as well. In real life, where time for philosophizing is limited and agreement seems always to lie over the next horizon, institutional design goes forward even without consensus on high-level commitments. I will attempt to show, throughout, that democrats of differing views can find common ground on institutional arrangements even under these daunting circumstances. 16

Mechanisms of Democracy

Democracy and Institutional Design

Mechanisms and Limiting Conditions Democracy-promoting mechanisms are not panaceas to be prescribed under any and all conditions. In what follows, the limits of the mechanisms I discuss are as important as their virtues. The basic project, in each case, is to identify the conditions under which particular democracy-promoting mechanisms can promote widely endorsed attributes of democratic institutions or, more ambitiously, can optimize across democratic values. Some of the mechanisms I discuss are more fragile than others, in the sense that they function well or at all only under a narrow range of conditions. In Chapter 2, for example, I suggest that the veil of uncertainty is a highly fragile mechanism for reducing self-interested behavior by government officials. Understanding the limiting conditions for democratic mechanisms is a worthy goal in itself. I have emphasized that the mechanisms of democracy do not replace or compete with large-scale democratic structures, but rather supplement and complement them. Of course the particular mechanisms I discuss hardly exhaust the universe of available supplements. Identifying the limits of the mechanisms introduced here underlines the need for further progress in the theory of institutional design writ small.

Institutional Variables and Their Values Another limitation of the project is that, although institutional, it is not empirical, at least not in a large-number way. Throughout, I draw on comparative case studies of democratic mechanisms and institutional structures, such as various types of alternative voting rules used in various polities or at various levels of American government (Chapters 3 and 4). Nowhere do I engage in large-number empiricism, complete with regression analysis. The aim is to identify institutional variables, to indicate their signs (the direction of the effects they produce), and to give a rough estimate of their magnitudes. Although this is a limitation, it is for the most part an unavoidable one. The chronic condition of institutional reform is that the effects even of small-scale proposals are somewhat unpredictable until they have been tried, perhaps repeatedly. As far as many of the mechanisms suggested here are concerned, there simply is no large-number data available. In other cases, although the facts are in principle lying out there to be collected, it is beyond my competence to do so. Institutional Design Writ Small

17

Introduction

Descriptive and Prescriptive Claims The book has dual aims: analytic and reformist. On the analytic side, I hope to explain the working of novel mechanisms that can, in principle or at the level of the first-best, promote agreed-upon democratic ends (under particular conditions). On the reformist side, I propose that practically workable versions of such mechanisms be adopted in real-world democratic institutions. In the latter vein, I attempt in every case to account for second-best problems, political constraints, and the costs of enactment and transition to the new rules. As will be explained shortly, it is a major virtue of institutional design writ small that small-scale proposals do not incur prohibitive costs of enactment and transition. Partly for that reason, small-scale proposals are more likely to slip past the political constraints on reform. By contrast, changes to institutional design writ large are unpredictable, rare, and rarely susceptible to deliberate institutional engineering. Where I do offer reform proposals, I do not mean to suggest that the mechanisms I suggest are entirely unprecedented. In most cases, I complement the prescriptive claims with descriptive claims that the existing landscape of law and institutions in democratic regimes already contains examples of the mechanisms I favor. In some cases the examples are not labeled as such by participants in the system; instead they represent my analytic interpretation or characterization of institutional rules. In Chapter 1, for example, I characterize many rules of American constitutional law as “veil rules” that produce uncertainty and thus, indirectly, produce a kind of impartial decisionmaking. The point of this work is to show that the mechanisms I suggest have real-world purchase. These mechanisms are in some sense feasible, at least under some conditions; the constraints on the range of institutional possibilities are not typically so tight as to rule them out, everywhere and right from the beginning.

The Demand Side, the Supply Side, and Constraints on Reform The topic just broached—feasibility constraints on democratic reform— deserves more extended comment. So far we have primarily addressed the demand side of mechanism design: the conditions under which various democracy-promoting mechanisms might be desirable, whether or not attainable. What of the supply side? Which democratic actors might adopt 18

Mechanisms of Democracy

Democracy and Institutional Design

new design mechanisms, assuming they are indeed desirable? Are those mechanisms even technically or politically feasible? Proposals for new institutional arrangements will only be adopted if they are compatible with the desires, beliefs and technical and political opportunities of actors within the system. Absent some account of the supply side, or of the political and technical constraints that limit opportunities for reform, perhaps proposals of the sort advanced here are so much whistling in the wind. This is an important theoretical point. Yet paradoxically enough, we must be cautious not to internalize it if our interest is reform, rather than objective analysis of the political order. In many settings, it is sensible to recognize a division of labor: theorists propose optimal institutional designs (on the level of ideals or first-best), while politics disposes (on the level of practical constraints or second-best). In this role, theorists should delineate schemes that would be desirable if implemented, while legal and political activists attempt to implement them. Politics itself will serve as the filter that sorts feasible proposals from others. Of course the theorist whose interest is itself the analysis of political constraints is exempt from this point. As to theorists whose interest is in the optimal design of institutions, however, it is best from the systemic point of view to ignore political constraints, thus avoiding a kind of selfcensorship that inflicts social harms by prematurely filtering out ideas that might, in fact, have been adopted. The proponent of optimal design mechanisms is typically not an expert in politics, so self-censoring in light of the reformer’s estimate of political feasibility carries the risk that the reformer will mistakenly filter out an ideal solution that would actually have been enacted. By contrast, the opposite error—the proposal of desirable solutions that are politically infeasible—is solved automatically by the operation of the political filter. In accordance with this principle, my main focus throughout is on the demand side of mechanism design. Certainly, for the familiar reasons outlined above, it is also desirable for optimal proposals to be feasible; proposals of that sort are even better than first-best proposals thrown hopefully against the political filter. I suggest, in other words, a three-tier framework for the evaluation of proposals. Worst of all, for systemic reasons, is self-censorship by theorists of optimal institutions. Rather than self-censor, it is better to put forward first-best proposals and let politics dispose of them, usually by rejection but occasionally by surprising acceptance. Best of all, however, are proposals that are both desirable on first-best grounds and also feasible. Institutional Design Writ Small

19

Introduction

I believe that the small-scale proposals offered here are both desirable on their own terms and quite thinkable, in a practical way, and thus fit into the third and highest tier. The comparisons to the experience of other polities that I draw throughout are helpful in this regard. In the case of the congressional Office of Constitutional Issues proposed in Chapter 7, for example, the fact that Parliament has set up a similar body —the Joint Committee on Human Rights — is a first step towards dispelling concerns about feasibility. There are many possible disanalogies between the two institutional changes, but their similar scale and purpose makes it implausible that the former would be politically out of bounds while the latter was actually adopted. More positively, feasibility in both the technical and political senses is an affirmative virtue of institutional design writ small. Precisely because small-scale proposals tend to incur lower costs of enactment and lower costs of transition from one institutional regime to another, they are also more likely to pass through the political filter. Adjustments to decisionmakers’ information (Part I), to voting rules (Part II), and to rules structuring transparency and deliberation (Part III), are not like proposals that the United States should switch to a parliamentary system,27 or the curious recent proposal for a new national holiday devoted to collective political deliberation.28 Feasibility is a central trait of the small-think proposals I offer here, whatever their other merits or demerits.

Self-Defeating Proposals Another aspect of the supply-side problem is that proposals must not be self-defeating. Proposals defeat themselves when the motives, beliefs or political opportunities ascribed to relevant actors by the theorist’s diagnosis are incompatible with the solution that the theorist offers. An inconsistency between the motives assigned to relevant actors in the theorist’s diagnosis and prescription is often called “incentive incompatibility”; in Chapter 2, I discuss several proposals that are incentive incompatible and thus self-defeating, most notably Bruce Ackerman’s scheme for a framework statute to govern emergency decisionmaking. 27

28

20

See, e.g., Woodrow Wilson, Congressional Government: A Study in American Politics (Dover 2006) (1885); see also Lloyd Cutler, To Form a Government, 59 Foreign Aff. 126 (1980). See, e.g., Bruce Ackerman & James S. Fishkin, Deliberation Day (2004).

Mechanisms of Democracy

Democracy and Institutional Design

To be clear, the problem with self-defeating proposals is not merely that they are ruled out by politics. I argued above that it is bad, from the systemic point of view, for proponents of novel arrangements to selfcensor in light of their estimate of the political constraints, although the external analyst may of course assess those constraints. Rather, the problem with self-defeating proposals is that the diagnosis and prescription make different assumptions about relevant features of the relevant actors. Such proposals are internally inconsistent from the inception, and thus incoherent, quite apart from whether they have any chance of being adopted. I do not think any of the proposals I offer elsewhere in the book stumble into this methodological trap, but the problem is very real, and my claim to have avoided it must be assessed after reading the whole. In outline, I propose to cope with the problem in the following ways. A proposal urging self-interested officials to subject themselves to a veil of uncertainty, in order to hamper their pursuit of their self-interested aims, would obviously be self-defeating. However, Part I makes no such proposal. Instead I suggest that various mechanisms for creating a veil of uncertainty might be and to a large extent already are embodied in constitutions, and in the United States Constitution turn out to be rather sensibly distributed across the branches of government. Of course there is no general reason to assume that constitutions are optimally designed, including the United States’ constitution; but it turns out that the framers of 1789 happened to adopt a number of veil-generating rules, in part to enforce impartiality and in part for other reasons. So the distribution of veil rules in the Constitution is best seen as a combination of intentional action and happy coincidence. Because the Constitution already embodies veilgenerating mechanisms, there is no need, at least in the American case, to propose a constitutional amendment, which would obviously raise serious feasibility concerns. In Parts II and III, by contrast, the argument is that the relevant proposals might straightforwardly be embodied in statutes or internal legislative or judicial rules, because they are compatible with the motives and incentives of the relevant institutional actors. Thus in Chapters 3, 4, and 5, for example, I suggest mechanisms that explain why and where legislative and judicial majorities might adopt and have adopted voting rules that give procedural power to legislative and judicial minorities. Institutional Design Writ Small

21

Introduction

Overview The book is organized into three parts. Part I takes up the problem of promoting impartial action by officials, especially elected officials, and the related goal of dampening self-interested official behavior. These problems are not exactly coterminous. It is possible to behave altruistically, yet not impartially, as by favoring the interests of some ideological or ethnic group both over one’s own interests and over those of the whole polity.29 Yet the indispensable first step towards genuinely impartial and publicspirited government is to suppress self-interested representation. Were official self-interest to be eliminated or sufficiently constrained, many other problems would remain; but those are the sort of problems many democracies would not mind having. Many of the large-scale structures of constitutional democracy are intended to either remove the causes of selfinterested representation or to reduce the harms that flow from it. Elections, the separation of powers, and judicial review have all been justified in these terms. Even given such structures, however, supplemental mechanisms are needed. The mechanism featured here is the veil of uncertainty, which suppresses self-interest by depriving officials of the information needed to gauge where their interests lie and which policies might promote those interests. Chapter 1 introduces the veil mechanism and suggests that a range of constitutional provisions already embody veil-producing devices or tactics. Chapter 2 then indicates the limits of the veil mechanism by discussing three tradeoffs: (1) between impartiality and information; (2) between impartiality and motivation; and (3) between impartiality and ex post enforceability. A heavily veiled government might possess too little information to act effectively. Less intuitively, its officials might lack the motivation to introduce many beneficial projects, and might thus display an insufficient level of activity or energy. Finally, the impartial enactments that did occur might more frequently be undone in later periods, after the veil is lifted. A premise of Part I is that elections are not sufficient, although certainly necessary, to eliminate self-interested representation. Part II begins from a similar premise, but moves in a different direction. Elections are necessary but not sufficient to supply democratic accountability, a value 29

22

See Albert O. Hirschman, The Passions and the Interests (1977); see also Stephen Holmes, The Secret History of Self-Interest, in Beyond Self-Interest (Jane Mansbridge ed., 1990).

Mechanisms of Democracy

Democracy and Institutional Design

that sweeps beyond the suppression of self-interested representation. In a well-functioning democracy, officials are not merely free from self-interest; they are also accountable even for the public-spirited choices they make. The best conception of accountability is a matter of large controversy in democratic theory, but I will suggest some accountability-promoting mechanisms that fit comfortably with the major normative views of accountability and representation. The introduction to Part II clears away some ground by rejecting one of the principal accountability-promoting mechanisms in American constitutional theory, especially theory of a textual or historical cast: the nondelegation doctrine, which places constitutional constraints on legislative delegations of discretion to the executive branch. I argue that the nondelegation doctrine cannot be derived from any high-level account of democracy or accountability. The more natural view, which I will defend, is that delegations from the legislature to the executive are fully satisfactory, on the score of accountability, so long as the executive is independently accountable and so long as legislative decision to delegate is itself made in an accountable way. Accordingly, Chapters 3 and 4 change the focus from the nondelegation doctrine to legislative accountability for all decisions, including the decision to enact delegating statutes themselves. Here my aim is to suggest mechanisms that enhance the accountability of legislative voting. Chapter 3 describes submajority voting rules that allow minorities to force majorities to take a stand on public issues. Chapter 4 describes absolute voting rules that aim to block strategic behavior by legislative minorities and to liberate legislators from accountability to narrow self-interested groups. Chapter 5 draws a corollary for judicial decisionmaking in the administrative state. It suggests a supermajority voting rule for courts under which executive agencies will prevail, against a claim that they have exceeded their delegated authority, unless a defined supermajority of the judges rejects the executive’s position. The point of the voting rule is to channel policymaking discretion from the judiciary to the executive branch, acting under a legislative delegation. Where statutes contain gaps that must be filled in, the executive rather than the judges should do so. It is not just that the executive is superior to the judiciary on the score of democratic accountability, although that is true. The further point is that such a mechanism enforces the legislature’s second-order decision—itself fully accountable—to delegate to the executive in the first place. Institutional Design Writ Small

23

Introduction

Part III offers suggestions for promoting both transparency and deliberation, and for managing the tradeoffs between them; as will be seen, these values also interact in complex ways with impartiality and accountability. Chapters 6 and 7 provide a two-pronged approach; the former focuses on the distinctly secular arena of the congressional budget process—a fine testing ground for high-minded democratic aspirations— while the latter focuses on legislative deliberation about constitutional questions. Chapter 6 argues for selective transparency, in which some decisionmaking processes are opaque, and in which some information is disclosed to some constituents at some times but not to others at other times. I suggest that selective transparency should be instituted through a regime of delayed disclosure of legislators’ actions, to promote sound legislative deliberation and to liberate legislators from undesirable accountability to narrow interests. Chapter 7 turns to the production of constitutional deliberation in legislatures. I argue that constitutional deliberation is a collective good that will be underproduced in legislatures, and suggest some incentive-compatible mechanisms for overcoming this problem, such as a congressional Office of Constitutional Issues designed to encourage more successful legislative deliberation on constitutional questions. Throughout, my principal aim is to be suggestive and creative about mechanism design—to expand our repertoire of institutional tools, rather than to arbitrate or finally resolve contested institutional issues, let alone contested high-level questions of democratic theory. As I have emphasized, the appeal of particular democracy-promoting mechanisms, relative to competitors, will turn on difficult empirical and predictive questions about which information is sparse and judgments are inevitably uncertain. I both record my own judgments on such matters, primarily with a view to provoking the reactions and judgments of others, and also suggest second-order decision procedures for designing institutions under conditions of normative disagreement and severe empirical uncertainty.

24

Mechanisms of Democracy

Part I

Impartiality and Uncertainty

This page intentionally left blank

Impartiality and Uncertainty

I

mpartiality is a protean concept. A basic distinction is between firstorder and second-order impartiality.1 First-order impartiality would require me to treat your children as well as my own, whereas second-order impartiality would allow me to give preference to my children and you to give preference to your children. The latter conception merely bars me from claiming that impartiality requires you to give preference to my children, just as I do.2 Another basic distinction is between self-interested partiality and other-regarding partiality. The former arises, for example, when I allocate a valuable resource to myself, ignoring the impartially superior claims of others. The latter arises when I allocate the valuable resource to, say, coreligionists or coethnics or compatriots, ignoring the impartially superior claims of others (perhaps including myself). As the latter type of case shows, partiality and altruism or other-regardingness can easily coexist; governmental decisionmaking that is free from selfinterest is a necessary but not sufficient condition for governmental impartiality. Impartiality is, in turn, a necessary but not sufficient condition for democracy.3 Later chapters will address other values that enter into a well-functioning democratic polity. In Chapters 1 and 2, I will be solely concerned with impartiality in the narrower sense of freedom from self-interest. Perhaps it would be simpler to choose a different term than impartiality to connote the value at stake—perhaps “public-spiritedness” or the like—but to do so would distort the substantive point. Although democracy goes beyond impartiality, and impartiality goes beyond the absence of self-interest, the first emphatically does include the second, and the second emphatically does include the third. A government whose officials systematically function in selfinterested ways is neither impartial nor democratic. The small-scale

1 2 3

See Brian Barry, Justice as Impartiality: A Treatise on Social Justice, Volume 2 11 (1995). See Jon Elster, Mimicking Impartiality, in Justice and Democracy 112, 112–13 (Keith Dowding et al. eds., 2004). See Robert E. Goodin, Democracy, Justice and Impartiality, in Justice and Democracy 97, 99–101 (Keith Dowding et al. eds., 2004).

Institutional Design Writ Small

27

Part i

mechanisms of democracy I urge are necessary to promoting the minimum conditions of democracy. Among these minimal conditions are that government officials not act in self-interested ways, whatever else we might desire of them. Promoting impartiality, even in the narrower sense of restricting self-interest in official decisionmaking, is thus a core democracypromoting project. How, then, can impartial government (in the narrower sense) be induced or promoted through institutional design? Typical answers center on institutional design writ large. Self-interested representation is to be checked by elections, which tie representatives to the wishes of their constituents, and by the separation of powers, which pits self-interest (Madison’s “ambition”) against self-interest. Chapter 1 proposes and scrutinizes a different sort of answer, one that is compatible with but that also goes beyond the baseline institutions of elections and the separation of powers. Self-interested decisionmaking is to be checked by mechanisms that generate a veil of uncertainty. Whereas elections and the separation of powers limit self-interested behavior by subjecting officials to political constraints, uncertainty-generating mechanisms suppress the information that self-interested officials would need to pursue their interests. Such mechanisms thus cause decisionmakers to act as though they are impartial, whatever their true motivations.

Veil Rules and Their Limits My project in this Part is twofold. The first aim is to introduce the veil of uncertainty as a small-scale mechanism for promoting impartiality. The second aim, pursued in Chapter 2, is to delineate the built-in limitations of the veil of uncertainty, the tradeoffs it produces. Throughout, I use the American federal constitution as a source of illustrations, although with lateral comparisons to other constitutions where useful. Does the Constitution make optimal use of veil rules? Is there good reason that it uses veil rules where it does use them, but not elsewhere? Chapter 1 suggests that Congress is subject to more constitutional veil rules, of wider scope, than is the President or the judiciary. Is there any good reason why should that be so, given that it would be perfectly possible to apply a range of veil rules to the latter institutions as well? It sounds paradoxical to move from an explication of existing veil rules in some settings to an explanation for their conspicuous absence in others, but that 28

Mechanisms of Democracy

Impartiality and Uncertainty

question illuminates the tradeoffs inherent in mechanism design. Having appreciated the power of veil rules to dampen self-interest, we should ask why the veil technique should not be ubiquitous. Chapter 2 thus offers an account of the limits of uncertainty as a technique of constitutional design. The centerpiece of the account is a crucial tradeoff between information and motivation, or (as the Constitution’s framers would have put it) institutional “energy.” Veil rules dampen both information and bias, including bias arising from self-interest; at the same time, however, they also suppress decisionmakers’ activity or energy. Removing the spur of self-interest threatens to reduce decisionmakers’ activity below acceptable levels, to the point where constitutional designers might plausibly prefer to lift the veil and spur more activity, even if the price is that some fraction of that increased activity is self-regarding. If we hope to spur energetic presidential activity, then something roughly like the current skewed distribution of veil rules suggests itself. Where the gains from executive energy outweigh the losses from self-interested executive action, unveiling the executive is justified overall, even if it produces executive abuses. On this view, the Constitution happens to adopt a rather sensible distribution of veil rules across branches. Methodologically, this conclusion is no more than a happy coincidence. There is no general reason to expect constitutions to be optimally designed, because the motives of constitutional framers and the political circumstances under which they work may often drive them to adopt suboptimal rules.4 My claim is just that the Framers of the American Constitution, for good reasons or bad, happened upon a sensible distribution of veil rules across institutions, as far as we can judge today. Moreover, whether this claim is correct or incorrect, it is intended only to illustrate the central tradeoffs inherent in veil mechanisms, and the limits of uncertainty as a technique of constitutional design. After identifying three important tradeoffs—between impartiality, on the one hand, and information, motivation, and ex post enforceability on the other—I critique recent institutional-design proposals that have appealed to uncertainty, in some cases erroneously.

4

See Jon Elster, Ulysses Unbound 172 (2000).

Institutional Design Writ Small

29

This page intentionally left blank

Chapter 1

The Veil of Uncertainty

A

veil rule, I shall say, is a rule that suppresses self-interested behavior on the part of decisionmakers by subjecting the decisionmakers to uncertainty about the distribution of benefits and burdens that will result from a decision. For a simple example, consider a rule that requires an official with regulatory authority over the stock market to place her assets in a “blind trust.” Even if the official is self-interested, she will be uncertain whether any given decision will increase or decrease the value of her portfolio. She will thus be unable to skew her decisions in order to promote her personal interests. In these examples and others, the crucial feature is that more information is not always better than less. From the social point of view, denying information to decisionmakers can produce better—in the sense of more nearly impartial—decisions. Veil rules suppress bias that arises from information. That bias need not be based on self-interest, although it may be. Where orchestras audition candidates from behind a screen, denying evaluators any information about the gender or attractiveness of the candidate, information is suppressed in order to suppress a type of bias; but self-interest is not in play here. Bias arising from self-interest is the major worry where government officials are concerned, and will be my focus. To the extent that veil rules

Institutional Design Writ Small

31

Chapter 

suppress both self-interested decisionmaking and other forms of biased decisionmaking, so much the better. A veil rule may produce distributive uncertainty by either of two methods. One method is to place decisionmakers under a constraint of ignorance about their own identities and attributes. Rawls’s famous phrase, “the veil of ignorance,” describes a hypothetical original position in which principles of justice are chosen under precisely this constraint.1 But that is a special case of veil rules generally, indeed a radical case. Rawls’s thought experiment introduces uncertainty by allowing the decisionmaker to know the distributive consequences of a decision on future citizens— call them A and B—but denying the decisionmaker the knowledge of whether she herself will occupy A’s position or B’s position. Where veil rules appear under historical rather than hypothetical conditions, however, the relevant decisionmakers will usually know their own identities and interests. Veil rules that appear in actual constitutions, then, more often adopt a second method for introducing uncertainty. Although the decisionmaker knows or can guess whether she will occupy A’s or B’s position, the rule introduces uncertainty about the payoffs to A or B—about what gains, if any, A or B will reap from the decision. In other words, constitutional design decisions give particular actors particular payoffs. The first type of veil rule suppresses the decisionmaker’s knowledge of the actors’ identities; the second type suppresses the decisionmaker’s knowledge of the actors’ payoffs. As we shall see, the second type of veil rule is more plausible or realistic. Such rules can take advantage of the inherent uncertainty that, in real polities, attends upon complex political decisions whose consequences extend into the future. Between these two types of veil rule, an intermediate case arises when a decisionmaker faces a decision that will principally affect not the decisionmaker herself, but her (genetic) descendants. If the time horizon over which the decision will matter is long enough, the attributes of descendants may be so difficult to predict that the decisionmaker will effectively be left ignorant of the identity of the persons whose interests she would favor, if she could. In this vein, participants at the Constitutional Convention argued that the Framers should impartially assess the interests of social classes, and of large and small states, because posterity 1

32

John Rawls, A Theory of Justice 118–23 (rev. ed. 1999).

Mechanisms of Democracy

The Veil of Uncertainty

would distribute their descendants throughout the strata and regions of society.2 The point of a veil rule is to prevent self-interested decisionmaking. Constitutional designers may plausibly fear that a decisionmaker who knows both her own identity and the distribution of future benefits and burdens that results from a decision will systematically favor her own interests, in ways detrimental to the quality of the decision. Self-interested decisionmaking need not arise from bad motivations. Cognitive mechanisms such as self-serving bias3 and the proclivity to engage in motivated reasoning4 can produce it even in public-spirited officials. Behind the veil, however, the decisionmaker afflicted with uncertainty will, for lack of ex post information about whose interests to favor, choose the option or rule that impartially promotes the good of all those affected in an ex ante sense. The dearth of information produces decisions that track the output of an impartially motivated decisionmaker. What constitutes “impartial” decisionmaking—the undistorted baseline from which to measure self-interested decisionmaking and suppress it by constitutional rules—is a more difficult question in democratic theory than in the legal enterprise mechanism design, simply because the law’s ambitions are far more modest. The constitutional choice literature debates whether decisionmakers in Rawls’s original position would or should choose rules to maximize average utility or total utility;5 should instead select a “maximin” criterion that works to the benefit of the leastwell-off, as Rawls advocates;6 or should implement some less formal 2

3

4

5

6

1 The Records of the Federal Convention of , 49 (Max Farrand ed., 1966). I owe this point to Jon Elster, Arguing and Bargaining in Two Constituent Assemblies, 3 U. Pa. J. Const. L. 345 (2000). See Linda Babcock & George Loewenstein, Explaining Bargaining Impasse: The Role of SelfServing Biases, 11 J. Econ. Persp. 109, 109, 111 (observing that self-serving bias causes decisionmakers to “arrive at judgments of what is fair or right that are biased in the direction of their own self-interests”). See Christopher H. Schroeder & Robert L. Glicksman, Chevron, State Farm, and EPA in the Court of Appeals During the 1990s, 31 Envtl. L. Rep. News & Analysis 10,371 (explaining “motivated reasoning”). John Harsanyi, for example, advocates a utilitarian account of choice in the original position and uses an “equiprobability postulate” that maximizes the mean utility level of all individuals in society. John C. Harsanyi, Morality and the Theory of Rational Behaviour, in Utilitarianism and Beyond 39, 44–46 (Amartya Sen & Bernard Williams eds., 1982). The maximin criterion instructs decisionmakers to choose, from among possible actions, the one whose worst possible consequences are better than the worst possible consequences of the alternative actions. See R. Duncan Luce & Howard Raiffa, Games and Decisions 278 (1957). Rawls uses the maximin criterion to derive his “difference principle,” which says that

Institutional Design Writ Small

33

Chapter 

interpretation of impartiality, such as decisionmaking that displays “equal concern and respect” for all parties.7 From the standpoint of mechanism design under historical circumstances, however, the alternatives are much less refined. The aim of real-world veil rules is simply to suppress the crudest instances of self-dealing and factional oppression—decisions that would be excluded on any plausibly impartial criterion for public choice, such as the retroactive punishment of political enemies, bills of attainder, or official self-dealing with regard to salaries and emoluments. This is an example of the modest aspiration, set out in the introduction, to suggest mechanisms that aim to secure the minimum conditions of democratic governance. Disagreement about the uniquely best definition of impartiality need not prove an embarrassment to the limited ambitions of real-world democratic design, which are fully satisfied by identifying a set of decisions that all competing definitions of impartiality condemn. For similar reasons, I shall subsume within the category of selfinterest a partial preference for decisionmakers’ relatives, descendants, political allies, and so forth. The justification for this treatment derives from the modesty of law’s ambitions and the crudity of its instruments. Although there may be conceptual reasons for treating these categories differently, as suggested above in the discussion of different forms of bias, uncertainty-creating mechanisms are not usually fine-grained enough to take such differences into account.

Constitutional Choice in Two Stages By speaking of veil rules in constitutions, I mean to pose a very different question than the one pursued in the standard discussions of the veil of ignorance. The constitutional choice literature stemming from James Buchanan, Gordon Tullock, and their successors conceives the uncertainty produced by the veil of ignorance as a mechanism for inducing hypothetical constitutional designers to approach the choice of the constitutional rules themselves in an impartial way.8 Decisionmaking by

7 8

34

allocative inequalities are permissible only to the extent they maximize the position of the least-well-off. Rawls, supra note 1, at 132–35. Ronald Dworkin, Taking Rights Seriously 272–78 (1978). See Geoffrey Brennan & James M. Buchanan, The Reason of Rules 28–31 (1985); James M. Buchanan & Gordon Tullock, The Calculus of Consent 77–80 (1962); Dennis Mueller, Constitutional Democracy 61–64 (1996); Jonathan Riley, Constitutional

Mechanisms of Democracy

The Veil of Uncertainty

legislators and other officials within the framework of the constitutional rules, by contrast, falls in the domain of “ordinary politics,” where selfinterested individuals and factions struggle for advantage. The constitutional designers’ self-interest is constrained by uncertainty;9 that of ordinary decisionmakers is constrained by voting rules (such as supermajority requirements), by substantive constitutional prohibitions on inefficient legislation, and by institutional competition resulting from the separation of powers. Here I ask whether and how constitutional rules might subject insystem decisionmakers to the same uncertainty constraint that governs the hypothetical stage of constitutional choice, and for similar reasons. To illustrate, I will survey a range of rules from the federal constitution of the United States. I will suggest that the federal Constitution itself contains a number of provisions best justified as (whether or not originally designed as) veil rules. Rules, provisions, structures, and practices as diverse as the Ex Post Facto and Bill of Attainder Clauses,10 the Emoluments Clause,11 the Twenty-Seventh Amendment,12 Article V’s procedures for constitutional amendment,13 the doctrine of precedent, the original mechanism for selecting senators (by vote of the state legislatures),14 and the rules governing presidential election and succession15 may all profitably be considered in this light, although not all of these should count as examples of veil rules rightly understood.

9

10 11

12

13 14 15

Democracy as a Two-Stage Game, in Constitutional Culture and Democratic Rule 147 (Ferejohn et al., eds., 2001). This claim is rarely accurate for actual constitutional designers, who are impelled by a variety of motives, including self-interested ones, and who are constrained by circumstances, including the need to secure popular ratification of the constitutions they propose. See Jon Elster, Ulysses Unbound 129–46 (2000). U.S. Const. art. I, § 9, cl. 3 (“No Bill of Attainder or ex post facto Law shall be passed [by Congress].”); id. art. I, § 10, cl. 1 (“No State shall . . . pass any Bill of Attainder [or] ex post facto Law. . . .”). Id. art. I, § 6, cl. 2 (“No Senator or Representative shall, during the Time for which he was elected, be appointed to any civil Office under the Authority of the United States, which shall have been created, or the Emoluments whereof shall have been encreased during such time. . . .”). Id. amend. XXVII (“No Law, varying the compensation for the services of the Senators and Representatives, shall take effect, until an election of Representatives shall have intervened.”). Id. art. V (specifying procedures for constitutional amendment). Id. art. I, § 3, cl. 1 (stating that federal senators shall be chosen by the state legislatures), amended by id. amend. XVII (replacing legislative selection with direct election). Id. art. II, § 1, cls. 3, 5 (providing rules for presidential election and succession); id. amend. XII (providing rules for presidential election); id. amends. XX, XXV (providing rules for presidential succession).

Institutional Design Writ Small

35

Chapter 

Self-Interested Decisionmaking: Two Strategies of Design The use of veil rules is not the principal means by which the Constitution constrains self-interested decisionmaking. The Framers’ simple theory of action held that decisions are the product of motives and opportunities, and described the principal motives as reason, interest, and passion.16 The theory suggests two different strategies for constraining the operation of self-interest. The first is to replace decisionmakers’ self-interested motives with the motive of impartial rationality, thus striking at the problem at its source. Veil rules aim to accomplish that end by interfering with decisionmakers’ information, introducing uncertainty about the distribution of benefits and burdens, which forces the self-interested decisionmaker to proceed in an impartially rational way. Madison, however, argued in The Federalist No. 10 that suppressing self-interest at its source is infeasible. The aim of “giving to every citizen the same opinions, the same passions, and the same interests”—precisely what a veil strategy does in an ex ante sense, not by enforcing homogeneity in fact but by reducing decisionmakers’ information—Madison rejected out of hand as an “impracticable” method for suppressing self-interested factions.17 He therefore preferred a second strategy to constrain selfinterest: leaving self-interested motives in place while constricting the opportunities available to self-interested decisionmakers. The “policy of supplying, by opposite and rival interests, the defect of better motives”18 is to parcel out decisions among different officials and institutions with different self-interested agendas, in the hope that competition among institutions will block self-interested action by any one of them. This is just the separation of powers, which controls the effects rather than the causes of self-interest by ensuring that ambition is made to counteract ambition across institutions. Madison did not wholly carry the day; existing constitutional veil rules are precisely the sort of “parchment barrier” that he rejected in favor

16

17 18

36

See Morton White, Philosophy, The Federalist, and the Constitution 102–28 (1987). In the present analysis, I will ignore the role of passion. Attempts by constitutional designers to prevent in-system decisionmakers from falling prey to their temporary passions is the subject of the theory of constitutional precommitments. See, e.g., Jon Elster, Ulysses Unbound 88–167 (2000); Jeremy Waldron, Precommitment and Disagreement, in Constitutionalism: Philosophical Foundations 271 (Larry Alexander ed., 1998). The Federalist Papers No. 10, at 123 (James Madison) (Isaac Kramnick ed., 1987). The Federalist Papers No. 51, supra note 17, at 320 (James Madison).

Mechanisms of Democracy

The Veil of Uncertainty

of institutional competition.19 Nonetheless, it is fair to say that the baseline constitutional strategy for suppressing self-interested decisionmaking is the separation of powers, and that the use of veil rules is a supplemental and exceptional strategy. One of my concerns here is to say something about where and why the Constitution uses, or declines to use, veil rules as supplements to the normal baseline.

Veil Mechanisms The veil of uncertainty is an institutional-design strategy that may be implemented by a variety of mechanisms. I will examine four particular mechanisms that all attempt to introduce uncertainty about the distribution of future benefits and burdens that will result from a decision. The mechanisms are prospectivity, by which constitutional rules require decisionmakers to create legal rules without knowing the identities of the rules’ violators; generality, which causes decisionmakers to anticipate that both the interests they favor and those they disfavor may be governed by a current decision; durability, by which rules that make decisions relatively durable cause decisionmakers to anticipate that a present decision will govern cases in the remote future, cases whose effect on the decisionmakers’ future interests is presently unpredictable; and delay—that is, delay of the effective date of a rule, which restricts the range of a decision’s future application to the long term, rather than the short term, in the hope that decisionmakers’ long-term interests are inherently unpredictable.

Prospectivity The simplest mechanism for introducing uncertainty is to entrench a constitutional requirement that rules be prospective—enacted in advance of the events they govern. The power of retroactive legislation, for example, enables legislators to identify the winners and losers from proposed policies—to know who will bear the costs and enjoy the benefits as well as what those costs and benefits will be. The opportunities for legislative selfdealing are obvious if legislators can match identified winners and losers with past or future friends and enemies, respectively. Under a prospectivity requirement, however, legislators are hard put to match consequences 19

The Federalist Papers No. 48, supra note 17, at 309 (James Madison).

Institutional Design Writ Small

37

Chapter 

with allegiances, because prediction is intrinsically more difficult and less certain than backward-looking observation, and because targets who know of the law will be able to steer clear of its prohibitions. A related point is that a prospectivity requirement reduces uncertainty by attaching clear legal consequences to events in advance of the events’ occurrence. Despite appearances, this is wholly consistent with the claim that prospectivity requirements, like other veil tactics, aim to produce beneficial uncertainty; the two observations address different issues. Consider, for example, the many instances in which the Constitution delegates authority to Congress to “provide by law” for the contingencies of presidential and vice-presidential succession in cases of death, disability, and so forth.20 These are simple examples of constitutional veil rules if they are interpreted to require that Congress legislate in advance of the relevant contingencies. Akhil Amar praises the veil-like effects of these rules on the ground that “[u]ncertainty, especially over so vital an issue as Presidential succession, is not, on balance, a virtue,” so that “our Constitution . . . has explicitly invited Congress to lay down clear succession rules in advance of a crisis.”21 But this runs together the issue of timing with the separate issue of the clarity of legal commands. Certainty in the latter sense is a function of the specificity and comprehensiveness of legal commands, and it can be attained after the relevant events’ occurrence as well as before. Once the President dies, nothing at all prevents Congress from legislating quite clearly about who his successor shall be. The real fear that provokes the demand for veil rules is that after a presidential death or disability, decisionmakers’ knowledge of the identities of the possible successors will provoke socially harmful, because self-interested, congressional decisionmaking, as legislators maneuver to enthrone themselves or their partisans in the White House. That fear is dampened not only by the clarity of previously enacted succession rules but also by uncertainty at the time of enactment about whose ox a given rule will gore. It is true, however, that later interpreters who know the effects of the previously enacted rule on their current interests may take advantage of any ambiguities in the rule by means of tendentious interpretation. I discuss that problem at greater length below. 20 21

38

See supra note 15. Akhil Reed Amar, Presidents, Vice Presidents, and Death: Closing the Constitution’s Succession Gap, 48 Ark. L. Rev. 215, 221, 227 (1994).

Mechanisms of Democracy

The Veil of Uncertainty

There is no general federal constitutional requirement that legislation (and a fortiori adjudication) be solely or even partially prospective. The most prominent prospectivity rules in the Constitution are the Ex Post Facto Clauses,22 which forbid federal and state legislatures from enacting retroactive punishments.23 The ex post facto prohibition is best explained as a veil of ignorance rule, albeit of limited scope. An older view held that the prohibition prevents unfair surprise to law-abiding citizens, but that is both circular (presumably, without the prohibition, fewer people would be surprised by retroactive criminal legislation) and morally unattractive (when, for example, the new rule punishes a defendant who previously tortured and murdered his own children,24 an act he ought to have known would upset the public’s representatives). It is a problem for the veil account that the ex post facto prohibition applies only to retroactive criminal laws and not retroactive civil laws. If identifiability is the key, why should not all retroactivity be suspect? One way out is to supplement the veil story with an interest-group tale of differential access to the political process, as does Harold Krent.25 Felons cannot vote, have no trade association and often no money, and the stigma of their past convictions or threatened retroactive convictions scares off potential political allies, so their interests will be underweighed in the legislative process. But the parties typically burdened by retroactive civil legislation have both cash and friends; they need no special judicial protection. This is a clever story, but in the end not a convincing one. For one thing, adding interest-group analysis might cut the wrong way, suggesting that no constitutional barrier to retroactive legislation is necessary. If the retroactivity of a pending bill makes winners and losers identifiable, 22 23

24 25

See supra note 10. E.g., Weaver v. Graham, 450 U.S. 24 (1981). The Supreme Court has occasionally read ex post facto principles into the Due Process Clause, with the result that a few cases hold unforeseeable judicial interpretations of statutes unconstitutional. E.g., Bouie v. City of Columbia, 378 U.S. 347 (1964) (holding that the South Carolina Supreme Court violated the petitioners’ due process rights by retroactively applying its new construction of a state statute). But the unforeseeability requirement is construed very grudgingly, and the official view remains that the ex post facto prohibition binds only legislatures, not courts. See Rogers v. Tennessee, 121 S. Ct. 1693 (2001). See Dobbert v. Florida, 432 U.S. 282 (1977) (upholding, against an ex post facto challenge, a retroactive change in death penalty sentencing procedure). Harold J. Krent, The Puzzling Boundary Between Criminal and Civil Retroactive Lawmaking, 84 Geo. L.J. 2143 (1996).

Institutional Design Writ Small

39

Chapter 

interest groups may lobby all the more fiercely to block the bill than they would to block prospective laws that are less clearly harmful to their interests.26 For another, the interest-group tale explains a problem at the margins of the prohibition rather than the problem at its core. Krent focuses principally upon the case in which legislation increases the punishment for previously convicted criminals. Sentenced offenders, as Krent argues, are hardly a politically appealing constituency. But the core concern of the prohibition is with wholly new retroactive crimes, and it is not clear that Krent’s analysis works in that case. New criminals, unlike previously sentenced criminals, have the formal right to vote and the informal access to political resources that sentenced felons lack. In this respect, the veil account fits the scope of the ex post facto prohibition better than does Krent’s account. If the prohibition is a veil rule, then it sensibly covers both new crimes and sentence enhancements. In either case, the key concern is legislators’ ability to identify and to specify the targets of the new criminal rule, rather than the content of the rule itself. Rather than invoke the uncertain aid of interest-group analysis, we should probably just say that the distinction between permissible civil retroactivity and impermissible criminal retroactivity is a pseudo-puzzle. As a historical matter, the ex post facto prohibition may have been understood to cover both categories during the founding era.27 As a matter of judicial doctrine, to say that civil retroactivity is permissible is an accurate description of the ex post facto prohibition standing alone, but it is an overstatement if applied to constitutional law generally. Other clauses sometimes bar civil retroactivity that impinges on particular constitutional interests such as property; the Supreme Court occasionally invalidates retroactive civil legislation under the Takings and Due Process Clauses. If there remains a slight mismatch between the veil account and the scope of constitutional antiretroactivity rules generally, we might justify it simply as the law’s rough attempt to adjust the direct costs and benefits of veil rules. As a rough empirical generalization, subject to pervasive 26 27

40

Cf. Saul Levmore, The Case for Retroactive Taxation, 22 J. Legal Stud. 265, 280 (1993). The Clause was held to cover only criminal legislation in Calder v. Bull, 3 U.S. (3 Dall.) 386, 390 (1798). But that holding was hotly contested at the time. Satterlee v. Matthewson, 27 U.S. (2 Pet.) 380, 414–16 (1829) (Johnson, J., dissenting). It is still controversial in originalist circles today. See, e.g., E. Enters. v. Apfel, 524 U.S. 498, 538 (1998) (Thomas, J., concurring) (expressing his willingness to overrule Calder).

Mechanisms of Democracy

The Veil of Uncertainty

exceptions and boundary problems, laws imposing criminal burdens are blunter instruments, which impose cruder, more fearsome, and more stigmatizing harms on those subjected to them than do civil or regulatory laws. Conversely, the loss in impartiality that arises when relatively lessharmful civil statutes are left outside the veiled domain is more than compensated by the corresponding gains in legislative information and flexibility. As Justice Iredell put the same point, “[w]ithout the possession of this power [to enact retroactive civil laws] the operations of government would often be obstructed.”28

Generality Closely entangled with the idea of prospectivity is the idea of generality.29 A variety of constitutional rules require generality, or more accurately, push official decisions in the direction of generality across a continuum between the highly specific and the highly general. The Bill of Attainder Clauses, applicable to both federal and state governments, prevent legislatures from inflicting criminal punishments on specific, identified individuals or groups (without a judicial hearing); “special legislation” clauses in state constitutions prohibit or hamper statutes that impose special burdens, or (in some states) statutes that confer special benefits, on particular individuals, groups, or geographic areas; various constitutional requirements of uniformity prevent disparate treatment of different states or regions;30 and the equal protection requirement serves as a constitutional backstop that may be invoked to scrutinize both federal and state official action that seems excessively targeted against identified individuals or classes. The generality requirement, like the prospectivity requirement, is said to produce veil effects that deprive decisionmakers of the information needed to pursue selfish or partial interests. “[G]eneralizations place people behind the veil of ignorance, depriving them of the knowledge of

28 29 30

3 U.S. (3 Dall.) at 400 (Iredell, J.). For the conceptual issues surrounding generality, see Frederick Schauer, Profiles, Probabilities and Stereotypes 251–77 (2003). U.S. Const. art. I, § 8, cl. 1 (“[A]ll Duties, Imposts and Excises shall be uniform throughout the United States. . . .”); id. art. I, § 8, cl. 4 (“Congress shall have power . . . To establish an uniform Rule of Naturalization, and uniform laws on the subject of Bankruptcies throughout the United States. . . .”); id. art. I, § 9, cl. 6 (“No Preference shall be given by any Regulation of Commerce or Revenue to the Ports of one State over those of another. . . .”).

Institutional Design Writ Small

41

Chapter 

whether [a rule’s] application will help or hurt themselves or their friends.”31 If legislators, for example, must frame statutes in general terms, rather than tailoring their proscriptions to identified or (at least) identifiable enemies, there will be some risk, however small it may be in practice, that the legislator himself (or his allies) will fall subject to the law. That risk will moderate the legislator’s hostility or indifference to disfavored individuals or groups, bringing decisions more nearly into line with impartial decisionmaking, however defined. Untangling the relationship between generality and prospectivity helps us to appreciate the work that each requirement does. Granting that “the legislature must prescribe penalties generally and prospectively, behind a suitably impersonal veil of ignorance,”32 we may wonder whether either generality or prospectivity does any veil-generating work on its own, absent the other. Consider the following three hypothetical cases:33 (1) A statute providing that “all those who hereafter steal from another shall be put to death.” (2) A statute providing that “all those who have previously stolen from another shall be put to death” (where either there was no preexisting theft statute or the preexisting statute imposed a lesser penalty). (3) A statute providing that “anyone named Vermeule who hereafter steals from another shall be put to death” (where there is no other, general theft statute in force, and only one person named Vermeule). The first statute is both general and prospective, the second general but retroactive, and the third prospective but extremely specific. Under current doctrine the first is clearly valid, while the second clearly violates the Ex Post Facto Clause. It might be said that the generality of the second statute imposes a sufficient veil of ignorance, because the legislators will 31

32 33

42

Richard A. Epstein, A Common Law for Labor Relations: A Critique of the New Deal Labor Legislation, 92 Yale L.J. 1357, 1364 (1983); see also Richard A. Epstein, Takings 240 (1985) (justifying general tort rules on veil of ignorance grounds). For a recent attempt to make generality the master principle of constitutional choice, see James M. Buchanan & Roger D. Congleton, Politics by Principle, Not Interest (1998). Akhil Reed Amar, Attainder and Amendment 2: Romer’s Rightness, 95 Mich. L. Rev. 203, 210 (1996). Adapted from a similar list in Amar, supra note 32.

Mechanisms of Democracy

The Veil of Uncertainty

be unable to identify the distribution of the law’s burdens with any precision, but that is only true in a very partial sense. Legislators enacting a general but retroactive statute may not know precisely who falls within the statute’s terms, but because the statute is retroactive they need only know whether they and their allies have already committed the relevant crime in the past; if they have not, then the statute’s burdens can only ever fall upon outsiders. The third statute presents a somewhat more difficult case than the first two. In a challenge to the statute brought under the Bill of Attainder Clause, the key question would be whether the statute is excessively specific, but the notion of “specificity” may itself be cashed out in different ways. One view has it that legislation displays unconstitutional specificity when, and perhaps only when, “the legislation defines a closed class, a class with a membership that is permanently fixed when the class is defined, from which members can never exit and into which nonmembers can never enter.”34 Yet, as the Supreme Court has noted, the closed-class test proves too restrictive if it supplies a sufficient condition for an attainder violation, because it would prohibit the imposition of seemingly benign burdens. Consider a statute that forbids epileptics from operating dangerous machinery without a license, on pain of criminal sanctions.35 So another view has it that the touchstone must be whether the legislative purpose in defining a suspiciously small or impermeable class is punitive or legitimately nonpunitive. On the latter view, the epileptic-licensing statute pursues a nonpunitive or essentially regulatory end and is therefore acceptable, while a statute making it a crime for any present or past member of the Communist Party to serve as a labor-union officer embodies punitive purposes and thus constitutes a forbidden attainder.36 Probably either view would condemn the third statute. Under the closed-class view, a statute subjecting a named individual (or an equivalent class of one) to a legal disability is a paradigmatic attainder. Even under the more permissive nonpunitive-purpose view, it is hard to imagine a nonpunitive rationale for legislating against a named individual that would not apply to some other potential malefactor as well. One could 34 35 36

Roderick M. Hills, Jr., Is Amendment 2 Really a Bill of Attainder? Some Questions About Professor Amar’s Analysis of Romer, 95 Mich. L. Rev. 236, 240 (1996). See United States v. Brown, 381 U.S. 437, 454 n.29 (1965); Laurence H. Tribe, American Constitutional Law 646 n.25 (2d ed. 1988). See Brown, 381 U.S. at 450–56.

Institutional Design Writ Small

43

Chapter 

label the latter point as an equal protection problem rather than an attainder problem,37 but the core concern about inadequate generality is the same under either rubric. Hence, prospectivity without generality is as condemnable as generality without prospectivity, and the upshot of the hypotheticals is that the twin requirements of prospectivity and generality both do some independent veil-generating work. The former prevents decisionmakers from reaching into the past to burden identified parties, while the latter prevents decisionmakers from loading down identified parties with special future burdens. This account may load the dice by using examples that impose burdens rather than benefits. Constitutional rules that restrain the retroactive imposition of burdens, such as the ex post facto prohibition, usually do not restrain retroactive benefits at all. Likewise, the Bill of Attainder Clauses do not prevent legislatures from singling out named beneficiaries for special largesse. It might be thought embarrassing for the veil of ignorance account of prospectivity and generality that these rules reach only burdens. If the core feature of a veil rule is that it denies decisionmakers the ability to act on information about identified or identifiable individuals or groups in order to promote impartial decisions, the restriction of the ex post facto and attainder prohibitions to burdens rather than benefits may suggest that the current constitutional rules are underinclusive. This objection, while not frivolous, is probably overblown. First, a doctrinal point in the federal system is that while general benefits are usually immune from constitutional challenge, specific benefits are always subject to an equal protection challenge for lack of a rational basis, and will be strictly scrutinized if the specific class of beneficiaries is defined along illicit lines. In state systems, “special legislation” clauses go even farther toward prohibiting targeted benefits. In terms of constitutional clauses, the generality and prospectivity requirements are better tailored to the veil rationale than a constricted focus on the ex post facto and attainder prohibitions might indicate. Second, the harms of self-interested legislative decisionmaking are plausibly higher where decisions selectively allocate burdens than where decisions selectively allocate benefits, despite the abstract economic equivalence between losses and foregone gains. Constitutional designers might intuitively grasp that, for good or bad cognitive reasons, the law’s subjects 37

44

See Vill. of Willowbrook v. Olech, 528 U.S. 562 (2000).

Mechanisms of Democracy

The Veil of Uncertainty

count the selective imposition of a burden, measured from some preexisting baseline, as a more grievous loss than the selective denial of an opportunity to receive some benefit, measured from the same baseline.

Durability A standard claim in the constitutional choice literature holds that as the durability of rules is increased toward the limit case of permanence, constitutional designers become increasingly uncertain about how the choice of rules will affect their future interests.38 Durability means that the designer must take into account long-term as well as short-term effects; the inherent unpredictability of long-term interests means that the designer can do no better than to choose impartially. An important special case of durability arises where the decisionmaker anticipates that she will occupy different, even opposed, roles in a series of future transactions that will be governed by a durable legal rule. In nonconstitutional contexts, a decisionmaker who must choose rules of contract law, for example, might be led to choose impartially by anticipating that (1) the rules will govern a long series of future contracts and that (2) she will sometimes occupy the role of buyer, sometimes the role of seller. The veil of ignorance arguments advanced at the Federal Constitutional Convention, noted by several commentators, have precisely this structure. In arguing for direct popular election of House members, rather than election by state legislatures, George Mason argued that: We ought to attend to the rights of every class of the people. He [Mason] had often wondered at the indifference of the superior classes of society to this dictate of humanity & policy, considering that however affluent their circumstances, or elevated their situations, might be, the course of a few years, not only might but certainly would, distribute their posterity throughout the lowest classes of Society. Every selfish motive therefore, every family attachment, ought to recommend such a system of policy as would provide no less carefully for the rights—and happiness of the lowest than of the highest orders of Citizens.39 38

39

See, e.g., Brennan & Buchanan, supra note 8, at 29–30. For a meditation on durability, with reference to Article V, see Lawrence Sager, The Birth Logic of a Democratic Constitution, in Constitutional Culture and Democratic Rule 130 (Ferejohn ed., 2001). 1 The Records of the Federal Convention of 1787, at 49 (Max Farrand ed., 1966).

Institutional Design Writ Small

45

Chapter 

Likewise, Gouverneur Morris decried the delegates’ attachments to the interests of their home states, on the ground that “after all how little can be the motive yielded by selfishness for such a policy. Who can say whether he himself, much less whether his children, will the next year be an inhabitant of this or that State.”40 Similar examples are strewn throughout the Convention records. Here our subject is not durability at the level of decisions about constitutional rules, but durability at the level of decisions made under constitutional rules. Perhaps rules increasing the durability of in-system decisions will produce veil-like effects, as long as decisionmakers are aware that their decisions will remain in place for long periods. Thus Article V’s relatively onerous procedures for constitutional amendment have been said to produce a veil effect because amenders have to abide by the amendment in remote future circumstances that they are unable to predict at the time of amendment.41 Procedures for constitutional amendment are a special case, because they govern decisions of a hybrid character; proposals for amendment are offered by within-system decisionmakers, but once effective the decision has constitutional status, and to that extent represents an exercise of constitutional choice within a restricted domain. A cleaner example is the set of rules that govern the durability of constitutional interpretations rendered by in-system decisionmakers, principally the doctrine of stare decisis for constitutional questions. Here the claim is that interpreters will reason impartially if they anticipate that the decision may be invoked in future cases whose valence in terms of the decisionmakers’ future interests is unpredictable. Upholding the free speech claim of a left-wing dissenter now may require upholding the free speech claim of a right-wing dissenter later. The point generalizes beyond judicial decisionmaking to any institution that follows a de jure or de facto norm of precedent. If Republican senators impeaching a Democratic president in 1998 know that their interpretations of the relevant constitutional provisions may govern the impeachment of a Republican president by 40 41

46

Id. at 531. See Michael A. Fitts, Can Ignorance Be Bliss? Imperfect Information as a Positive Influence in Political Institutions, 88 Mich. L. Rev. 917, 967 n.173 (1990); John O. McGinnis, The Inevitable Infidelities of Constitutional Translation: The Case of the New Deal, 41 Wm. & Mary L. Rev. 177, 209 (1999); John Ferejohn & Lawrence Sager, Commitment and Constitutionalism, 81 Tex. L. Rev. 1929, 1958 (2003).

Mechanisms of Democracy

The Veil of Uncertainty

Democratic senators on some unpredictable future occasion, the claim runs, they will be pushed toward impartial interpretations, or at least moderate ones.42 Durability is analytically distinct from generality. A fact-specific decision may be quite durable in the empty sense that it governs no future cases and thus provides no occasion for reconsidering its force, while general rules often prove short-lived. But there is a plausible empirical or causal connection between the two features. Constitutional designers might require in-system decisions to be durable as well as general in order to prevent decisionmakers from circumventing generality requirements. If generality requirements are only imperfectly enforceable, a decision cast in adequately general terms that favors decisionmakers’ short-term interests will prove more attractive if the decision can be rescinded once it has done its work, less so if it will remain in force for the remote future. As I discuss shortly, however, the converse point is that durability without generality will not succeed in producing a veil effect. A durability requirement can be circumvented if the original decisionmakers can issue a decision of excessive specificity, or if subsequent decisionmakers can eviscerate a prior decision by narrow interpretation. Despite the frequent invocation of durability as a veil-producing mechanism, it is hardly clear that durability successfully dampens decisionmakers’ self-interest. First, the mechanism supposes, implausibly, that decisionmakers care equally about both the present and the remote future. If decisionmakers substantially discount the future, however, it is perfectly rational for them to choose rules according to their short-term interests. If the rule indicated by short-term expediency is sufficiently beneficial, the discount rate will ensure that its expected value overwhelms other candidates even if that rule has predictably adverse effects on long-term interests. Introducing uncertainty about long-term interests does nothing to dilute the effect of the discount rate; indeed it strengthens it. Furthermore, when the decisionmaker has decent information about the short-term effects of the rule, as is often the case, ordinary principles of decisionmaking under risk or uncertainty will counsel that the unpredictable long-term effects of the rule should be ignored; the decisionmaker should take the bird in hand without worrying about what is 42

See Neal Kumar Katyal, Impeachment as Congressional Constitutional Interpretation, 63 Law & Contemp. Probs. 169, 186 (2000).

Institutional Design Writ Small

47

Chapter 

left in the bush. The veil arguments advanced at the Federal Constitutional Convention proved largely inadequate to overcome this effect. In many cases, although not all, delegates from small and large states pursued their own and their principals’ short-term interests, making it necessary to settle the major choices not only by rational argument but also by frank bargaining.43 A second problem is that rules that make decisions relatively durable may be quite difficult to enforce. Decisionmakers who anticipate the durability of rules may opt for fact-specific standards, precisely in order to circumvent the veil effect of durability. A highly specific decision guts durability of any substance; the initial decision will govern any future cases to which it applies, but it won’t apply to any future cases.44 By making a decision in the form of a ticket good for one day only, courts may sidestep the veil effect that would otherwise arise from before-the-fact uncertainty about whether the decision might rebound in harmful directions in future cases. Of course, even that uncertainty might prove a feeble constraint where a majority of the court prefers to take the gains from deciding a current case in a manner conformable to the majority’s preferences, at the price of expected losses in future cases; this is the point about discount rates again. Durability rules are also extremely porous after the fact. Subsequent decisionmakers will attempt to undo prior decisions that constrain their current interests, and for two reasons they will often succeed. First, no durability rule does or could make an initial decision permanent. Under current American law at least, there is no absolute rule requiring adherence to past precedent, by contrast to the rule that used to govern the decisions of the House of Lords. The normative case for rejecting an absolute rule of precedent is that later decisionmakers have better information than initial decisionmakers about the current effects of prior decisions in light of changing social circumstances. Here again, hoped-for veil effects are compromised by the need to ensure that decisionmakers possess adequate information. But the revisability of decisions also means that subsequent decisionmakers may opportunistically overturn prior decisions 43

44

48

Jon Elster, Arguing and Bargaining in Two Constituent Assemblies, 2 U. Pa. J. Const. L. 345 (2000); Robert A. McGuire, Constitution Making: A Rational Choice Model of the Federal Convention of 1787, 32 Am. J. Pol. Sci. 483 (1988). See Bush v. Gore, 531 U.S. 98, 109 (2000) (per curiam) (“Our consideration is limited to the present circumstances, for the problem of equal protection in election processes generally presents many complexities.”).

Mechanisms of Democracy

The Veil of Uncertainty

that disfavor them (at present), and if their information is also better than that of contemporaneous observers, it will be difficult for those observers to sort out warranted from unwarranted revisions. Second, even short of an outright overruling or revision of an earlier decision, the earlier decision is only as robust as whatever theory of interpretation is subsequently in place. If the interpretive rules are loose, clever readings can narrow or distinguish inconvenient decisions after the fact almost to the same extent that decisions can be limited at the outset. As a conceptual matter, this sort of subsequent gamesmanship does not detract from the initial veil effect produced by a durability requirement, but the twist is that the subsequent unenforceability of durability rules can be anticipated by the initial decisionmakers themselves. Those decisionmakers may reason that moderating their current decisions in order to forestall the possibility that the decision will rebound harmfully in future cases is foolish. If future interpreters of the current decision are hostile to the future interests of the current decisionmakers, and if the boundaries of fair interpretation are wide, then the future interpreters will not be constrained to abide by the moderate character of the initial decision in any event. This dynamic undercuts Neal Katyal’s claim that if there is a norm of legislative precedent, Republicans interpreting the constitutional standards for impeachment in the case of a Democratic president will moderate their decisions to forestall a future impeachment in the opposite direction.45 Current Republicans may reason that future Democrats can and will distinguish or redescribe the current decision in any event, so that current restraint would amount to nothing more than unilateral disarmament.

Delayed Effectiveness The pervasive weakness of the durability mechanism is that decisionmakers remain aware of their short-term interests, and if they discount sharply enough, or have decent information only about the short term, they will rationally ignore the long-term uncertainty produced by the durability requirement. A mechanism that attempts to lengthen the time horizon is to delay the effectiveness of decisions, either for some fixed period or until after some future event. An example might be a constitutional rule providing that all ordinary legislation will take effect only X years after the 45

See Katyal, supra note 42, at 186.

Institutional Design Writ Small

49

Chapter 

date of enactment. Many constitutions stipulate that constitutional amendments or certain classes of statutes take effect only after being approved by two successive sessions of the legislature, and this has something of the same effect. A delayed-effect rule produces uncertainty only in an indirect sense. Such rules impose no direct restriction on decisionmakers’ information, nor do they introduce a new source of uncertainty. Instead, delay rules take advantage of a preexisting uncertainty—the inherent unpredictability of the decisionmakers’ long-term interests—that would otherwise be overwhelmed by the incentives to focus on short-term considerations. By confining the range of the decision’s application to cover only the period (beyond the delay period) in which decisionmakers’ interests are unpredictable, a delay rule ensures that the only period current decisionmakers can affect is one that is, from their ex ante standpoint, subject to a veil of uncertainty. The example of a general constitutional rule that explicitly delays the effective date of ordinary legislation is obviously fanciful, and that is interesting in itself. The closest analogue in the federal Constitution is a matched pair of rules in Article I and in Article V. The Article I provision forbade Congress to prohibit the “Migration or Importation of such Persons as any of the States now existing shall think proper to admit” (read: slaves) until the year 1808. The latter provision made the former provision unamendable before 1808 (at which point any amendment would be moot). But the analogy is not very close. These rules prohibited action until a particular time certain; Congress could and did legislate in 1807 with no uncertainty at all about the distribution of benefits and burdens from the decision. A delayed-effectiveness rule, by contrast, produces veil-like effects only because it builds in a delay period that attaches no matter when the legislation is enacted or the decision made. Nonetheless, there are constitutional rules that might be explained or rationalized as incorporating a delay strategy of a sort. I will discuss two examples: the Twenty-Seventh Amendment and the Emoluments Clause. The Twenty-Seventh Amendment provides that “[n]o law varying the compensation for the services of the Senators and Representatives shall take effect until an election of Representatives shall have intervened.”46 The Amendment was not formally ratified until 1992, but that is irrelevant 46

50

U.S. Const. amend. XXVII.

Mechanisms of Democracy

The Veil of Uncertainty

for my current purpose, which is to use the Amendment as an example of the delaying tactic, rather than to make claims about its historical significance. The purpose and effect of the provision is less obvious at the second glance than at the first. Certainly the target is legislative self-dealing over salary and other compensation, a serious concern given the Constitution’s vesting of appropriations power in the legislature. But the question is why the ordinary operation of the election cycle will not prevent legislators from conspicuous feeding at the public trough. On standard public-choice premises, constitutional designers should worry more about less visible forms of self-dealing, such as kickback schemes and in-kind exactions from citizens affected by legislation. Put differently, the Amendment cannot be explained as an attempt to restrain legislators by subjecting them to the anticipation of a post-hoc electoral check. That motive would operate with full force even in the absence of the Amendment as long as the pay raise became public knowledge, and publicity is guaranteed elsewhere by constitutional rules that require each house of Congress to publish the journal of its proceedings, that require appropriations to be made by (published) statutes, and that require regular publication of the expenditures of all public money. The key to the Amendment’s design is that it delays the effective date of a legislative pay raise until after the next election (of representatives). The delay prevents legislators from benefiting during the interim period, and thus ensures that legislators will not consider the question of appropriate pay under the distorting influence of short-term personal interest. As Madison said while introducing the Amendment in the First Congress: Perhaps of all the powers granted, it [i.e.] the legislative power to set legislators’ salaries] is least likely to abuse; but there is a seeming impropriety in leaving any set of men without control to put their hand into the public coffers, to take out money to put in their pockets; there is a seeming indecorum in such power. . . . I have gone, therefore, so far as to fix it, that no law, varying the compensation, shall operate until there is a change in the Legislature; in which case it cannot be for the particular benefit of those who are concerned in determining the value of the service.47 47

1 Annals of Cong. 457–58 (Joseph Gales ed., 1789) (emphasis added).

Institutional Design Writ Small

51

Chapter 

Madison’s explanation was not perfectly tailored to the scope of the rule proposed. To ensure that a salary increase could not operate for the “particular benefit” of the legislators who voted on it would instead require an eligibility restriction, either one that made legislators ineligible for election to a new Congress after voting for a (delayed) salary increase, or one that denied legislators a component of salary equivalent to the increase for which the legislator previously voted. But the Twenty-Seventh Amendment achieves a similar effect by introducing uncertainty about personal benefit, through the strategy of delay. Marginal legislators who are unsure whether they will maintain their seats after the next election will also be unsure whether a current vote for a delayed salary increase will benefit themselves or instead benefit future legislators who will have replaced them, perhaps political rivals from their home states or districts. Without the delay provision, legislators who anticipate a risk of losing the next election would be especially likely to vote for the short-term benefit of an immediate pay raise; with the delay provision, they are especially unlikely to do so. A more complicated and historically more important example is the Emoluments Clause, which provides that “[n]o Senator or Representative shall, during the Time for which he was elected, be appointed to any civil Office under the Authority of the United States, which shall have been created, or the Emoluments whereof shall have been encreased during such time.”48 Justice Story’s view was that the Clause attempts to suppress “venality.” As Story wrote, “[t]he reasons for excluding persons from offices, who have been concerned in creating them, or increasing their emoluments, are, to take away, as far as possible, any improper bias in the vote of the representative, and to secure to the constituents some solemn pledge of his disinterestedness.”49 Absent the Clause, not only would legislators have the power to vote on the existence or perquisites of offices they would be eligible to fill, but they could strike collusive deals with the President, distorting policy in other areas in order to obtain a presidential promise of future appointment to a new or enhanced office.

48 49

52

U.S. Const. art. I, § 6, cl. 2. Joseph Story, Commentaries on the Constitution of the United States at 311 (Carolina Acad. Press 1987) (1833). A somewhat different account of the purpose of the Clause emphasizes the fear that legislative-executive collusion would create an excessive number of new federal offices, expanding the federal government at the expense of the states. John F. O’Connor, The Emoluments Clause: An Anti-Federalist Intruder in a Federalist Constitution, 24 Hofstra L. Rev. 89, 164–68 (1995).

Mechanisms of Democracy

The Veil of Uncertainty

Here too, however, the precise scope of the Clause seems poorly tailored to the aim of suppressing venality. Subjecting each legislator to a permanent ineligibility to federal offices that were created or whose emoluments were increased during the legislator’s term would detach selfinterest from decisions about the existence and compensation of federal offices. This is something of an exaggeration, because even a permanent disability would leave scope for legislators to advance friends and relatives to federal offices, to create a new office in the hope of assuming an old office after its current occupant has moved to the new one, or to carry out other subtle schemes of evasion. But the Framers argued that suppressing the core case of biased decisionmaking would be worthwhile even if other (but costlier) substitutes remained available.50 Instead, the Clause as enacted only makes the legislator ineligible for appointment during the term for which the legislator was elected. The historical explanation for this slight mistailoring is that the current form of the Clause was a compromise among the Convention delegates. An initial proposal would have made legislators ineligible for any federal office for up to one year after the legislator’s term ended. Madison’s eventual proposal, which he described as a “middle ground,”51 restricted the Clause along both dimensions to palliate other delegates’ concerns that a broad ineligibility would discourage too many talented but ambitious candidates from seeking federal legislative office. Purely as a textual matter, however, the actual wording of the Clause might be justified as a delay rule aiming to produce uncertainty and consequent veil-like effects. Consider the choices facing a legislator who must vote up or down on a bill to create a new office or increase the emoluments of an old one, and who has some hope or expectation of obtaining executive nomination and confirmation to the office. Absent the Clause, the legislator may strike a deal to obtain executive appointment as the price of supporting the bill, vote for the bill, and assume the office, all during the current electoral term—although a different constitutional rule that prohibits simultaneous service in the legislative and executive branches requires him to surrender his current legislative office as the price of moving into the executive. With the Clause, the legislator knows that if the bill is enacted, he will be ineligible to be nominated or confirmed to the 50 51

1 The Records of the Federal Convention of 1787, 387–89 (Max Farrand ed., 1966). Id. at 308.

Institutional Design Writ Small

53

Chapter 

new or enhanced office until after the end of his current term. Note, however, that in this century the Clause has been partially diluted by the so-called Saxbe fix, in which Congress attempts to remove a legislator’s disability under the Clause by reducing the compensation of the office to the level prevailing at the beginning of the legislator’s current electoral term. Therefore, if Congress raises the salary of the Secretary of the Treasury, and the President nominates a sitting senator to the post, the Saxbe fix purports to remove the senator’s ineligibility.52 Nothing I say here turns upon the much-debated question whether the Saxbe fix is constitutional. The delay created by the Clause increases the legislator’s uncertainty about whether he will benefit from the creation or enhancement of the office. For one thing, the postponement decreases the credibility of a presidential promise, given at the time of the vote, to nominate the legislator in the future. The legislator will anticipate that if he loses the intervening election, the President may decide that he can renege without incurring the wrath of any current legislator (one of whom may be offered the job in place of the former legislator), and will also anticipate that if the intervening election coincides with a presidential election, the President may not be around to fulfill the promise anyway. For another thing, the delay takes advantage of the inherent unpredictability of future politics. The legislator may, for example, anticipate or fear that the intervening election will change the composition of the Senate in such a way as to preclude a realistic chance of obtaining senatorial confirmation to a principal office. Legislative salaries and appointment to federal office were two areas of paramount concern to Framers who feared self-interested legislative decisionmaking. The corresponding provisions, the Twenty-Seventh Amendment and the Emoluments Clause, are structured around delay mechanisms that introduce uncertainty about the distribution of benefits from the relevant decisions. The veil of ignorance strategy, then, is an important feature not only of negative restrictions on legislative action, such as the Ex Post Facto and Bill of Attainder Clauses, but also of structural constitutional law. Veil rules are a central feature but not a pervasive one; there remains a set of puzzles about why the veil rules we have are the only ones we have, and why the ones we have are located where they are. The next chapter turns to those puzzles. 52

54

See Michael Stokes Paulsen, Is Lloyd Bentsen Unconstitutional?, 46 Stan. L. Rev. 907, 908–09 (1994).

Mechanisms of Democracy

Chapter 2

The Limits of Uncertainty

I

begin with a general consideration of the limits of uncertainty and the distribution of veil rules across legislative, executive and judicial institutions. Although the examples are drawn from American constitutionalism, the tradeoffs I identify have broader application. Three tradeoffs are central, and explain the appreciable constraints on law’s use of uncertainty-generating mechanisms. First, uncertainty reduces self-interest, but has the obvious cost of reducing the decisionmaker’s information. Under some circumstances, impartial but ignorant decisions will be worse, even from the social point of view, than decisions that are self-interested but well-informed. Second, and less obviously, impartiality trades off against motivation. Self-interest is often the spur to action. An impartial decisionmaker who can select his own activity level and agenda may often be an excessively passive decisionmaker from the social point of view, acting too rarely. Where a decisionmaker has control of his own agenda, yet cannot pursue self-interested projects, the decisionmaker may pursue too few projects, according to some given normative theory of proper activity levels. Third, impartial enactments face more serious commitment problems than self-interested enactments. The former are more likely to be undone after the veil of uncertainty has lifted, whereas self-interested Institutional Design Writ Small

55

Chapter 

enactments will tend to have an ongoing constituency who will fight fiercely in their defense. The tradeoffs between impartiality on the one hand, and information, institutional motivation or energy, and ex post enforceability on the other, help to justify the constraints on uncertainty as a democratic mechanism for enforcing impartiality. To illustrate these constraints, I examine some recent proposals that in some way or other trade on veil-generating mechanisms to produce impartial reforms, in the areas of judicial tenure, the emergency powers of government, and electoral redistricting. I conclude that these proposals are less likely to succeed when they are addressed to decisionmakers who need take no action at all, and when the commitment problem is most serious, as in the case of statutes intended to govern future wars and emergencies. They are more likely to succeed when they are addressed to decisionmakers who lack agenda control (at least with respect to the subject of the proposal), and when commitment problems are overcome by the routine status quo bias of the legislative process.

The Distribution of Veil Rules A striking feature of the veil rules explicated in Chapter 1 is their skewed distribution across the Constitution. The Ex Post Facto and Bill of Attainder Clauses constrain Congress, but there are no similar checks on independent executive action—meaning executive action taken pursuant to a freestanding grant of constitutional authority, such as the pardon power, rather than action taken in the President’s capacity as a participant in the process of statutory enactment, such as the signature or veto of a bill. Moreover, there is only a weak form of ex post facto check on judicial decisionmaking, as discussed above. The Twenty-Seventh Amendment and the Emoluments Clause constrain legislative self-dealing and collusion with the executive, but there are no equivalent checks on independent executive or judicial action. If this unequal distribution does not seem puzzling, it is only because it is so familiar to American constitutional lawyers. There is no reason rooted in the nature of things why constitutional rules might not aim to produce much more pervasive veil-like effects on executive and judicial decisionmaking. Start with the executive branch. Is presidential decisionmaking necessarily retrospective and specific because it is by some essential definition an exercise of the power to execute rather than to legislate? 56

Mechanisms of Democracy

The Limits of Uncertainty

The pardon power, for example, could easily be subjected to requirements of prospectivity and generality; it would then resemble the powers of suspension and dispensation—prospective waivers of general laws—that were long claimed by the English crown.1 So, too, with the power to prosecute. Under the current constitutional rules, Congress may not enact a retroactive criminal statute that identifies and punishes individuals for actions undertaken in the past, but the executive may wait until identified individuals have violated the statute and then select particular violators for prosecution. A more veiled approach would require prosecutors to enact and abide by general and prospective guidelines for exercising prosecutorial discretion—akin to requiring that the Justice Department’s prosecution guidelines be given legal force. The point holds for judicial decisionmaking as well. Judges could be subjected to veil effects by anticipated durability, as a less malleable doctrine of precedent might accomplish, or by pushing courts to decide cases prospectively in order to suppress bias-inducing information. Federal courts, for example, might be permitted or even required to issue advisory opinions in advance of controversies, without knowledge of the identity of particular litigants or of particular facts; courts in several states and in other countries are subject to a variety of similar requirements. Many of those strategies could be adapted to executive-branch settings as well. It is something of a puzzle, therefore, that the Constitution uses veil rules principally to control legislative action, but not executive or judicial action. I will ask whether there is any justification for this pattern. I do not assume that the justification, if there is one, explains the pattern. There is no reason to presume, in the absence of argument, that the constitutional design is optimal in this (or any other) respect. Nonetheless, I believe that there happens to be a good justification for the pattern we observe; a wellmotivated constitutional designer could well approve of the distribution of veil rules across the American constitution, given certain empirical conditions I will indicate. I begin with some extant accounts of the skewed distribution of veil rules. One such account discernible in the literature, although more as an implicit assumption than as an articulated theory, is that veil rules are unnecessary where there are alternative institutional-design features that check self-interested decisionmaking. Legal scholar Rick Hills offers an 1

See W.S. Holdsworth, A History of English Law 217–25, 240–41 (1927).

Institutional Design Writ Small

57

Chapter 

elegant account of the Bill of Attainder Clause, for example, under which the Clause enforces legislative impartiality—a condition secured in the adjudicative context by design features such as life tenure, precedent, the ban on ex parte contacts, and so on. “The veil of ignorance in the legislative chambers, in effect, replaces the blindfold on the face of Justice in the courtroom, and creates a different sort of blindness that accomplishes the same sort of effect—a minimal degree of impartiality.”2 This resonates with important constitutional themes, but it supplies only an incomplete justification for the distribution of veil rules in the American Constitution. We would want to know what, on this account, secures impartiality in the executive branch, given that the presumed source of legislative partiality—responsiveness to self-interested factions— is also present in the executive, while the claimed guarantors of judicial impartiality are notably absent. Nor can we simply say that the executive and judiciary are both constrained by the obligation to enforce statutes that must themselves be enacted by legislatures subject to pervasive veil constraints. That would fail to justify the absence of veil rules governing independent, nonexecutory presidential and judicial powers, such as the pardon power and constitutional judicial review. Perhaps unique institutional-design features of the presidency might be said to promote executive impartiality. Madison argued that groups are more likely to indulge in self-interested decisionmaking than single individuals, since each member of the group incurs only a fractional share of the public opprobrium resulting from the self-interested decision.3 On this view, perhaps the relatively unitary character of the presidency partially forces the President to internalize the reputational costs of (visible) self-dealing. But then we would also be entitled to point to unique institutional-design features of the Congress as guarantors of legislative impartiality; examples are bicameralism, long Senate terms, and selection from large districts—all of which Madison argued would dampen self-regarding action by legislative groups. In that case the puzzle would be not why the executive and judiciary are not subject to 2

3

58

See Roderick M. Hills, Jr., Is Amendment 2 Really a Bill of Attainder? Some Questions About Professor Amar’s Analysis of Romer, 95 Mich. L. Rev. 236, 242 (1996); see also Harold J. Krent, Should Bouie Be Buoyed?: Judicial Retroactive Lawmaking and the Ex Post Facto Clause, 3 Roger Williams U. L. Rev. 35, 41 (1997) (arguing that “legislative retroactivity in the criminal context is more disfavored than lawmaking by judges” because institutional-design features such as life tenure insulate judges from interest-group pressures). The Federalist No. 10, at 125–26 (James Madison) (Isaac Kramnick ed., 1987).

Mechanisms of Democracy

The Limits of Uncertainty

veil-producing rules, but why the legislature is. The account based on alternative checks against self-interest either leaves unexplained the absence of executive-branch veil rules or else poses the question why veil rules are necessary or desirable for any branch. A different sort of account would invoke the pervasive tradeoff between information and neutrality. In many settings, supplying decisionmakers with more information produces more decisional bias; restricting information in order to reduce bias produces impartial but poorly informed decisions. The regulatory specialist or expert understands an industry because he worked in it and has lots of friends who still do, while the administrative or judicial generalist who is free of similar bias-producing ties does not understand the subject as well.4 We might trade on this insight in the present context to argue that veil rules are most likely to be found where the costs of poorly informed decisions are plausibly lower than in settings where veil rules are not found. Perhaps the direct informational costs of throwing veils over congressional decisionmaking are lower than in executive and judicial settings; perhaps the legislature’s information will remain in some sense adequate to its tasks even with veil rules in place, while that of other branches will not. The basic insight is powerful, and it does much to justify the fact that constitutional rules do not impose far more stringent veil strategies than they actually do. A rule delaying the effective date of all legislation for ten years, for example, might produce enormous veil effects, yet prove socially disastrous because of the rapid pace of social change; the legislation in effect at any time would have been enacted by the ill-informed legislators of a decade ago. However, the information-neutrality tradeoff is general across institutions, and thus lacks bite. It offers no account of the distribution of veil rules that the Constitution actually embodies. I suggest the following account. The key is to focus not on the direct informational costs of veil rules, but on their indirect and secondary effects on decisionmakers’ motivations and activity levels—on what the Framers called institutional “energy.”5 The Framers’ Humean theory of political psychology described self-interest as the principal spur to action.6 4 5 6

See Saul Levmore, Efficiency and Conspiracy: Conflicts of Interest, Anti-Nepotism Rules and Separation Strategies, 66 Fordham L. Rev. 2099 (1998). See The Federalist Papers No. 70, supra note 3, at 402 (Alexander Hamilton) (discussing “[e]nergy in the executive”). Morton White, Philosophy, the Federalist, and the Constitution (1989).

Institutional Design Writ Small

59

Chapter 

Impartial reason was normatively superior but practically feeble; the pallid claims of reason could never provoke the same degree of energy and activity as the prospect of selfish gain. A constitutional designer who subscribes to this political psychology might see the distribution of veil rules not only as a direct strategy for checking decisional self-interest, but also as an indirect means for allocating or adjusting energy or activity levels across competing institutions. A decisionmaker who controls his own activity level or agenda, and is subject to a complex of veil rules, would predictably undertake fewer projects and make fewer decisions (and fewer selfinterested decisions) than would a decisionmaker enjoying full information about the distribution of benefits and burdens from its decisions. Conversely, the constitutional designer might sensibly decline to impose veil rules on an institution intended to display great energy; the price of that energy, more self-interested decisions, would be worth paying in order to secure more projects and activities overall. On this picture, the skewed distribution of veil rules in the Constitution—with legislative action heavily veiled and presidential action veiled hardly at all—indirectly encourages executive activity relative to legislative activity, even if the price is that a greater fraction of executive than of legislative activity will be selfinterested, because that price is worth paying for overall gains. Although it is not essential to the argument, which is justificatory rather than explanatory, I add that this rationale accords well with some of the Framers’ principal aims in constitutional design. The story is familiar. The excesses of populist legislatures in the period of the Articles of Confederation caused the Framers to regard legislatures as the “most dangerous branch”; the one that was most likely to usurp the powers of other branches, disturb vested property rights, and in general provide scope for unrestrained factionalism. Perhaps Madison’s central concern, both during the Convention and in the ratification debates, was to dampen the “restless energy” of faction-driven legislatures, and we may interpret veil rules as devices for suppressing that energy by removing the impetus of self-interest. To be sure, as previously discussed, Madison’s principal strategy for checking self-interest was the institutional competition arising in a system of separated powers. But it is noteworthy that when Madison sought to justify the constitutional requirements of prospectivity and generality embodied in the Ex Post Facto and Bill of Attainder Clauses, his defense sounded precisely this concern about the activity-dampening effects of veil rules: 60

Mechanisms of Democracy

The Limits of Uncertainty

Bills of attainder, ex post facto laws, and laws impairing the obligation of contracts, are contrary to the first principles of the social compact and to every principle of sound legislation. . . . The sober people of America are weary of the fluctuating policy which has directed the public councils. They have seen with regret and indignation that sudden changes and legislative interferences, in cases affecting personal rights, become jobs in the hands of enterprising and influential speculators, and snares to the more industrious and less informed part of the community. They have seen, too, that one legislative interference is but the first link of a long chain of repetitions, every subsequent interference being naturally produced by the effects of the preceding.7 Madison’s point here is not that enactments by legislative factions free to implement self-interested plans are necessarily substantively bad enactments (although we have other grounds for inferring that he held that view as well). Rather, the argument is that excessive legislative activity taken under the spur of self-interest produces instability that is socially damaging independent of the content of enactments. Self-interested factions will change property and contract rules too frequently, retarding productive investments and long-term commercial planning. Prospectivity and generality would dampen legislative turbulence and “give a regular course to the business of society.”8 The concern for property and settled expectations is why Madison ties prospectivity and generality to the Contracts Clause; although the Supreme Court has today largely abandoned that link, the specific character of the harm flowing from excessive levels of legislative activity is essential to the theory. Conversely, some of the Framers hoped to imbue the executive branch with the “vigor” it lacked in state constitutions built around legislative supremacy. Hamilton’s series of essays on the presidency has as its guiding thread the premise that “[e]nergy in the executive is a leading character in the definition of good government,” energy that Hamilton contrasts with the “deliberation and wisdom” that a (suitably impartial) legislature would display.9 Far from wishing to devise rules that would 7 8 9

The Federalist No. 44, supra note 3, at 287–88 (James Madison). Id. at 288. See The Federalist No. 70, supra note 3, at 402–03 (Alexander Hamilton).

Institutional Design Writ Small

61

Chapter 

suppress self-interested motives for presidential action, Hamilton argued for structuring the office in ways that would encourage the President’s personal interest in and attachment to his station and the exercise of its powers. That personal interest would in turn supply the President with a motive to govern actively, to undertake publicly beneficial projects (from which a unitary executive would reap all the reputational gains), and to resist legislative encroachments.10 By combining self-interest with opportunity, precisely the combination that veil rules prevent through their effects on information, the Framers hoped to promote executive activity of the very sort that they sought to discourage in legislatures. There are many historical controversies about whether Hamilton’s view of the executive was shared by other Framers. And, of course, the power of the modern executive would have been unimaginable and perhaps horrific to Madison, or even to Hamilton. Again, however, it is inessential to the account I offer that the foregoing historical suggestions are correct. If the current distribution of veil rules across the Constitution is justifiable, given current conditions, then from our standpoint it is irrelevant how it came about. We cannot assume that current justifications also explain the constitutional design we now have, but we are under no compulsion to assume that anyway.

Uncertainty, Delay and Motivation I now turn to some applications of these ideas with contemporary relevance. The tradeoff between impartiality and motivation or energy becomes most critical where proposals trade on the mechanism of delayed effectiveness, discussed above. The delay mechanism is among the most appealing and broadly useful veil-generating mechanisms, as pointed out in important work by legal scholars Ariel Porat and Omar Yadlin.11 By diluting current decisionmakers’ ability to assess where their own interests lie, delayed enactments might be thought to ease the path of reform in a range of settings. In fact, however, successful delayed enactments are uncommon, as Porat and Yadlin acknowledge. Their explanation involves the 10 11

62

See The Federalist No. 71, supra note 3 (Alexander Hamilton) (arguing for four-year presidential terms on this basis). See Ariel Porat & Omar Yadlin, Promoting Consensus in Society Through DeferredImplementation Agreements, 56 Univ. of Toronto L. J. 151 (2006).

Mechanisms of Democracy

The Limits of Uncertainty

commitment problems that afflict impartial enactments; I examine these shortly. Another major reason, the one I will turn to first, is the tradeoff between impartiality and motivation. Delay may buy off opposition, but it also eliminates the short-term, self-interested benefit that would accrue to the political actors who would otherwise support the proposal and incur the costs of shepherding it through the legislative process. Delay replaces self-interested motivation with impartial reason, but the latter motivation is frequently too feeble to produce action; the pallid claims of reason rarely provoke the same degree of energy and activity as the prospect of selfinterested gain. Political action oriented toward the long run tends to occur only where the political valence of the proposal is clear ex ante, as in proposals for environmental regulation with long-term effects. Even in such cases, the long-run character of the relevant concerns saps most peoples’ motivation to act; for both good and bad reasons, many think it acceptable to let future generations take care of themselves. To be sure, another possibility is that a political coalition may be organized around one issue and, on a separate issue, hold only weak preferences or even be indifferent. In this scenario, farsighted reform is possible because the coalition’s leaders enjoy a certain degree of slack or freedom from constituent pressure with respect to that issue; elected on another program entirely, they may seize the opportunity to implement reform if they happen to favor it, without fracturing their political base.12 But this is usually possible only with proposals that provoke no strong opposition, which will also tend to be low-stakes proposals for marginal reforms. Where the governing coalition is largely indifferent to the reform because the benefit to its members is small, but there is strong opposition because the costs to others are high, the coalition’s leaders, however devoted to the reform, will be unable to mobilize enough support to secure enactment. More generally, the higher the stakes in a reform proposal, the more opposition it will tend to provoke. Reform can and does occur, but this tendency means that most of the reforms that do succeed will be marginal accomplishments. For concreteness, I will consider several proposals that attempt to trade on uncertainty generated by a delayed onset date, but that run aground on the motivational tradeoff I have sketched. The first is Paul Carrington and Roger Cramton’s recent proposal for reforming the 12

Thanks to Mark Tushnet for emphasizing this possibility.

Institutional Design Writ Small

63

Chapter 

Justices’ tenure to a fixed term of years,13 which in effect delays the proposal’s onset date by grandfathering currently sitting Justices. The grandfathering presumably buys off self-interested opposition on the part of sitting Justices.14 The safe prediction is that it will never be enacted (for reasons over and above the high base rate of failure for all public-policy proposals). The delayed onset of the proposal makes its political valence uncertain, which in turn means that no current political actor will be strongly motivated to shoulder the burdens of enactment. Another example is Bruce Ackerman’s proposal for a framework statute regulating emergency powers, which he hopes will be enacted “before the next attack.”15 In this proposal, the delayed-onset is de facto, not de jure. Ackerman’s framework statute would go into effect immediately as a nominal matter, but its operation would not be triggered until some point in the indefinite future, when a large-scale terrorist attack occurs. But legislators are unlikely to enact a framework statute to regulate emergencies in the hazy future. Although in such a position legislators would act impartially, behind a veil of uncertainty that suppresses knowledge of the statute’s short-run political payoffs, that very uncertainty saps legislators’ motivation to act, and thus makes it less likely that any legislation will be enacted in the first place. The high opportunity costs of political action, constricted agenda space in Congress, the horizon of reelection, and the tendency to discount the future all push legislators to rank projects by the amount of benefit they produce in the near term. Projects that will produce large collective benefits in the long run, but whose distributive valence is uncertain, will generally be subordinated to projects that produce larger factional benefits in the short run. Legislating for the remote future replaces self-interested motivation with impartial reason, but impartial motives are often too weak to produce action. As we will see in Chapters 6 and 7, quasi-constitutional framework statutes sometimes do slip between these two opposing forces in moments of (what Ackerman would call) “higher lawmaking.” But to date, at least, Ackerman’s own proposal does not seem to be of this kind. 13

14 15

64

See Roger C. Crampton & Paul D. Carrington, Reforming the Court: An Introduction, in Reforming the Court: Term Limits for Supreme Court Justices (Roger C. Crampton & Paul D. Carrington eds. 2006). This is my interpretation of the proposal’s effect; I do not assert that Carrington and Cramton intend this effect. See Bruce Ackerman, Before the Next Attack: Preserving Civil Liberties in the Age of Terrorism (2006).

Mechanisms of Democracy

The Limits of Uncertainty

The civil-libertarian constituencies, of right as well as left, that would provide the core support for Ackerman’s proposal are most likely to be mobilized and highly motivated where there is a threat to civil liberties; and in such cases there will usually also be a corresponding or offsetting threat to security that will mobilize resistance on the other side. The special conditions required for impartial framework statutes to slip through seem least likely to occur during the very wars or emergencies that would be necessary to give Ackerman’s proposal any real base of support. A structurally similar problem afflicts John Hart Ely’s proposal that Congress should rouse itself to enact a strengthened version of the War Powers Resolution16 in order to place constraints on presidential warmaking before the fact—the same type of suggestion that Ely elsewhere mocks as “a halftime pep-talk imploring [Congress] to pull up its socks and reclaim its rightful authority.”17 Ely recognizes that political constraints prevent Congress from challenging the executive in times of emergency (even if legislators desire to do so), and he thus expresses the hope that “stepping up and taking responsibility on the question whether to fight a particular war for which the president is beating the drums, and binding oneself in advance and outside any particular context to do so, are at least potentially different matters.”18 Of the many possible objections to this reasoning, only one is relevant here: the very same abstraction from context on which Ely relies to loosen the political constraints also saps legislators’ motivation to take any action, given the opportunity cost of foregoing other projects with concrete short-term payoffs. The War Powers Resolution itself might be depicted as the rare moment of “higher lawmaking” in which legislators overcame this motivational deficit, but feasible proposals cannot count upon the repetition of an exceptional event. Overall, the tradeoff between impartiality and motivation forces reform proposals to slip between Scylla and Charybdis. On the one hand, stipulating that a proposal should have a delayed effective date helps to remove the taint of self-interested or partisan motives and avoids triggering the heuristic that one should not try to change the rules in the middle of the game. On the other hand, delaying the proposal’s effect means that 16

17 18

John Hart Ely, War and Responsibility: Constitutional Lessons of Vietnam and Its Aftermath 63–66 (1993). I am simplifying Ely’s position here, because he also hopes that the courts may prod Congress to act. See id. at 54–60. Id. at 52. Id. at 65.

Institutional Design Writ Small

65

Chapter 

concrete problems of the here and now, the sort of problems most likely to motivate reform, cannot be addressed. The delaying tactic makes reform possible by producing a kind of impartiality and buying off current opposition, but the tactic also makes the reform less likely to be proposed and pursued. Conversely, the short-term interest that motivates reform itself tends to create self-limiting political constraints. Proposals that produce short-term benefits for particular groups will attract motivated supporters but will also provoke opposition. This is not to say that generating uncertainty through delay is never a successful strategy of institutional design. It is to say that it is most likely to be successful when the decisionmaker or institution that is placed behind a veil of uncertainty lacks agenda control or cannot select its own activity level, at least with respect to the veiled subject matter. In such cases, where the veiled decisionmaker can be forced to decide, there is no alternative to deciding impartially. Consider Adam Cox’s proposal that legislatures should be required—perhaps by courts—to enact redistricting plans with delayed effect, perhaps as much as two or three election cycles. The benefit of the proposal is that legislators who would otherwise gerrymander districts for some mix of personal and partisan advantage will be pushed, by uncertainty, closer to the type of plan that an impartial designer would select.19 Whatever other costs such a proposal might have, it is not open to the objection arising from the tradeoff between impartiality and motivation. “[I]n the redistricting context . . . the motivational slack that veil rules can produce is not a problem. There is no concern that political actors will simply fail to undertake redistricting if they cannot as effectively pursue their self-interest through the process. This is because constitutional doctrine requires the political actors to redistrict regularly.”20 Institutions that lack agenda control in relevant respects may be subjected to a veil of uncertainty without collateral damage to activity levels or institutional energy.

Uncertainty, Impartiality and Commitment Porat and Yadlin state that Delayed Implementation Agreements (DIA)— a subcategory of veil rules—are rare, and explain this quite persuasively by 19 20

66

See Adam B. Cox, Designing Redistricting Institutions (U. of Chicago, Public Law Working Paper No. 131, 2006). Id. at 14.

Mechanisms of Democracy

The Limits of Uncertainty

reference to problems of commitment. In general, unless the enacting majority in the first period undergoes the very costly process of encoding the DIA in a binding constitutional commitment, there is a high risk that a later majority will repeal the DIA, for reasons I will discuss shortly. Anticipating these commitment problems, the first-period majority will refrain from incurring the costs of enactment. This important point generalizes to all impartial enactments (at least nonconstitutional ones) passed behind the veil of uncertainty, in the following way. In the first period, behind the veil, suppose that a majority coalition can assemble to enact an impartial proposal. In the second period, when the veil has been lifted, the distributive implications of the enactment will become clear; some group or other will realize that its ox has been gored. Importantly, that group may well turn out to be a majority or even a supermajority. There is nothing at all that prevents the median voter, in the second period, from being in the class of those harmed rather than of those benefited by the earlier impartial enactment. By hypothesis the earlier enactment was impartially desirable, but that need not be at all the same thing as beneficial to a subsequent majority. Where the impartial enactment turns out to enhance overall welfare by making transfers from a (later) majority to a (later) minority, or by protecting (later) minority interests or rights at some cost to the (later) majority, the enactment may face a majority coalition agitating for its repeal in the second period. It is true that the first-period enactment will now represent the legal status quo, and will be protected by the ability of intensely interested minorities to block repeals in the legislative process. But that is true of all enactments. In one respect at least, the commitment problems are more serious for impartial enactments than for self-interested ones. Enactments that are rammed through by a self-interested constituency will, for some time at least, have a core group of intensely interested defenders around to defend them from repeal. Of course social circumstances may change, so that the self-interested constituency supporting the prior enactment may disappear. But the coalition that supported a prior impartial enactment may disappear as well; that point is on both sides of the comparison. Subtracting it out, self-interested enactments systematically tend to enjoy a protection against subsequent repeal that impartial ones do not possess. To illustrate these points, let us return briefly to Ackerman’s proposal for a framework statute governing emergencies. Suppose it were Institutional Design Writ Small

67

Chapter 

enacted, and suppose that an impartial decisionmaker would approve of its being enacted, because the large gains in civil liberties for the few would outweigh the slight reduction in security for the many. The very circumstances that make the statute impartially desirable also make it especially vulnerable to being either formally repealed or effectively ignored in subsequent periods. Quasi-constitutional statutory frameworks are at their weakest during wars and emergencies. The War Powers Resolution, which limited the circumstances under which the president could use military force and imposed various reporting requirements when the president did use force, has been ignored—as shown by President Clinton’s clear violation of the Act during the Kosovo conflict.21 The National Emergencies Act similarly imposed restrictions and reporting requirements on the president’s power to declare emergencies, and the International Emergency Economic Powers Act limited the president’s power to impose economic sanctions during emergencies. These statutes have failed, in the end, to constrain presidents very tightly.22 Statutory frameworks governing war and emergencies tend to fail because, when a crisis arises, the executive needs flexibility to protect the subsequent majority from harm, or so the subsequent majority often believes. Let us suppose, compatibly with Ackerman’s view, that the majority is perfectly willing to reduce the civil liberties of a minority, because it does not bear the costs of doing so, and its own security will be enhanced. (This common picture is deeply confused,23 but I will assume it here to elicit the logic of the proposal.) Because statutory limitations determined in advance reduce flexibility, in a way that does not anticipate the particular requirements of a new emergency, the majority will usually have little interest in insisting that these limitations be respected. Ackerman provides no valid reason for thinking that his emergencyregulating framework statute—which is far more ambitious than the other ones—might fare differently. It is true that Ackerman’s framework statute 21 22 23

68

Major Geoffrey S. Corn, Clinton, Kosovo, and the Final Destruction of the War Powers Resolution, 42 Wm. & Mary L. Rev. 1149 (2001). For documentation of these claims, see Adrian Vermeule, Posner on Security and Liberty: Alliance to End Repression v. City of Chicago, 120 Harv. L. Rev. 1251 (2007). See Eric Posner and Adrian Vermeule, Terror in the Balance: Security, Liberty, and the Courts (2006).

Mechanisms of Democracy

The Limits of Uncertainty

arranges the status quo differently than does, for example, the National Emergencies Act.24 Under the latter, Congress must take affirmative action to override a presidential declaration of emergencies, whereas under Ackerman’s proposal the President’s emergency powers will lapse automatically unless Congress votes to extend the emergency.25 But this is to confuse the legal status quo with the factual status quo; the latter is set by presidential action on the ground, whatever the law may say. The War Powers Resolution, which Ackerman barely mentions, sets the status quo in the same way that his framework statute would, by requiring the President to obtain congressional approval for deployments of U.S. forces after the initial sixty-day period has passed. And the War Powers Resolution is utterly defunct, as shown by the Kosovo episode, a 78-day conflict conducted with only a frivolous claim to congressional authorization.26 The President’s central power is to move first, in the world beyond the statute books, and thus confront Congress with a fait accompli.27 The ill-motivated President that Ackerman pictures will do just that, and the passive Congress that history reveals during times of war and emergency will be both unwilling and largely powerless to do anything about it, however the legal status quo is nominally set. In short, the problem with Ackerman’s proposal is that of the timeinconsistency of emergency policymaking, or the demonstrated inability of Congress to effectively bind future Congresses and future Presidents where emergencies and war are concerned.28 The point here is not that framework statutes enacted at one time never constrain legislators or other actors at a later time; above, I have indicated some mechanisms by which they might do so. The point is that they are least likely to constrain in the settings of war and emergency that Ackerman is discussing, and 24 25 26

27 28

See Ackerman, supra note 15, at 123. See id. at 4. Congress failed to vote formal authorization for the war; the Clinton administration then tried to locate such authorization in appropriations measures, despite the longstanding interpretive principle that appropriations measures are not to be understood as making substantive policy. For an overview, see Richard F. Grimmett, The War Powers Resolution: After Twenty-Eight Years, Congressional Research Service Report for Congress (Nov. 15, 2001), available at http://www.globalsecurity.org/military/library/report/crs/RL31185.pdf. For the Clinton administration’s position, see Office of Legal Counsel Opinion “Authorization for Continuing Hostilities in Kosovo” (December 19, 2000), available at http:www.usdoj.gov/ olc/2000opinions.htm. Terry M. Moe & William G. Howell, The Presidential Power of Unilateral Action, 15 J.L. Econ. & Org. 132 (1999). For skepticism about the constraining force of legal precommitments during subsequent emergencies, see Martha Minow, What Is the Greatest Evil?, 118 Harv. L. Rev. 2134, 2166–67 (2005).

Institutional Design Writ Small

69

Chapter 

given the conditions he diagnoses. Where emergency measures would benefit majorities at the expense of minorities, earlier framework legislation is most likely to be circumvented, violated or repealed outright.

Impartiality, Motivation and Self-Defeating Proposals The most general problem with Ackerman’s proposal is that it is selfdefeating: the diagnosis that Ackerman offers itself rules out the prescription that he suggests. Here we need only focus on the problem of motivational inconsistency, which arises when the institutional actors who have the ability to adopt and enforce the theorist’s prescription have no desire to do so. In the general case, the tradeoffs between impartiality, on the one hand, and motivation and commitment, on the other, mean that self-interested decisionmakers with agenda control will not push impartial projects to the top of the agenda or will not protect them from repeal if they were somehow enacted in the past. In the special case of Ackerman’s framework statute for emergencies, the motivational problem is that those who might adopt the framework will have no incentive to do so, nor any incentive to enforce it once adopted. We might, with some charity, interpret Ackerman as simply offering his views about optimal institutional design, regardless of politics. But if his diagnosis and prescription make different assumptions about the motives, beliefs or opportunities of political actors, then the two parts of his proposals are mutually incompatible and incoherent, whether or not they take account of political constraints. And in any event, charity stretches only so far. Ackerman’s proposals are clearly not intended as ideal or first-best schemes, that might or might not pass through the political filter. They are already nonideal or second-best schemes designed to be enactable; that is the reason Ackerman gives for proposing a framework statute rather than a constitutional amendment.29

The Limits of the Limits of Uncertainty I have emphasized cases—the proposals by Carrington and Cramton, by Ely, and most of all by Ackerman—where reform proposals that attempt 29

70

See Bruce Ackerman, Before the Next Attack: Preserving Civil Liberties in the Age of Terrorism 77 (2006).

Mechanisms of Democracy

The Limits of Uncertainty

to trade on uncertainty fall flat, by ignoring the basic tradeoffs between impartiality and other political goods. But we should not end on too negative or skeptical a note. As Cox’s ingenious proposal demonstrates, welldesigned mechanisms can harness uncertainty to suppress official self-interest, thereby promoting (one component of) impartiality, in turn promoting (one component of) well-functioning democracy. Overall, the limits of uncertainty are real, and rather tight, because of the pervasive tradeoffs between impartiality, on the one hand, and information, motivation, and commitment on the other. Yet they are not so tight as to make uncertainty-based proposals of low value as a class. And, of course, other mechanisms of democracy can also help to promote impartiality by suppressing self-interested representation. In Chapter 6, I suggest that this is a major benefit of transparency.

Institutional Design Writ Small

71

This page intentionally left blank

Part II

Accountability and Voting

This page intentionally left blank

Accountability and Voting

F

or democrats of all stripes, it is not enough that officials be impartial (in either the broader or narrower sense). A benevolent social planner might be utterly impartial, yet the resulting government would not be recognizably democratic. Officials, to govern democratically, must also be accountable in some fashion to the broader citizenry for the policies they choose. Precise specifications of accountability differ, and are one of the main subjects of this chapter. In broad outline, however, there is democratic consensus on the principle itself. In this Part, I suggest mechanisms that generate accountability in democratic institutions, mechanisms that are for the most part agnostic as between and compatible with a broad range of specific and controversial conceptions of accountability. It will be one of my claims that the differences between or among such conceptions do not often make a difference on the ground; and it will be my hope that democrats of many varieties, all committed to accountability, can favor the mechanisms I suggest. In some domains, tradeoffs arise between accountability and other democratic values, or even within accountability itself. In such cases, a further hope is that the mechanisms examined here succeed as democratic optimizers— devices that strike a desirable balance, at the margin, between or among competing democratic goods.

Democracy and Delegation: A Preface Before proceeding, it is necessary to clear some thorny ground. For some democratic theorists, and for many legal theorists, any mention of accountability immediately brings to mind some version of the “nondelegation doctrine,” which places constitutional constraints on legislative delegation to executive officials and agencies.1 So far as relevant here, the main concern behind the doctrine is that delegation is said to undermine democratic accountability, by transferring policymaking to unelected bureaucrats. 1

For an overview of the doctrine, see Stephen G. Breyer et al., Administrative Law and Regulatory Policy 36–40 (6th ed. 2006).

Institutional Design Writ Small

75

Part ii

Although rarely enforced by American courts as a constitutional matter,2 the doctrine has many supporters who hope for a revival, and in Germany and other polities similar doctrines are enforced more vigorously.3 I shall argue that the nondelegation principle has no appeal as a matter of democratic theory; it is irrelevant to accountability, or accountability is irrelevant to it. The executive to whom delegations are made is also democratically accountable; even more important, legislators can themselves be held accountable for the very decision to delegate.

Nondelegation: The Standard View In a number of polities, a central constitutional device said to promote accountability is (some version of) the principle that legislatures may not delegate their power to the executive. In America, the “nondelegation doctrine” holds that Congress must supply an “intelligible principle” to guide the exercise of statutory authority by executive officials and agencies.4 In 1935, the United States Supreme Court, enforcing the doctrine for the first and only time, invalidated two federal statutes on nondelegation grounds.5 The doctrine has gone unenforced ever since, although it persists as a principle of statutory interpretation.6 Many legal scholars—typically historically-minded “originalists” or libertarian opponents of the administrative state, or both—hope to revive it in one form or another. Among their central claims, and the one of principal interest here, is that the nondelegation doctrine promotes democratic accountability.7 The main argument to this effect runs this way. Suppose that Congress delegates certain powers to an agency. Those who want to influence policy will make arguments to the agency, not to Congress. Those who are disappointed by the agency’s decisions will appeal to the agency, not to Congress. Members of Congress are thus not accountable for political outcomes under their responsibility. But in a well-functioning 2 3 4 5

6 7

76

See id. at 38. See David P. Currie, The Constitution of the Federal Republic of Germany 125–134 (1995). J.W. Ham Co. v. United States, 276 U.S. 394, 409 (1928). A.L.A. Schechter Poultry Corp. v. United States, 295 U.S. 495 (1935); Panama Refining Co. v. Ryan, 293 U.S. 398 (1935). The Court invoked the doctrine as an alternative holding in Carter v. Carter Coal Co., 298 U.S. 238 (1936). See Cass. R. Sunstein, Nondelegation Canons, 67 U. Chi. L. Rev. 315 (2000). See David Schoenbrod, Power Without Responsibility (1993).

Mechanisms of Democracy

Accountability and Voting

democracy those who are ultimately responsible for policy should be directly accountable to those who are affected by it.8 However, this core claim of nondelegation proponents—that delegation dilutes accountability, and is therefore undemocratic—fails on two grounds: the executive also has democratic credentials, and legislators can be held fully accountable for the decision to delegate. The first point is that executive delegates are also accountable. In parliamentary systems with an executive who is elected by the elected legislators themselves, the executive is doubly accountable, both to the parliamentary majority or coalition, and to the voters in the executive’s constituency. In systems with an independently elected executive, agencies are accountable to the President, who is more or less accountable to the voters—more anyway than are the courts who are supposed to enforce the nondelegation doctrine. (In the United States and elsewhere, there are also independent agencies, not formally accountable to the President, but critics of delegation do not limit their arguments to those agencies.) When American critics of delegation equate congressional accountability with democratic accountability more generally, they are covertly assuming that Congress somehow has democratic credentials superior to those of the President. The basis for this assumption is obscure. Congress is a summation of local majorities or pluralities; the President is usually elected by a national winner-take-all majority or plurality (putting aside the Electoral College, which does not usually determine outcomes, although it does shape campaign tactics).9 It is not obvious, in the abstract, which form of democratic accountability is superior, and on what grounds one would make such a judgment. The point is not that presidential accountability is superior; it is that the critics of delegation offer no basis for assuming that the congressional form of accountability is preferable. In an illuminating treatment of the issues, Jide Nzelibe suggests, for the American case, that “the president will often have an incentive to cater 8 9

See Theodore J. Lowi, Two Roads to Serfdom: Liberalism, Conservatism, and Administrative Power, 36 Am. U. L. Rev. 295, 296–98 (1987). Many point out that the Electoral College sometimes produces plurality winners. See, e.g., Sanford Levinson, Our Undemocratic Constitution (2006) 81–101. Of course, legislators are sometimes elected by plurality as well. In this setting, the right comparison is not between the outcomes produced by the Electoral College and some majoritarian ideal; it is between the Electoral College and actual outcomes under the first-past-the-post voting rule used in most American legislative elections. The Electoral College very occasionally hands victory to a candidate who was not even the plurality winner, see id., but this has happened only a few times in American history.

Institutional Design Writ Small

77

Part ii

to a narrower geographic and population constituency than the median member of Congress.”10 This is less consequential than it might sound. The median member of Congress is a modeling fiction, because the two Houses of Congress vote separately. In many cases, of course, the median member of either house does not control outcomes in that house because the pivots or veto gates are defined by supermajority rules, filibuster rules, and so forth. The pivotal player in a given house, however, need not correspond to the fictional congressional median either. Nzelibe’s treatment illustrates the complexity of these issues; comparisons of congressional and presidential accountability are sensitive to the precise political model at issue. It thus emphasizes that the typical critiques of delegation are, at a minimum, simplistic. Overall, critics of delegation arbitrarily equate democracy with legislative or congressional forms of accountability, ignoring that the executive is itself accountable in different ways. Perhaps each form of accountability is best for particular purposes. Perhaps legislative institutions are superior deliberators, while presidential institutions are superior at taking action and getting things done. An overall set of democratic institutions needs both the capacity for deliberation and the capacity for action; neither is obviously superior from the standpoint of democratic values. I return to the democratic accountability of the executive in Chapter 5. For now, let us put aside these points; suppose the democratic credentials of the relevant legislature, Congress or Parliament, are in some way superior, perhaps along the lines that Nzelibe suggests. Still, there is a second major problem with the democratic critique of delegation. The problem is that legislators are accountable when they delegate power—they are accountable for their second-order decision to delegate power to the executive or the agencies. I will illustrate by focusing on the American case, although similar arguments would hold, with appropriate modifications, in other polities. If Congress’s particular structure and organization give it a superior form of democratic accountability, or make it a superior democratic deliberator, then the decision to delegate, or not, will be made in an especially accountable and deliberative way, and will have superior democratic 10

78

Jide Nzelibe, The Myth of the Nationalist President and the Parochial Congress, 53 U.C.L.A. L. Rev. 1217 (2006).

Mechanisms of Democracy

Accountability and Voting

credentials in its own right. Conversely, if an agency exercising delegated power performs its function poorly, citizens will hold Congress responsible for the poor design of the agency, or for giving it too much power or not enough, or for giving it too much money or not enough, or for confirming bad appointments, or for creating the agency in the first place. And, as noted above, agencies are accountable in a more direct sense as well. Congress retains the power to interfere with new laws when agencies make bad decisions; indeed, it does so frequently. A variant of the nondelegation argument emphasizes the ability of Congress to claim victory over some difficult problem (for example, pollution) by enacting a vague statute that seems to tackle the problem (the Clean Air Act) but in fact delegates authority to an agency that will have to make the difficult compromises (between clean air and economic growth).11 But Congress can engage in happy talk about nondelegating statutes as well.12 Either citizens and the media believe it or not; there is nothing special about delegation per se, as opposed to direct and very specific legislation. If members of Congress can be taken to task at the next election for saying that welfare reductions will help poor people, then they can be taken to task for saying that the Consumer Product Safety Commission will help small businesses. Critics of delegation believe that this response makes unrealistic assumptions about the psychology and capacities of the electorate. Anecdote and intuition tell us that a member of Congress is more vulnerable to criticism for voting in favor of a law that, say, shuts down a local polluter who employs many constituents, than for voting for a law that authorizes an agency to shut down polluters, or for not voting for a law that stops an agency that already has that authority. This argument overlooks the price that the member of Congress pays for delegating power. The member of Congress receives less criticism because he has less control over the outcome. The legislator shares the criticism with the president who appointed the agency officials, and the criticism is diluted to the extent that the agency otherwise does a good job. Moreover, if citizens have the capacity to sanction politicians who make bad policy in direct, fully specified statutes, they should also have the capacity to sanction politicians who fail to punish agencies that make bad policy, or who 11 12

See Schoenbrod, supra note 7, at 90–93. See Jerry L. Mashaw, Greed, Chaos and Governance at 147 (1997).

Institutional Design Writ Small

79

Part ii

delegate authority to such agencies in the first place.13 Or at least they will develop such a capacity; the techniques of public monitoring of agencies are endogenous, not fixed, and should improve as agencies acquire more power through delegations.14 Accountability is not lost through delegation; it is transformed. Congress is accountable for the performance of agencies generally, and people properly evaluate the agencies’ accomplishments as well as failures when deciding whether to hold legislators responsible for authorizing the agency, or for failing to curtail its power, fix its mistakes, or eliminate it altogether. It is important, in this connection, that the nondelegation doctrine restricts delegation to executive agencies but not to other agencies within the legislative branch. Political scientists David Epstein and Sharyn O’Halloran argue that if Congress could not delegate broad powers to agencies, it would delegate more power to congressional committees and staff members.15 Whether this substitution would occur, and whether it would reduce accountability if it did, are difficult questions, but ones that the critics of delegation have not adequately addressed.

Delegation, Deliberation and Constitutionalism Many views attempt to tie nondelegation to the forms and values of democratic constitutionalism. These views emphasize the deliberative virtues of forcing lawmaking to proceed through the hurdles of the legislative process, including passage by two houses and (in the American case) approval by the President, rather than through the brisker procedures of agency rulemaking. As legal scholar John Manning puts the argument (without necessarily endorsing it): By dividing legislative power among three relatively independent entities, that intricate and cumbersome process serves several crucial constitutional interests: it makes it more difficult for factions (or, as we would put it, “interest groups”) to capture the legislative process for private advantage, it promotes caution and restrains momentary passions, it gives special protection to 13 14 15

80

Mashaw makes essentially this argument. See Mashaw, supra note 12, at 146–47. See Daniel A. Farber & Phillip P. Frickey, Law and Public Choice 81–82 (1991). See David Epstein & Sharyn O’Halloran, Delegating Powers: A Transaction Cost Politics Approach to Policy Making Under Separate Powers (1999).

Mechanisms of Democracy

Accountability and Voting

residents of small states through the states’ equal representation in the Senate, and it generally creates a bias in favor of filtering out bad laws by raising the decision costs of passing any law.16 Again, however, it is a full response to these points that delegatory legislation will itself pass through the full legislative process, and will thus enjoy all the deliberative advantages that first-order legislation would enjoy. If the legislative process reduces the influence of interest groups, enabling the legislature to serve the public interest, then legislators will have no incentive to create agencies that are themselves vulnerable to these same interest groups. If the legislative process prevents legislators from being carried away by momentary passions, then legislators will, quite calmly, burden agencies with procedural requirements that prevent the agencies from being carried away by momentary passions. Residents of small states can be expected to exercise their political power by blocking delegations that will disproportionately harm them or by insisting that the authorizing statutes enlarge the agencies’ ability to do good, and limit their ability to do harm, in small states. Bad delegations will be filtered out because of the cost of enacting the delegating statute. All of the deliberative and constitutional values mentioned by Manning are protected at one remove, because the delegatory statute itself must go through bicameralism and presentment. The scale of the policy choices that Congress must make is itself an appropriate topic for congressional deliberation. If Congress concludes that in light of its limited resources it should deliberate more about the general structure of government, including the delegation of powers, and less about policy minutiae, then constitutional interference with this choice can hardly be thought to promote deliberation.

Delegation, Accountability and Democracy: A Summation However the argument from democracy is understood, it cannot be used to justify the nondelegation doctrine in the absence of any grounds to believe that delegation interferes with accountability, deliberation, and other democratic values. The critics of delegation load the dice against 16

John F. Manning, The Nondelegation Doctrine as a Canon of Avoidance, 2000 Sup. Ct. Rev. 223, 239–40.

Institutional Design Writ Small

81

Part ii

Congress by arguing that the absence of constant supervision of agencies by Congress shows that agency lawmaking is unaccountable, when the absence of constant supervision is the intended consequence of delegation. Constant interference with an agency would be evidence of delegatory failure.17 Besides, Congress does enact laws that reverse the decisions of agencies, weaken their powers, and even abolish whole agencies. And Congress often takes care to draft statutes to confine discretion, and give interest groups the ability to alert Congress by pulling a fire alarm when the agency goes astray.18 One cannot in the abstract say whether the observed level of monitoring and punishment shows that Congress remains accountable, but it suffices to undermine any basis for confidence in the argument of nondelegation proponents. To show that delegation has reduced accountability, one must provide a more particular kind of evidence, namely, that the agency regulates against the interests of Congress and the public or the interest groups that have influence with Congress, and that Congress does not attempt to discipline the agency. And that is only the first step. To show that this lack of accountability is due to the delegation rather than to the nature of Congress itself, one must show that when Congress does not delegate, the rules it creates reflect more closely the desires of the public or of interest groups. What systematic evidence there is tends to suggest the opposite: that the amount of delegation appears to reflect legitimate policymaking concerns, such as the similarity of the preferences of Congress and the agency.19 From the standpoint of democratic values, there is nothing special about delegation. A legislature can use its lawmaking powers for good or evil: to single out one of these powers for restriction one needs a theory that explains why this power is subject to greater abuse than others. Critics of delegation have made their case by pointing to the potential ill effects of delegation, but they have shown only that delegation is a powerful tool, and so its effects—whether good or bad—can be substantial. But the same can be said about any legislature’s power to tax. If one focuses only on the bad uses of the taxation power, taxation would seem an instrument of 17 18

19

82

See Barry R. Weingast & Mark J. Moran, Bureaucratic Discretion or Congressional Control?: Regulatory Policymaking by the Federal Trade Commission, 91 J. Pol. Econ. 765, 768–70 (1983). See D. Roderick Kiewiet & Matthew D. McCubbins, The Logic of Delegation: Congressional Parties and the Appropriations Process 206–10 (1991); see also Matthew D. McCubbins & Talbot Page, A Theory of Congressional Delegation, in Congress: Structure and Policy 402 (Mathew D. McCubbins & Terry Sullivan eds., 1987). See Epstein & O’Halloran, supra note 15; see also Kiewiet & McCubbins, supra note 18, at 235–37.

Mechanisms of Democracy

Accountability and Voting

evil, but no one argues that legislatures should not have the power to tax. Both taxation and delegation can be put to good and bad uses; what makes delegation especially bad? Someone might yet come up with a theory that explains why delegation should be restricted, but it is unlikely that any such theory will be found. A public-choice theory that envisions the political process as interest group warfare will not be able to explain why delegation is worse than any other tool at Congress’s disposal. A more subtle theory that focuses on the differential costs of monitoring political actors will have trouble explaining why those who monitor political actors face greater difficulties evaluating statutes that delegate, and the regulations they authorize, than statutes that do not delegate; why the technology and costs of monitoring are not endogenous to the amount of delegation that occurs, so that the costs will adjust over time; and why the additional costs of monitoring delegations, if they exist, are likely to exceed the gains from institutional division of labor. Those gains are the whole point of delegation, and it is not plausible to say that they are democratically suspect.

Accountability, Delegation, and Lawmaking in General The discussion so far merely aims to clear away some confusions. On the conventional nondelegation doctrine—or rather, the view that would hold were the nondelegation doctrine enforced, which it rarely is as a constitutional matter—accountability is to be enforced by constitutional constraints that require specific policy choices by Congress. On the view offered here, by contrast, accountability is equally present in delegatory legislation and direct legislation, because legislators are accountable for the choice to delegate itself. The consequence is that all legislative lawmaking, direct or indirect, is on the same footing; delegation is not special. Rather than trying to carve out special classes of legislative action for special concern, then, the project should be to adopt mechanisms that enforce accountability across all classes of lawmaking. The next two chapters turn to this task, suggesting legislative voting rules that can improve accountability across the board. Of course, accountability is one democratic value among others; and as discussed in subsequent chapters, there are internal conflicts between different forms of accountability itself. These points mean that accountability faces both external and internal tradeoffs, meaning in turn that accountability is to be optimized, not maximized. Institutional Design Writ Small

83

Part ii

What I have argued, however, is that the distinction between delegatory lawmaking and other forms of lawmaking is not relevant to the project of optimizing accountability through institutional design.

Optimizing Accountability Having cleared away the nondelegation idea, we should move on to the constructive work of improving accountability—regardless of the form of lawmaking, delegatory or direct. Part II attempts to do so. Chapters 3 and 4 each suggest a voting mechanism intended to improve upon existing structures of accountability, although in both cases the mechanism is already used in some legislatures, some of the time, and also in other democratic institutions. Chapter 3 suggests that submajority voting rules can improve accountability—both within and without legislatures—by allowing democratic minorities to force democratic majorities to squarely confront important public issues. Chapter 4 suggests that absolute majority voting rules, which require a majority of all eligible members or voters to take institutional action, can promote accountability by insuring legislative and democratic majorities against strategic behavior by minorities. Moreover, such rules simultaneously promote the accountability of legislators and other officials to broad electorates and minimize the accountability of officials to narrow interest groups and other democratically suspect third parties. These two effects are complementary—the first promotes accountability where it is desirable, the second dilutes it where it is undesirable. Absolute majority rules, which can be used in conjunction with submajority rules, thus provide a tool for optimizing accountability. Chapter 5 turns to judicial voting and legal doctrine. I suggest a supermajority voting rule that weights law in favor of the exercise of delegated lawmaking authority by government officials. Legal doctrine already requires this, but does not enforce the requirement with a voting rule. The suggestion here is that a small-scale institutional mechanism can better enforce the current law, promoting the democratic accountability of administrative law and policy at lower cost than the current regime. Chapter 5 thus connects delegation with voting rules geared to promoting accountability, the theme of Chapters 3 and 4. Throughout the chapters in this part, I begin to elicit the connections between accountability, transparency, and deliberation—the central topics of Part III.

84

Mechanisms of Democracy

Chapter 3

Submajority Voting Rules: Forcing Accountability

L

egal and political theory have paid a great deal of attention to supermajority rules,1 which require a fraction of votes greater than 1/2+1 to change the status quo, and thus empower a minority to block change. In this chapter I consider the opposite deviation from simple majority voting: submajority rules, under which a voting minority is granted the affirmative power to change the status quo. My central analytic claim is that institutions use submajority rules only for preliminary and procedural questions, and to set agendas—for voting on the question of which issues are to come up for a substantive vote by simple or special majorities. My central normative claim is that this empirical pattern can be justified by good reasons (whether or not it is caused by those reasons, an issue of functional explanation that I do not discuss). Submajority rules enable minorities to force public accountability and a kind of transparency upon majorities.2

1 2

See, e.g., James M. Buchanan & Gordon Tullock, The Calculus of Consent, Logical Foundations of a Constitutional Democracy (1962). Compare Michael Suk-Young Chwe, Minority Voting Rights Can Maximize Majority Welfare, 93 Am. Pol. Sci. Rev. 85 (1999). Chwe’s argument is not based on accountability-forcing, but

Institutional Design Writ Small

85

Chapter 

Submajority rules are rarely discussed,3 either because they are assumed not to exist,4 or because they are assumed to lack any institutional virtues, or because submajoritarian decisions are assumed to be chronically unstable in light of the risk that subsequent majorities will reverse or undo the submajority’s decision. I will dispute all three assumptions. As to the first, submajority rules are to be found in a range of legislatures, courts, international bodies, and other democratic institutions. Among the examples I will consider are:

• •

• • • • •

3

4

5

86

The Journal Clause, which allows one-fifth of the legislators present in either House of Congress to force a roll-call vote; The “discharge rule” in the House of Representatives, which (at various points, although not today) has permitted a specified minority of legislators to force bills out of committee for consideration on the floor; Senate Rule XXII, under which a cloture petition is valid when signed by sixteen Senators;5 The “Seven Member Rule,” under which a minority of designated committees in the House and Senate can require the executive branch to divulge information; House Rule XI, which entitles committee minorities to call witnesses at hearings; The famous “Rule of Four” that allows four Justices to grant a writ of certiorari and thereby put a case on the Supreme Court’s agenda. Rules governing direct democracy that permit a defined minority of a state’s electorate to place a question on the ballot, or to force a recall election.

on the idea that where minorities have special information, self-interested majorities might grant them special voting power in order to elicit that information. A useful exception is Richard L. Revesz & Pamela S. Karlan, Nonmajority Rules and the Supreme Court, 136 U. Pa. L. Rev. 1067, 1067–1133 (1988). But Revesz and Karlan confine themselves to courts (in fact the United States Supreme Court), whereas I shall incorporate submajority rules that arise in legislatures and other nonjudicial institutions. See Dennis C. Mueller, Public Choice III 77 n.7 (2003) (denying that either House of Congress uses a less than 50 percent majority rule). It is unclear whether Mueller means to confine his remark to final passage of legislation; if so, but only if so, the remark is correct. A motion for cloture must be carried by a Senate supermajority. In the other examples I give, the ultimate disposition, such as the final passage of a bill discharged from a House committee or the ultimate judgment in a judicial case, is by simple majority only.

Mechanisms of Democracy

Submajority Voting Rules: Forcing Accountability

•

Rules governing international organizations, which frequently allow a defined minority to call an emergency session or to force a roll-call vote.

As to the second assumption, submajority rules have important procedural and epistemic virtues, I will argue. Their principal benefit is to enable a minority to force a kind of public accountability upon the majorities who would otherwise prefer to sweep minority views and desires under the rug. Democrats of many stripes—pluralist democrats who favor the aggregation of preferences through bargaining as well as deliberative democrats committed to an ideal of public reason—can converge to approval of submajority rules. Pluralist democrats can approve of them as institutional mechanisms by which groups excluded from the majority coalition can force their way to the bargaining table; deliberative democrats can approve of them as mechanisms that force majorities to take account of the views of all relevant parties and that encourage majorities to engage in reasoned deliberation, rather than governing through power. As to the third assumption, the reversibility problem can be and is dampened either through collateral institutional rules and norms that protect submajoritarian decisions, or by the simpler expedient of adopting submajority rules only for decisions that are inherently irreversible or costly to reverse, such as decisions that release information into the public domain. The irreversibility of public disclosure connects the positive and normative features of submajority rules: such rules are best used to set procedures and agendas for public deliberation and voting precisely because the costs to majorities of reversing submajoritarian decisions are highest in that domain. Conversely, restricting submajority rules to preliminary and procedural matters best enables institutions to cope with reversal and other problems.

Submajority Rules and Near Relations A submajority rule is a voting rule that authorizes (i) a predefined numerical minority within a designated voting group (ii) to change the status quo (not merely to prevent change) (iii) regardless of the distribution of other votes. The “status quo” to which I refer is always defined by the rules of the relevant institution. There is no claim here that the status quo would be well defined in some apolitical sense, apart from the institution’s rules, Institutional Design Writ Small

87

Chapter 

but on the other hand there is no conceptual problem with identifying the status quo when it is so specified. This definition excludes majority rule, which trivially fails condition (i), and which also fails condition (iii) because any given individual (and thus any minority) can be decisive only given some particular distribution of other votes—in the individual case, where the other voters are equally divided. Condition (ii) excludes supermajority rules; in their usual asymmetrical form,6 supermajority rules afford decisive veto power to submajorities—typically 2/5 + 1, 1/3 + 1, or even 1/5 + 1—who are empowered to block alteration of the status quo, but not to alter it themselves. In any event supermajority rules often apply to substantive matters, whereas the “status quo” referenced in condition (ii) is typically procedural rather than substantive (as shown below). House Rule XXVII, which requires a two-thirds supermajority to agree to a suspension of the House Rules themselves, allows a minority to block change in the procedural status quo, whereas the Supreme Court’s Rule of Four authorizes a submajority of four Justices affirmatively to change the procedural status quo. The only voting procedure that directly determines substantive outcomes by submajoritarian standards is substantive lottery voting, under which the votes for and against a proposal become tickets in a random draw.7 In the most radical versions, a bill favored by only 10% of the legislature nonetheless has a 10% chance of being enacted. But lottery voting does not empower a predefined minority, so it fails condition (i). More substantially, I will suggest below that reversal problems are far more severe for submajoritarian decisions on substantive questions than on preliminary questions of process and agenda-setting—which may help to explain why substantive lottery voting, although occasionally defended in theory, appears so rarely in practice. Condition (i) excludes institutions, like states or provinces, legislative committees, or juries, that provide a submajority of some larger body 6

7

88

Throughout, I bracket the complications that arise from symmetrical supermajority rules, which can be used where the status quo is not the default choice; then no decision is made unless it gains the requisite supermajority. See Robert Goodin & Christian List, Special Majorities Rationalized, 36 Brit. J. Pol. Sci. 213 (2006). Because there is no status quo, symmetrical supermajority rules are thus excluded by condition (ii) in a literal, albeit vacuous, sense. These complications are tangential to my line of argument, so I shall use “supermajority rules” strictly to refer to asymmetrical supermajority rules. See Richard Zeckhauser, Majority Rule with Lotteries on Alternatives, 83 Q. J. Econ. 696 (1969).

Mechanisms of Democracy

Submajority Voting Rules: Forcing Accountability

(such as the whole nation or the whole legislature or the jury-eligible population) with decisionmaking authority or agenda control over designated issues. Such bodies can in effect grant minorities the de facto power to set the agenda or otherwise change the status quo within a certain domain,8 and are thus relatives of the institutions I will discuss. Nonetheless state legislatures, committees in the national legislature, juries and the like typically vote according to an internal majority rule or supermajority rule, not a submajority rule. My focus here is on formal or de jure submajority rules, which have the strikingly countermajoritarian property that, even within the designated voting group, the few may formally change the (procedural) status quo over the objection of the many. Condition (ii) is the most critical, for the following reason. Putting aside the location of the status quo, any rule that licenses a submajority of n/x to reach a decision D might be redescribed as a converse rule licensing a supermajority of (x-n+1)/x to reach a decision ~D.9 “[T]he rules ‘Four votes are required to grant certiorari’ and ‘Six votes are required to defeat certiorari’ on a nine-judge court are distinguished only by the impact of abstentions.”10 The difference is small in the judicial setting, but in institutions with many members, such as legislatures, the location of the status quo will often matter precisely because abstentions are predictable and important. As we will see, a prime reason for choosing a submajority rule over the converse supermajority rule is that institutional designers might justifiably prefer one status quo position over another, either because the supermajority reciprocal is conceptually ill-defined, or because the larger transaction costs of assembling a supermajority coalition in the face of abstentions make the default rule sticky.

Submajority Rules and Accountability A successful submajority vote emphasizes the countermajoritarian features of the decision, especially but not only in binary decisions. If one hundred Senators are present and twenty desire to record the Yeas and Nays then, by force of the Journal Clause, they are recorded; it is irrelevant 8

9 10

On the affirmative agenda-setting power of legislative committees (as opposed to their negative gatekeeping power), see David P. Baron & John Ferejohn, The Power to Propose, in Models of Strategic Choice in Politics 343 (Peter Ordeshook ed., 1989). Again, I bracket here the possibility of asymmetrical supermajority rules. Lewis Kornhauser & Lawrence Sager, Unpacking the Court, 96 Yale L.J. 82, 99 (1986).

Institutional Design Writ Small

89

Chapter 

whether the various alternative methods of vote-counting are all on the table and the remaining Senators split their votes among them, or instead vote en masse, eighty-twenty, for some one other method and against the submajority. Submajority decisions thus contradict, flagrantly, the usual idea that “the minority gives way not because it is convinced that it is wrong, but because it is convinced that it is a minority.”11 In more formal terms, submajority rules violate the neutrality feature of majority rule, which requires that no choice be preferred over another by the voting rule itself.12 “Where submajority rules are in play typically one outcome is favored, in the sense that it will be adopted if it receives k votes, whether or not some other outcome also receives k.”13 The Rule of Four, for example, violates neutrality because four votes for certiorari prevail, whereas four votes against certiorari lose. Supermajority rules (of the asymmetric sort, the only sort I discuss here) also violate neutrality. Thus, in the Senate, forty votes to continue debate succeed in defeating a cloture motion, whereas forty votes to cut off debate fail. Yet supermajority rules allow a minority only to block changes in the status quo, not to change the status quo in the face of a contrary majority view. The most striking feature of submajority rules is this flaunting of countermajoritarianism, allowing a minority to effect an affirmative change to the status quo in the teeth of majority opposition. Why, and under what circumstances, would a well-motivated institutional designer wish to adopt a rule of this sort? The key point, or so I argue, is that submajority rules are rarely or never used directly for final substantive decisions, such as the passage or defeat of legislation. Instead they are used for procedural and collateral matters: to set institutional agendas, to decide what information the institution will gather before reaching a decision, and to make transparent the decisions that are reached by majority or supermajority vote. Decisions on these matters will, of course, have indirect effects on outcomes. But submajority rules are best understood in procedural terms, as devices that empower minorities to force public accountability and transparency on the majority. An institution that is committed to making final substantive 11 12 13

90

James Fitzjames Stephen, Liberty, Equality, Fraternity 70 (Stuard D. Warner, Liberty Fund 1993) (1873). See Kenneth O. May, A Set of Independent, Necessary and Sufficient Conditions for Simple Majority Decision, 20 Econometrica 680 (1952). Kornhauser & Sager, supra note 10, at 99.

Mechanisms of Democracy

Submajority Voting Rules: Forcing Accountability

decisions by majority or supermajority vote, for the standard reasons, might work better if minorities have the power to force accountability upon the majority; and submajority rules are a useful way to confer that power. Accountability-forcing is accomplished by empowering minorities, through submajority rules, to force the majority to make a highly visible, ultimate substantive decision on a given question, rather than disposing of the issue in some less prominent fashion, including simple inaction.

Distributing Agenda-Setting Power Submajority rules often have the effect of distributing agenda-setting power away from majorities to minorities. The leading examples in American law are the Rule of Four,14 which authorizes any four Justices to put a case on the Supreme Court’s calendar for plenary hearing and disposition, and the rule governing discharge petitions in the House of Representatives, which, at various points between 1910 and 1935, authorized 145 or 150 legislators to force a bill out of committee.15 There are similar rules governing international organizations as well.16 Why, if at all, should agenda-setting power be distributed in this way? In the judicial case, the Justices first articulated the Rule of Four to assuage two related congressional fears: first, that granting the Court discretionary jurisdiction over (most of) its docket would result in arbitrary selection; second, that “too few” cases would be granted for plenary hearing, while most cases would be disposed of through summary action or simple denial.17 Yet these concerns are underdeveloped. As to the first, the 14 15

16

17

See Robert L. Stern et al., Supreme Court Practice, For Practice in the Supreme Court of the United States (8th ed. 2002). See Sarah A. Binder, Minority Rights, Majority Rule: Partisanship and the Development of Congress 136–53 (1997). If the requisite number of signatures were present, the resulting floor vote would not be on final passage but on the motion to discharge the committee. Both the discharge motion and final passage would, of course, require a majority of a quorum. Such rules typically empower submajorities to convene special sessions of international bodies. See, e.g., UNESCO, Manual of the General Conference and Rules of Procedure of the Executive Board, Section D, Art. 9, http://www.unesco.org/confgen/en/articles/article4.htm (last visited Nov. 10, 2006) (authorizing 1/3 of members to call a special session); WIPO, Convention Establishing the World Intellectual Property Organization, July 14, 1967, Stockholm, Art. 6(4)(b) http://www.certh.gr/cordis/t_en/i/i_410_en.asp-adt_id=6&ads=0. htm (visited Nov. 10, 2006) (authorizing 1/4 of members to call an emergency session). See John Paul Stevens, The Life Span of a Judge-Made Rule, 58 N.Y.U. L. Rev. 1, 14 (1983).

Institutional Design Writ Small

91

Chapter 

Justices’ criteria of fair selection are vague,18 and in any event it is unclear why a submajority voting rule would implement those criteria more accurately than an ordinary majority rule. The second concern articulates no theory of optimal docket size that would tell us how many cases are “too few,” and it seems to assume, oddly, that the Justices are docket-minimizers, perhaps from a desire to maximize leisure. The subsequent history of the Court’s discretionary docket, however, suggests if anything that the Justices will often push the Court’s agenda capacity to the limit, so no simple docket-minimizing picture is persuasive. A better justification is that a submajoritarian agenda rule can force majoritarian accountability. In these settings, we can interpret “accountability” in no fewer than three ways: (1) as accountability to the intrinsic claims of reason; (2) as accountability motivated by the desire to appear reasonable; or (3) as accountability in the sense that all affected groups are given a share of agenda control, and thus empowered to bring their preferred issues to the table for pluralist bargaining. Either of the first two interpretations are congenial to deliberative democrats, who see arguing on the basis of discursively justifiable reasons as the democratic ideal. The third is congenial to pluralist democrats, who see the aggregation of preexisting preferences as the central function of democratic institutions. We need not choose among these to approve of submajority rules. Within the bounds set by the collateral institutional costs of such rules, a subject I take up later, they exemplify the possibility of overlapping consensus, by democrats of differing high-level commitments, on small-scale mechanisms of institutional design. On the first interpretation, we assume that members of the majority will deliberate sincerely, on the basis of publicly justifiable reasons, if only the relevant questions can be forced upon their attention. The majority’s preferences, or its views, are endogenous to the quantity and quality of deliberation that the institution gives to a particular issue. The submajoritarian agenda rule then allows the minority to focus the majority’s attention by putting the decision under the most intensive form of scrutiny the institution affords—floor debate (in legislatures), full public argument on the merits (in courts), or some other equivalent. Applying this picture to the judicial setting, the five-member majority is especially accountable 18

92

See Supreme Court Rule X (certiorari is discretionary; relevant factors are conflict between lower federal courts or state courts and the importance of the federal issue presented).

Mechanisms of Democracy

Submajority Voting Rules: Forcing Accountability

to the claims of legal reason advanced by the minority during full-dress argument, and may change its views accordingly. On the second interpretation of accountability-forcing, we assume that members of the majority act strategically to maximize the satisfaction of fixed preferences. The underlying preferences themselves will not be changed by group deliberation, although deliberation might provide useful information. Nonetheless, a submajority agenda rule can still promote majoritarian accountability by raising the political visibility of the relevant decisions. If members of the majority are concerned for their public reputations, visibility can force majority-bloc members to behave as though their actions are motivated by publicly justifiable reasons. I will illustrate the second interpretation in both the certiorari setting and the legislative setting. In the former setting, on this account, the Rule of Four serves to prevent an entrenched five-Justice majority from simply disposing of disfavored claims through low-visibility procedures.19 At time one, let us suppose, the majority upheld Claim C by articulating Principle P. Suppose also that in a different case, to be decided at time two, Principle P underwrites not only Claim C but also Claim D—a claim that the majority dislikes and wishes to deny. A low-visibility disposition of the second case, as by a denial of certiorari or a summary disposition, would enable the majority to sweep D under the rug at time two. A highly visible decision on the merits, by contrast, would force the majority either to uphold D on the basis of the previously articulated Principle P, or else be exposed as opportunists; this is “the civilizing force of hypocrisy.”20 On this view, the submajoritarian agenda rule works to counteract the “subtle vices of the passive virtues”21—the ability of majorities to exploit various low-visibility techniques for disposing of cases in unprincipled ways. This account assumes that certiorari denials or summary dispositions are less visible than merits decisions. The assumption may be objectionable on methodological grounds. It casually posits that the audience for judicial opinions is differentially ignorant, either because it is costlier to monitor summary dispositions or certiorari denials than to monitor merits decisions, or because the audience is subject to some form of flawed 19 20 21

See Revesz & Karlan, supra note 3, at 1108. Jon Elster, Arguing and Bargaining in Two Constituent Assemblies, 2 U. Pa. J. Const. L. 345, 349 (2000). Gerald Gunther, The Subtle Vices of the “Passive Virtues”: A Comment on Principle and Expediency in Judicial Review, 64 Colum. L. Rev. 1 (1964).

Institutional Design Writ Small

93

Chapter 

cognition that sophisticated interest groups cannot wholly dispel. But perhaps we ought not be too impressed by the methodological infirmity of the assumption, if it has the ring of truth about it. The analysis is similar in the legislative setting. In the House of Representatives, proponents of the submajoritarian discharge rule justified it by arguing that the rule empowered minorities to force majorities to “put up or shut up” on the floor.22 When a submajority requirement of 150 was adopted in 1924, Mr. Crisp of Georgia . . . stated that he was a believer in control by the majority party, but continued: “I believe the minority party has the right to smoke out the majority and make them face issues, make them vote on great public questions.”23 This argument assumes that floor disposition is more visible to interested publics than killing a bill in committee; if so, then the discharge rule forces public justification of the majority’s preferred disposition. And if the reasons that enacting majorities or their agents gave, on previous occasions, impose consistency constraints even on rationally self-interested legislators—as the civilizing force of hypocrisy supposes—then the requirement of public justification may indirectly constrain or alter legislative outcomes. So far we have spoken of public reason-giving as the democratic ideal—although the first and second interpretations of the accountabilityforcing effects of submajority rules differ about whether reason-giving is typically a sincere exercise or a product of reputational pressure and the civilizing force of hypocrisy. The third interpretation, however, abandons the deliberative ideal altogether. On this view, submajority rules promote accountability by empowering democratic minorities to force their way to the bargaining table. Pluralist democrats typically hope that competition among organized groups, aggregating their preferences through bargaining within 22

23

94

Discharge is an issue only in the House, because only the House rules bar nongermane amendments on the floor. Senate rules allow nongermane amendments, so that any bill squelched by a committee can be proposed, by any Senator, as an amendment to unrelated legislation under debate. Paul DeWitt Hasbrouck, Party Government in the House of Representatives 153 (1927).

Mechanisms of Democracy

Submajority Voting Rules: Forcing Accountability

democratic institutions, will approximate some overall public good. This picture is exposed to two lines of criticism relevant here. First, deliberative democrats emphasize that preferences are themselves, at least in part, endogenous products of the processes that democratic institutions use to aggregate differing views; deliberation and reason-giving can alter preferences and perhaps even channel preferences in public-spirited directions. On this view, the first and second interpretations of accountability are preferable, but we have already seen that they support the submajoritarian mechanism. Bracketing the deliberative critique of pluralism, then, the second major line of critique is that pluralism fails because many groups are left out of democratic bargaining. This is the public choice concern: differential costs of organization across interests—broad or diffuse interests find it harder to organize than concentrated interests with high per capita stakes—mean that the bargaining table only seats a few. In legislatures, a “majority” coalition can be composed of a group of representatives beholden to a narrow range of narrow interest groups. Let us also bracket the broad issue of whether this public-choice criticism of pluralism is correct. Suppose that it is. From a pluralist perspective, submajority rules can be viewed as an institutional fix for the public-choice concern. Such rules improve the workings of pluralism by allowing underrepresented interests to force their issues onto the institutional agenda and thus force their way to the bargaining table. One possibility is that, once the minority’s issue is put before the whole body, majority leaders will find that including the minority at the bargaining table is the path of least resistance. Furthermore, because agenda space— time—is itself a highly valuable resource in legislatures and other aggregating institutions, the mere fact of being awarded a measure of agenda control by the submajority voting rule gives underrepresented minorities a valuable resource—one that they may trade away, if they like, for a measure of satisfaction on other issues or policy dimensions. The institutional fix is, of course, not perfect, because of the sort of tradeoffs and optimization problems that are ubiquitous in institutional design. As we will see, submajority rules must be calibrated to ensure that very small minorities are not empowered to hold up institutional action. If the fix is not perfect, however, it is still, from a pluralist perspective, an improvement on the baseline of ordinary voting rules. The upshot is that we need not, in this domain anyway, settle all high-level theoretical differences about the best specification of common Institutional Design Writ Small

95

Chapter 

democratic commitments. Democrats of many stripes can converge, from their different premises, to approve of an institutional design mechanism that promotes accountability of several different types and along several different margins.

Information, Agendas and Publicity A closely related function of submajority rules is to allow minorities to collect and publicize information that the majority would prefer not to admit into the public record. Examples are the “Seven Member Rule,”24 allowing a designated minority of designated committees to force formal disclosure of executive-branch documents, and House Rule XI, which entitles committee minorities to call witnesses at investigatory and oversight hearings.25 In some cases the power to generate information of record is just agenda-setting power in a de facto sense. Legislative majorities set their agendas in light of the information known to them and the information known to relevant publics; by changing the latter, submajorities may force a new agenda item upon the majority. Here too we may interpret the resulting accountability in three ways. On the sincerely deliberative interpretation, the power to put information or arguments on the public stage may force the majority to respond in kind, by generating better data in support of a proposed action, or better arguments. On the strategically deliberative interpretation, the pressure to respond to information or arguments generated by submajorities may force majorities who are subject to the civilizing force of hypocrisy not only to talk the language of public reason, but actually to act as if motivated by public principle. On the pluralist interpretation, the minority’s ability to threaten public embarrassment for the dominant majoritycoalition, itself a pluralist assemblage, can force the dominant coalition to 24

25

96

5 U.S.C. §2954 provides: An Executive agency, on request of the Committee on Government Operations [now the Committee on Government Reform] of the House of Representatives, or of any seven members thereof, or on request of the Committee on Governmental Affairs of the Senate, or any five members thereof, shall submit any information requested of it relating to any matter within the jurisdiction of the committee. See House Rule XI (j)(1) (minority members shall be entitled to call witnesses upon request by a majority of the minority). House Rule XI (2)(c)(2) allows three members of a standing committee to file a written request that the chair call a special meeting of the committee, but if the chair takes no action a majority is ultimately required to force a meeting, so this is not strictly a submajority rule.

Mechanisms of Democracy

Submajority Voting Rules: Forcing Accountability

bring outsider minorities inside the tent, for fear of what they may do if left outside. High-level differences about democratic theory are not implicated, and we could imagine submajority rules winning a consensus or at least a supermajority in an assemblage of democratic theorists of various sorts.

Voting, Agency and Transparency The common theme of the foregoing is that submajority rules subject the majority to public accountability that improves majoritarian decisionmaking. Although the immediate valence of the rules is dramatically countermajoritarian, their downstream effects may be justified in majoritarian terms. On this account, the paradigmatic submajority rule in the U.S. Constitution is the Journal Clause, which provides: Each House shall keep a Journal of its Proceedings, and from time to time publish the same, excepting such Parts as may in their Judgment require Secrecy; and the Yeas and Nays of the Members of either House on any question shall, at the Desire of one fifth of those Present, be entered on the Journal.26 This provision makes a number of fundamental design choices; most critically, for open voting rather than the secret ballot in Congress, at least as to some matters and on the request of a minority of legislators. Also important are the design possibilities the clause rejects, and that are present in constitutions of other jurisdictions, such as constitutionallymandated roll-call voting in legislative committees and a public right of physical access to legislative proceedings. The stakes are high in these matters: throughout most of its history the English Parliament operated in secrecy and indeed punished attempts to publish records of its proceedings; the Continental Congress initially closed its proceedings to outsiders and the Constitutional Convention did so throughout; and even today most legislatures use secret ballots to select their officers while some, like the Italian Parliament, have until quite recently used them for final voting on legislation. The transparency of legislative deliberation and voting is in broad historical compass a recent and problematic design innovation. 26

U.S. Const. art. I, § 5, cl. 3.

Institutional Design Writ Small

97

Chapter 

Transparency is a solution to an agency problem. Voters are the principals, legislators are the agents, and constitutional provisions that force agents to publicize their actions lower the monitoring costs that principals must incur, thereby making principals and trustworthy agents better off. Importantly, this picture is agnostic across theories of representation and across the divide between pluralist and deliberative accounts of democracy. Whether representatives are trustees representing the common good, or delegates from defined (perhaps geographic) constituencies of voters, we can still identify a common set of cases in which they act as unfaithful agents—say, by taking bribes from narrow subsets of the constituency in return for acts that impose net harms both on the broader constituency and on the polity as a whole. Likewise, voter-principals might want their representatives to bargain, or to deliberate. In either case, voters might rationally believe that increased transparency might help enforce their preference by allowing them to observe what their representatives do—although one must hastily add the important caveat, which I take up in Chapter 6, that transparency might itself dampen bargaining and have effects on how much deliberation, and of what kind, representatives conduct. At the Constitutional Convention, Framers spoke in general terms about the agency problem for which transparency is a cure. James Wilson argued that “The people have a right to know what their Agents are doing or have done, and it should not be in the option of the Legislature to conceal their proceedings.”27 We may thus understand the one-fifth trigger for roll-call voting as a means by which constitutional designers join forces with future legislative minorities to control future legislative majorities. We may surmise that the Framers anticipated that competition between legislative factions would routinely produce public voting, as indeed it has done. Legislative majorities (although not voters) might be better off if legislators could agree to enforce strict secrecy provisions, but legislative minorities armed with the submajoritarian roll-call power produce democratically beneficial transparency. By enlisting the interests of future legislative minorities, constitutional framers force accountability upon future legislative majorities, in the higher interests of the electoral or popular majorities whose agents the legislators are.

27

98

1 The Records of the Federal Convention of 1787, 260 (Max Farrand ed., 1966).

Mechanisms of Democracy

Submajority Voting Rules: Forcing Accountability

Problems with Submajority Rules I have argued so far that submajority rules allow minorities to force public accountability upon unwilling majorities; and that the pressure of public accountability pushes decisionmaking, by legislatures, courts and other institutions, in directions that both deliberative and pluralist democrats might favor, from their differing perspectives. Let us turn to the downside, to some problems with or costs of submajority rules. Of these, some are pseudo-problems, while some represent real costs of using submajority rules in institutional design. The most important example in the latter category is the reversibility problem, which can, however, be dampened by collateral institutional rules or avoided altogether by using submajority rules only for decisions that are costly to reverse or intrinsically irreversible.

Minoritarian Holdup We may illustrate the problem of minoritarian holdup by returning to the submajoritarian trigger for roll-call voting in legislatures. Constitutional designers in the American states28 and in other polities, such as Canada and Japan,29 have tied submajority rules to transparency, especially through rollcall voting, in similar ways. But no polity (of which I am aware) allows a single legislator to force a roll-call vote. In some international organizations a single member can force a roll-call,30 but these organizations typically have far fewer members than a domestic legislature. In a decisionmaking body with a small number of members, the optimum might also be equivalent to the minimum—one member.

28

29

30

See, e.g., Cal. Const. art. IV, § 7B (stating “Each house shall keep and publish a journal of its proceedings. The rollcall vote of the members on a question shall be taken and entered in the journal at the request of 3 members present”); see also Haw. Const. art. III, § 12 (stating “The ayes and noes of the members on any question shall, at the desire of one-fifth of the members present, be entered upon the journal.”). International Centre for Parliamentary Documentation of the Inter-parliamentary Union, Parliaments of the World: A Comparative Reference Compendium Vol. I 480 (1986) (in Canada, all votes are voice votes unless five MP’s request a roll call); Japan Constitution Chapter III Article 57 Section 3 (“Upon demand of one-fifth or more of the members present, votes of the members on any matter shall be recorded in the minutes.”). See, e.g., Rules of Procedure of the Human Rights Committee, UN Doc CCPR/C/3/Rev.6 (2001) (Rule 52), http://www.unhchr.ch/tbs/doc.nsf/ (last visited Jan. 20, 2004); see also Rules of Procedure of the Committee on the Rights of the Child, UN Doc CRC/C/4 (1991) (Rule 54), http://www.unhchr.ch/tbs/doc.nsf/ (last visited Jan. 20, 2004).

Institutional Design Writ Small

99

Chapter 

The general pattern is that the submajority voting rule requires some threshold, and nontrivial, number of members to achieve a submajoritarian outcome. This reflects a systemic tradeoff. On the one hand, submajority rules allow minorities to force accountability on majorities. On the other hand, roll-call voting creates transaction costs and opportunity costs: [In the Journal Clause] [t]he restriction of call of the yeas and nays to one fifth is founded upon the necessity of preventing too frequent a recurrence to this mode of ascertaining the votes, at the mere caprice of an individual. A call consumes a great deal of time, and often embarrasses the just progress of beneficial measures. It is said to have been often used to excess in the congress under the confederation; and even under the present constitution it is notoriously used, as an occasional annoyance, by a dissatisfied minority, to retard the passage of measures, which are sanctioned by the approbation of a strong majority.31 From the standpoint of institutional design this tradeoff creates an optimization problem, one that was debated quite explicitly at the Constitutional Convention. At one extreme, Gouvernor Morris “urged that if the yeas and nays were proper at all any individual ought to be authorized to call for them”; he feared that “the small States may otherwise be under a disadvantage, and find it difficult to get a concurrence of one-fifth.”32 At the other extreme, several members complained that roll-calls had been abused, in the states, by “stuffing the journals with them on frivolous occasions.”33 George Mason spoke for the silent majority on this issue: he praised the one-fifth rule as “a middle way between two extremes,”34 presumably the point at which the net benefits of the rule reach an internal maximum. Another submajority rule also illustrates this optimization problem. Under Senate Rule XXII (2), a cloture petition (a petition to vote to 31 32 33 34

Joseph Story, Commentaries on the Constitution of the United States 842 (Carolina Acad. Press 1987) (1833). Jonathan Elliot, Debates on the Adoption of the Federal Constitution 407 (1881). Id. at p. 407 (quoting Mr. Gorham). Id. at p. 407.

100

Mechanisms of Democracy

Submajority Voting Rules: Forcing Accountability

cut off debate) can be lodged with the signatures of sixteen Senators, and is given priority in the order of business. A much lower numerical threshold would permit harassing petitions by outlying Senators on either extreme, or by senators representing a small number of states. A much higher one would collapse the decision to file a cloture petition with the vote on the merits of the petition. There is nothing general to say about the solution of optimization problems such as this; the threshold of Senate Rule XXII(2) plausibly exemplifies Mason’s middle way.

Reversibility The most conspicuous problem with submajority rules is that submajoritarian decisions are exposed to reversal by subsequent majorities, and might thus be chronically unstable. If one hundred-odd members of the House may discharge a bill from committee, why may not two hundred and eighteen just send it back again? If four Justices may grant certiorari, what happens if five dismiss certiorari as improvidently granted? Majority rule is the lowest-decision-cost voting rule that guarantees stability.35 A great deal of work, however, slides towards the casual assumption that submajority rules are therefore infeasible or even nonexistent. But instability is just another institutional problem to be managed; it is a cost to be weighed against the benefits of submajority rules, rather than a hard constraint on the very possibility of such rules. The tempting mistake here is to assume that an institutional problem with a given voting rule can be solved only by adjusting the institution along the very same margin, by using a different voting rule. In fact there are at least two other margins on which institutions may cope with instability arising from the threat or fact of reversal. Most directly, institutions may employ formal collateral rules, or may develop informal collateral norms, that bar defeated majorities from reversing or undermining a submajoritarian decision—at least not right away, or not without intervening changes. Less directly, institutions may use submajority rules only for decisions that are intrinsically costly or impossible to reverse. I will take up these possibilities in turn. Institutions that use submajority rules may develop written rules or unwritten norms that preclude or limit reversals by subsequent majorities.

35

See Mueller, supra note 4, at 76–77.

Institutional Design Writ Small

101

Chapter 

A judicial example involves the conventions surrounding the Rule of Four: The Court has construed and applied the Rule of Four so as to require all Justices to consider the case on the merits, following the grant of certiorari on four votes. Most members of the Court have felt that the other five Justices who did not vote to grant are thereafter precluded from voting to dismiss the petition as improvidently granted in the absence of additional intervening factors which were not known or fully appreciated at the time certiorari was granted.36 The requirement of “intervening factors” is vague, and has occasionally given rise to intracourt controversies where five Justices dismissed a case over the dissents of four Justices, with the latter group complaining that no intervening factors warranted the action. At various points, therefore, some Justices have claimed that the Court does or should adhere to a stricter practice in which no petition can be dismissed as improvidently granted, even given changed circumstances, unless one of the four Justices in the granting bloc switches her vote.37 There is inconsistency and variation in these practices over time, and all of the relevant practices are politically fraught. But the general picture is that strong normative rules dampen the potential instability of the submajoritarian agenda rule; marginal fluctuations in the rules or norms should not impress us too much. It is quite clear that the ability of subsequent majorities to reverse a grant of certiorari is sharply constrained, relative to a hypothetical baseline in which no such rules or norms exist. Legislatures may also develop rules or norms that dampen or control reversals. Consider the problems arising when a legislative submajority’s discharge petition is brought to the floor. The majority that controls the bill’s subsequent fate may well be the same majority that controls the committees whose jurisdiction has been ousted (depending upon the degree of agency slack between committees and floor); so a danger is that the floor majority will block the discharged bill through procedural subterfuges, or simply by refusing to take up the bill for a final substantive 36 37

Robert L. Stern et al., supra note 14, at 297. Id. at 298.

102

Mechanisms of Democracy

Submajority Voting Rules: Forcing Accountability

vote if it is not privileged business. Thus, under the submajority discharge rules obtaining in 1924, the Howell-Barkley bill for settling railway labor disputes was discharged by a submajority, but eventually died after its initial consideration; its supporters could never obtain a majority to move the bill out of the pile of “unfinished business” in order to force final consideration.38 In 1931, therefore, a near-unanimous House adopted new rules that “explicitly provided for the consideration of discharged measures as privileged business after their initial consideration,” effectively ensuring discharged bills an up-or-down vote.39 Comparing Congress to the Supreme Court, the 1924 disposition of the Howell-Barkley bill is analogous to a dismissal of certiorari as improvidently granted over four dissents, while the 1931 rule is analogous to the rule, or norm, that bars dismissals by a five-Justice coalition who lost at the certiorari stage. Institutional rules and norms are not the only way to safeguard submajoritarian decisions from reversal. Alternatively, some decisions might by their nature be costly to reverse or even intrinsically irreversible, and institutions might sensibly restrict submajority rules to that class of issues. We have seen that, as an empirical regularity, submajority rules often permit minorities to throw information into the public domain, by requiring roll-call votes that may be witnessed by spectators and later published in the legislative journals, by forcing other judges to hear public argument and dispose of the case with a full written opinion, by calling witnesses at committee hearings who will submit facts or arguments that undermine the majority’s preferred narrative, or by extracting documents or information from the executive. If there is a deep connection between submajority rules and information, the key to explaining it is that decisions about information are in some cases intrinsically irreversible. The basic asymmetry arises when a decision to reveal information today precludes a decision to conceal the same information tomorrow. Once published, perhaps by a submajoritarian decision, the information circulates beyond the power of subsequent majorities to suppress, whether or not they possess legal authority to do so. Thus Tom Paine mocked 38

39

Historically, this is a rather ambiguous case, because it is not clear whether a majority of the whole House would have supported the bill on final passage. Although the motion to discharge the bill from committee prevailed 194–181, a later majority voted 144–134 to strike out the enacting clause (rendering the bill inoperative); a yet later majority defeated a motion to refer the bill back to committee, 201–181. See Hasbrouck, supra note 23, at 160–61. Binder, supra note 15, at 149 n.28.

Institutional Design Writ Small

103

Chapter 

Burke’s futile attempt to check the spread of popular enlightenment: “Ignorance is of a peculiar nature; once dispelled, it is impossible to re-establish it.” This is the “irreversibility thesis”40: once the veils have been ripped away from obscurantist traditionalism, they can never be restored. Although the irreversibility thesis is pitched at the level of high political theory, the same mechanism operates at the lower level of institutional design. Once the opposing witnesses have been heard in public session, or their testimony placed in the record; once the executive branch has been forced to divulge information to committee minorities; once legal arguments have been ventilated in a public hearing; once the roll-call votes have been published in the newspapers; what can subsequent majorities really do? A natural asymmetry obtains, such that initial decisions to disclose cannot easily be undone. This point connects the positive and normative features of supermajority rules. A well-motivated institutional designer can most easily apply submajority rules to preliminary and procedural decisions that force public accountability upon majorities. Because such decisions typically produce public revelation of information, submajoritarian decisions in that domain are costlier for majorities to reverse. In that class of cases, the decisionmaking benefits of submajority rules reach their zenith, while the prospects of reversal and instability reach their nadir.

Cycles, Agenda-Setting and Manipulation A standard social-choice concern is that a submajoritarian rule for agendasetting combined with a majority voting rule for substantive decisions enables submajorities to manipulate outcomes for countermajoritarian ends, by exploiting latent voting cycles. (This problem cuts across the difference between rules that empower submajorities to add agenda items, on the one hand, and supermajoritarian agenda rules, or equivalent structures like the committee system, that empower minorities to affect agendas by blocking items that a majority might wish to consider.) Two points alleviate this concern. Submajority rules do not inherently privilege any particular minority over any other in agenda-setting.41 On the Supreme Court, any four of the nine Justices may form an 40 41

Don Herzog, Poisoning the Minds of the Lower Orders 86 (1998). Thanks to Robert Goodin for this point.

104

Mechanisms of Democracy

Submajority Voting Rules: Forcing Accountability

agenda-setting coalition in particular cases. Thus the distribution of agenda-setting power by submajority rules has a kind of impartiality about it, at least when viewed ex ante, from behind the veil of uncertainty. The contrast is with institutional rules that concentrate agenda-setting power in the hands of an identified subset of the whole body, such as a Rules Committee. More broadly, the manipulation of outcomes by agenda-setters is not clearly a real-world problem. The massive literature on cycling and agenda-setting has achieved great technical refinement but few compelling results for students of lawmaking institutions.42 At the level of theory, exploitation of the voting many by the agenda-setting few is extremely difficult to achieve, and requires strong modeling assumptions even to become minimally plausible. At the level of fact, the cycling literature has produced few confirmed claims. The theoretical problems are legion. Some of the relevant cycling stories arbitrarily assume that the agenda-setter is strategic while the voters are sincere, or that the agenda-setter is fully informed about the voters’ rankings while the voters do not understand the agenda-setter’s aims.43 Real institutions never employ the arbitrarily long agenda chains necessary to move outcomes to the agenda-setter’s preferred point.44 In models that take a slice or snapshot of ongoing institutions, agenda-setters can affect outcomes in limited ways, as in the standard models of committees acting under a closed rule. Such models, however, just pose the familiar questions about whether and why floor majorities permit committees with outlying preferences, relative to the floor median, to exploit closed rules or to exist in the first place.45 Where Congress (rather than some abstract “legislature” or “committee”) is the subject, there is all the less reason to think that agenda manipulation is an important concern. Legislators are above all members 42

43 44 45

Here I essentially follow the lead of Gerry Mackie, Democracy Defended (2003), and Bernard Grofman, Public Choice, Civil Republicanism, and American Politics: Perspectives of a “Reasonable Choice” Modeler,  Texas L. Rev. 1141 (1993). See Mackie, supra note 42, at 171–72. See Grofman, supra note 42, at 1569; see also Scott Feld et al., Limits on Agenda Control in Spatial Voting Games, 12 Mathematical & Computer Modeling 405 (1989). On the principal-agent problem between floor and committees, see Keith Krehbiel, Information and Legislative Organization (1992); Barry R. Weingast & William J. Marshall, The Industrial Organization of Congress; or, Why Legislatures, Like Firms, Are Not Organized as Markets, 96 J. Poli. Econ. 132 (1988); Kenneth A. Shepsle & Barry R. Weingast, The Institutional Foundations of Committee Power, 81 Am. Poli. Sci. Rev. 85 (1987).

Institutional Design Writ Small

105

Chapter 

of political parties, and a two-party system dampens the possibility of the latent voting cycles that are a prerequisite for agenda manipulation. “[A] two-party system inevitably creates a single-dimensional competition along the ideological spectrum. The effect of this single-dimensional competition reduces the number of likely preference orderings.”46 And, in fact, about 85% of roll-call votes in the Congress are predicted by a simple model that uses a single dimension (left-right) to measure legislators’ ideologies.47 We may add epicycles about cycling, such as the claim that latent or hidden cycles occur within the two major parties and are thus important even if the observable action within the legislature occurs on a single dimension. This seems unpersuasive, however. Intraparty conflict is centrally about money—which factions can bring the most money to the table—and this sort of willingness-to-pay to influence political outcomes effectively cardinalizes preferences, whereas many cycling results hold only with ordinal preferences. Finally, and of critical importance here, agenda manipulation requires sophisticated voting. But legislators will find it difficult to justify clever votes to constituents who may see only that the legislator has conspicuously voted against their interests and against the legislator’s professed commitments.48 Here is another link between submajority rules and publicity or transparency: provisions like the Journal Clause increase the number of public roll-call votes, in turn dampening agenda manipulation. Empirically, it has been notoriously difficult for cycling theorists to demonstrate clear cases of agenda manipulation in legislatures. (Judicial decisionmaking displays better examples,49 in part because legislatures enjoy greater scope than courts for logrolling and other practices that measure the intensity of participants’ preferences and thus avoid the standard cycling conditions.)50 As far as Congress goes, early claims that 46 47 48 49 50

Bernard Grofman, Public Choice, Civil Republicanism, and American Politics: Perspectives of a “Reasonable Choice” Modeler, 71 Texas L. Rev. 1541, 1557 (1993). See Keith T. Poole & Howard Rosenthal, A Spatial Model for Legislative Roll Call Analysis, 29 Am. J. Pol. Sci. 357, 368 (1985). See Arthur T. Denzau et al., Farquharson and Fenno: Sophisticated Voting and Home Style, 79 Am. Poli Sci. Rev. 1117 (1985). See Maxwell Stearns, Standing and Social Choice: Historical Evidence, 144 U. Penn. L. Rev. 309 (1996). See Maxwell Stearns, The Misguided Renaissance of Social Choice, 103 Yale L.J. 1219, 1276–80 (1994).

106

Mechanisms of Democracy

Submajority Voting Rules: Forcing Accountability

congressional history was replete with exploitative agenda-setting have been grievously undermined by subsequent work.51 Many studies have failed to find much or any evidence of the sophisticated voting that would be necessary to exploit cycles within Congress.52 Even if one or two examples are confirmed, still the frequency of cycles is probably trivial— too small a tail to wag the gigantic dog that is the legislative cycling literature. This apparent lack of cycling and sophisticated voting in Congress should not be too surprising. One point, already mentioned, is that it will be hard to explain sophisticated or strategic votes to constituents. Moreover, the danger for unscrupulous agenda manipulators is that those who have been duped in the short run eventually discover the bad faith, and band together to punish their overly cunning colleagues. So the possibility of sophisticated manipulation by agenda-setters looks theoretically abstruse, and empirically difficult either for participants to achieve or for analysts to confirm.

Submajorities and the Jury Theorem A different critique is that submajority rules may produce inaccuracy, or less accuracy than majority rule would attain, in situations where the conditions for the Condorcet Jury Theorem hold. This point does not hold uniquely for submajority rules; it is equally true for many deviations from simple majority voting, including supermajority rules, although the Theorem does hold for plurality voting.53 In the following discussion I shall address only versions of the Jury Theorem in which the “correct” answer is exogenously chosen or determined. In another, thinner version the Jury Theorem merely captures the probability that a majority of the voting pool has correctly chosen the result that a majority of voters would consider best promotes their individual interests, rather than the common 51

52

53

Compare William H. Riker, The Art of Political Manipulation (1986) (purporting to identify examples of agenda manipulation in congressional history) with Gerry Mackie, Democracy Defended (2003) and Keith Krehbiel & Douglas Rivers, Sophisticated Voting in Congress: A Reconsideration, 52 J. Pol. 548 (1990) (both works criticizing Riker’s evidence and conceptual premises). See Keith T. Poole and Howard Rosenthal, Congress: A Political-Economic History of Roll-Call Voting (1997), John D. Wilkerson, “Killer” Amendments in Congress, 93 Am. Pol. Sci. Rev., 535 (1999). See Christian List and Robert Goodin, Epistemic Democracy: Generalizing the Condorcet Jury Theorem, 9 J. of Pol. Phil. 277 ().

Institutional Design Writ Small

107

Chapter 

interest somehow defined.54 In the latter form, however, the Theorem lacks the special epistemic credentials that make it normatively challenging for nonmajoritarian decision rules. If all the Theorem is aggregating is the probability that the majority has correctly determined where its interests lie, minorities or submajorities with different interests owe it no normative respect, and the Theorem underwrites no objection to submajority or supermajority rules. To the extent that the sorts of decisions covered by submajority rules fall within the Jury Theorem’s scope, there is an epistemic cost to submajoritarianism. Then the only thing left to say would be that submajoritarian inaccuracy at the predecisional or agenda-setting stage might trade off against increased jury-theoretic accuracy at the stage of substantive decisionmaking. That tradeoff would obtain if, for example, the increased public accountability accompanying an up-or-down decision on the merits of a bill reduces the tendency of legislator-voters to follow the lead of party mavens, and thus increases the effective number of independent votes, with a resulting increase in the voting group’s overall accuracy. This is of course a contingent empirical conjecture. Public accountability might instead enhance the control of party leaders, by making commitments to vote in certain ways more easily monitored, and might thus reduce the effective number of independent votes, reducing in turn the accuracy of the whole voting group. (Of course if party leaders have a very high voting competence then group accuracy might possibly increase, despite the reduction in the effective number of votes,55 but under plausible assumptions a larger number of moderately competent voters quickly outperforms a single voter of even superhuman accuracy.) More importantly, however, most submajoritarian decisions probably fall outside the Theorem’s scope in the first place. Even with wellmotivated participants, agenda decisions are typically value choices about how the institution’s time is best spent, not judgments that we might label accurate or inaccurate. Participants in agenda decisions are not, or not commonly, trying to converge upon a correct judgment on the same 54

55

See Nicolas R. Miller, Information, Electorates and Democracy: Some Extensions and Interpretations of the Condorcet Jury Theorem, in Information Pooling and Group Decisionmaking 173–192 (Bernard Grofman and Guillermo Owen eds., 1986); see also Paul Edelman, On Legal Interpretations of the Condorcet Jury Theorem, 31 J. Legal Stud. 327, 337–39 (2002). David M. Estlund, Democratic Theory and the Public Interest: Condorcet and Rousseau Revisited, 83 Am. Pol. Sci. Rev. 1317 (1989).

108

Mechanisms of Democracy

Submajority Voting Rules: Forcing Accountability

question; they are expressing different preferences about how a fixed resource, time, should be expended, and each participant may be right (about how best to satisfy their own preferences) even if they reach different answers. The same point holds for other collateral decisions typically subject to submajority rules, such as the decision whether to publicize legislative votes. Recall that to publicize voting is always an agenda choice in itself, since a principal cost of roll-call voting is time, and the framers’ principal fear was that roll-call voting would be abused to delay legislative proceedings. The Jury Theorem typically fails to bite on the preliminary and procedural matters that submajority rules typically regulate.

Selecting the Status Quo Point We have seen that for every submajority rule there is a reciprocal (asymmetric) supermajority rule, with a mirror-image status quo point. Why do those rules not exist, while the corresponding submajority rules do? Institutional designers might say that the legislative roll must be called for every vote on final passage unless 4/5 + 1 opt for a voice vote; is there any reason to think that rule a bad one? There is a mirror-image question about supermajority rules. Why should not the three-fifths rule for tax increases be a rule requiring 2/5 + 1 to reject any proposed tax increase? In some circumstances, submajority rules are undoubtedly equivalent to their supermajority reciprocals. It is hard to believe that a Rule of Six—all certiorari petitions are on the docket unless six Justices vote to deny them—would produce any different results than the Rule of Four; abstentions are sufficiently rare on the Court, and the alternatives are sufficiently similar, that nothing turns on the location of the status quo point. But this equivalence does not hold generally. First, in some situations the supermajority reciprocal is conceptually ill-defined. Under standard procedures for recall elections, some submajority fraction—say, 15% of the number of voters in the last regular statewide election—are empowered to alter the status quo by putting the recall question on the ballot in a special statewide election. Here the idea of a supermajoritarian reciprocal is obscure; what would it mean to say that a recall question is always on the ballot (every day?) unless 85% + 1 affirmatively vote not to hold the recall election? The supermajority reciprocal is well defined in the certiorari setting only because an outside actor—the litigant—is authorized to file a certiorari petition in the first place. In the direct-democracy setting, however, there is no exogenously defined pool Institutional Design Writ Small

109

Chapter 

of potential questions that can be put on or off the agenda by supermajority voting. Second, even where the supermajority reciprocal is conceptually well defined, transaction costs may make the status quo sticky, thereby making consequential the choice between a submajority rule and its supermajority reciprocal. A submajority rule will be preferable to the reciprocal supermajority rule where, and to the extent that, the costs of assembling the necessary submajority are appreciably lower than the costs of assembling the mirror-image supermajority. In small-group settings like the Supreme Court, these two costs will rarely diverge to any appreciable degree. But in large-number settings, the costs of assembling a submajority will be far lower than the costs of assembling a reciprocal supermajority. The extreme case involves mass democracy; for concreteness, I will use the recall example again. Under current law, recalls occur when a submajority puts the question on the statewide ballot. Suppose that, contrary to what I have said above, we could imagine a conceptually coherent reciprocal rule under which a recall election would be held (at stated intervals?) unless recall opponents could muster a statewide supermajority to block the recall question from the ballot. But this scheme is silly; the difference between the two rules matters only if there are a sufficient number of abstentions, but it is predictable that there will be many. Put differently, if abstentions could be reduced to such a degree that a large supermajority of the whole statewide electorate could be mustered to exclude the recall question from the ballot, the recall issue would in substance already have been decided on the merits; a preliminary vote on the agenda, rather than the merits, would be otiose. Numerically, and in terms of the costs of assembling requisite fractions of the whole voting pool, legislatures are an intermediate case between courts and direct democracy. Submajority rules will sometimes matter in legislatures, because the difference between the lower-bound cost and the upper-bound cost will sometimes be appreciable, in turn because absenteeism or abstention is a chronic condition both in committee and on the floor. In a legislature that, like Congress at the beginning of the last century, is plagued by absenteeism, the difference between obtaining 145 signatures in favor of a discharge petition (under the pre-1935 submajority rule) and obtaining 291 signatures against a discharge petition (under the reciprocal supermajority rule) is highly consequential. Under the latter rule, the greater costs of assembling the requisite coalition would create greater scope for strategic behavior. Which bills would 110

Mechanisms of Democracy

Submajority Voting Rules: Forcing Accountability

be subject to possible discharge (unless a contrary supermajority could be assembled)? Note that a smaller minority than the fraction required by the submajority rule (here, one-third of the House) must, in such a regime, be entitled to force a supermajority to coalesce against discharge on any bill. In that case the opportunity for strategic abuse will be greater, and the resulting costs of delay higher, than under the submajority rule. If the submajority rule already strikes the right balance between the costs of delay, on the one hand, and the benefits of forcing accountability upon the majority, on the other, then the combination of a reciprocal supermajority rule with a lowerthan-submajoritarian trigger will be nonoptimal. Finally, we may connect these points to the earlier discussion of reversal and instability. I have suggested that submajority rules are formally equivalent to supermajority reciprocals (abstracting from the location of the status quo and the transaction costs of assembling coalitions in the face of abstentions). I have also suggested that submajority rules are used exclusively for preliminary and procedural matters, including agenda-setting. Perhaps there is a tension between these claims. Supermajority rules are frequently used for ultimate substantive decisions; on the account here, why would they not be confined to procedural matters, like their submajority reciprocals? But in fact this pattern makes perfect sense, precisely because the risks of reversal and instability reach a nadir in the procedural and agendasetting domains where submajority rules are usually found. In such domains, well-motivated institutional designers can have their cake and eat it too: they may conserve on the transaction costs of assembling coalitions by selecting a submajoritarian status quo point instead of a supermajoritarian one, while avoiding large collateral costs arising from the threat of majoritarian reversal. Conversely, because reversal is a more serious problem for substantive decisions than for preliminary and informational decisions, the higher transaction costs of supermajority rules are more frequently worth incurring where substantive decisionmaking is concerned.

Supply-side Issues: Majority Motivations and Political Constraints I have offered a strictly evaluative and prescriptive account of submajority rules, not a positive account of how submajority rules come to be. Where submajority rules should be used but currently are not, who will have an Institutional Design Writ Small

111

Chapter 

incentive to supply them? Why would a legislative majority ever commit itself to conferring accountability-forcing power on the minority, by adopting submajority rules? But of course, majorities often have done exactly that—both in Congress, where majorities created the submajoritarian committee discharge rule in place before 1935, and on the Supreme Court, where majorities created the Rule of Four. Apart from constitutional submajority rules such as the Journal Clause, it is always true that the submajority enjoys its special voting power only by virtue of rules created by the majority itself; in general, submajority rules are not constitutionally entrenched. Thus a majority of the Supreme Court could, today, doubtless change the Rule of Four, and a majority of Congress could certainly do so. Likewise a majority of the whole Congress has delegated submajoritarian voting power by statute (in the case of the Seven Member Rule), while a majority of the House has delegated power by internal rule (in the case of House Rule XI); both delegations are clearly revocable. Several mechanisms might cause majorities to confer accountability-forcing power on minorities.56 One possibility is that submajority rules come into existence when the majority that creates the rule fears that it will soon become a minority, and thus creates submajoritarian procedural rights to improve its lot when in opposition.57 Alternatively, in line with the veiling effect of uncertainty developed in Chapter 1, where the majority is uncertain about its future status, the uncertainty may cause it to adopt submajority rules that impartially benefit all parties. To be sure, these mechanisms operate only under special conditions. On one account,58 small majorities must often make procedural concessions to minorities, as a few defections from the majority to the minority can turn the tables. On the other hand, however, small majorities are sometimes more internally cohesive, perhaps because they incur lower costs of monitoring and disciplining their members; small majorities may thus be more likely to curtail

56

57

58

Michael Suk-Young Chwe, Minority Voting Rights Can Maximize Majority Welfare, 93 Am. Pol. Sci. Rev. 85 (1999) discusses a mechanism that might cause majorities to grant minorities special voting power on information-eliciting rather than accountability-forcing grounds. Similar mechanisms of partisan entrenchment are discussed in Howard Gillman, How Political Parties Can Use the Courts to Advance Their Agendas: Federal Courts in the United States, 1875–1891, 96 Am. Pol. Sci. Rev 511 (2002). Binder, supra note 15.

112

Mechanisms of Democracy

Submajority Voting Rules: Forcing Accountability

minorities’ procedural rights.59 The first point emphasizes a factor that reduces the capacity of small majorities to have their way, while the latter emphasizes an offsetting factor that increases the same capacity. The net effect is unclear. Overall, there is no reason to think that the only submajority rules we can have are the ones we actually observe, where we actually observe them. The political and motivational constraints are not so tight as that. Even self-interested majorities have perfectly sensible incentives to create and respect submajority rules. And, of course, sometimes majorities are not self-interested, and will adopt such rules if and when their public virtues outweigh their vices.

59

Douglas Dion, Turning the Legislative Thumbscrew: Minority Rights and Procedural Change in Legislative Politics (1997).

Institutional Design Writ Small

113

This page intentionally left blank

Chapter 4

Absolute Majority Rules: Optimizing Accountability

C

hapter 3 introduced an institutional-design mechanism, the submajority voting rule, that enables democratic minorities to force accountability upon majorities, and examined how submajority rules can be tailored to maximize their benefits while minimizing the risk that very small minorities will clog up the working of democratic processes. This chapter introduces another voting mechanism, the absolute majority rule, and explains how it too can contribute to optimizing accountability in democratic institutions. Absolute majority rules are best introduced by illustration. In the United States Congress, the votes of a majority of those present and voting are necessary to approve a law.1 In the legislatures of California and Minnesota,2 however, the votes of a majority of all elected members are required. All these legislatures use “majority rule.” All respect equality 1 2

Subject to the requirement of a quorum, as discussed below. Cal. Const. art. IV, § 8B; Minn. Const. art. IV, § 22. For other examples, see International Centre for Parliamentary Documentation of the Inter-parliamentary Union, Parliaments of the World: A Comparative Reference Compendium Vol. I 528 (1986).

Institutional Design Writ Small

115

Chapter 

among the group of legislators; no one’s vote counts for more than one, and there are no supermajority requirements. What is the difference between these schemes, and is it consequential? Similar examples are found in many polities. In each house of the Parliament of India,3 to enact a constitutional amendment requires (for most subjects) both (1) a majority of the total membership of the house and (2) two-thirds of the members of the house who are present and voting. How, if at all, do these requirements differ? In Iraq in 2005, the Transitional Law stated that “[a] general referendum will be successful and the draft constitution ratified if a majority of voters in Iraq approve and if two-thirds of voters in three or more [provinces] do not reject it.” The Iraqi Parliament interpreted this clause to mean that “two-thirds of voters” meant “two-thirds of registered voters,” not of those voting, even though “a majority of voters” in the same passage required only a majority of votes cast.4 What is at stake in this decision? In these examples, the issue is whether the majority is calculated with reference to the number of votes cast, the whole number of eligible voting members of the institution, or on some other basis. A fully specified voting rule must state both a multiplier and a multiplicand. To say that the voting rule should be “a majority” is an underspecified statement, like saying “X is more than” or “three multiplied by.” If a voting rule is to be coherently stated, one must ask “a majority of what”? The choice of multiplicand is at least as important as the choice of a multiplier. Suppose a voting body with one hundred members, a quorum rule under which at least a majority of fifty-one must be present and vote in order to take decisions with legal effect, and a standard multiplicand under which only those present and voting are counted. The choice of a multiplier makes some difference. With these specifications, a minimum simple majority is twenty-six; varying the multiplier to two-thirds requires a minimum supermajority of thirty-four. Varying the multiplicand, however, makes a much larger difference. With a multiplicand of all members, a minimum majority is fifty-one—unanimity if only a minimum quorum is present—and a minimum two-thirds supermajority is sixty-seven. In this and other examples, what makes the multiplicand important is that 3 4

Const. of India Part XX, http://indiacode.nic.in/coiweb/coifiles/part.htm (last visited Nov. 10, 2006). See Election Change Seems to Ensure Iraqis’ Charter, New York Times, Oct. 4, 2005, at A1.

116

Mechanisms of Democracy

Absolute Majority Rules: Optimizing Accountability

the difference between one-half and two-thirds of a small number may be much less than the difference between one-half of a small number and one-half of a larger number. Everything depends upon what the numbers actually are, but clearly the choice of the multiplicand is consequential. In what follows I will try to show that the design of voting multiplicands is just the sort of small-scale institutional feature that can have important effects of democratic accountability and other values.5 An absolute voting rule is a rule whose multiplicand is all members eligible to vote in the institution. The antonym of an absolute voting rule is a simple voting rule, under which the multiplicand is all those present and voting. An absolute voting rule can take any multiplier, as can a simple voting rule. To narrow the issues, I confine the discussion to a comparison between absolute majority rules and simple majority rules. Under plausible conditions, I will argue, absolute majority rules prove democratically superior to simple majority rules. Two claims are central. First, under certain conditions absolute majority rules will insure legislative majorities against a loss of control, and against strategic behavior by minorities, that can occur under simple majority rule. Democratic accountability, on any plausible conception, requires not just a certain sort of relationship between voters and legislators; it also requires a certain sort of relationship between legislative majorities and legislative outcomes. By making it less likely that legislative majorities will lose control of legislative outcomes on democratically suspect grounds—such as strategic behavior by minorities— the absolute majority rule promotes this crucial aspect of accountability. Second, by counting abstention as a negative vote, absolute majority rules permit strategic ambiguity on the part of legislators and other

5

For discussions en passant in the literature, see, among others, Amartya Sen, Collective Choice and Social Welfare (1979); Dan S. Felsenthal and Moshe Machover, “Ternary Voting Games,” 26 Int’l J. of Game Theory, 355 (1997); Keith L. Dougherty and Julian Edward, The Pareto Efficiency and Expected Costs of k-Majority Rules, 3 Pol., Phil. & Econ. 161 (2004). An excellent unpublished treatment of absolute majority rules from a socialchoice perspective is Keith L. Dougherty & Julian Edward, Simple vs. Absolute Majority Rule, University of Georgia Working Paper (Sept. 26, 2006) (on file with author). Dougherty & Edward evaluate simple majority rule and absolute majority rule by reference to a range of Paretian and ordinal-utilitarian welfare criteria, concluding that by most criteria simple majority rule is superior. The authors limit their treatment in three ways: (1) transaction costs are ignored; (2) all voting and nonvoting is assumed sincere; (3) there are no political constraints on the choice of voting rules. I relax these assumptions here. When transaction costs, political constraints, and strategic behavior are taken into account, absolute majority rules are superior to simple majority rules under plausible conditions.

Institutional Design Writ Small

117

Chapter 

institutional voters, who are empowered to contribute to a measure’s defeat without openly opposing it. In this way, absolute majority rules allow strategic nonparticipation that liberates institutional voters from accountability. Under some circumstances this liberation from accountability can itself be desirable on democratic grounds, because a broad range of democratic theories would count the relevant forms of accountability as bad—where, for instance, the accountability runs to narrow interest groups rather than to the electorate or to society generally. Putting the two claims together, the suggestion is that absolute majority rules can serve as optimizing mechanisms that promote good forms of accountability (the first claim) while suppressing bad forms (the second claim). Theorists with differing high-level commitments can each find one or the other of these rationales attractive; my aim is to be catholic among different abstract approaches to democratic theory, which converge to support absolute majority rules under the conditions I identify. As throughout the book, the analysis is normative and prescriptive. I make no attempt to offer a positive theory that would explain why absolute majority rules appear where they do and do not appear where they do not.

Preliminaries I will begin by briefly setting out the mechanics of majority rule with various multiplicands; especially important is the differing effect of abstention under different rules. The inquiry is limited by the following assumptions. (1) An absolute majority rule requires a majority of all members. (2) I discuss only cases in which voters must choose between two options, such as enacting a measure or adhering to the status quo, so I will ignore the plurality variant of majority rule. (3) Throughout, I assume that the legislative process is fully transparent, both in the sense that legislators can observe each others’ votes, and in the sense that outsiders can observe all legislators’ votes. This is the standard case for modern legislatures deciding on sub-constitutional issues. The optimal design of the transparency regime, and the interactions between transparency and voting, are deferred to Part III. Absolute majority rules are common. Even more common, however, is the scheme of simple majority rule favored by Robert’s Rules. Under this scheme, (1) a quorum is a majority of all members; (2) a simple majority is needed to change the status quo; and (3) only those present 118

Mechanisms of Democracy

Absolute Majority Rules: Optimizing Accountability

and voting are counted in calculating whether requirement (2) is satisfied.6 I will compare this scheme to the following absolute majority voting scheme, which I take to be standard: (1) a quorum is a majority of all members and (2) the affirmative votes of a majority of all members are required to change the status quo. Absolute majority rules and simple majority rules perfectly converge where all duly qualified members of the institution are present and cast votes. A major difference between the simple majority scheme and the absolute majority scheme is the effect of abstention from voting or physical absence from the voting institution. Under any absolute voting scheme, the requisite majority needed to enact a proposal is a fixed quantity, whereas under a simple majority scheme the requisite majority is variable. The consequence is that under the absolute majority scheme, abstention or absence from the legislature in effect counts as a negative vote, because it reduces the pool from which the requisite fixed number of affirmative votes can be drawn. Conversely, under a simple-majority scheme, “[e]liminating ‘no’ votes reduces the number of supporters needed to create a majority.”7 In a twelve-person voting body, under a simple majority scheme, six affirmative votes fail to enact a measure if six voters are opposed (assuming ties are broken in favor of the status quo), but the switch of even one vote from opposition to abstention changes the outcome. Under an absolute majority rule, the switch makes no difference. In general, this feature of absolute multiplicands is highly consequential: it entails that absolute majority rule produces a status quo bias.8 Suppose that a voting body with one hundred members must vote on a proposition.9 Forty members vote yes, twenty vote no, and forty either abstain or are not present. There is a standard quorum rule and (at least) sixty participants present, so action can be taken. Simple and absolute majority rules yield different outcomes, however. Under simple majority 6 7 8

9

See Henry M. Robert, Robert’s Rules of Order: Modern Edition (Darwin Patnode ed., Thomas Nelson Publishers 1989) (1876). Cary R. Covington, Building Presidential Coalitions Among Cross-Pressured Members of Congress, 41 W. Pol. Q. 47, 50 (1988). See Dougherty & Edward, Simple vs. Absolute Majority Rule, supra note 5, at p. 18–19. I will use the term “status quo” to refer strictly to the legal status quo, which may or may not correspond to any particular status quo for policy choices. The legal status quo may itself permit change in the policy status quo, as when an administrative agency is given, by current law, delegated power to decide whether or not to adopt a new policy. The example is adapted from Wikipedia, Absolute Majority Rule, http://en.wikipedia.org/ wiki/absolute_majority.

Institutional Design Writ Small

119

Chapter 

rule, only those present and voting are counted, so the measure passes forty-twenty. The measure would also pass even under a two-thirds supermajority multiplier with the ordinary present-and-voting multiplicand. Under absolute majority rule, however, the measure fails, since the fiftyone-vote absolute threshold is not reached. In this example, then, the status quo effect produced by changing the multiplicand is larger than the status quo effect produced by increasing the multiplier to two-thirds. What of the nonvoters?10 If they did not exist, so that the body had only sixty eligible voters, the proposal would pass under an absolute majority rule, indeed even under an absolute majority rule with a twothirds supermajority multiplier. “The effect of the [absolute majority rule] is to count every member of the body that does not vote affirmatively as voting against the passage of the act.”11 In other words, merely by declining to cast a vote in favor of the proposal, nonvoters can defeat the measure under an absolute majority rule. Under either the simple or the absolute majority rule, the abstainers could also defeat the measure simply by casting their votes against the measure. The absolute majority rule gives the abstainers another option, however: defeat the measure while failing to participate. Members who have other reasons for abstaining or being physically absent, including other pressing business or a desire to suspend judgment on the measure, contribute to its defeat. There is a complication here about the precise definition of the alternatives. The most common simple-majority scheme requires a majority of those “present and voting.” Legislators who abstain may still count towards a quorum, which means that under a simple-majority scheme, the winning majority may be far less than a majority of quorum. In the purest version of the rule, a statute may be enacted (with a quorum present) by the affirmative vote of only one legislator, if no one opposes.12 Yet there is an intermediate possibility. Consider the following discussion, from a recent case in the Alabama Supreme Court: Section 63 of the Alabama Constitution of 1901 provides that “no bill shall become a law” unless “a majority of each house 10 11 12

120

I am eliding here a possible distinction between abstention and physical absence from the legislature. The distinction is discussed below. Bd. of Supervisors v. Heenan, 2 Minn. *2 (Minn. 1858). Id. (“Previous to the constitution [which instituted an absolute majority rule] . . . laws could be passed by a single member voting in the affirmative, if no one voted against him.”)

Mechanisms of Democracy

Absolute Majority Rules: Optimizing Accountability

be recorded thereon as voting in its favor. . . .” This provision could be interpreted in at least three ways: 1. An absolute majority of each house—that is, 53 members of the 105-member House of Representatives and 18 members of the 35-member Senate—must vote in favor of a bill for the bill to pass; 2. Assuming a quorum of 53 House members and 18 Senate members are present when a vote is taken, a majority of those 53 House members (at least 27) and a majority of those 18 Senators (10) must vote in favor of the bill; or 3. Assuming a quorum of each house is present, a majority of those present and voting in each house must be in favor of the bill (e.g. if 53 House members are present, and 10 vote ‘yes’ and five vote ‘no’ and the remaining 38 do not vote, the bill passes the House; similarly, if 18 Senators are present, of which 5 vote ‘yes’ and 3 vote ‘no’ and the remaining 10 do not vote, the bill passes the Senate).13 The first interpretation is an absolute majority rule, under which either abstention or absence from the legislature is in effect counted as a negative vote. The second and third interpretations differ in the following way: the second requires a majority of those present, whereas the third requires only a majority of those present and voting. Under the second option, as under an absolute majority rule, an abstention by a legislator who is physically present in effect counts as a negative vote by reducing the pool from which the necessary majority may be drawn. Only by physically absenting herself from the legislature, so as not to count among those present, may the legislator ensure that she has no effect on the vote. Under the third option, by contrast, a legislator may abstain while remaining present, and the abstention is not counted. In other words the first option, absolute majority rule, treats both absence and abstention as a negative vote; the second treats abstention but not absence as a negative vote; and the third treats neither absence nor abstention as a negative vote.14 13 14

Civic Center Authority v. City of Birmingham, 2005 WL 1023157 (Ala. 2005) (Parker, J., concurring specially). The same three possibilities exist where the multiplier is a supermajority rather than a majority. In the United States Senate, for ordinary motions a majority of those present and voting are required for passage, assuming the presence of a quorum (as per option 3 in the text).

Institutional Design Writ Small

121

Chapter 

In terms of the three options listed by the Alabama Supreme Court, I am principally interested in the choice between the first and the third. The intermediate second option presents the issues less crisply. For simplicity, I will generally treat physical absenteeism and abstention as equivalent, except in those cases where the difference between the two makes a difference. With these preliminaries in place, I will offer a normative and prescriptive analysis of absolute majority rules. Absolute majority rules can simultaneously promote majoritarian control and can liberate voters from democratically undesirable forms of accountability. In this double-sided way, absolute majority rules promise to optimize accountability.

Majoritarian Control Absolute majority rules promote majoritarian control of institutional outcomes. The central concern here is what Jeremy Bentham called the tactic of “intentional surprise”15: under a simple majority scheme, minorities can exploit high majority absenteeism to outvote a rump majority. In legislatures, absolute majority rules block intentional surprise by strategic minorities, and thus ensure that the legislative majority retains a veto over outcomes even with a given level of absenteeism or abstention. I will begin by describing this effect, and then explain its connection to democratic accountability.

The Insurance Effect of Absolute Majority Rules Consider the ordinary legislative baseline of simple majority voting with a standard present-and-voting multiplicand. Where there is asymmetrical absenteeism between minority and majority, a group that is a minority in the chamber as a whole may prevail over the votes of a rump majority. Suppose a one hundred member body, a standard simple-majority scheme, and a quorum of fifty-one attendees, all of whom vote. The vote is twentysix in favor of a measure and twenty-five against. Because of the standard

15

122

However, two-thirds of the Senators present are required to ratify a treaty. See U.S. Const. art. II, § 2, cl. 2. The effect of the latter rule is that abstention by a Senator who is present is equivalent to a no vote, but absence is not. See Joseph Freixas and William S. Zwicker, Weighted Voting, Abstention, and Multiple Levels of Approval, 21 Soc. Choice & Welfare 399, 413 (2003). Jeremy Bentham, Political Tactics (Michael James et al. eds., Oxford University Press 1999).

Mechanisms of Democracy

Absolute Majority Rules: Optimizing Accountability

multiplicand, the twenty-six prevail even if, with full participation, they would have lost by a massive supermajority of seventy-four to twenty-six. Under an absolute majority rule, by contrast, there are not enough votes in favor to enact the measure. This is an extreme case16 that illustrates the more general point. An absolute majority rule prevents minorities from concentrating their forces at a decisive time and place, à la Napoleon, in order to overwhelm the few voters who are present from a larger but dispersed majority. More precisely, an absolute majority rule safeguards the majority against affirmative surprise attacks by Napoleonic minorities; it does not prevent the minority from ganging up to block a majority afflicted with absenteeism from taking affirmative steps to enact its preferred measures. This emphasizes that absolute majority rules protect the status quo, just as does raising the voting multiplier from majority to supermajority. A minority may employ Napoleonic tactics only if, and when, there is asymmetrical absenteeism or abstention, such that a disproportionate number of the majority do not contribute to the present-and-voting multiplicand. Asymmetrical participation may arise from several causes.17 In the most appealing case, the minority may care more intensely about a particular issue than do some or many members of the majority. As I will discuss below, this is the best argument against absolute majority rules from a democratic perspective. However, there are many other cases as well, ones that put legislative minorities in a less appealing light; and there is no reason to believe they are any less common. For one thing, even if policy preferences are equally intense all around, members decide whether to absent themselves or abstain based on random factors—political campaigning or personal business in the home district, travel problems, illness, and mandatory disqualification due to a conflict of interest. Sheer statistical flux will then 16

17

Note, however, that the risk of Napoleonic minorities increases when the quorum rule is set lower than a simple majority, as in the United Kingdom, Austria and Ireland. See Bjorn Erik Rasch, Parliamentary Voting Procedures, in Parliaments and Majority Rule in Western Europe 498 (Herbert Döring ed., 1995). Rational and public-spirited legislators might abstain if they are uninformed and are aware that they are uninformed. In such cases, the worst outcome for the uninformed voter is to cast the decisive or swing vote. Putting aside any desire to take a position for its own sake, the best course is then to abstain in deference to informed voters. See Timothy J. Feddersen & Wolfgang Pesendorfer, The Swing Voter’s Curse, 86 Am. Econ. Rev. 408 (1996). However, there is no reason to think that this sort of abstention is markedly asymmetric across chamber majorities and minorities.

Institutional Design Writ Small

123

Chapter 

ensure that alert minorities will sooner or later have an opportunity for action.18 Moreover, entrepreneurial minorities can and do search for and hit upon issues that put majority voters to a political Hobson’s choice, such that neither a vote for nor a vote against a measure is palatable.19 In such cases, many members of the majority may abstain, potentially giving the minority voting control. In the Philippines Senate in 2004, the minority bloc introduced a resolution indicting a politically controversial figure for fraud. The resulting vote was nine in favor, none against, and nine abstentions. “The minority bloc immediately declared victory. . . . [However, the Senate President ruled that because] the ayes and the abstentions are even, the result is a tie and the resolution has not been adopted.”20 In effect, the Senate President reinterpreted the voting rule as an absolute majority rule to protect the majority bloc. Finally, asymmetrical participation may arise precisely because a majority tries to maximize its power by spreading itself over more legislative terrain than the minority. This tactic can backfire if minorities are strategic. Thus, in committee voting in the United States House of Representatives before 1995, majorities faced a chronic problem: the multiplicity of committees meant that several committees or subcommittees would often meet at the same time. Simultaneous scheduling meant that “political control [might] slip away to a well-organized minority that might concentrate its strength at a single location for a ‘sneak attack’ on the majority.”21 The strategic threat of minority action may itself force the majority to maintain a lower level of absenteeism than it would otherwise choose, and this is a cost. The absolute majority rule avoids that cost, but exacts a price in turn by raising the cost of changing the legal status quo. Under the 18

19

20 21

124

See Lawrence S. Rothenberg & Mitchell S. Sanders, Legislator Turnout and the Calculus of Voting: The Determinants of Abstention in Congress, 103 Pub. Choice 259 (2000); Rothenberg & Sanders, Rational Abstention and the Congressional Vote Choice, 11 Econ. & Pol. 311 (1999). These studies find that in the 104th Congress, the timing of a vote during the week and during the session was the most important determinant of abstention; campaigning in the home district and (presumably) the demands of travel have a larger effect on the decision whether to vote than does the likelihood of being the pivotal voter. Martin Thomas, Issue Avoidance: Evidence from the U.S. Senate, 13 Pol. Behav. 1 (1991), finds that “proactive issue avoidance” through abstention, by members who do not want either to support or oppose a measure, occurs with nontrivial frequency in the U.S. Senate. Editorial, 112th Congress Has Nothing to Show, Manila Times Internet Edition (Feb. 9, 2004). Guide to Congress 552 (5th ed. 2000).

Mechanisms of Democracy

Absolute Majority Rules: Optimizing Accountability

absolute majority rule, a majority of the whole body cannot succeed if its participation falls below the absolute threshold, no matter what supermajority results. Under absolute majority rule in a one hundred member body, a vote of fifty for and twenty against fails to enact a measure, so the coalition or party that holds an absolute majority must ensure participation by at least a chamber majority to enact legislation. Minorities who are obstructionist, rather than affirmatively Napoleonic, benefit from the status quo effect of an absolute majority rule, so long as a decisive fraction of the majority is absent; and this raises the cost of majority absenteeism, the very cost the absolute majority rule reduces in other situations. The net effect may still be desirable from the majority’s point of view. The threat posed by strategic minorities forces legislative majorities acting under a simple-majority scheme to maintain a sufficient presence at all times. If there is some rate of absenteeism that the majority would like to permit among its membership, an absolute majority rule ensures that the majority can do so without exposing its flank to the minority’s concentration of force. The accompanying cost is that the absolute majority rule makes it harder to overcome obstructionist minorities who wish only to block legislation. Under an absolute majority rule, however, the chamber majority at least obtains the benefit of certainty: knowing the vote threshold it must meet to enact legislation, it can at least plan the occasions on which it is feasible and desirable to assemble in full to pursue its program, while retaining the assurance that absenteeism cannot be exploited by the minority on other occasions. Generally speaking, an absolute majority rule has the costs and benefits of an insurance policy. It protects the absolute majority from a minority’s tactic of concentrating forces, but forces the absolute majority to pay a premium in the form of higher transaction costs of organizing to enact legislation in other cases.22 Relative to simple majority rule, an absolute majority rule reduces the cost of majority absenteeism or nonparticipation, at least insofar as the threat is affirmative minoritarian opportunism rather than minoritarian obstruction. Majorities will favor this arrangement when the benefits of paying the insurance premiums exceed the costs. 22

Richard Forgette & Brian R. Sala, Conditional Party Government and Member Turnout on Senate Recorded Votes, 1873–1935, 61 J. of Pol. 467 (1999), shows that party leaders have substantial power over the rate of participation by members in the U.S. Senate.

Institutional Design Writ Small

125

Chapter 

Alternatives? There are other institutional arrangements that also tend to promote majoritarian control. I will examine several candidates: (1) the ability of voting majorities to reverse countermajoritarian results in later voting; (2) quorum-breaking by majorities and adjustments to the quorum rule; (3) adjustments to the voting multiplier; and (4) proxy voting. In general, although these devices qualify the risk of strategic minorities, they do not eliminate it. Absolute majority rules will often be the least costly and most effective means of controlling the risk of countermajoritarian results that arises under a simple-majority scheme.

Reversal by Voting I will begin with the ability of the absolute majority to reverse countermajoritarian outcomes in subsequent voting. Bentham expressed skepticism about the need for quorum rules on the following grounds: It might be apprehended, that where parties existed, those who found themselves one day in superior force, would abuse this superiority to the production of a decree contrary to the will of the majority. But this danger is not great; for the majority of to-morrow would reverse the decree of the past day, and the victory usurped by the weaker party would be changed into disgraceful defeat.23 The legislative majority that would have prevailed with full attendance may, on this view, simply repeal the minoritarian enactment the next time it assembles, and the minority, anticipating this, will refrain from the useless exercise. The ability to reverse minoritarian legislative action functions as an ultimate constraint that reduces the importance of the quorum minimum, a point missed by George Mason when he argued to the Philadelphia Constitutional Convention that without a quorum minimum in Congress, “the U[nited] States might be governed by a Juncto.”24 Yet this is a soft constraint; it dampens but does not eliminate the risk of Napoleonic minorities. Two points are important. First, some measures are costly to reverse in fact, even if they can be easily reversed in law. 23 24

126

Jeremy Bentham, The Collected Works of Jeremy Bentham: Political Tactics 62 (Michael James, et al. eds., 1999). 1 The Records of the Federal Convention of 1787, 252 (Max Farrand ed., 1966).

Mechanisms of Democracy

Absolute Majority Rules: Optimizing Accountability

A measure that reveals information to the public may be impossible to undo in any practical sense. So too, a measure that makes a payment of money or property out of the treasury may be legally difficult to undo if constitutional protections of property are triggered, or simply because it is politically costly for government to demand recoupment from parties who have relied to their detriment. Second, a minoritarian enactment may decisively change the status quo point. It may then be more difficult for the legislative majority to repeal an earlier minoritarian enactment than it would have been to vote it down in the first instance, even if the enactment has only been law for a brief period. The change in the status quo point may affect outcomes if some legislators support neither the enactment nor its repeal. Suppose that under a simple majority scheme the status quo is no law, that a bill is proposed, that it would be defeated seventy-four to twenty-six with full participation, and that it is enacted twenty-six to twenty-five with asymmetrical absenteeism. It does not follow that the new status quo will be reversed, and the old one reinstated, by a vote of seventy-four to twentysix when the legislature next convenes with full participation. If twentyfour of the forty-nine absent voters—less than a majority of the absentees—are opposed to both the enactment and its repeal, the vote on the reversal will be fifty-fifty. If so, the repeal will fail, because in most legislatures (other than those who give the presiding officer a tiebreaking vote) a tie defeats a measure.25 Opposition to any change in the status quo on the issue at hand might result from a concern for the stability of legal rules, a wish to use the constrained legislative agenda for other pressing business, or simple status quo bias. Even bracketing such cases, Bentham’s assumption is seemingly that a minority will not strike if the threat of future reversal would make the strike futile. This is erroneous, however; it overlooks the collateral costs that the minority may inflict. Even if the rules allow eventual reversal by an absolute majority, the sequence of minoritarian enactment and majority reversal imposes opportunity costs on the majority. Time, the most precious resource in legislatures, must be spent merely to restore the status quo ante the minority’s Napoleonic victory. This may be a large cost to the majority if, as often happens, the constitution or some other legal

25

See Robert, supra note 6, at 79.

Institutional Design Writ Small

127

Chapter 

rule mandates a cut-off date for the legislative session—in which case the minority wins, so long as it can delay adverse action long enough. Thus in the United States House of Representatives, as mentioned above, minorities have often attempted to concentrate their forces at a particular committee meeting for an attack on overstretched majorities. In some cases the minority’s goal is simply to block committee approval of the majority’s program, but in others it is to secure affirmative (albeit minoritarian) action from the committee, such as an amendment to a bill or the initiation of a hearing or investigation. In such cases, I conjecture that the minority is under no illusion that the full House would enact a minority-favoring bill, or would otherwise assent to the committee’s action. The point of the attack, however, is presumably just to disrupt the ordinary schedule of business, and in general to throw sand into the gears of the House’s majoritarian machine.

Quorum Rules and Quorum-breaking Although quorum rules do partially promote the aim of majoritarian control, they do so only in a sharply limited fashion. As indicated by the previous example, in which twenty-six legislators outvote twenty-five representatives of an overwhelming but dispersed majority, the standard majority quorum requirement is simply too weak to prevent the concentration tactic. Likewise, although Rule XI of the U.S. House of Representatives provides that “[a] measure or recommendation may not be reported by a committee unless a majority of the committee is actually present.”26 that majority may be composed mostly of minority members. The basic problem is that under a simple majority voting scheme with a standard majority quorum minimum, absenteeism or abstention by members of the absolute majority does not contribute to defeating a minoritarian proposal; a quorum rule does nothing to address this problem. To be sure, standard quorum rules do allow quorum-breaking by the rump majority as an ultimate tactic of majoritarian control. Where Napoleonic tactics are feared, all the majority members who are present may physically absent themselves or abstain to ensure that the quorum minimum is not satisfied. In the running example of a twenty-six–twentyfive vote, if the quorum is defined by votes cast rather than by those physically present, the twenty-five members of the dispersed absolute majority 26

128

House Rule XI (h)(1).

Mechanisms of Democracy

Absolute Majority Rules: Optimizing Accountability

may simply decline to vote rather than vote in the negative, and can thus shut down the body altogether until reinforcements arrive. Even if the quorum is defined by those present, the rump majority can leave the building altogether. This tactic—postponement of the legislative session to defeat a concentrated minoritarian attack—presumably also underlies the rule, recently adopted in the United States House of Representatives, that a committee chair may postpone a vote until a future date certain.27 The tipoff here is the further provision that, when the committee reconvenes, the proposal will be open for amendment by majority vote, even if a majority of those attending the first session already ordered the previous question.28 A committee chair who realizes that her troops from the majority side have failed to turn out can thus choose to fight another day. Quorum-breaking is a sword with two very sharp edges, however. Historically, majorities have benefited by the standard quorum rule, which counts all those physically present as contributing to the quorum. In the United States House of Representatives before 1890, the quorum rule was defined by the number of votes cast, rather than by physical presence. When the majority party could not muster enough members, this allowed minorities to employ the “disappearing quorum,” remaining silent in order to break the quorum and thus to block action. Speaker Thomas Reed ruled that those present but not voting counted towards the quorum, thus putting the minority to the much less appealing choice between being outvoted or physically absenting themselves from Congress.29 Once put in place, however, the Reed rule binds the majority as well as the minority. Where the majority puts the Reed rule in place at a given time, majoritarian quorum-breaking at a later time requires that the rump members of a dispersed absolute majority not merely abstain, but actually stage a public walkout.30 This version of the quorum-breaking tactic is politically more costly, because more obviously strategic; I will give an 27 28 29

30

House Rule XI (h)(4)(A). House Rule XI (h)(4)(B). See Sarah A. Binder, Minority Rights, Majority Rule: Partisanship and the Development of Congress 125–28 (1997). The celebrated first example of a disappearing quorum occurred when Rep. John Quincy Adams and supporters refused to answer the roll-call during voting on a pro-slavery bill, precluding a quorum. Roger H. Davidson & Walter J. Oleszek, Congress Against Itself 23 (1977). Note that, when concentrated minorities have the upper hand because of majority absenteeism, it is too late for the majority to attempt to change or suspend the Reed rule itself—even apart from the blatantly strategic character of such a course.

Institutional Design Writ Small

129

Chapter 

example shortly. The walkout tactic is also risky, because it requires high coordination and compliance among majority members. If even some of the majority’s members defect (by participating in the minority-dominated vote), the result will be the worst-case scenario from the majority’s point of view: a quorum will be established, but with the minority in voting control. Similar issues are presented by a closely related tool of majoritarian control, which is to adjust the quorum rule upwards while retaining a simple-majority voting rule. Consider a scheme, used in a few legislatures, that combines (1) a simple majority voting rule with a standard presentand-voting multiplicand and (2) a supermajority quorum requirement, such as the presence of two-thirds of the membership. Under this scheme, the minority’s concentration of force in our running example would be inadequate. The cost of this hybrid scheme, however, is that minorities may exploit the disjuncture between the quorum rule and the voting rule to block action by an absolute majority or even a supermajority. Under this scheme, if sixty-six legislators favor a measure, and thirty-four oppose it, the measure will not be defeated if the minority votes against it but will be defeated if the minority withdraws from the legislature to defeat the quorum. Thus in a legislative struggle over redistricting in the state of Texas, a legislative chamber with one hundred fifty members, a quorum minimum of one hundred and a majority party of eighty-eight was temporarily paralyzed by the walkout of fifty-one dissenters.31 Importantly, however, the high political costs of this conspicuously strategic behavior soon forced the minority’s return32—underscoring the limits of the walkout tactic as a tool of majoritarian control.

Supermajority Multipliers If a standard quorum rule and standard quorum-breaking tactics are not well suited to block Napoleonic tactics by minorities, another option is to make an upward adjustment in the voting rule multiplier. One might, that is, adopt (1) a standard quorum minimum, (2) a supermajority multiplier, 31

32

130

See Texas House Paralyzed by Democratic Walkout, Redistricting at Issue, May 19, 2003, http://www.cnn.com/2003/allpolitics/05/13/texas.legislature. For similar European cases, see Rasch, supra note 16, at 497. See Michael Martinez, Dems Flight Was Wrong, Many El Pasoans Say, El Paso Times, Nov. 20, 2003 at 1A; GOP Gets Quorum in Texas Senate: Democrat Breaks Ranks, Facilitating Remapping Effort, Chicago Tribune, Sept. 16, 2003, at 101.

Mechanisms of Democracy

Absolute Majority Rules: Optimizing Accountability

and (3) a standard present-and-voting multiplicand. Many legislatures use such a scheme for some issues, although few use it generally for all issues. The larger questions about such a scheme, of course, involve the costs and benefits of supermajority rules. A chamber majority that creates a supermajority multiplier to protect itself from affirmative minority action will pay a price when obstructionist minorities block action that the majority would desire to take. But the same is true of an absolute multiplicand, which dampens the risk of affirmative minority opportunism while increasing the costs of organizing to overcome obstructionist minorities. The narrow point of interest here is this: Insofar as the majority’s aim is just to minimize the risk of affirmative Napoleonic minorities, upward adjustments to the multiplier are generally inferior to adjustments to the multiplicand, on essentially mathematical grounds. The effect of large changes in the multiplicand will frequently swamp the effect of small changes in the multiplier. The difference between one-half of all those present and voting and two-thirds of the same multiplicand, on the one hand, will often be much less than the difference between one-half of those present and voting and one-half of all members, on the other. With one hundred members and a standard quorum rule, as we have seen, only a voting rule with unanimity in the multiplier will duplicate the effect of an absolute voting rule if the number of those present-and-voting is equal to the quorum minimum of fifty-one. (If any of those present do not vote, the absolute majority rule cannot be satisfied even by unanimity.) This is not a realistic case, because participation will usually not be so low. The larger point, however, is that if the asymmetries in absenteeism are serious, one may need a very high multiplier to duplicate the majorityprotecting effect of an absolute majority rule. In a legislature of one hundred members divided fifty-fifty on the relevant issue, where seventyfive participate and all absentees are from one group, even a two-thirds supermajority rule would permit an enactment by the fifty voters from the other group (i.e., by a vote of fifty–twenty-five), although the enactment would fail under full participation, a simple majority rule, and standard tiebreaking rules.

Proxy Voting Finally, there is the possibility that majorities can use proxy voting to maintain control while permitting their members the desired rate of absenteeism. In the House of Representatives before 1995, “proxy voting Institutional Design Writ Small

131

Chapter 

was simply a means of ensuring majority control over committees as subunits of the House and preventing such control from succumbing to the whims of committee scheduling or flukes of members’ absences.”33 When in 1995 the Republican majority abolished proxy voting in committees, [T]he Republicans endured immediate political pain for their decision. They were forced to conduct numerous simultaneous committee meetings and House floor votes to ensure prompt passage of the legislative agenda promised in the election campaign, while struggling to maintain voting control with the narrowest House majority in forty years. . . . [T]he Republicans eventually loosened some of the other restrictions they had passed on the number of committee and subcommittee assignments members might hold, so they would be able to place enough majority members in the right places to ensure control.34 This point dramatizes that, in the absence of control devices such as proxy voting or an absolute majority rule, majorities will be forced to maintain a lower rate of absenteeism than they would otherwise desire. Proxy voting lowers the cost of majority absenteeism, as legislative minorities often point out in high-minded fashion. Why then is proxy voting not more common? Although some legislatures permit it at the committee level, most do not, and it is allbut-universally prohibited at the level of voting in the whole legislature (the main exceptions being Brazil, France and some of the francophone nations of Africa).35 The standard conjecture, which I accept here, is that there are tight political constraints on proxy voting. On this view, the politically active public understands that full participation in every case is an unattainable and indeed undesirable goal, and will thus tolerate some level of absences and abstentions. But representatives who both absent themselves and conspicuously cast a proxy vote will be seen as irresponsible. If the proxy was given before the relevant proposal was placed on the 33 34 35

132

Guide to Congress, supra note 21, at 552. Id. at 552. See Parliaments of the World, supra note 2, at 479–93.

Mechanisms of Democracy

Absolute Majority Rules: Optimizing Accountability

agenda, the proxy vote will be and will seem uninformed, rather than being a ministerial arrangement. The overall claim is that absolute legislative majorities will favor absolute majority rules where the insurance benefits of absolute majority rules outweigh the greater organizational demands and transaction costs. An absolute majority rule raises the costs of assembling a coalition to enact laws, and thus makes it harder for majorities to overcome obstructionist minorities, but it also provides protection against tactics that might allow opportunistic minorities to obtain affirmative enactments. No alternative measures for securing majoritarian control are equally effective. These are ceteris paribus points. They neither show nor are intended to show that absolute majority rules are superior to simple majority rules in all circumstances. One must net out the costs and benefits indicated above as facts dictate in particular settings, and compare them to the net costs of the alternatives. But the prospect of affirmative strategic behavior by minorities counts against the standard simple-majority scheme and makes absolute majority rule a real alternative.

Legislative Majorities and Democratic Accountability Democrats of many stripes can converge on the view that legislative majorities deserve protection against strategic behavior by minorities, and that such protection is justified by standard conceptions of democratic accountability. At a minimum, accountability has two elements: (1) a particular relationship between voters and legislators; (2) a particular relationship between legislators and legislative outcomes. The second element is just as important as the first, on any conception of democracy. Representatives who do not control outcomes are in a practical sense not representatives at all. By ensuring that legislative majorities can maintain control of outcomes, in the face of Napoleonic minorities, absolute majority rules promote this form of accountability. Of course, democratic accountability need not entail majority voting on outcomes. (In Chapter 3, I suggested a form of nonmajority voting on preliminary and procedural questions; here the subject is the final enactment of legislation.) There are familiar democratic arguments both in favor of and against majority voting. Many of the competing views were introduced in Chapter 3. Where value choices are involved, May’s Theorem shows that majority rule has powerful egalitarian appeal. Institutional Design Writ Small

133

Chapter 

With two options, it is the only voting rule that simultaneously (1) treats votes neutrally, without regard to whether they are cast for or against a proposition; (2) treats each voter anonymously, without regard to identity; and (3) is positively responsive to shifts in voter preferences.36 A supermajority multiplier, by contrast, violates neutrality. Votes against a proposition can defeat it even if the same number of votes in favor cannot enact it—producing a status quo bias.37 Where right answers rather than value choices are involved, the Condorect Jury Theorem entails that majority voting is most likely to identify those answers so long as voters are on average more likely correct than incorrect, and so long as errors are uncorrelated. Against these points, however, advocates of supermajority rules emphasize that most legislative decisions involve value choices rather than Condorcetian aggregation; that supermajority rules may prevent welfarereducing exploitation of minorities who hold especially intense preferences on certain issues;38 that supermajority rules may induce broader consensus; that under narrow circumstances, supermajority rules can actually improve the majority’s own information by encouraging minorities to reveal information;39 and that (tangentially to my interests here) supermajority rules dampen cycling in cases where three or more options are present.40 On the other hand, it has been pointed out that majority rule gives minorities the best chance to join a coalition sufficiently large to overturn unfavorable decisions in the future.41 So democrats disagree on whether majority voting on outcomes is always or necessarily desirable. I make a narrower point, however. The claim is not that legislative majorities should always be in control; it is that they should not lose control except for democratically permissible reasons, and in democratically permissible ways. If minorities have especially 36 37

38 39 40 41

134

The literature stems from Kenneth O. May, A Set of Independent, Necessary and Sufficient Conditions for Simple Majority Decision, 20 Econometrica 680 (1952). Again, I am bracketing the case of asymmetrical voting rules, which do not favor the status quo. See Robert Goodin & Christian List, Special Majorities Rationalized, 36 Brit. J. Pol. Sci. 213 (2006). See James M. Buchanan & Gordon Tullock, The Calculus of Consent (1962). See Compare Michael Suk-Young Chwe, Minority Voting Rights Can Maximize Majority Welfare, 93 Am. Pol. Sci. Rev. 85 (1999). See Joseph Greenberg, Consistent Majority Rule over Compact Sets of Alternatives, 47 Econometrica 627 (1979). See A.J. McGann, The Tyranny of the Supermajority: How Majority Rule Protects Minorities, 16 J. of Theoretical Pol. 53 (2004).

Mechanisms of Democracy

Absolute Majority Rules: Optimizing Accountability

intense preferences, or special information on a certain matter, or are especially subject to exploitation, then there are good democratic arguments for a voting rule that takes these circumstances into account. The problem with simple majority rule, however, is that it permits a form of strategic behavior by minorities that need not correspond to any of these democratically sensible grounds for departing from majoritarianism. As explained above, the sort of Napoleonic tactics that minorities can and do employ under simple majority rule have no necessary connection to, or even systematic correlation with, situations in which minority preferences are unusually intense, or minorities are especially susceptible to exploitation, or minorities have special information that majorities would like to elicit. This point can be approached from another direction as well. Is it necessarily desirable, democratically speaking, to give majorities iron-clad insurance against affirmative minoritarian lawmaking? Some democratic theorists might deny that it is.42 Perhaps minorities should win some of the time, especially when their preferences on a given issue are particularly intense. Perhaps a scheme of affirmative minority lawmaking ensures a kind of “second-order diversity,”43 in which diversity of views or preferences in the voting body also translates into diversity of results, rather than invariable victory for the majority. As with probabilistic voting—in which a vote of ninety legislators against and ten legislators for yields a 90% chance that the enactment will be defeated and a 10% chance that it will succeed—the standard simple-majority scheme ensures that minorities will sometimes win, and this may be valued as producing a more diverse, less monolithic body of law.44 The comparison with probabilistic voting cuts the other way, however. Probabilistic voting can produce minority-preferred policies, but does not reward strategic behavior by minorities.45 By contrast, as noted above, the threat posed by strategic minorities may force legislative majorities acting under a simple-majority scheme to maintain a sufficient presence at all times. Given that there is some socially beneficial rate of 42 43 44 45

See, e.g., Lani Guinier, The Tyranny of the Majority (1994). See Heather Gerken, Second-Order Diversity, 188 Harv. L. Rev. 1099 (2004). Dennis C. Mueller, Probabilistic Majority Rule, in Readings in Public Choice Economics 296 (Jac C. Heckelman ed., 2004). At least in the last round of voting, a probabilistic voting rule makes sincere voting a dominant strategy. See id. at 300.

Institutional Design Writ Small

135

Chapter 

absenteeism, this represents a social cost that is not present under probabilistic voting and other schemes that are second-order diverse. Praise for probabilistic voting is not a fully adequate response to the idea that minorities should sometimes win, however. Perhaps the idea is that, for expressive reasons, minorities should sometimes win by a majority vote among those present and voting. Majorities should sometimes know what it feels like to be outvoted in a majoritarian order, as will occur under simple majority voting when absenteeism among the majority hands tactically alert minorities a victory. That sort of win for minorities does not occur under pure probabilistic voting; there, a small minority of the present-and-voting group might prevail on any given issue by a lucky draw, but it does not prevail by a majority vote. It might be nice for legislative minorities to sometimes win by majority vote, but is it good overall to adopt a voting rule that allows such results? Here too, the crucial question is whether the asymmetrical absenteeism needed to produce minoritarian enactments under simple majority rule is a plausible proxy for asymmetries in intensity of preference, or vulnerability to exploitation, or any other democratically respectable reason for supporting minority victory. The answer is probably no, because the strategic behavior of Napoleonic minorities need not correlate with any of those things. That a minority has concentrated its forces at a given time and place to take advantage of the majority’s disproportionate absenteeism may reflect nothing more than canny tactics, or statistical flux, as opposed to any underlying difference in views, or preferences, or judgments, or information, or anything else of interest to democratic theorists. Of course this does not show that absolute majority rules are democratically superior to simple majority rule in all circumstances; one must take other considerations into account, particularly the tradeoff between the ability of absolute majority rules to insure against affirmative lawmaking by strategic minorities, on the one hand, and their tendency to increase the blocking power of minorities, on the other. The precise circumstances under which absolute majority rules are or are not preferable will vary with the facts, as the values of the relevant variables change over time and across institutional contexts. But we have made progress by identifying the conditions under which absolute majority rules are superior, and the grounds on which they might be so. From the standpoint of democratic accountability the prospect that legislative majorities will lose control of 136

Mechanisms of Democracy

Absolute Majority Rules: Optimizing Accountability

outcomes due to strategic behavior, and the costs of preventing such behavior, count against the standard simple-majority scheme to which absolute majority rule is an alternative.

Strategic Nonparticipation I will now turn to the second major justification for absolute majority rules. In the first justification, absolute majority rules are devices for protecting legislative majorities from strategic behavior by minorities, which in turn indirectly enforces an important aspect of democratic accountability. The reverse of the coin is that absolute majority rules can be seen as devices for optimizing accountability under conditions where some forms of accountability are democratically undesirable. We have seen that absolute majority rules treat both absence and abstention as equivalent to a negative vote. Standard simple-majority schemes, on the other hand, treat absence or abstention as a nullity. The incentive effect of standard simple-majority schemes is that opponents of a measure must attend and vote negatively in order to contribute to the defeat of a measure. Under an absolute majority rule, opponents may contribute to the defeat of a measure either by a negative vote, or by not voting or attending at all. Merely by declining to cast a vote in favor of the proposal, nonvoters can defeat the measure under an absolute majority rule. Under either the simple or the absolute majority rule, the abstainers could also defeat the measure simply by casting their votes against the measure. The absolute majority rule gives the abstainers another option, however: the strategic ambiguity of nonparticipation. Any particular member may claim other reasons for abstaining or being physically absent, including other pressing business or a desire to suspend judgment on the measure. It will often be unclear whether the abstention was genuinely motivated by these reasons or by opposition to the measure on the merits. The strategic ambiguity of nonparticipation is seemingly objectionable from a democratic point of view. In the simple majority case, Bentham condemned abstention or absence as an evasion of accountability: [Where legislators may be absent,] there is entire security, not for complete prevarication, but for demi-prevarication. Suppose a measure so bad that a deputy, if he were present, could not in honour refrain from voting against it. Does he fear to offend a Institutional Design Writ Small

137

Chapter 

protector, a minister, or a friend? He absents himself: his duty is betrayed, but his reputation is not compromised.46 Superficially, the charge holds even more strongly for strategic abstention under an absolute majority rule. In the latter case, unlike the simple-majority case Bentham is discussing, the abstainer can both vote against the measure (in effect) and also take no part; rather than confronting the choice between “honour” and “reputation,” the abstainer may contribute to the measure’s defeat while retreating into ambiguity to shield her reputation. I will argue, however, that this additional effect can make abstention under an absolute majority rule better, from a range of democratic perspectives, than abstention under a simple majority rule. The core idea is that nonparticipation is ambiguous, and that strategic voters can exploit this ambiguity to liberate themselves from accountability. Under some circumstances, such as where the relevant accountability would run to partial interest groups rather than constituents or society at large, liberating voters from accountability can actually be desirable on democratic grounds. Why, if at all, should opponents of a measure enjoy a choice between outright opposition (voting no) and opposition by default (abstention under an absolute majority rule)? Perhaps simple-majority rules are preferable because they force both supporters and opponents to stand and be counted. This is the core of Bentham’s complaint that abstention and absenteeism amount to a type of “demi-prevarication.” On Bentham’s analysis, the member who would vote in the negative, if forced to vote sincerely on a measure stipulated to be bad, may simply absent herself or abstain; typically this may be because she “fear[s] to offend a protector, a minister, or a friend,”47 or (we may add) an interest group or a President. Under a simple-majority scheme the effect of the absence or abstention is to “deprive one party of [her] vote,”48 that is an affirmative vote in favor of the measure, but the absence does not amount to a negative vote, and is thus less objectionable to the third party who is pressuring or coercing the member for support. As we have seen, under a simple-majority scheme, in

46 47 48

138

Jeremy Bentham, The Collected Works of Jeremy Bentham: Political Tactics 57 (Michael James et al. eds., 1999). Id. Id.

Mechanisms of Democracy

Absolute Majority Rules: Optimizing Accountability

contrast to an absolute majority rule, “[e]liminating ‘no’ votes reduces the number of supporters needed to create a majority.”49 When pressuring vacillating legislators for support, if President Lyndon Johnson “could not obtain a favorable vote, he might ask, in effect, for an abstention.”50 The crucial point, however, is that under a simple-majority scheme the opponent of a bad measure, who is assumed subject to pressure or coercion from third parties to vote in the affirmative, faces a Hobson’s choice: either “betray her duty” or “compromise her reputation.” Absenteeism or abstention amounts to choosing reputation over duty. Absolute majority rules solve this problem, in the sense that under an absolute majority rule, the opponent of a bad measure can have her cake and eat it too, opposing a measure simply by failing to attend or to vote in its favor. This allows the opponent both to do her duty, from an impartial point of view, and to avoid compromising her reputation with the thirdparty coercer by casting an openly negative vote. In short, absolute majority rules liberate voters from accountability in certain cases. One might reason that the third party will also understand the negative effect of absence or abstention under an absolute majority rule, and will thus punish the member’s absence or abstention in precisely the same way, or to the same degree, that the third party would punish a formal negative vote. This is erroneous, however. Some absences or abstentions occur from reasons or causes other than opposition on the merits, causes such as illness, involuntary absence, other business either within or without the legislature, and so on. The third party knows this, and if the third party cannot reliably distinguish the different motives for absence, the member may plausibly lie about her motives. Here the member takes refuge in prevarication that promotes her duty rather than betraying it. This is the strategic ambiguity of nonparticipation: if members who are opposed on the merits can mimic members who have good grounds for abstention or absence, a type of pooling equilibrium arises that helps to shield all nonvoting members from third-party coercion. In principle, the third party may punish abstainers or absentees to hold them accountable for their abstention or absence. But it is common ground that some nonvoting is excusable, even praiseworthy. Members who can gin up a 49 50

Covington, supra note 7, at 50. Id. (quoting Edwards (1980).

Institutional Design Writ Small

139

Chapter 

plausible pretext for nonparticipation may thus take refuge in the inability of third parties to sort good reasons from excuses. Consider an example from procedure in the United States Senate. When Senate Democrats filibustered a controversial nominee for U.N. Ambassador, John Bolton, the final vote on cloture was fifty-four in favor, thirty-eight against and eight abstentions.51 By virtue of the absolute voting rule for cloture, which requires sixty of the Senate’s one hundred voters to succeed no matter how many participate, the cloture motion was unsuccessful. The abstentions effectively counted as no votes; had six of the eight abstainers voted in favor, cloture would have been obtained. I conjecture that some fraction of the abstainers, marginal members of the Republican party, opposed the nomination but took refuge in abstention in order to deflect political pressure from the administration. But are such tactics democratically desirable? Sometimes the answer is yes, whatever one’s conception of democracy. Absolute majority rules work well in the situation Bentham posits, where the accountability that creates pressure to approve the underlying measure is itself assumed to be objectionable. A broad range of democratic theories, with differing conceptions of accountability, can converge on the idea that liberating voters from public accountability can be good as well as bad. That is because accountability can be bad as well as good, depending upon whom the voter is accountable to and whether that party desires legislation that is good or bad. Accountability to third parties or subgroups who have interests different from the aggregated interests of the represented class as a whole— such as ill-motivated executives or narrow but well-organized rent-seeking groups—is bad representation under any plausible view. In the case of the Bolton nomination, conditional on believing the nomination disastrous, the possibility of abstention-cum-opposition was a useful way to shield principled action from adverse political retaliation by the White House. No controversial view of accountability is at issue here. Of course differing theories of representation suppose differing views of accountability. On a loosely Burkean view of representation, the legislator is a trustee for the common good.52 On a competing view, the legislator is a 51 52

140

Charles Babington & Jim VandeHei, Democrats Block Vote on Bolton, Wash. Post, June 21, 2005, at A01. Edmund Burke, Speech to the Electors of Bristol, in The Founder’s Constitution (Philip B. Kurland & Ralph Lerner eds., 1987), http://press-pubs.uchicago.edu/founders/documents/ v1ch13s7.html.

Mechanisms of Democracy

Absolute Majority Rules: Optimizing Accountability

delegate from a particular constituency, typically defined in geographic terms. Both views, however, can agree that some types of accountability are objectionable. Suppose that a narrow interest group presses for legislation that would impose net harms on the larger constituency and also on the common good, however defined; that there are asymmetries in the costs of organization or elsewhere in the political environment that make it impossible for other groups to oppose the measure successfully; and that the strategic ambiguity of nonparticipation would enable legislators to defeat the measure without fear of political punishment from the interest group. Then there is a kind of tradeoff between accountability to the interest group and accountability more broadly, such that protecting the latter type of accountability (however it is specified) requires suppressing the former type. I will return to this point at length in Chapter 6. To be sure, nothing in the strategic ambiguity of nonparticipation guarantees that its effect will be good. If we take Bentham’s third party to be the whole class of those whose interests are properly taken into account, given the underlying theory of representation—all voters from a legislative district, or all citizens of the polity—then this mechanism might allow members to liberate themselves from accountability in a way that can be normatively improper. Recall the case from the Philippines Senate in 2004, where a politically motivated resolution of indictment received nine votes for, none against, and nine abstentions, a tally that defeated the resolution under the prevailing absolute majority rule. Were these abstentions good, because they amounted to strategically sophisticated votes against a bad proposal, or were they bad, on the opposite grounds? One simply cannot tell from the voting alone. But this is just to say that circumstances will vary. Under the conditions I have outlined—where accountability to some subset of third parties will undermine broader and more desirable forms of accountability— then whatever normative theory of representation is assumed to govern will imply that liberating members from accountability is desirable. We will see an extended example of such a domain in Chapter 6, where the process of legislative budgeting is examined; there accountability to narrow interest groups often has democratically undesirable effects, on any plausible conception of democracy. The further questions are essentially empirical. In some times and places, under some political conditions, the facts will be such that the small-scale mechanism of absolute voting rules will enhance desirable forms of accountability overall. Such rules should have Institutional Design Writ Small

141

Chapter 

a place in the toolkit of institutional design. In particular domains, where the influence of narrow interest groups is greatly to be feared, and where legislators are often subject to the pressure, coercion, and bad accountability that Bentham feared, absolute majority rules will (I have argued) often prove superior to alternative tools.

142

Mechanisms of Democracy

Chapter 5

Delegation, Accountability and Judging

S

o far the focus has been on voting in legislatures, a natural starting point for democratic theory. But voting is a central feature of other institutions as well, particularly the judiciary, where multimember panels are the norm at the higher levels of the judicial system. In this chapter, I turn to the following question: In a democratic polity, how (if at all) can judges promote accountability and allied democratic virtues? In the introduction to Part II, we saw one answer: democracy requires, and judges should enforce, a “nondelegation doctrine” that prohibits “excessive” delegations to the executive and administrative agencies, thereby requiring that the legislature fully specify the goals and principles of legislative programs. On this view, the legislature is assumed to be the democratically accountable branch par excellence. I argued against that doctrine, suggesting that it lacks any connection to or basis in democratic accountability. On the competing view I have offered, two institutional structures enforce accountability in the modern administrative state. First, the legislature is in principle fully accountable for all lawmaking, including the very act of delegating by statute. Chapters 3 and 4 thus turned to

Institutional Design Writ Small

143

Chapter 

the task of creating democratic mechanisms to promote accountability in legislative lawmaking (including delegatory lawmaking). Second, the executive branch and regulatory agencies are accountable—in part by virtue of the executive’s electoral connection, where that exists, as in the American presidency—for the policies they choose to implement broad statutory delegations. What of the judges? In a world without a nondelegation doctrine, the role of judges is to enforce congressional instructions, including instructions to delegate. The principal rule the American legal system uses to implement this scheme is the so-called Chevron doctrine, after an allimportant Supreme Court decision in 1984.1 Under Chevron, judges are to enforce clear congressional instructions, but where statutes are unclear or silent, judges presume that Congress has delegated policymaking authority to the executive. The Chevron rule promotes accountability by ensuring that clear congressional instructions are honored, while also ensuring that, where Congress has not spoken clearly, the politically accountable executive rather than the judiciary chooses policies in the relevant domain. There are analogous decisions in many other polities—the most important being the famous Wednesbury case of 1948 in the United Kingdom.2 So far, so good—if Chevron works. If it does not work, in the sense that Chevron’s own goals of democratic accountability might be attained at lower cost through alternative means, then the alternatives dominate. This chapter argues that Chevron is a poor means for promoting accountability in the world without a nondelegation doctrine. Chevron fares poorly because it attempts to allocate policymaking authority through soft-edged legal doctrine, making it vulnerable to a range of problems: conceptual imprecision, cognitive burdens that afflict boundedly rational judges, and manipulation on the part of biased judges. The alternative suggested here is a democratic mechanism aimed at promoting accountability while sidestepping these problems. The suggestion is to implement Chevron through voting rules. A supermajority voting rule for multimember judicial panels could state that on the nine-member 1 2

Chevron U.S.A. v. Natural Res. Def. Council, 467 U.S. 837 (1984). Associated Provincial Picture Houses v. Wednesbury Corp. 1 KB 223 (1948), where the court stated that it would not intervene to correct a bad administrative decision on grounds of its unreasonableness unless the decision was “so outrageous in its defiance of logic or accepted moral standards that no sensible person who had applied his mind to the question to be decided could have arrived at it.”

144

Mechanisms of Democracy

Delegation, Accountability and Judging

Supreme Court, agencies will win against any claim that they have exceeded the bounds of their delegated authority, unless six Justices vote against the agency. The details can vary, and the proposal is fleshed out below. The basic idea, however, is to find voting mechanisms that implement Chevron’s own goals of accountability at lower cost. In so doing, the hope is that small-scale changes in the design of judicial voting rules can produce large improvements in the system of democratic lawmaking.

Background—and a Claim Of central importance to administrative law and theory, in any democratic polity, is the question whether, and when, courts will defer to agency interpretations of law. In Chevron v. Natural Resources Defense Council, the United States Supreme Court replaced earlier answers to that question in American law with a new framework: courts should defer to an agency interpretation unless the relevant statute is clear or the agency interpretation is unreasonable.3 In the past two decades, however, the Chevron framework has come under increasing strain. As a matter of legal doctrine, there are many ambiguities and uncertainties about the nature of the inquiry under Chevron, including questions about the admissibility and weight of various legal sources.4 In practice, it is unclear whether Chevron has increased overall deference to agencies, but recent evidence makes it clear that Chevron’s application does vary markedly with the ideological and political preferences of the judges who apply it.5 3 4 5

See Chevron, 467 U.S. at 865–66. For an overview of the doctrinal puzzles, see Stephen Breyer et al., Administrative Law and Regulatory Policy 319–413 (5th ed. 2002). Thomas J. Miles & Cass R. Sunstein, Do Judges Make Regulatory Policy?, 73 U. Chi. L. Rev. 823 (2006); Frank B. Cross & Emerson H. Tiller, Judicial Partisanship and Obedience to Legal Doctrine, 107 Yale L. J. 2155 (1998); Richard L. Revesz, Environmental Regulation, Ideology, and the D.C. Circuit, 83 Va. L. Rev. 1717 (1997). There is conflicting evidence about whether Chevron has increased deference to agencies. An early study found a significant effect. See Peter H. Schuck & E. Donald Elliott, To the Chevron Station: An Empirical Study of Federal Administrative Law, 1990 Duke L.J. 984, 1057–59 (1990). However, other studies either suggest either that Chevron has had little effect overall effect on deference, or that its effects are unclear. See Linda R. Cohen & Matthew L. Spitzer, Judicial Deference to Agency Action: A Rational Choice Theory and an Empirical Test, 69 S. Cal. L. Rev. 431, 474–75 (1996) (arguing “[o]ur tests show that the Court does not uniformly endorse judicial deference, but rather does so discriminately in the years where the doctrine yields policy outcomes more to the Court’s liking”); Thomas W. Merrill, Textualism and the Future of the Chevron Doctrine, 72 Wash. U. L.Q. 351, 373 (1994) (asserting that Chevron had a limited impact on the rate of affirmance of agencies’ interpretations in the Supreme Court).

Institutional Design Writ Small

145

Chapter 

These problems arise, in part, from a dubious premise of the Chevron enterprise, one that should be rethought. The dubious premise is that the legal system should adopt a doctrinal solution—the Chevron rule—for what is, after all, an institutional problem: the allocation of interpretive authority between agencies and courts when congressional instructions are silent or ambiguous. I suggest an alternative, which is to adopt an institutional solution to the institutional problem. The institutional solution is to cast Chevron as a voting rule, thereby institutionalizing deference to administrative agencies. Where a litigant challenges agency action as inconsistent with an organic statute, the agency would prevail unless the judges, asking simply what the best interpretation of the statute is, vote to overturn the agency by supermajority vote—say, by a six-three vote on the Supreme Court, or by a three-zero vote on a court of appeals panel. A voting rule of this sort would capture the benefits of the doctrinal version of Chevron while generating fewer costs. In the doctrinal version, judges must develop and internalize a legal distinction between the best interpretation of the statute and a reasonable interpretation of the statute. For conceptual, psychological and motivational reasons, this distinction is tenuous, even unstable. This should be unsurprising; the doctrinal solution requires judges to internalize a legal norm of deference, but it is accompanied by none of the traditional mechanisms law uses to force decisionmakers to internalize the consequences of their choices. Conversely, the principal advantage of institutionalizing Chevron as a voting rule is that it makes agency deference an aggregate property that arises from the whole set of votes, rather than an internal component of the decision-rules used by individual judges. Casting Chevron as a voting rule has other benefits as well. A voting-rule version of Chevron would allow more precise calibration of the level of judicial deference over time; and holding the level of deference constant, a voting rule of agency deference would produce less variance in deference across courts and over time, yielding a lower level of legal uncertainty than does the doctrinal version of Chevron.

Legal Doctrine and Institutional Problems In many domains, legal doctrine is developed by judges, lawyers and commentators with a view to solving an institutional problem—for example, the allocation of power across different institutions or among different officials within the same institution. Thus the presumption of 146

Mechanisms of Democracy

Delegation, Accountability and Judging

constitutionality for legislation allocates a measure of interpretive authority over the Constitution to legislatures;6 the “clear error” standard of review allocates fact-finding competence to trial courts; and the legal norm of precedent or stare decisis allocates decisionmaking authority from present judges to past judges, whose views will control the judgment of the present on some questions. Lawyers, however, frequently overlook the possibility that legal doctrine is rarely the only type of solution to an institutional problem, and may not always be the best solution. An alternative is to change the rules that govern the composition, powers, or voting rules of the relevant institutions. We may call these “hard” solutions, in contrast to “soft” doctrinal solutions. The relative costs and benefits of soft solutions, on the one hand, and hard solutions, on the other, vary across contexts and over time. Of course hard solutions are not always superior. Yet consider some examples of legal problems where there is an important choice between soft and hard solutions, especially changes to relevant voting rules, and where the hard solutions are at least plausible alternatives.

Deference to Legislatures In American constitutional law the so-called “presumption of constitutionality” for legislation, according to which courts should uphold legislation if there is a reasonable argument for constitutionality, was advocated by James Bradley Thayer7 and many later judges and commentators. The strength of the presumption, however, has waxed and waned over the course of American constitutional history. Today, many believe that the presumption has withered away, at least in certain contexts. A hard alternative periodically surfaces in the form of proposals that the Supreme Court, and perhaps lower courts, should (be required to) use a supermajority rule to invalidate statutes on constitutional grounds.8 A voting rule of this sort builds Thayerian deference into judicial decisionmaking at the aggregate level of the whole court, rather than urging individual Justices or judges to internalize deference as a legal norm. 6 7 8

See United States v. Morrison, 529 U.S. 598, 607 (2000). James Bradley Thayer, The Origin and Scope of the American Doctrine of Constitutional Law, 7 Harv. L. Rev. 129 (1893) Evan H. Caminker, Thayerian Deference to Congress and Supreme Court Supermajority Rule, 78 Ind. L. J. 73 (2003); Jed Handelsman Shugerman, A Six-Three Rule: Reviving Consensus and Deference on the Supreme Court, 37 Ga. L. Rev. 893 (2003).

Institutional Design Writ Small

147

Chapter 

Precedent Just as judges might defer to legislatures where there is reasonable doubt about the legal issues, so too judges might and do defer to past judges, where there is reasonable doubt about the legal issues. The doctrine of stare decisis has many formulations and complexities, but a simple version requires judges to follow horizontal precedent—the previous decisions of the same court—unless the precedent is “clearly erroneous.” This is a soft solution; a hard alternative would be to say that the precedent decision must be followed unless overruled by a supermajority vote or even a unanimous vote of the later court.

The Rule of Four The previous examples involved legal norms that have been embodied in doctrine, but that might also be embodied in voting rules, with a different set of costs and benefits. The converse example is a voting rule that might be recast as a legal doctrine. Consider the Rule of Four, according to which the votes of any four of the nine Justices of the Supreme Court are sufficient to grant certiorari for a full hearing on the merits of a case.9 A soft analogue of the Rule of Four would be an ordinary majority vote on the decision to grant certiorari, accompanied by an internalized legal norm that Justices would follow in casting their individual votes. The content of the internalized norm might be something along the following lines: Justices should vote for certiorari if a reasonable Justice could believe that the petition warrants a full hearing. Whatever its precise content, the internalized norm might yield roughly the same number of certiorari grants, all else equal, as a nonmajority voting rule under which each Justice directly asks whether, in his or her judgment, the Court’s criteria for granting certiorari are satisfied.

Appointments and Senatorial Deference So far the examples have focused on courts, but the distinction between hard and soft rules is relevant to other institutional contexts as well. Doctrinal norms exist outside of courts; many institutions have moreor-less explicit or articulate systems of precedent and more-or-less canonical verbal formulations of norms that are embodied in the institution’s past decisions. Consider the view, advanced by many Senators of both 9

See generally Richard L. Revesz & Parmela S. Karlan, Nonmajority Rules and the Supreme Court, 136 U. Pa. L. Rev. 1067 (1988).

148

Mechanisms of Democracy

Delegation, Accountability and Judging

parties at many times, that Senators should give some degree of deference to the President’s appointments, especially for executive offices but for judges as well. The level of deference is hard to capture in any single verbal formula, but many Senators say something to the effect that they will defer unless a nominee is “clearly” unsuitable. More recently, a decisive fraction of the Senate—the Gang of Fourteen—agreed to defeat filibusters unless “extraordinary circumstances” exist.10 One might imagine a system of presidential appointments that embodied these ideas in a hard voting rule, something like a reverse filibuster: unless a supermajority votes to defeat a nominee, the appointment will be deemed to have been confirmed. It is hardly clear that such a rule would be constitutional, because the best reading of relevant constitutional provisions might be that an affirmative majority vote is necessary to approve a nominee (whether or not it is sufficient to do so). For present purposes, it is irrelevant whether the hard solution would indeed be constitutionally permissible; as a conceptual matter it is an entirely viable alternative to the soft norm of deference to the President in appointments matters. These examples could easily be multiplied. The central point is that the bare specification of a problem in the legal system does not entail that the solution should be doctrinal, rather than institutional in a hard sense. Lawyers frequently overlook this choice among solutions, perhaps because their training focuses on legal doctrine and because some hard solutions can be supplied only by nonjudicial institutions—as when a statute or constitutional amendment is necessary to change voting rules. As a general matter, hard solutions are not the dominant solution, but they are not rare either.

General Tradeoffs With these examples in hand, we can sketch the general tradeoffs that determine the choice between soft and hard solutions—at least by indicating some frequently encountered variables, together with a rough estimate of their signs. Of course the magnitudes of the relevant variables will differ greatly across settings, and in many settings there will be special considerations 10

Charles Babington & Susan Schmidt, Filibuster Deal Puts Democrats in a Bind; Pact May Hinder Efforts to Block High Court Nominee, Wash. Post, July 4, 2005, at A1; Charles Babington & Shailagh Murray, A Last-Minute Deal on Judicial Nominees, Wash. Post, May 24, 2005, at A1.

Institutional Design Writ Small

149

Chapter 

that do not generalize. These qualifications notwithstanding, there are regularities that make some tentative generalizations possible.

Aggregate Norms versus Individual Norms Consider the problem of partial deference—the recurring situation in which it would be good or right for institution A to defer to institution B, but only if institution B’s decision is not clearly wrong or unreasonable. Institution A might be a court, and institution B an agency (or a legislature); A might be a higher court, and B a lower court, and so on. Let us bracket, for now, the question why deference would be good or right in this situation; the focus is on the choice between means for attaining a posited goal, not on the theory that makes the goal desirable. Let us also assume that institution A is a multimember decisionmaking body, such as an appellate court. Single-member bodies like district courts present distinct problems; in any event, single-member bodies rarely make final decisions in modern legal systems. A problem that arises in many situations of partial deference is that the triggering conditions for deference are vague or imprecise. What is “clear” or “unreasonable” to one judge is not “clear” or “unreasonable” to another. Conceptually, it is not clear what “clear” means. Psychologically, it is difficult for judges and other decisionmakers to avoid collapsing their conception of a “reasonable” legal answer into their conception of the best legal answer, thereby defining all second-best answers as unreasonable. Motivationally, deference rules based on vague triggering conditions allow scope for ideological and even partisan biases. I will return to these points below, with specific illustrations from the law of the administrative state. The basic problem in such cases is that individual decisionmakers are charged with internalizing a legal norm of deference that is conceptually ill-defined and that cuts against both their individual judgments of what is best and their biases and prejudices. And this duty of internalization is not aided by any of the usual mechanisms by which law encourages or forces actors to internalize legal rules, principally material rewards or penalties. The rewards and sanctions that affect judicial behavior are weak, secondorder forces like professional reputation.11 Of course legal doctrine requires 11

150

See “Lawrence Baum, Judges and Their Audiences” (2006); Frederick Schauer, Incentives, Reputation, and the Inglorious Determinants of Judicial Behavior, 68 U. Cin. L. Rev. 615 (2000); Richard A. Posner, What Do Judges and Justices Maximize? (The Same Thing Everybody Else Does), 3 Sup. Ct. Econ. Rev. 1, 5–6 (1993).

Mechanisms of Democracy

Delegation, Accountability and Judging

judges to internalize norms in many settings; but the burden of internalization is especially heavy in those cases in which judges are required to make second-order judgments about deference, as illustrated below. A shift from a soft legal norm to a hard institutional solution would solve these problems. Imagine a supermajority voting rule under which institution A defers to institution B unless two-thirds of the members of A believe that B is wrong on the merits. Each decisionmaker asks simply what legal answer is best and votes accordingly. Deference is an emergent property of the aggregate vote, rather than of individual decisions. Conceptually, there is no need for decisionmakers to develop a theory about what counts as a “clear” (as opposed to merely correct or incorrect) legal answer. Psychologically, there is no requirement that the decisionmaker simultaneously hold in her mind two conflicting legal standards, so the cognitive load is greatly reduced. Motivationally, each judge may be biased or prejudiced in some sense, yet assuming some diversity of preferences, the biases will be washed out at the aggregate level, with deference enforced by the voting rule. Moreover, it is easier for voters or other principals to monitor the behavior of their agents, the decisionmakers in institution A, because those agents have less room to maneuver in a supermajority voting rule system. Rather than concealing their biased votes by claiming that institution B’s decision is or is not clearly wrong or unreasonable—claims whose second-order character makes them inherently costly for outside observers to evaluate—the decisionmaker must now simply state his or her understanding of the correct legal answer. There is one less degree of freedom for bias to operate. These points abstract from the level of deference that actually results from either the soft or hard solution. It is an empirical question whether or not a six-three supermajority rule, on a nine-member court, produces more, less, or the same level of deference as an internalized legal norm of deference. Holding the level of deference constant, the suggestion is that a shift from deference as an internalized norm to deference as an aggregate property can produce a given level of deference at lower total cost to decisionmakers themselves and to other actors in the system.

Calibration versus Fuzziness Suppose, in operation, that an internalized norm of deference produces too much or too little deference from institution A to institution B, with “too much” or “too little” defined by some extrinsic theory. How can the Institutional Design Writ Small

151

Chapter 

level of deference be adjusted up or down? If a soft solution is in place, adjustment is difficult and imprecise. The problem is that verbal formulae are typically too crude to capture the fine shades of difference that are needed to tweak deference rules in either direction. As Judge Richard Posner suggests, “the cognitive limitations that judges share with mere mortals may constitute an insuperable obstacle to making distinctions any finer than that of plenary versus deferential review.”12 Legal language can capture the idea of the best legal answer, and can indicate the idea of deference where decisions under review are not clearly erroneous. More finegrained standards of deference, however, are difficult to express; cognitive capacities are typically inadequate to sustain a fine-grained schema of deference standards.13 Law is replete with arguments about the precise meaning of different standards of review, such as the idea that judges should uphold factual findings of trial courts unless there has been “clear error,” the requirement that agencies show “substantial evidence,” for their decisions, and so on. Such arguments are interminable and, for the most part, unilluminating. These attempts at verbal distinction often, perhaps always, collapse in practice.14 A voting rule can be more precisely calibrated. If a six-three supermajority rule produces too little deference, a seven-two rule can be substituted. If a submajority rule such as the Rule of Four produces too few grants of certiorari, a Rule of Three might be adopted instead. The calibration will still be imperfect, because the voting rules are still slightly lumpy; perhaps the optimal level of deference, according to the extrinsic theory, lies just in between the level produced by a six-three rule and that produced by a seven-two rule. Comparatively, however, adjustments in voting rules will be less lumpy, and more fine-grained, than slight manipulations in the wording of judicial doctrine.

Certainty versus Variance A corollary of the last point is that hard solutions increase legal certainty. Suppose that a particular doctrinal formulation produces a particular level of deference, D. This level, D, is a kind of expected value, with variance around that value. Some courts applying the doctrine will exceed D; some 12 13 14

152

School District of Wis. Dells v. Littlegorge, 295 F.3d 671 (7th Cir. 2002). United States v. McKinney, 919 F.2d 405, 423 (7th Cir. 1990) (Posner, J., concurring). See Paul R. Verkuil, An Outcomes Analysis of Scope of Review Standards, 44 William & Mary L. Rev. 679 (2002).

Mechanisms of Democracy

Delegation, Accountability and Judging

will fall short; and some courts will exceed D in one period while falling short in another period. A voting rule will produce greater certainty about deference, holding the expected level of deference constant. To the extent that reducing variance is a benefit for actors in the legal system, a voting rule may well be preferable to a doctrinal solution. In part, this is a result of the calibration point. If the category of deference is broad, individual judges will likely select different deference points within the broad category. Variation in individual-level deference translates into confusion about the overall level of deference. By making deference a characteristic of the aggregation mechanism, the voting rule removes this source of uncertainty. Beyond the calibration point, so long as a voting rule that builds in deference is more difficult for biased judges to thwart, there will be greater certainty that deference will be applied. For reasons relating to clarity, predictability, and stability, the voting rule generates greater certainty.

Chevron: Democracy, Accountability and Expertise Let us now turn from the abstract to the concrete. Chevron sets out a framework for judicial review of agencies’ statutory interpretations. The framework applies when parties claim that agencies have overstepped the bounds of their delegated statutory authority. At Step One of Chevron, judges ask whether the statute speaks to the “precise question at issue”; if so, then the judges simply enforce its commands. If the statute contains a gap—if, in other words, it is silent or ambiguous on the relevant question—then judges are to proceed to Step Two, at which they ask whether the agency interpretation of the statute is “reasonable,” or, in other words, whether the agency interpretation falls within the scope of the statute’s ambiguity.15 Chevron’s theoretical rationale is contested, although there are important areas of agreement. In the subsequent Mead decision,16 the Court suggested that Chevron rests on Congress’s implicit delegation of law-interpreting authority to agencies. In this view, Chevron’s global interpretive default rule—namely, that statutory silence or ambiguity confers law-interpreting authority on agencies—derives from an implicit general 15 16

See Chevron, 467 U.S. at 845. 533 U.S. 218 (2001).

Institutional Design Writ Small

153

Chapter 

instruction by Congress. As both judges and commentators have noted, however, this rationale is a fiction.17 The Administrative Procedure Act (APA), which is the nearest Congress has come to providing a general instruction on the allocation of law-interpreting authority, says that courts are to decide all relevant questions of law.18 Some scholars argue that deference to agencies is itself among the legal rules that courts are to apply.19 Even if true, that argument still provides no affirmative evidence whatsoever that Congress intended the APA to be a general meta-instruction on the delegation of law-interpreting authority. The implied-delegation rationale for Chevron risks treating Congress as a ventriloquist’s dummy, into whose mouth may be inserted whatever fictional legal meta-instructions are necessary to square agency deference with the conviction that courts must say what the law is. The Chevron opinion itself did not adopt this approach. In a crucial passage, the Court explicitly rejected the idea that statutory gaps necessarily represent a congressional meta-instruction to delegate law-interpreting authority to the agency: Congress intended to accommodate both [economic and environmental interests], but did not do so itself on the level of specificity presented by these cases. Perhaps that body consciously desired the Administrator to strike the balance at this level, thinking that those with great expertise and charged with responsibility for administering the provision would be in a better position to do so; perhaps it simply did not consider the question at this level; and perhaps Congress was unable to forge a coalition on either side of the question, and those on each side decided to take their chances with the scheme devised by the agency. For judicial purposes, it matters not which of these things occurred.20

17 18 19 20

154

See Antonin Scalia, Judicial Deference to Administrative Interpretations of Law, 1989 Duke L.J. 511, 517 (1989). 5 U.S.C. § 706 (2000) (“To the extent necessary to decision and when presented, the reviewing court shall decide all relevant questions of law. . . .”). See Henry P. Monaghan, Marbury and the Administrative State, 83 Colum. L. Rev. 1, 25–28 (1983). Chevron, 467 U.S. at 865 (emphasis added).

Mechanisms of Democracy

Delegation, Accountability and Judging

On this picture, a congressional meta-instruction might exist (as in the first scenario the Court gives), but it might not (as in the second or third scenarios), and the issue whether it does exist lacks the paramount importance that the Mead decision tried to give it. The real basis for agency deference, according to the Chevron opinion itself, was not an implicit congressional meta-instruction, but a candid recognition, by judges, of the virtues of democratic accountability and of the limits of their own institutional capacities. “Judges are not experts in the field, and are not part of either political branch of government.”21 Overall there is broad agreement about this fundamental accountability-promoting goal of Chevron. Those who see Chevron as resting on a higher-order implicit delegation from Congress think that the best reconstruction of Congress’s intentions is a default rule favoring agency authority, precisely because of agencies’ democratic virtues and institutional capacities. Agencies’ superior expertise and democratic accountability better position them to fill statutory gaps. The democratic accountability of agencies arises, in part, by virtue of their connection to the President, who is typically elected by a national majority or plurality (ignoring the small chance that the Electoral College will make any difference to the outcome, a point discussed at the beginning of Part II). Of course some agencies, especially independent agencies, have a great deal of autonomy from the President, and also from Congress as well. And of course Congress too is democratically accountable. But a comparison between the President and Congress, or between presidential and congressional control, is never the relevant comparison for Chevron purposes. If Congress has spoken clearly in statutes, that is the end of the matter, under the first of the two Chevron rules. Chevron bites only where Congress’s instructions are unclear, in which case the only choices on offer are that agencies or courts should fill in the statutory gaps. The relevant comparison is not between the President and Congress, but between the President and the judges. In that comparison, even if the President’s democratic credentials are imperfect, there can be no doubt that they are superior to those of the judiciary. Of course Chevron also rests on a vision of administrative expertise; how does this relate to democratic accountability? It is sometimes said that these two components of Chevron’s rationale are in tension with each other, 21

Chevron, 467 U.S. at 865.

Institutional Design Writ Small

155

Chapter 

because political accountability distorts expertise. But this too overlooks that Chevron’s claim is a strictly comparative one about the relative institutional capacities of agencies and courts. By analogy, even if speed and power trade off against one another at the outer margins of athletic performance, still one baseball player might be both speedier and more powerful than another. So too, even if there is a marginal tradeoff between expertise and political accountability, agencies might be both systematically more expert and systematically more accountable than are generalist judges.

Permissible Interpretations and Best Readings For present purposes, the most important feature of Chevron is that—in the doctrinal version—it requires individual judges to develop a theory of reasonable or permissible statutory interpretation, distinct from the judge’s theory of what interpretation is best. Under pre-Chevron doctrine, many decisions suggested that the judicial task is to identify the statute’s “true meaning” or “best reading.”22 This is the standard conception of the judicial task where no agency is in the picture, or where the relevant agency does not receive deference at all—as in criminal cases, in which the nominal rule is that judges do not defer to prosecutors’ legal interpretations. To be sure, judges seeking the best legal answer might take agency interpretations or prosecutorial interpretations as persuasive guidance, depending upon the agency’s accumulated experience, “the thoroughness evident in its consideration, the validity of its reasoning, its consistency with earlier and later pronouncements, and all those factors which give it power to persuade, if lacking power to control.”23 Yet in principle at least, these are just helpful pointers to the best interpretation. Chevron departs from this baseline in a crucial respect. Under the doctrinal version of Chevron, the judge must be able to say or think, in some set of cases, that the agency’s interpretation of a statute is “reasonable” or “permissible,” even though it is not the one the judge herself would deem best, were law-interpreting authority to be exercised by the court de novo. The judges must, in effect, add to her first-order theory of statutory interpretation a second-order theory that identifies some first-order interpretations as reasonable, whether or not correct. 22 23

156

See, e.g., NLRB v. Hearst Publications, 322 U.S. 111 (1944). Skidmore v. Swift & Co., 323 U.S. 134 (1944).

Mechanisms of Democracy

Delegation, Accountability and Judging

The point of this innovation is to open up space for discretionary policy judgments by democratically accountable agencies. Chevron doctrine does this by distinguishing between the judges’ views about the statute’s best reading, on the one hand, and a range of permissible agency interpretations, on the other. Instead of an interpretation that is a “point estimate,” Chevron aims to open up a “policy space” that gives agencies breathing room to pursue policies based on technocratic judgments or democratic politics.24 I take this goal as given, putting aside—for the reasons given in Chapter 2—root-and-branch criticisms of Chevron based on the nondelegation doctrine. Still, Chevron’s second-order approach to interpretation has proven problematic on conceptual, psychological and motivational grounds, as I will detail below. There is a better way to achieve Chevron’s goals. Democratically accountable agencies should be given breathing room by means of voting rules rather than legal doctrine.

Deference Regimes So far we have examined two different distinctions: de novo judicial interpretation, on the one hand, versus Chevron doctrine on the other; majority rule, on the one hand, versus a supermajority rule, on the other. Correlating these two distinctions, four deference regimes are possible.

Regime (1): De novo Judicial Interpretation with a Majority Rule This is a regime of no deference at all. Before Chevron, many cases suggested that this was the law, although there were contrary decisions as well.25 Starting from this baseline, there are two ways to push the law towards greater deference: by adopting a legal doctrine of deference, and by adopting a voting rule weighted in the agency’s favor. Let us take up these possibilities in turn. 24

25

E. Donald Elliott, Chevron Matters: How the Chevron Doctrine Redefined the Roles of Congress, Courts, and Agencies in Environmental Law, 16 Vill. J. Envtl. L. 1, 11–12 (2005); Cass R. Sunstein, Beyond Marbury: The President’s Power to Interpret the Law, Yale L.J. (forthcoming 2006). Compare NLRB v. Hearst Publications, 322 U.S. 111 (1944) (suggesting de novo judicial interpretation of “pure” questions of law), with Rochester Tel. Corp. v. United States, 307 U. S. 125, 146 (1939), and Gray v. Powell, 314 U.S. 402, 411 (1941) (suggesting judicial deference to agency interpretations).

Institutional Design Writ Small

157

Chapter 

Regime (2): Chevron with a Majority Rule This is the solution the Court adopted in 1984, and it is the current law. In the terms used here, it is a soft rather than a hard solution. I suggest below that this solution produces a range of costs and problems that might be avoided by adopting a hard solution instead. Of course, if (1) and (2) were the only options, (2) might well be thought preferable. But a different regime is better still.

Regime (3): De novo Judicial Interpretation with a Supermajority Rule (Weighted in Favor of Agency Interpretations) This is the regime I will denote by the label “Chevron as a voting rule.” I will suggest that it is superior to (2), because it captures the benefits of (2) at lower cost. Starting from the current regime—Chevron combined with majority rule—a move to Chevron as a voting rule requires two legal changes, not just one. Both the legal standard and the voting rule are changed, the former by abolishing Chevron doctrine in favor of de novo judicial interpretation, the latter by adopting a supermajority rule.

Regime (4): Chevron with a Supermajority Rule In principle, there is no reason to think that soft and hard solutions are mutually exclusive; the two might be combined so as to capture the distinctive benefits of both regime (2) and regime (3). In practice, however, this regime should suffer from the same problems as regime (2), problems detailed in full below. The defects of regime (2) arise from the two-level structure of Chevron reasoning and the conceptual, psychological, and motivational burdens that structure imposes upon judges; merely grafting a supermajority rule on to Chevron leaves the two-level structure in place, and thus fails to get at the root of the problem. Moreover, the pure cases— regimes (2) and (3)—illustrate the relevant considerations more cleanly. Subject to that caveat, Chevron with a supermajority rule could provide some of the benefits of a hard voting rule, albeit without avoiding the costs of a soft doctrinal approach. Against this background, I will compare regimes (2) and (3). The basic argument is that regime (3) captures the benefits of regime (2) at lower cost.

158

Mechanisms of Democracy

Delegation, Accountability and Judging

Aggregation: Solving Chevron Problems Chevron doctrine requires judges to distinguish between first-order interpretation, namely, finding the best reading of the statute, and secondorder interpretation, which supposes a theory of permissible or reasonable interpretation. In operation, this requirement produces a range of problems, many of which can be avoided or ameliorated by adopting a votingrule solution. For simplicity, I assume throughout that the relevant case is binary—an agency offers one interpretation of the statute, a challenger offers another, and the judicial task is to choose between them. Some cases are not like this, but most cases are, and the costs of the extension would outweigh its marginal intellectual benefits.

Conceptual Problems What exactly does it mean to say that an agency’s interpretation, although not best by the judge’s lights, is nonetheless reasonable or permissible? What does it mean to say that a statute “clearly” means X, as opposed to saying that the statute is “best read” to mean X? The answers to these questions are themselves hardly clear. Under Chevron doctrine, the hardest question is where, even in principle, the bounds of permissible interpretation should be taken to lie. Consider the following construal, which is quite artificial but useful for expository purposes. Under standard first-order interpretation, the judge (let us suppose) is considering two different readings of the statute, X and Y. Consulting all legal materials that are relevant under the judge’s first-order interpretive theory (text, perhaps legislative history, perhaps various interpretive default rules, and so on) the judge decides that reading X is, all things considered, somewhat superior to Y—that X is 65% likely to be correct, while Y is only 35% likely to be correct. Under de novo interpretation, the judge votes in favor of X. Under Chevron, however, another layer of decisionmaking is required. The judge must ask whether X, at the 65% level, is “clearly” correct as opposed to simply better; alternatively, the judge could ask whether Y, at the 35% level, is reasonable or permissible. The example suggests that first-order interpretation is strictly comparative—the judge simply decides which interpretation is better—while a theory of permissible interpretation must build in an absolute threshold

Institutional Design Writ Small

159

Chapter 

above which interpretations may, but need not, be adopted by agencies. The problem is that nothing in Chevron tells judges, even in principle, where the threshold should be located, and the metric for setting the threshold is obscure, even in principle. Perhaps an interpretation that is plausible at the 35% level suffices; perhaps it does not. Perhaps an interpretation that reaches the 65% level is “clearly” correct, or perhaps not. And in both these cases, it is hardly clear what or where the relevant “level” is. With judges of reasonably diverse preferences and psychologies, thresholds may vary widely, and it will not be possible to say that any threshold is conceptually preferable to any other. The switch to regime (3), Chevron as a voting rule, avoids the problem of an absolute threshold altogether. Each judge now asks, simply in comparative terms, what the better reading of the statute is. The purpose of Chevron’s second-order approach to interpretation—providing space for agency policy judgments—is still fulfilled, just at the aggregate level of the whole court rather than at the level of individual judicial judgments. Assuming reasonable and predictable diversity of first-order interpretations across judges on multimember courts, breathing room for agency policy judgments will arise from the operation of the voting rule itself. In the example above, some judges will decide that the agency’s interpretation Y reaches only the 35% level, and will vote to overturn its interpretation (no consideration of thresholds is here required). Other judges will decide, by contrast, that the agency’s interpretation reaches the 65% level of likelihood. Unless a supermajority of the judges adopts the former view, the agency will prevail. Across cases, as agency interpretations become less and less plausible, it is more and more likely that a supermajority will be found to overturn them. Crucially, however, none of the judges in any case need wrestle with the conceptually obscure problem of where to locate the threshold of permissible interpretation. A nuance in this picture is worth emphasizing: to say that under de novo judicial interpretation judges search for the “best” reading is imprecise, although useful as a shorthand. Where the case is binary, as I assume throughout, judges need only decide which of the two readings offered by the parties is the better one. This comparative judgment is, plausibly, much easier than the judgment that Chevron requires about the location of the reasonableness threshold. The latter judgment is an absolute one, with no clear metric against which to make the judgment in the first place. 160

Mechanisms of Democracy

Delegation, Accountability and Judging

It is familiar that absolute judgments are often more difficult than comparative ones. The question “Was the Duke of Wellington tall?” is harder than the question “Was the Duke of Wellington taller than Napoleon?” People can get confused when they make implicitly categorybound judgments, as when they say that an eagle is “big” while a cabin is “small”; but presumably few people would say that an eagle is bigger than a cabin.26 Voters who know little about candidates’ views or positions, in absolute terms, are very good at knowing which candidate is farther to the left or right, in relative terms.27 Assessing the plausibility of an agency’s interpretation has similar features. Of course, there may well be settings in which a judgment about plausibility is easier or less psychologically costly than a judgment about optimality. The suggestion here is mainly local rather than global. Under Chevron doctrine, a judge consults the full panoply of sources to derive statutory meaning at Step One. Although it is possible to imagine a parallel world in which administrative law judges quickly glance at the statute to make sure an agency’s interpretation is not implausible, that is not our world; currently, Chevron doctrine requires a far more elaborate inquiry. These points bracket any questions about outcomes—about the rate at which agency interpretations are upheld. It is possible, in particular cases, that a shift from doctrinal Chevron to Chevron as a voting rule will result in invalidations of agency interpretations that would not have occurred under doctrinal Chevron. Consider a case in which (1) a supermajority of judges believe that the agency’s interpretation is reasonable (or, equivalently, that the statute does not clearly rule it out) but also believe that (2) the agency’s interpretation is not the better reading. Under doctrinal Chevron the agency would win if judges have internalized the doctrinal norm, because a (super)majority would vote in the agency’s favor, whereas under Chevron as a voting rule the agency will lose, because a supermajority will vote against it. This is merely one possible case, however. Consider the opposite possibility: any case in which a bare majority of the multimember court 26 27

See Cass R. Sunstein et al., Predictably Incoherent Judgments, 54 Stan. L. Rev. 1153, 1171–77 (2002). See, e.g., Stuart Elaine Macdonald, George Rabinowitz & Ola Listhaug, Political Sophistication and Models of Issue Voting, 25 B.J. Pol. S. 453, 458 (1995) (“It is easier to know the side a party takes on an issue than its particular policy proposals, and it is especially easy to know the side a party takes when it is intense on an issue.”).

Institutional Design Writ Small

161

Chapter 

believes that the statute clearly rules out the agency’s interpretation, or equivalently that the agency’s interpretation is impermissible. In such cases, the agency would lose under doctrinal Chevron but will win under Chevron as a voting rule. It is unclear, before the fact, which type of case is more frequent, and thus unclear what the outcome effects of the change in regime would be. Moreover, if Chevron as a voting rule produces too little deference, according to some extrinsic theory, the supermajority requirements can be calibrated upwards. A requirement of unanimity on the Supreme Court to overturn agency decisions would either produce massively increased deference in decided cases, or would produce a selection effect such that litigants would challenge only the most egregious agency decisions. It follows that a shift to Chevron as a voting rule might change the mix of cases in which deference occurs, even if the level or frequency of deference is held constant. Low-intensity cases—in which a large supermajority of judges think that the case is close, but that the agency’s reading is worse—will now be decided against the agency; high-intensity cases— in which the judges are polarized into a bare majority and a minority with sharply opposed views—will be decided in the agency’s favor. Intuitively, this switch seems perfectly consistent with the rationales for Chevron. If the goal is to provide breathing space to agencies to make technocratic or democratic judgments in contested domains of law and policy, then this change in the incidence of deference is either an improvement or neutral. Whether the justification for deference hinges on agency expertise or democratic pedigree, Chevron as a voting rule performs equally well. Rather than individual judges setting aside their own views in favor of more expert or more democratic judgments, the voting mechanism enforces respect for agency views at the aggregate level.

Psychological Burdens In an influential early discussion of Chevron and agency deference, Justice Stephen Breyer of the United States Supreme Court touched on the problems inherent in judging under a second-order theory of permissible interpretation: A . . . reason why neither a strict view of Chevron nor any other strictly defined verbal review formula requiring deference to an agency’s interpretation of law can prove successful 162

Mechanisms of Democracy

Delegation, Accountability and Judging

in the long run, is that such a formula asks judges to develop a cast of mind that often is psychologically difficult to maintain. It is difficult, after having examined a legal question in depth with the object of deciding it correctly, to believe both that the agency’s interpretation is legally wrong, and that its interpretation is reasonable. More often one concludes that there is a “better” view of the statute for example, and that the “better” view is “correct,” and the alternative view is “erroneous.”28 The post-history of Chevron provides some evidence for this view. Consider the decision in MCI v. AT&T.29 The Supreme Court, in an opinion written by Justice Antonin Scalia, concluded that an agency rule exempting nondominant long-distance telecommunications carriers from a regulatory requirement exceeded the agency’s statutory authority. The statute gave the agency authority to “modify” the requirement. In part, Justice Scalia argued that the plain meaning of “modify” encompassed only small changes, not large changes, pointing to dictionaries supporting his reading and discounting a prominent dictionary that said the contrary. In cases like MCI, it looks as though the distinction between reasonable views and correct views has collapsed altogether. If Justice Breyer is right that there is a general tendency of this sort, then the requirements of Chevron doctrine are unstable, because second-order interpretation is psychologically too demanding for judges. Perhaps Justice Breyer’s claim proves too much. In many areas of law, judges are asked to distinguish between their first-order judgments about what is correct, and their second-order judgments about what is permissible or reasonable. But the fact that judges make similar distinctions in other areas of law does not show that they do so successfully, or that Breyer is incorrect about Chevron. It is certainly imaginable that, in those other areas, the distinction tends to collapse as well, just as it does under Chevron (on Breyer’s view). The same social-scientific tools that have usefully exposed decisionmaking distortions under Chevron might well be applied in other areas, with similar debunking effect. That judges think the distinction between correctness and reasonableness works in 28 29

Stephen G. Breyer, Judicial Review of Questions of Law and Policy, 38 Admin. L. Rev. 363, 379 (1986). 512 U.S. 218 (1994).

Institutional Design Writ Small

163

Chapter 

some area of law is, of course, neither here nor there. Breyer’s critique itself suggests that judges will think they are faithfully distinguishing the correct from the reasonable, while in fact self-serving bias, motivated reasoning, and other mechanisms cause them to think that the only reasonable view is the one they happen to find correct. Absent some evidence about whether and when judges make such distinctions successfully, the objection that “judges do it all the time” assumes away Breyer’s argument rather than undermining it. Perhaps judges do not really do it all the time, at least not successfully; perhaps they do it successfully in some areas and not others. Suppose Justice Breyer is right, about Chevron at least. Then it is straightforward that the shift to Chevron as a voting rule will eliminate the psychological burdens of second-order interpretation, removing a major source of instability in the law of agency deference. The distinction between agency interpretations that are correct and agency interpretations that are reasonable will still exist. But the distinction need not be internalized by individual judges. Instead, it will arise from the operation of the voting rule itself. Agency interpretations that are reasonable will be more likely to attract the votes necessary to block formation of a contrary supermajority.

Bias Justice Breyer’s psychological conjecture about Chevron is plausible, but unproven. By contrast, there is ample evidence that a related sort of slippage occurs under doctrinal Chevron: judges tend, systematically, to uphold agency interpretations that accord with their political preferences and to invalidate agency interpretations that do not. On both the Supreme Court and the lower federal courts, judges are more likely to uphold liberal agency action if they are liberal, and vice-versa if they are conservative.30 The difference between this point and Justice Breyer’s claim is clear, at least in principle. The problem here is not that judges’ first-order legal views and their second-order legal views collapse into one another. It is that their views are not being driven by legal materials at all, or at least not wholly. To political scientists, the so-called attitudinal model of judging, under which judges vote their attitudes or policy preferences rather than their legal views, describes judicial behavior in many settings, not just the 30

164

Miles & Sunstein, supra note 5.

Mechanisms of Democracy

Delegation, Accountability and Judging

law of agency deference.31 The Chevron doctrine is not the cause of biased judging. Yet a plausible conjecture is that Chevron doctrine provides greater scope for the operation of biased judging than would Chevron as a voting rule. Under the former regime, judges acting in bad faith or in the grip of bias have, in effect, two margins on which to advance their agendas: the first-order interpretive question and the second-order interpretive question. A judge who wishes, for quite extrinsic reasons, to uphold the agency interpretation may claim either that the agency interpretation is correct or that it is reasonable, whether or not correct. Under Chevron as a voting rule, by contrast, the judge who wishes to uphold the agency interpretation must argue that the agency’s reading is correct. To be sure, that determination may be made in a biased fashion, just as determinations of the best reading of the Constitution may be made in the same fashion. Yet it is plausible that removing one margin on which bias can operate will improve matters. Removing a degree of freedom for the biased judge, all else equal, should lower the costs of monitoring by other judges, legislators, litigants and interested publics. What about the opposite case, where the political judge wishes to invalidate agency action? Under doctrinal Chevron, the judge must say that the agency’s interpretation is clearly incorrect. Under Chevron as a voting rule, more simply, the judge must say that the agency’s interpretation is incorrect. In either case, the judge has only one degree of freedom; the difference between them is that in the former case the legal inquiry is weighted in the agency’s favor, while in the latter case the voting rule is weighted in the agency’s favor. The upshot is that Chevron as a voting rule provides no more scope for biased judges who wish to invalidate agency action, provides less scope for biased judges who wish to validate agency action, and yet weights the scales in the agency’s favor to the same degree across all cases (or can be calibrated to do so, as I will discuss shortly). If reducing biased judging is desirable, then this is a pure improvement, whatever level of deference to agencies one desires and whatever particular conception of biased judging one holds.

Internalization versus Aggregation There is a common thread running through the foregoing points. Doctrinal Chevron imposes greater demands on the individual judge than does 31

The literature is vast, but for a canonical overview, see Jeffrey A. Segal & Harold J. Spaeth, The Supreme Court and the Attitudinal Model (1993).

Institutional Design Writ Small

165

Chapter 

Chevron as a voting rule. Doctrinal Chevron requires the individual judge to internalize a complex, two-tier legal structure whose conceptual foundations are unclear, whose maintenance is psychologically burdensome, and which provides multiple degrees of freedom for the operation of bias. This demand for costly internalization is accompanied by none of the incentives for internalization that appear elsewhere in law. Against this background, it is hardly surprising that Chevron often fails, in the sense that the Chevron two-step does not seem to be fairly applied; what is surprising is that it often succeeds. Chevron as a voting rule, by contrast, makes deference an aggregate property of the voting group, rather than a norm to be internalized by the individual judge, and thus alleviates these burdens.

Calibration A critical advantage of Chevron as a voting rule is the ability to better calibrate the level of deference given to agency decisions. If, in practice, a deference norm produces too little deference, then the voting rule allows deference to be ratcheted up or ratcheted down with greater precision. On a nine-member court, if a five-four majority rule generates too little deference to agencies, then a six-three, seven-two, eight-one, or ninezero requirement can be adopted. Of course too much deference is as bad as too little, from the standpoint of democratic accountability, if it means that Congress’s clear instructions would be ignored in some case. That point is tangential to calibration, however. No matter what the optimal level of deference from the standpoint of democratic accountability, the voting rule approach allows for more fine-grained calibration to that goal. Such gradation is, of course, possible with doctrine as well. A doctrine might command that agency action only be overturned if the decision is (a) unreasonable, (b) clearly unreasonable, or (c) implausible. There is no shortage of linguistic variants in theory, and as a result soft doctrinal rules are capable of calibration too. The trouble is that, as detailed above, historical experience suggests that legal language cannot capture real distinctions in a fine-grained way; lawyers’ attempts to fine-tune verbal formulations merely produce greater uncertainty. Indeed, the more linguistic variants one uses, the greater a morass the doctrinal solution becomes. 166

Mechanisms of Democracy

Delegation, Accountability and Judging

Certainty A final corollary is that the shift from a soft internalized norm of deference to deference as an aggregate property also reduces costly legal uncertainty, of several sorts. Voting rule deference increases predictability and stability, and reduces subjective uncertainty. To the extent that clarity, precision, and the ability of potential litigants to confidently predict judicial analysis are desirable, Chevron as a voting rule is preferable to doctrinal deference. To say that there is uncertainty in a legal regime is really to say that it is difficult to predict either the content of legal rules, the likelihood that a given rule will be applied in a specific case, or how such rules will translate into legal outcomes. One of the key reasons for the apparent failure of Chevron to eliminate if not significantly reduce uncertainty about deference is that the framework makes deference an individual rather than aggregate property of the judicial system, and relies on underspecified norms that are imperfectly internalized by judges. Here two points are important. First, doctrinal Chevron contains inherent ambiguity about what it commands of judges. All agree that judges should defer to reasonable, that is to say, permissible agency interpretations. But, as discussed above, what permissible means is itself highly uncertain. How unlikely must an agency interpretation before it is impermissible? Ninety percent unlikely? Forty-five percent unlikely? The Court has not said, and the views of individual judges will vary. This inter-judge variation generates an additional dimension of uncertainty. Second, suppose compressing Chevron into a single step inquiry, thereby removing one degree of freedom for bias to operate, reduces the ability of biased judges to manipulate the outcome of cases. So long as the potential injection of bias is a source of uncertainty about how judges will behave, the reduction of that source should produce a corresponding gain along the certainty dimension. The core intuition is simply that less discretion and more parsimonious decision processes should reduce uncertainty. Within the doctrinal Chevron framework, agencies and regulated parties know that as a formal matter, deference will be given to agency interpretations. But whether deference will be given in practice is a function of heterogeneous interpretive methods used by individual judges and divergent views about the degree of clarity in statutes. The more heterogeneity at the level of the individual judge, the less certainty there is about how (and even whether) doctrinal Chevron will function in practice. Institutional Design Writ Small

167

Chapter 

Chevron as a voting rule does not face these same dangers. The agency’s decision will be upheld unless a supermajority of the panel concludes that the agency’s interpretation of the statute was incorrect. Different judges may well have different thresholds for identifying the “right” answer in statutory interpretation. But doctrinal Chevron produces uncertainty at an additional stage of the analysis as well—the determination whether the agency’s interpretation is permissible.

Costs and Objections So far, the suggestion has been that switching from doctrinal Chevron to Chevron as a voting rule would promote democratic accountability, just as doctrinal Chevron does, but would do so at lower cost, avoiding the burdens that Chevron places on judges, parties, agencies and the legal system. Are there offsetting costs that would arise from switching to the new regime? Let us examine some objections to Chevron as a voting rule. These objections either are invalid, or apply equally to doctrinal Chevron, and thus afford no basis for preferring one regime to the other.

Voting Theorems Under Chevron as a voting rule, a supermajority of a multimember judicial panel would be necessary to overturn agency interpretations of law. Two well-known voting theorems introduced in previous chapters—May’s Theorem and the Condorcet Jury Theorem—support majority rule. Neither theorem, however, supplies a cogent objection to a supermajority rule in this setting.

May’s Theorem In the simplest version, where two options are involved, May’s Theorem says that only majority rule satisfies a stipulated set of conditions, including neutrality (neither option is preferred by the voting rule), anonymity (the outcome does not depend upon which voter ends up on which side), and two more technical conditions.32 The force of May’s Theorem is that if the conditions are attractive, majority rule should also be attractive. 32

168

Kenneth O. May, A Set of Independent, Necessary and Sufficient Conditions for Simple Majority Decision, 20 Econometrica 680 (1952).

Mechanisms of Democracy

Delegation, Accountability and Judging

Conversely, if one rejects majority rule in some setting, one should also be willing to explain why one or more conditions of the Theorem are unattractive.33 In this setting, the nub of the argument is that neutrality should be rejected. The outcome in which the agency interpretation prevails is more desirable than the outcome in which the agency interpretation is rejected, and the voting rule version of Chevron merely reflects this. The point of Chevron is to put a thumb on the scales in favor of agency interpretations of law, in order to allocate interpretive authority between agencies and courts. A Chevron supermajority rule does so formally, through the aggregation mechanism; doctrinal Chevron with a majority rule does so as well, just informally, through legal doctrine that individual judges are required to internalize. The move to doctrinal Chevron in 1984 already rejected neutrality, in substance if not in form. My suggestion is that the rejection of neutrality is best done explicitly in the voting rule itself. Whether or not the suggestion is persuasive on other grounds, May’s Theorem supplies no valid objection to it.

Supermajority Rules and the Status Quo Related to the foregoing is a point about agencies and policy change. A standard observation in voting theory is that supermajority rules, by violating neutrality, place a thumb on the scales in favor of the status quo.34 One might worry that using a supermajority rule in place of Chevron deference will produce too much status quo bias. However, the status quo must be understood, here, in a legal rather than factual sense. Suppose, as is usually the case, that the agency moves first by issuing an interpretation of the statute; this interpretation then becomes the new legal status quo. A supermajority rule in favor of the agency’s interpretation protects the new status quo as defined by the agency. This approach does not protect the policy status quo, however. To the contrary, freeing up agencies to change policies, as Chevron does in either the 33

34

I bracket here recent extensions of May’s Theorem to multiple options. See Robert Goodin & Christian List, Special Majorities Rationalized, 36 Brit. J. Pol. Sci. 213 (2006). I have assumed, throughout, that in most cases judges will face a choice between the agency’s interpretation and one offered by a litigant, and in any event the argument holds, with appropriate modifications, when extended to the multiple-option case. This is (by definition) only true of “asymmetrical” supermajority rules, not of symmetrical ones; the latter respect neutrality but yield nontrivial ties. For an explanation, see Goodin & List, supra note 33.

Institutional Design Writ Small

169

Chapter 

doctrinal version or the voting-rule version, works to prevent regulatory policy from becoming obsolete. Under either version of Chevron, it is entirely legitimate for agencies to update policies in light of changing circumstances or changing democratic preferences.35

The Jury Theorem The Condorcet Jury Theorem says that where right answers exist, and where the average competence of the voting group exceeds .5, then the probability that majority voting will hit the right answer increases as the group’s size increases and as its average competence increases.36 Perhaps the Jury Theorem suggests that majority rule is preferable, because a simple majority of judicial votes is most likely to get the answer right. It is important to be clear that this point does not support regime (2)—Chevron plus majority rule—as compared to regime (3)—de novo judicial interpretation plus a supermajority rule. The only plausible candidate for a “right answer,” here, is that there might be a right answer about what the relevant statute means. To ask that question is to engage in de novo judicial interpretation, not Chevron deference. The Jury Theorem objection is in effect an argument for regime 1 (de novo judicial interpretation with majority rule). As my main concern is to compare alternative regimes of deference to executive, in particular regimes (2) and (3), the objection is somewhat tangential. The regime I suggest is not Condorcetian, but the current regime is not either. In any event, the objection is also dubious on its own terms, for two reasons. First, de novo judicial interpretation requires judges to decide which party offers the better interpretation of the statute, after considering all permissible sources. It is not at all clear that this sort of legally better answer counts as a “right answer” within the terms of the Jury Theorem. Even if the better legal answer is quite indisputable relative to the rules, conventions and practices of the law, it is a separate and complex jurisprudential question whether that answer can be right or wrong in the same way that a guess about the number of beans in a jar can be right or wrong. These are deep waters, which I will quite simply skim across. But it is not obvious that the Jury Theorem even gets purchase in settings like these. 35 36

170

Motor Vehicle Mfg. Assoc. v. State Farm Mut. Aut. Ins. Co., 463 U.S. 29 (1983). The Jury Theorem can be extended to multiple options, in which it case plurality rule is preferred. See Christian List & Robert E. Goodin, Epistemic Democracy: Generalizing the Condorcet Jury Theorem, 9 J. Pol. Phil. 277, 283–88 (2001).

Mechanisms of Democracy

Delegation, Accountability and Judging

The second and more critical point is that even if the legally best answer counts as a “right answer” in the required sense, the Jury Theorem does not at all support a simple majority voting rule among judges. The confusion here is a common one in discussions of the Jury Theorem. The Theorem itself says nothing at all about the composition of the group that should be governed by majority rule. As shown at length in Chapter 4, it is always necessary to ask: “a majority of what”? Arguments from the Jury Theorem to judicial majority rules are often flawed because they assume, arbitrarily, that only the votes of a majority of judges should be decisive, when in fact the votes of agency decisionmakers are also useful inputs for Jury Theorem purposes. Here is a deliberately artificial example, for clarity. Suppose a multimember administrative commission, like the Federal Communications Commission, votes four-one to resolve a certain question of statutory interpretation, on which (let us suppose) there is a right answer within the terms of the Jury Theorem. Suppose also that review goes directly to the nine-member Supreme Court, which splits five-four against the rule. The Jury Theorem, taken by itself, does not in the least suggest that the court’s view should trump the agency’s. If all voters possess the same average competence, it is arbitrary to exclude the voters who happen to sit on the commission; with their votes, there is an eight-six majority in the rule’s favor. Even if the judges have higher average competence, including the agency voters can only improve the group’s aggregate accuracy, so long as the agency voters’ competence exceeds .5. At a minimum, even if courts should not look behind the agency decision to count the votes cast inside the agency, the agency should at least be counted as having cast one vote. From this perspective, a supermajority rule in the agency’s favor is in effect a way to ensure that the agency has at least one virtual “vote” in the judicial proceedings. The objection assumes, without foundation, that majority rule among the judges alone should be decisive. The Jury Theorem, rightly understood, does not require this; indeed it supports the expansion of the group whose views are aggregated to include agency officials, on plausible assumptions about the competence of those officials, and the supermajority rule in effect does just that. This exemplifies a point made in the introduction: epistemic arguments for majority voting need not entail majority voting in every institutional group, taken one by one. Here, using a nonmajority rule among judges plausibly proves superior on epistemic grounds. Both epistemic democrats and other types of Institutional Design Writ Small

171

Chapter 

democrats can, under these conditions, support nonmajority voting on judicial panels.

Decision Costs Perhaps the switch from doctrinal Chevron to Chevron as a voting rule would increase the costs of decisionmaking, at least for judges. Suppose that under doctrinal Chevron, judges sometimes decide that the agency’s proffered interpretation is reasonable, in the absolute sense required by Chevron, without deciding which party’s interpretation is better (the comparative judgment required by de novo interpretation). In those cases, the judges have avoided a further inquiry that is required under Chevron as a voting rule, and thus economized on the burdens of decision. Judges can use the reasonableness inquiry to avoid the harder problem about what the statute, correctly interpreted, should best be taken to say. The argument fails, on several grounds. First, it is not the case that regime (3) requires judges to decide everything they must decide in regime (2), and then adds more. Rather, regime (3) replaces the reasonableness question with the question of the better interpretation. If the latter question is less costly to answer than the former, then the decision costs of regime (3) are lower, not higher, than the decision costs of regime (2). We suggested above that the particular question whether an agency interpretation is reasonable is indeed harder, on average, than the question which party has the better interpretation. The former requires an absolute judgment that must be made without any theory of what counts as reasonable, and without any clear metric. The latter requires a relative judgment, which is often easier. Second, it is erroneous to say that under regime (2), the judge need only decide reasonableness. Given the current law, particularly the Supreme Court’s recent Brand X decision,37 regime (2) requires the judge to go further, sooner or later, and decide whether the agency interpretation is reasonable because it is the only permissible answer or is instead reasonable because it is permissible but not required. Under Brand X the two types of “reasonable” agency interpretation have very different legal consequences; the former is frozen in place, while the latter can be changed by the agency. Thus the determination whether the agency has offered one type of reasonable 37

National Cable & Telecommunications Association et al. v. Brand X Internet Services et al., 545 U.S. 967 (2005).

172

Mechanisms of Democracy

Delegation, Accountability and Judging

interpretation or the other cannot be postponed forever. Of course that further determination need not be made in Case 1; it can be postponed to Case 2, or Case N. But the only sensible question from the perspective of institutional design is which regime produces higher decision costs across the complete array of cases. The objection rests on the erroneous premise that reasonableness is, under the current law, all judges have to decide. Because the law requires them to go further, doctrinal Chevron has no advantage over the regime we suggest as far as decision costs are concerned. Third, the objection focuses too narrowly on the decision costs incurred by the individual judge, as opposed to the net decision costs of the overall system of litigation. Suppose, contrary to our earlier suggestion, that it is easiest—perhaps all too easy—for individual judges to say that one agency interpretation is reasonable, another is not, and so on, without making relative judgments about better answers. Still, we have emphasized that different judges will have different thresholds of reasonableness, and will reach different conclusions in ways that agencies, litigants and others will find it difficult to predict. In that case, deciding on the basis of ungrounded judgments of “reasonableness” may minimize decision costs for individual judges, but inflict larger decision costs on others, who will find it difficult to decide what to do when the aggregate behavior of a set of judges is unpredictable. We might even object that under the reasonableness approach, judges are exporting decision costs to other actors in the legal system, simplifying their own task while creating net systemic harms. If these matters are unclear, provoking different intuitions in different observers, the most fair-minded conclusion is that the issue of aggregate decision costs probably does not cut strongly in one direction or the other. In an individual case, decision costs might increase or decrease, but any net shift is, plausibly, a second-order consideration. The definite advantages of switching to Chevron as a voting rule are unlikely to be swamped by ambiguous, and plausibly minor, considerations of this sort.

Voting Rules and the Internal Point of View The broadest challenge is that a switch to a supermajority voting rule on a multi-judge panel would cast aside the judicial obligation to reason about law from an internal point of view. Supermajority voting rules, the suggestion runs, shift law to an external standpoint; they stack the deck in favor of one decision over another, independently of the specific arguments for Institutional Design Writ Small

173

Chapter 

or against.38 To some, this represents a basic change in the judicial role. Litigants told that they have lost their case even though a majority of judges voted in their favor might have genuine grievances about the very process, beyond disappointment over the outcome. All this is mistaken, however. Under a supermajority voting rule, each judge approaches law from an internal point of view, just as under a majority voting rule. The difference in the voting rules has no effect on what judges do. The voting rule affects only what the court does; it aggregates individual-level votes into a collective outcome in one way or another. But this has no jurisprudential significance. When the Justices of the Supreme Court vote on certiorari petitions under the nonmajority rule that four of nine votes are necessary for a grant, they are not engaging in a different sort of reasoning (from the internal point of view) than they would if a majority were required. They may of course behave strategically, under any voting rule, but that is another matter. An internal perspective on law does not dictate any particular voting rule; it would be odd to say that a specific institutional practice like majority voting is required by a high level jurisprudential commitment to internalist legal analysis. The oddity of that view is underscored by noting that, if it is taken seriously, the current certiorari rules, which are challenged by no one, would be stamped as unacceptable.

Political Constraints and the Supply Side Even if one accepts all the foregoing, there is a separate issue about which institution(s) can or will supply the change to a supermajority judicial voting rule for the administrative state. Is a transition from doctrine to voting rule simply infeasible? While the obstacles to implementation are not trivial, they are not insurmountable either. In many settings, voting rule solutions have been proposed and adopted as alternatives to doctrinal solutions. Sometimes the adoption has occurred by congressional mandate, sometimes by judicial fiat. No feature of the Chevron context suggests implementation would be harder here than elsewhere, and there are significant reasons to think implementation would be far easier. Whereas other proposals for supermajority voting rules have been associated with efforts to demand judicial deference against judicial wishes, Congress and the courts apparently agree that deference to agencies is a desirable goal. 38

174

Thanks to John Ferejohn for this formulation of the argument.

Mechanisms of Democracy

Delegation, Accountability and Judging

There is therefore good reason to think judges would be less resistant to the imposition of Chevron as a voting rule by statute; perhaps judges might even adopt voting rule Chevron without a congressional dictate. In general, Congress has proven almost completely mute about when courts should defer to administrative agencies. This may show that the Supreme Court accurately selected the interpretive default rule in Chevron: statutory silence evidences congressional satisfaction with the current soft deference framework. Of course, congressional silence before Chevron was decided in 1984 might just as convincingly be taken to have signaled support for the rather different deference regime that then obtained. Current congressional silence no more demonstrates satisfaction with the Chevron doctrine than did congressional silence before 1984 demonstrate congressional satisfaction with the lack of a Chevron doctrine. One possibility is that the issue of deference simply lacks political salience with constituents and therefore also lacks payoffs for legislators— the motivational problem discussed in Chapter 2. If so, Congress is unlikely to adopt the proposal; but the judicial route remains, and congressional passivity means that a legislative override of a judicially adopted change would itself be unlikely. The flipside of the foregoing point, however, is that the proposal is impartial in the sense also discussed in Chapter 2: it is not related to a contested institutional fight about if and when courts should show deference to executive or legislative judgments on any particular policy or in particular, hotly controversial areas. While the Chevron doctrine has received its fair share of criticism in the commentary, and peripheral questions about scope and intensity of review remain, the core doctrine of Chevron deference is now almost universally embraced by Congress, courts, and agencies. In this sense, the task is quite different from that of those who, say, urge that Congress foist a supermajority voting rule on the Supreme Court in any case involving abortion rights. Of course the proposal would face resistance, but the nature of the resistance would be fundamentally different, and its intensity much lower, simply because the judiciary already holds the view that deference is appropriate.

Judges, Democracy and Accountability The Chevron doctrine is a pillar of American administrative law. I have argued that, once the bogus nondelegation principle is cleared away, Institutional Design Writ Small

175

Chapter 

democratic accountability requires that courts should defer to the democratically superior judgment of administrative agencies, where Congress has not spoken clearly. Given that premise, a voting rule deference framework equals or exceeds a doctrinal deference framework on virtually every dimension of comparison. Rather than impose upon judges the awkward task of developing and executing a second-order theory of permissible interpretation, the voting rule approach asks judges to do what they do best: identify the best interpretation of a statute. This shift reduces the conceptual, psychological and motivational burdens that the Chevron doctrine places on judges. Such burdens are not just benign academic quandaries. They result in confusion and uncertainty about the manner in which deference will be applied. Such confusion is costly in its own right, but it also helps mask other potential biases in judicial decisions. Chevron as a voting rule avoids these problems by making deference an aggregate rather than individual feature of judicial review. Combined with the far more powerful ability to calibrate deference levels, the analysis suggests the hard voting rule is a superior mechanism for promoting the democratic values of accountability at the core of the modern administrative state.

176

Mechanisms of Democracy

Part III

Deliberation and Transparency

This page intentionally left blank

Deliberation and Transparency

S

uppose, following Parts I and II, that the background democratic institutions of the relevant polity manage both to enforce impartiality (at least in the narrow sense that officials are not systematically selfinterested) and also accountability to the public as a whole (that is, accountability in its democratically laudable as opposed to damaging variant). Is that enough, from the democratic point of view? Surely not. These would be worthy accomplishments, but it is perfectly possible for an impartial and accountable government to be dysfunctional, democratically speaking. Part III examines two further elements of a well-functioning democratic polity: deliberation and transparency. In Chapter 6 transparency takes center stage, while deliberation is the focus of Chapter 7. Democratic polities decide issues of low policy as well as high principle, and democratic mechanisms must be capable of cutting across that distinction. Accordingly, Chapter 6 analyzes transparency in the setting of the national budget process in the United States Congress, while Chapter 7 analyzes deliberation in the rather more elevated setting of legislative resolution of constitutional issues. Why exactly are deliberation and transparency important? Imagine that a purportedly democratic government did its business by simply aggregating pre-existing preferences, perhaps by voting in a multimember assembly, but without any prior exchange of views among officials or the citizenry. On most accounts of the democratic ideal, there is something deeply wrong with this procedure. The missing element is deliberation. The deliberation at issue need not be very elevated, or principled, or public-spirited, or directed towards ends rather than means. It may be hot rather than cool, messy rather than tidy, rhetorical rather than analytic. It may be simply deliberation that represents an exchange of information and instrumental views—although even deliberation of that sort will often end up changing at least the legislators’ derived preferences, even if not their ultimate ones, as legislators realize the consequences of pursuing the goals they had previously set. The important point is that some sort of deliberation there must be. In either a thinner or a thicker

Institutional Design Writ Small

179

Part iii

sense, deliberation is central to the common public life that is at the core of the democratic aspiration. At the same time, just as from a democratic point of view there are good and bad consequences of uncertainty (Part I), and good and bad consequences of accountability (Part II), so too there are good and bad consequences of deliberation. As against the most highminded versions of the deliberative ideal—many here imagine the American Framers at the Philadelphia Constitutional Convention, although the reality was quite different—we may juxtapose the rhetorical intemperance, posturing, implicit threats, self-censoring by frightened dissenters, and inattention to constitutional rules that pervaded discussion in the French National Assembly of the early 1790s. These too are, predictably, part of what happens when democratic representatives exchange views on their common life. We may if we like define away these vices as not truly “deliberative,” but from the perspective of institutional design that is a merely a verbal solution. The tradeoffs are real, not verbal. Any set of institutions that allows for the good and highminded deliberation will produce the bad kind as well. The optimization problem is to design mechanisms that advance the former while suppressing the latter to the extent possible and desirable— always recognizing, first of all, that no realistic set of institutions will be able minimize the deliberative vices to zero, and second of all that extreme views and hot rhetoric may themselves make an affirmative contribution to the group’s deliberative success by introducing fresh perspectives, shattering false consensus, or providing the motivational punch that cool reason lacks.1 Chapter 7 is an extended attempt to show how the resulting problems might be approached, in this case by suggesting a set of institutions designed to optimize constitutional deliberation in the United States Congress. In a democratic polity, the hinge that connects accountability and deliberation is transparency. Transparency in and of itself is plausibly a democratic virtue. Citizens who are entitled at least to an indirect share in the making of the laws, and who will be bound by the laws, should also be able to observe their delegated representative-lawmakers at work. To be sure, transparency is always a defeasible virtue. It is true, in many settings, that transparency is overridden, as when government sensibly conceals 1

See Bryan Garsten, Saving Persuasion: A Defense of Rhetoric and Judgment (2006); Don Herzog, Romantic Anarchism and Pedestrian Liberalism (unpublished draft).

180

Mechanisms of Democracy

Deliberation and Transparency

the identity of spies or the location of troops abroad; but that does not mean transparency counted for nothing in the first place. Insofar as no countervailing considerations prevail, or even if they do, it is fully coherent to say that a more transparent government—one that allows citizen-subjects to observe decisions being made—is a more democratic government. Whatever the intrinsic weight of transparency as a democratic good, however, it is certainly an indispensable institutional precondition to the achievement of other democratic goals. Accountability requires some degree of transparency with regard to legislators’ behavior; pure retrospective voting, in which voters observe only the government’s policy choices (but not how their representative voted) and the state of the world since the last election, gives voters too little information about what their representative has done. On any sensible conception of representation, I shall argue, there is a type of agency relationship between voters and representatives, and the transparency of government action helps the voter-principals to police the agent’s actions. Relatedly, transparency deters officials from engaging in self-interested bargaining, which means that transparency helps to enforce impartiality in one of its senses. The dilemma, however, is that transparency can itself exacerbate (the risk of) the deliberative vices that I have mentioned. Knowing that a large political audience is at hand, representatives will more often tend to play to the galleries, stick to positions once taken even if evidence and good reasons accumulate against their views, and generally place vanity above reasonable practices of belief formation and decisionmaking. Moreover, transparency can silence dissenters by forcing them to take costly public risks rather than less costly private ones. Overall, making government a fishbowl can itself make government officials poor deliberators. Chapter 6 therefore attempts to optimize transparency by suggesting mechanisms—such as secrecy for the early stages of policymaking, and delayed disclosure of legislators’ decisional processes—that promote the desirable effects of transparency while suppressing its undesirable ones. Things are even more complex than that, however. Even insofar as accountability alone is concerned, transparency is not an unmixed blessing, as Chapters 3 and 4 began to suggest. There is good and bad accountability, and transparency that promotes the former also promotes the latter. The very transparency that allows voter-principals to hold their representatives accountable also allows third parties, such as narrow interest groups, to hold the same representatives accountable if they do not violate Institutional Design Writ Small

181

Part iii

the terms of the agency relationship by providing the third parties with democratically indefensible benefits. The same institutional arrangement that helps some to keep representatives on the straight and narrow path also helps others to lure representatives off the path of duty. Another of the principal aims of Chapter 6 is to ameliorate this tension by proposing design mechanisms that increase good accountability while reducing bad accountability. All this means that tradeoffs abound. Transparency is necessary for accountability, and helps to promote impartiality by suppressing self-interested official behavior. At the same time, however, transparency produces bad forms of accountability as well as good forms; moreover, transparency can harm deliberation by encouraging posturing and other deliberative vices. Transparency, accountability, and impartiality are all intimately connected, and the institutional designer must simultaneously consider all these democratic goods, in both their desirable and undesirable forms. Because tradeoffs abound, the search for optimizing mechanisms becomes all the more important. In this sense, Part III incorporates all that has gone before in Parts I and II, attempting to weave together impartiality, accountability, transparency, and deliberation in an institutionally sensible way.

182

Mechanisms of Democracy

Chapter 6

Optimizing Transparency: The Budget Process

T

he notion of “transparency” in government is very much in vogue both in the United States and worldwide, particularly in the arena of fiscal policy. The emphasis on openness is sensible, because transparency serves crucial objectives in democracies. Transparency can promote public-spirited behavior by constraining bargaining based on self-interest, promoting principled deliberation instead. Even where self-interest is universal, providing information to voter-principals about the action of their agents—here, elected officials—reduces the costs of monitoring, thereby promising to improve governance in a representative democracy. However, adherents of transparency are often insufficiently attentive to the costs of disclosure.1 Transparency is sometimes in tension with other important democratic values and may even, in some cases, be self-defeating. “Transparency” always has the rhetorical upper hand; it connotes the

1

For overviews of the costs of transparency, see Mark Fenster, The Opacity of Transparency, 91 Iowa L. Rev. 885 (2006); see also David Luban, The Publicity Principle, in Theory of Institutional Design 154, 183–87 (Robert E. Goodin ed., 1996).

Institutional Design Writ Small

183

Chapter 

opposite of secrecy and skullduggery. But this is misleading, because transparency has democratic costs as well as benefits. In this chapter I will use the federal budget process in the United States as a case study in optimizing the democratic costs and benefits of transparency. Money is central to what government does in any polity, and the budget process is central to what government does in the United States. The budget process is fundamental to any institutionally informed account of the workings of democratic institutions; as we will see, it is an arena in which all the themes of the book—impartiality, accountability, deliberation, and transparency—intersect in complex ways. Democratic theory ought to prove itself useful in mundane policy contexts, as well as elevated ones; the budget is an ideal arena in which to test this aspiration. The principal focus, however, is on transparency; the chapter offers a series of suggestions for small-scale mechanisms intended to optimize the transparency of the budget process, and many of these mechanisms generalize easily to other contexts.2 Under some circumstances transparency suppresses self-interested deals, and thus encourages a kind of impartiality; but under some circumstances transparency also diminishes the quality of democratic deliberation. In the face of these tradeoffs, the hope is to suggest mechanisms capable of producing large incremental gains in deliberation and impartiality at acceptable cost to transparency or other democratic values. A central design idea here is that transparency might be selective— not only in the sense that transparency applies only to certain classes or subjects of government decisionmaking, but also in the less familiar sense that disclosure might be made only to some people at some times and not others at other times. The basic reason to make transparency selective is that there is a tension between good and bad accountability, arising from the presence of multiple audiences. On the one hand, there are general constituents and voters whose attempts to hold their agent-representatives accountable is democratically desirable. On the other hand, there are narrow interest groups and other third-party interests whose attempts to 2

On another tack, Adam Samaha suggests that judges might use existing transparency-related statutes, such as the Freedom of Information Act, as platforms for judicial interventions aimed at optimizing or at least improving the transparency regime. See Adam Samaha, Government Secrets, Constitutional Law, and Platforms for Judicial Intervention, 53 UCLA L. Rev. 909 (2006).

184

Mechanisms of Democracy

Optimizing Transparency: The Budget Process

hold officials accountable exacerbate agency problems between representatives and their constituents, and are thus democratically undesirable. The basic solution to this problem is a system of selective disclosure of governmental decisionmaking to the former audience but not the latter one. Of course, the hard part of this trick is carrying it off. The suggestion here is that a system of delayed disclosure of governmental deliberations can go some way towards accomplishing selective transparency.

Overview There are several distinct arenas of budget policymaking, including decisionmaking purely within the executive branch; policy recommendations by federal advisory committees; the legislative process including committee consideration, floor deliberation and conference committee decisions; and interbranch decisionmaking in occasional budget summits. In each arena, there is a mixture of openness and secrecy; political players find private space for some deliberation, even in the face of aggressive open meeting requirements, by taking advantage of gaps in any rules or statutes. Is the current mix optimal, in the sense that there are no adjustments that could produce gains in impartiality and the quality of deliberation, while not doing great harm to the values underlying transparency? I will claim that the current arrangements are not optimal, and that there is no theoretical reason why they should be. Even without legal requirements for transparency, promising transparency can be in the agent-lawmakers’ interest, if it encourages voters or constituents to give the lawmakers authority that would otherwise be withheld.3 However, since there is no general mechanism that aligns this process of political bidding with social interests, there is no reason to believe that the degree of transparency reached by political actors on their own would necessarily be optimal from the democratic point of view. Furthermore, self-interested agents also have incentives to keep secret some aspects of budgeting that their principals would be eager to monitor—namely, the use of budgets to provide 3

See John Ferejohn, Accountability and Authority: Toward a Theory of Political Accountability, in Democracy, Accountability and Representation 131 (A. Przeworski et al. eds., 1999). Note, however, that the only extant test of this model reaches inconclusive results. See James E. Alt, Three Simple Tests of Ferejohn’s Model: Transparency and Accountability in US States (2005) (paper prepared for NYU Colloquium on Law, Economics and Politics).

Institutional Design Writ Small

185

Chapter 

benefits to well-funded and well-organized special interests that will reward lawmakers with campaign funds.4 The upshot of these cross-cutting forces is unclear, but it is unlikely that the status quo, where some deliberation occurs in secrecy notwithstanding open meeting requirements, precisely optimizes transparency. The chapter begins with a brief map of transparency in the federal budget process, as it is currently structured. I then identify some key tradeoffs that determine the costs and benefits of transparency. One tradeoff is that secrecy promotes good deliberation, while transparency deters self-interested bargaining. A second tradeoff is that transparency ensures both accountability to voters, which is good, and also accountability to narrow interest groups, which is bad. In light of these tradeoffs, I sketch an optimal transparency structure for budgeting, a structure with two crucial features. First, the early stages of the budget process, including the formulation of a concurrent budget resolution by budget committees, will be secret or opaque, while the later stages will be transparent. Here the dual aim is to encourage good deliberation where that is possible and to hamper self-interested bargaining where that is likely. Second, even for later stages of the budget process, disclosure will be delayed, perhaps until well into the election cycle. Here the aims are, first, to deny interest groups immediate access to the details of ongoing decisionmaking; and, second, to sort good from bad accountability by giving voters information when they need it while denying information to interest groups when they want it.

Transparency in the Federal Budget Process We must begin with some premises and a brief orientation tour, with a special focus on the current mix of transparency and opacity in the federal budget process.

4

See Alberto Alesina & Roberto Perotti, Fiscal Discipline and the Budget Process, 86 Am. Econ. Rev. 401, 403 (1996) (noting that “politicians do not have an incentive to adopt the most transparent practices” because they want to maintain their informational advantage). Ferejohn’s work suggesting that legislative agents have incentives toward some amount of transparency to allow audit of their activities by principals does not suggest that these incentives will produce the optimal mix of transparency and opacity, but that “the agent has an interest in offsetting the effects of some of those [informational] asymmetries.” Ferejohn, supra note 3, at 149.

186

Mechanisms of Democracy

Optimizing Transparency: The Budget Process

Two Senses of “Transparency” The transparency of the budget process is only one sense of “transparency.” The literature addresses both the transparency of the process, or of the inputs into the budget, and the transparency of the output itself, or of the budget documents produced by government officials.5 The latter concern, which might better be labeled “intelligibility,” emphasizes the ease with which people, the press, financial markets, and others can understand budgetary decisions reached by policymakers. Are the tradeoffs involved adequately described? Are the economic and accounting assumptions clearly set forth? Are the methods of accounting for expenditures and revenues likely to provide a basis for an accurate assessment of the country’s fiscal health? Are budgetary decisions made in one or a few documents so the entirety of the budget can be understood? Can outside, nonpartisan experts analyze the fiscal condition of the country on the basis of the budget documents? Indeed, much of the literature on fiscal transparency seems more concerned with intelligibility rather than the openness of the deliberations themselves.6 These two aspects of transparency are related. The more transparent the output, the more likely it is that outsiders can reason backwards to develop a sense of the inputs, without actually witnessing the process as it occurs. In the fiscal arena, however, the process of using the output to discover the inputs can be challenging. Budgets are complex, so it can be difficult to separate all the strands and exceedingly difficult to trace decisions back to particular actors and particular motivations. Moreover, certain outputs can be the product of various inputs—some public-regarding, others the result of the influence of private-regarding behavior by narrow interest groups. Even if the outcome is socially desirable, it may still be important in a democracy to understand whether it is the result of particular interest group pressure brought to bear on key legislators. Finally, if the process is open, outsiders can more easily provide helpful information as deliberations occur. Thus, the issue considered here is the transparency of the decisionmaking process in a budget system featuring substantial transparency of outputs. 5

6

For an overview of the political economy literature on fiscal transparency, which emphasizes transparency of outputs rather than of process, see Timothy Besley, Principled Agents? The Political Economy of Good Government 203–06 (Oxford 2006). See, e.g., James E. Alt et al., Fiscal Transparency, Gubernatorial Approval, and the Scale of Government: Evidence from the States, 2 State Pol. & Pol’y Q. 230, 235–36 (2002).

Institutional Design Writ Small

187

Chapter 

I will use secrecy or opacity as antonyms of “transparency”. In general, I will not discuss the differences between a regime of strong secrecy and one of weak secrecy. Under strong secrecy, the very existence of secret proceedings is itself secret. Under weak secrecy, it is known that such proceedings are occurring, but their content is not transparent.7 Only the latter type of secrecy is implicated here.

Transparency in the Budget Process: Current Rules There are degrees of openness in the federal budget process, varying across different arenas of decisionmaking. Generally, the process is an open one, particularly when it reaches Congress. In all stages of the process, however, players can find ways to bargain in private through informal interactions or negotiations with a subgroup of relevant decisionmakers that result in a deal later made public. Some of the least transparent stages of the budget process are those that occur entirely within the executive branch. Most opaque are policy discussions that take place wholly within the executive branch among administrators, who are protected by the deliberative process component of executive privilege. This protection from disclosure is not absolute.8 Nevertheless, many budget discussions do not take place at this level, and those remain confidential. The White House avidly guards the privacy of this kind of policy deliberation. Once the budget reaches Congress, however, it is shaped by internal rules in the House and Senate which generally require that committee hearings and meetings to transact legislative business, including markups, be conducted in open session. Although the rules allow meetings to be closed, they limit executive sessions to circumstances where disclosure would “endanger national security, would compromise sensitive law enforcement information, [or] would tend to defame, degrade, or incriminate any person.”9 As with all internal rules, these are enforced within the body so that members can raise points of order and other objections if they believe a meeting has been closed contrary to the rules. All the 7 8 9

Thanks to Fred Schauer for alerting me to this distinction. See National Regulatory Comm’n v. Common Cause, 674 F.2d 921 (1982); Federal Communication Commission v. ITT World Communications, Inc., 466 U.S. 463, 471 (1984). House Rule XI, cl. 2(g)(1); see also Senate Rule XXVI, para. 5(b) (providing similar reasons for executive session with some additional justifications such as relating “solely to matters of committee staff personnel or internal staff management”).

188

Mechanisms of Democracy

Optimizing Transparency: The Budget Process

relevant committees in the budget process—appropriations, budget, and tax-writing committees—operate under similar open meeting requirements.10 Thus, meetings where testimony is taken or business is conducted can be closed only in limited circumstances, such as committee meetings concerning the budget of intelligence agencies where discussions implicate national security. In particular, committee markups and votes on legislation relating to the budget must occur in public. Similarly, House and Senate rules require that conference committees conduct their business in open meetings.11 Notwithstanding open meeting rules in Congress, a great deal of the bargaining related to the federal budget occurs in private. Congressional parties discuss broad policy issues in party caucuses that are not subject to open meeting requirements. Increasingly, party leaders have used task forces made up of selected lawmakers—sometimes only from the majority party—to formulate legislative proposals, some of which may be incorporated into budget reconciliation bills.12 Once the proposal goes to a committee or the full body, the meetings to debate and amend it are open to the public, but the preliminary work of the task force occurs largely in private and may crucially shape the ultimate product. Even during committee deliberations and conference committee negotiations, discussions occur between the committee leadership and individual members, among co-partisans on the committee, between staff of members and the committee, or in other groups of lawmakers and staff that are not considered official “meetings” or “hearings” subject to the transparency provisions of the internal rules. Conference committees in particular routinely hold key gatherings behind closed doors, sometimes only among members of the majority party in both houses or, more frequently, just their leaders who negotiate deals in private before the overall bargain is presented to the whole committee in a pro forma public meeting.13 10 11 12

13

See, e.g., Rules of the House Committee on the Budget, Rule 5; Rules of the House Committee on Appropriations, sec. 4(d)(1); Rules for the Senate Committee on Finance, Rule 2. House Rule XXII, cl. 12(a)(1); Senate Rule XXVIII, para. 6. These informal task forces are different from task forces appointed by some committees, e.g., Rule of the House Committee on the Budget, Rule 12, which are the equivalent of subcommittees. For a discussion of the use of task forces, see Elizabeth Garrett, Attention to Context in Statutory Interpretation: Applying the Lessons of Dynamic Statutory Interpretation to Omnibus Legislation, Issues in Legal Scholarship, Dynamic Statutory Interpretation 9 (2002), http://www.bepress.com/ils/iss3/art1 (last visited Nov. 10, 2006). See, e.g., 16 Deschler’s Precedents of the United States House of Representatives, H.R. Doc. No. 94–661, ch. 33 (1976). It was not until the 94th Congress that the House and

Institutional Design Writ Small

189

Chapter 

Budget deliberations on the floor of the House and Senate are public and televised. It is highly unusual for either house to go into a closed session, and consideration of the various budget vehicles—the concurrent budget resolution, appropriations bills, budget reconciliation acts, and the various conference reports—are not likely to trigger an executive session. Again, although the deliberations, amendments, and votes are public, and constitutionally required to be memorialized in a journal of proceedings,14 many of the key deals are made in the cloakrooms or in members’ Capitol hideaways and not made public except as they can be discerned from the provisions of the law ultimately enacted and any public claims of responsibility from the lawmakers involved. Because the open meeting requirements affecting congressional committees are contained in internal rules of the bodies or of the committees themselves, enforcement is a purely internal matter. As long as the members are content with the mix of transparency and opacity in the congressional budget process, no outside watchdog can challenge either the decision to hold some meetings behind closed doors or the validity of the legislation that emerges from the process on the basis that transparency rules were violated. In the post-reform Congress, more of the work of committees has been done in public, and committee hearings and markups are broadly publicized through television and other press coverage. The force of the congressional rules is not clear, for several reasons. First, legislative rules requiring transparency are often partially evaded by members who need to negotiate in private. They take advantage of gaps in the rules’ coverage to discuss important issues informally behind closed doors. However, requiring transparency in formal rules is a signal to the public of lawmakers’ commitment to openness and may increase the political cost of defecting from the rules if intermediaries like the press or challengers successfully make secret deliberations an issue for voters. Second, lawmakers might carry out a great deal of their work in public even without rules requiring them to do so. As we have seen, elected officials have an incentive to conduct a significant portion of their activities in public as a way to signal to their principals—the voters—that they

14

Senate adopted rules requiring that conference committee meetings be held in open session. Id. at § 5.2. However, all that is required for a valid conference is a quorum on the signature sheet of the conference report and a public meeting of the conferees. Id. at § 5.8. The result is that a great deal of preliminary deliberations can be held behind closed doors. U.S. Const., § 5, cl. 3.

190

Mechanisms of Democracy

Optimizing Transparency: The Budget Process

are trustworthy. By adopting transparent practices, lawmakers offer voters better tools to monitor their responsiveness; voters are therefore more willing to trust officials with control over more resources.15 However, empirical tests of this suggestion have not been conclusive, in either direction.16 One final stage in the budget process has an entirely different transparency profile. In some years, the legislative and executive branches clash over budget policy, and interbranch negotiations are required to pass the appropriations and reconciliation bills necessary to implement budget objectives and to keep the federal government operating. This informal device—the budget summit—has been used on several occasions when interbranch compromise was difficult but necessary. Because summits are not formal entities with rules governing their conduct and because the only people involved are elected officials from Congress and executive branch officials from the White House and departments, open meeting rules do not apply. When it occurs, this final stage of deliberations concerning the federal budget is entirely opaque. Given all this, the difficult question is whether the equilibrium with respect to openness that has been reached in the federal budget process— whether by explicit rule or by practice—is optimal. There are two basic dimensions along which the degree of transparency can be evaluated. The first question is the effect of transparency on deliberation and discussion— how transparency affects the mix of arguing and bargaining in the budget process, and which type of deliberation is most suited to making decisions about the various aspects of budgeting. The second question is how transparency affects the power of organized interest groups to monitor and punish lawmakers, rather then empowering voters to hold legislators accountable on Election Day. One of the main goals of the chapter is to suggest small-scale design mechanisms that can restructure disclosure and transparency in order to benefit voters without similarly benefiting narrow special interests. I shall also suggest that this goal need not be controversial as a matter of democratic theory. Even pluralist accounts of democracy, which favor the contest of organized interests, look askance on differentially-organized groups whose grip on lawmakers cannot plausibly be defended as part of a larger democratic system. As we will see, the budget process is rife with such groups. 15 16

See Ferejohn, supra note 3. See Id.

Institutional Design Writ Small

191

Chapter 

Transparency, Accountability and Deliberation: Effects and Tradeoffs In any arena, the mix of transparency and opacity has three principal effects. One effect is that transparency can alter, at the margin anyway, the extent to which relevant actors engage in principled argumentation, on the one hand, or overt bargaining, on the other hand.17 Both arguing and bargaining are indispensable processes for aggregating judgments or preferences into collective decisions, so optimal structures will combine both in some mix. The second effect is that the glare of transparency may reduce the quality of democratic deliberation,18 causing representatives to engage in posturing, extremism, and unproductive rhetoric. As we have discussed, however, not all extremist speech or hot rhetoric is unproductive; where such tactics bring new perspectives to light, they can improve group decisionmaking overall. A third effect is that transparency promotes accountability. This common idea is surprisingly slippery. The crucial questions are accountability to whom, and for what? As we saw in Chapter 4, accountability can be good, when it runs from agents to principals, such as voters or constituents generally; it can also be bad, when it runs from agents to third parties such as transfer-seeking groups. As before, all standard accounts of representation can converge on this fear of bad accountability. On one account, legislators act as trustees to promote the general interest of the polity as a whole; on another, legislators act as delegates who are charged with promoting the interests only of their constituency, usually defined along geographic lines. The difference between these accounts, while important, is immaterial to this discussion. That is because both accounts condemn legislative capture by organized interest groups. Here “capture” can, but need not, be interpreted in standard welfarist terms; more generally, it denotes systematic legislative behavior that exclusively promotes the interests of narrowly defined groups (much smaller than even the smallest constituency). Such groups seek transfers that, while beneficial to themselves, inflict larger harms on disorganized constituents and society generally. That sort of representation is objectionable whether we hold some version of a trusteeship account of representation or some version of a delegate account, so we need not engage the deeper issues here. 17 18

This distinction is derived from Jon Elster, Arguing and Bargaining in Two Constituent Assemblies, 2 U. Penn. J. Constl. Law 345, 405–15 (2000). See Simone Chambers, Behind Closed Doors: Publicity, Secrecy and the Quality of Deliberation, 12 J. Pol. Phil. 389 (2004).

192

Mechanisms of Democracy

Optimizing Transparency: The Budget Process

This picture assumes that not all interest-group activity is good from the democratic point of view; to that limited extent I reject the most optimistic versions of pluralist theory. The most optimistic pluralists tend to argue that interest-group activity, even if harmful when taken in isolation, can be defended as part of a larger system that is democratically desirable overall.19 This view, even if correct, need not rule out institutional proposals that aim to improve upon the workings of the invisible hand by producing the same benefits at lower collateral cost. Those are the sort of proposals offered here. Furthermore, and perhaps more contentiously, I believe that the pluralist view does not jibe with the realities of the budget setting, where democratically harmful transfers and wasteful competition to obtain such transfers are not merely collateral costs of a desirable system, but rather the routine stuff of policymaking. Earmarks and deals that result in “bridges to nowhere”20 are themselves the very substance of the system, rather than byproducts of optimal arrangements. Of course some interestgroup activity is also beneficial, as when organized groups monitor each other, offset each other’s influence, or supply voters and constituents with useful information. But nothing in these points is inconsistent with the proposals discussed below, which are designed to capture these incidental pluralist benefits while suppressing interest-group activity that cannot be justified in systemic terms. By “democratically harmful” transfers, I do not just mean economically inefficient transfers, although the transfers to which bad accountability gives rise are often that as well. Legal scholar Einer Elhauge points out that concern about interest groups presupposes some normative baseline or other in democratic theory.21 But nothing in Elhauge’s argument requires that any particular normative baseline be chosen. Rather, a broad range of high-level views can converge to condemn the sort of transfers that the budget process routinely witnesses. Such transfers are typically motivated by self-interest among the participants, rather than by impartial reasons 19

20

21

See James Wilson, Democracy Needs Pork to Survive, Wall Street Journal, August 14, 1997 (pork is the “necessary glue that holds political coalitions together”); cf. Gary S. Becker, A Theory of Competition Among Pressure Groups for Political Influence, 98 Q. J. Econ 371 (1993). See Michael Grunwald, Pork by Any Other Name, Wash. Post, April 30, 2006, at B01; available at http://www.washingtonpost.com/wp-dyn/content/article/2006/04/29/AR2006042900141. html. See Einer Elhauge, Does Interest Group Theory Justify More Intrusive Judicial Review?, 101 Yale L.J. 31 (1991).

Institutional Design Writ Small

193

Chapter 

arrived at under conditions of well-functioning deliberation; as such they are condemned by deliberative views of democracy. As we have seen, pluralist views of democracy will not support them; they can be dispensed with under superior institutional arrangements, and in any event are not plausibly byproducts of some overall system of healthy democratic competition. They are also inefficient in any of the usual economic senses of that term: they reduce aggregate wealth and aggregate utility, and would be prohibited by hypothetical bargainers seeking Pareto-optimal arrangements ex ante. Thus they would be condemned by a whole cluster of consequentialist and welfarist accounts of democratic institutions. Finally, although this is now a point about theories of justice rather than theories of democracy, budget-process transfers are often distributively unjust, as they transfer resources from the relatively worse-off to the relatively betteroff without colorable justification. Accountability to a broader public for transfers of this sort is democratically desirable, but accountability to narrow groups that results in transfers of this sort is bad in an uncontroversial sense. It should be a major goal of any transparency regime to minimize the latter sort of accountability. Having laid out the major concepts, let us now take up their mutual causal relationships and the resulting tradeoffs.

Arguing and Bargaining Arguing, let us say, is deliberation (whether or not sincere in some subjective sense) that is pragmatically constrained to rest on impartial and internally consistent reasons—deliberation that is either sincere or at least constrained by the “civilizing force of hypocrisy.”22 The stuff of bargaining, by contrast, consists of credible threats and promises, usually though not inevitably in the service of the self-interest of the bargainers. Transparency dampens overtly self-interested bargaining and pushes officials in the direction of principled argumentation. In the glare of transparency, people tend to offer neutral principles related to the public good,23 not bargains based solely on private interests. 22 23

These definitions, and the terms “arguing” and “bargaining,” are adopted from Elster, supra note 17, at 372–3. An interesting attempt to study the facts of deliberation in the national legislatures of Germany, Switzerland, the United Kingdom, and the United States found that “[p]ublic debates also contained significantly larger numbers of appeals to the common good than non-public debates.” Jurg Steiner et al., Deliberative Politics in Action: Analysing Parliamentary Discourse 129 (2004).

194

Mechanisms of Democracy

Optimizing Transparency: The Budget Process

This marginal effect can either be good or bad, depending on the setting. Bargains may represent corrupt deals by which agents enrich themselves at principals’ expense, but bargains also permit logrolls that may allow the legislative process to register the intensity of constituents’ preferences,24 and that help to appease policy losers by giving everyone something. Argument by reference to neutral principles, by contrast, often pushes policy towards the extremes, resulting in total victory or total defeat. Transparency also subjects public deliberation to reputational constraints. One risk is that officials will stick to initial positions, once announced, for fear of appearing to vacillate or capitulate, and this effect will make deliberation more polarized and more partisan. In America, the Framers of the 1787 Constitution closed the Philadelphia convention to outsiders precisely to prevent initial positions from hardening prematurely.25 Another risk is that transparency will silence potential dissenters, who fear to take public risks and thus go along with the crowd or with apparent experts;26 the tenor of nonpublic deliberation, by contrast, can be more respectful of all concerned.27 In a related vein, transparency can produce ideological or moral posturing that is at odds with the sincerity that is part of ideal deliberation. Finally, transparency may simply drive decisionmaking underground, creating “deliberations” that are sham rituals while the real bargaining is conducted in less accessible and less formal venues, off the legislative floor. 24

25

26

27

Logrolling may, of course, either permit socially beneficial trades or inflict socially harmful externalities on nontraders. Much depends on the details of the situation. “Today, no consensus exists in the normative public choice literature as to whether logrolling is on net welfare enhancing or welfare reducing, that is, whether logrolling constitutes a positive- or a negative-sum game.” Thomas Stratmann, Logrolling, in Perspectives on Public Choice: A Handbook 322, 322 (D.C. Mueller ed., 1997). At the Federal Convention, the sessions were closed and secret. As Madison said later: “Had the members committed themselves publicly at first, they would have afterwards supposed consistency required them to maintain their ground, whereas by secret discussion no man felt himself obliged to retain his opinions any longer than he was satisfied of their propriety and truth, and was open to the force of argument.” The Records of the Federal Convention of 1787 (Max Farrand, ed.) 479 (Yale 1966) (remarks of Madison). See Ellen E. Meade & David Stasavage, Publicity and the Incentive to Dissent: Evidence from the U.S. Federal Reserve (Draft of March 2006), http://cep.lse.ac.uk/pubs/download/dp0608.pdf (last visited Nov. 10, 2006). For a contrasting model in which secrecy induces group members to vote in accordance with the group’s prior biases, see Gilat Levy, Decision Making in Committees: Transparency, Reputation and Voting Rules (LSE Online Research Working Paper, Nov. 16, 2005). An implication of this model is that “when a committee becomes transparent, it is more likely to accept reforms or radical decisions.” Id. at 2. See Steiner et al., supra note 23, at 129–30.

Institutional Design Writ Small

195

Chapter 

Although the relevant questions and answers relating to deliberative transparency vary from setting to setting, it is possible to offer a general guide to the relative costs and benefits of arguing and bargaining across contexts. The principal benefit of transparency is to dampen self-interested bargaining—including in this category bargaining by legislators who act as tightly constrained agents for self-interested private groups, a topic we take up shortly. The principal cost of transparency is that the glare of the public spotlight can produce bad deliberation that is not conducive to necessary compromise—deliberation infected by posturing and ideological polarization.28 This suggests a rule of thumb: in general, the less information legislators have about how decisions will affect their interests, the less selfinterested bargaining is possible in any event; the deliberative distortions produced by transparency are then all cost for no benefit. On this logic, to the extent that legislators must necessarily act behind a partial veil of uncertainty of the sort discussed in Chapter 1, then opacity is better all else equal. By the same logic, transparency is better in settings where legislators have rich information about the effects of particular decisions on their interests.29 As discussed below, this suggests that earlier stages in the budget process should be more opaque than later stages, because earlier stages occur behind a partial veil of uncertainty.

Accountability: Good and Bad One of the primary rationales for increased transparency in government is to reduce the monitoring costs borne by the principals (voters) to ensure that their agents (lawmakers) pursue their objectives. To be sure, even in the absence of open meeting requirements, one would expect to observe some level of transparency in democratic governments as lawmakers seek to increase voter confidence and therefore cause voters to allocate more resources to lawmakers’ control, but there is no reason to think this 28

29

See Jon Elster, Deliberation and Constitution Making, in Deliberative Democracy 97 (Jon Elster ed., 1998). Steiner and his co-authors find that (1) public debate produces more appeals to the common good than nonpublic debate and that (2) deliberative quality decreases as issues become more ideologically polarized. See Steiner et al., supra note 23, at 128–35. This suggests that public debate on divisive issues may be a particularly toxic combination, producing demagogic appeals to spurious common interests. Cf. Jon Elster, Legislatures as Constituent Assemblies, in The Least Examined Branch: The Role of Legislatures in the Constitutional State 191 (R. Bauman & T. Kahana, eds. 2006).

196

Mechanisms of Democracy

Optimizing Transparency: The Budget Process

mechanism will produce optimal transparency.30 Many who advocate strong transparency thus appeal to explicit or implicit principal-agent models. They argue that transparency will increase the ability of voters to hold elected officials accountable for their decisions because monitoring costs are reduced. Knowing that their behavior is observable, agents act differently, and principal-agent slack is reduced. Corrupt deals—meaning self-interested deals among agents, or between agents and third parties— will be deterred or chilled. All this assumes that legislators respond to electoral incentives, but even if one imagines legislators as good types or bad types in some fixed and exogenous sense, one might still defend transparency by reference to political selection effects. Putative candidates who value public office merely to further their self-interest might decide that a relatively transparent environment is not conducive to achieving their goals; thus, if transparency works as intended, only more trustworthy agents will seek elected office. Here too, characterizing the relationship between voters and elected officials as one of agency does not take sides on contested issues in the theory of representation; even on the Burkean or trustee account of representation (discussed in Chapter 4), it remains true that the representative is an agent, just one whose task it is, as an agent, to make decisions in light of the common good. However, there is an intrinsic cost to transparency, even from an agency perspective, and even subject to the point that not all bargaining is bad. To illustrate the cost concretely, consider the dual effects of open roll-call voting within legislatures. Open voting allows legislators to give third parties credible, because verifiable, commitments to vote in particular ways in return for bribes or in response to threats. With secret voting, by contrast, legislators cannot strike credible vote-selling bargains with the executive or interest groups, so the value of legislators’ votes to those groups declines. From the standpoint of voter-principals, the ability of legislators to credibly commit to sell votes to interest groups represents an agency cost insofar as the interest groups’ goals differ from the voters.’31 30

31

See Ferejohn, supra note 3; see also Alt, supra note 3. Alt et al., supra note 3, find that fiscal transparency in states increases the scale of government and gubernatorial approval, suggesting voters are willing to trust politicians with more resources when fiscal policy is transparent. It is a separate question whether an open market in votes provides legislators themselves any benefit. As Ferejohn points out, ex ante competition between candidates for legislative office

Institutional Design Writ Small

197

Chapter 

Even with secret voting, interest groups may pay for outcomes rather than actions, offering legislators payments conditional on favorable legislative decisions. Yet interest groups can always pay for outcomes, even with open voting, so secret voting at least reduces the value of the legislator’s vote by removing one dimension over which bargains can be struck. And paying legislators for legislative outcomes is senseless unless interest groups can identify the swing or marginal legislators, who alone control outcomes anyway. But the interest groups’ ability to identify swing legislators is endogenous to the voting practice; with secret voting, any legislator may claim to be marginal in order to win an interest-group payment, but no such claims will be credible. The upshot of these points is that open voting has cross-cutting or ambiguous effects on voters’ control of their legislative agents. On one hand, a switch from secret to open voting reduces agency costs by reducing the voters’ costs of monitoring their legislative agents. On the other hand, a switch from secret to open voting also creates an agency cost by creating an open market for legislative votes, allowing interest groups to divert legislators from voters’ goals. These two variables move in opposite directions, so the institutional-design question is how the two costs net out. The general point is that transparency produces not only good accountability that promotes informed democratic participation and voting, but also bad accountability that promotes interest-group capture—in the sense described above that is condemned by all varieties of democratic theory. Once decisionmaking is open to public view, it is not only voters in the lawmaker’s district or state who observe the behavior. Sophisticated and organized interest groups seeking undeliberated, pluralistically unjustified, and inefficient transfers, and their lobbyists, can also monitor legislative behavior that occurs in transparent institutions.32 Organized interests seeking transfers can weigh in while the legislation is being written; indeed, in a world of instant communication, lobbyists not only provide legislative language and expertise before mark-ups, they also

32

may dissipate the rents that legislators could otherwise obtain from vote-selling. Ferejohn, supra note 3, at 140 n.6. This effect merely reallocates rents from legislators to their interestgroup supporters; it does nothing to alleviate the agency loss to voters of legislative voteselling, and indeed exacerbates it insofar as increasing expenditures on (rent-dissipating) competition between candidates is itself socially wasteful. For other discussions of this cost to transparency, see Cass R. Sunstein, Government Control of Information, 74 Cal. L. Rev. 889, 896 (1986); Joseph E. Stiglitz, On Liberty, the Right to Know, and Public Discourse: The Role of Transparency in Public Life at 22 (1999) (Oxford Amnesty Lecture).

198

Mechanisms of Democracy

Optimizing Transparency: The Budget Process

email input to staff via BlackBerry while the meetings are occurring, ensuring that lawmakers hear their views at crucial moments and that they are aware that they are being watched closely. In a world of closed meetings, interest groups must rely on second-hand accounts and often cannot be sure what negotiations occurred. Contrast this with the situation of the ordinary voter. Although voters generally can now monitor legislator behavior more easily, because of open meeting requirements, any particular voter faces a significant collective action problem. The voter bears all the costs of monitoring, yet can do little, unilaterally, to punish a wayward representative. If one voter or group of voters succeeds in changing legislative behavior through timely input or other intervention, all voters in similar circumstances benefit without helping defray any of the activists’ monitoring costs. The chance that any will succeed in changing legislative outcomes is low. Of course voters organize themselves into interest groups, including good-government groups as well as transfer-seeking groups; moreover, democratic intermediaries like the press and would-be electoral challengers can use transparency to monitor lawmakers on behalf of voters and bring egregious examples to their attention near Election Day. But lawmakers know that it is not certain voters will pay attention to press stories, and ordinary citizens are not organized to mount effective action even if they do notice the news. Certainly, lawmakers will be concerned that particular kinds of behavior will affect voters’ decisions,33 but most decisions made in a budgeting context are not sufficiently noteworthy to produce electoral consequences. This contrast between good and bad accountability, to voters generally and to transfer-seeking interest groups, is deliberately overdrawn to clarify the issues. Some interest groups are good, either because they supply pluralistic political competition that increases efficiency and overall accountability, or because they act as informational intermediaries. An example of the latter case involves good-government groups who use their access to supply voters with cues about the position of particular legislators, intelligible accounts of budgeting decisions, or other useful information. By and large, however, the threat in the budget setting is that total transparency will increase damaging special-interest transfers, by exposing 33

See R. Douglas Arnold, The Logic of Congressional Action 82–87(1990). Empirical work suggests that voters do make electoral decisions based on budget policy. See, e.g., Robert C. Lowry et al., Fiscal Policy Outcomes and Electoral Accountability in American States, 92 Am. Pol. Sci. Rev. 759 (1998).

Institutional Design Writ Small

199

Chapter 

decisionmakers to intensive scrutiny and threats of electoral retaliation. In the budget setting, the groups that benefit most from fishbowl transparency and immediate access to details of ongoing negotiations are tightly organized groups seeking transfers to particular economic interests, rather narrowly defined. As I have argued, such transfers cannot be justified by any plausible conception of democratic accountability.

Two Proposals In light of these tradeoffs, the transparency of the budget process can be restructured to optimize official impartiality and the quality of budget deliberations, producing the minimum necessary harm to good accountability and the other values that transparency promotes. The guiding principles of design are (1) opacity becomes more costly as the budget process proceeds and legislators gain more specific information about how decisions will affect their interests; (2) where disclosure does occur, it should be delayed if doing so will maximize the benefits of information to voters and constituents generally while minimizing informational benefits to organized interest groups. Given these principles, decisionmaking at the stage of overall budget allocation should be opaque, while decisionmaking at later stages should be made transparent, but only through delayed disclosure.

First Proposal: Overall Allocation Should be Opaque Within Congress, the first basic stage of the budgetary process is the allocation of overall spending levels across budget categories or functions. The allocation is developed by the budget committees in both houses and then formalized through a concurrent budget resolution, which sets out a five-year budget plan. The concurrent budget resolution is the internal congressional vehicle that sets forth macro-budgetary objectives: it sets spending limits for discretionary programs; determines the amount of revenue that should be raised in taxes; reveals congressional priorities by dividing resources among various budget functions, which are the major categories of governmental activities; and provides for the debt limit. It does not require that any particular programmatic changes be made to achieve its broad goals, although the budget committees often include nonbinding recommendations for specific changes or members add such suggestions on the floor. Filling in the details of the budget 200

Mechanisms of Democracy

Optimizing Transparency: The Budget Process

resolution is the province of the appropriations committees, the substantive committees with jurisdiction over entitlement programs, and the taxwriting committees.34 This concurrent resolution stage of the budget process is in some ways analogous to the process of constitution-making. It puts in place a framework within which bargaining can occur at subsequent stages; in effect, the resolution sets out constraints and then delegates authority to tax-writing committees, appropriations committees and some other committees to make specific policy choices within those constraints, just as a constitutional convention sets out constraints within which later legislatures and agencies work. The constraints of the budget resolution are enforced through internal parliamentary devices such as points of order, some of which can be waived only with supermajority votes in the Senate. The analogy is contestable; perhaps the real analogue to the constitution-making stage is the enactment of framework statutes such as the 1974 Budget Act.35 The only point of the analogy, however, is that the overall budget allocations are established behind a partial veil of uncertainty about how macro-level decisions will affect legislators’ specific interests. At the stage of overall allocation, it is simply unclear what particular programs and appropriations will emerge from the later stages of the budget process, and hence unclear exactly how legislators’ interests will be affected by large-scale choices. The veil of uncertainty here is only of medium thickness, of course; certainly, members think about the particular programs and tax provisions that will emerge from the entire budget process when they make macro-budgetary decisions in the concurrent budget resolution.36 However, the overall allocations are reminiscent of constitution-making because they make value choices at a relatively high level of abstraction, choosing overall priorities—more for guns or butter?—and then leave it to later periods to connect those priorities to particulars. Moreover, there is a separation of powers or responsibilities here: large-scale allocative decisions and priority-setting are done by budget committees, whereas specific spending decisions at later stages are made by different committees with different memberships. This difference 34 35 36

See Allen Schick, The Federal Budget: Politics, Policy, Process 110–17 (rev. ed. 2000) (describing the structure of the concurrent budget resolution). Pub. L. No. 93–344, 88 Stat. 297 (1974). See Elizabeth Garrett, Rethinking the Structures of Decisionmaking in the Federal Budget Process, 35 Harv. J. on Legis. 387, 409–15 (1998).

Institutional Design Writ Small

201

Chapter 

in control contributes to the uncertainty afflicting members at the earlier stage, thickening the veil. At the overall allocation stage, then, legislators will be somewhat uncertain about how decisions affect their interests, which reduces the risk of self-interested bargaining and thus in turn reduces the principal benefit of transparency, which is to deter self-interested bargaining. The remaining risk, however, is that transparency will induce bad deliberation, through posturing, premature hardening of positions, and excessively principle-ridden debate. For the same reasons that closing off the federal constitutional convention was plausibly a successful decision—the Framers lacked full information about how to pursue their own interests, while the lack of transparency produced better deliberation—so too it seems plausible that large-scale allocative choices might best be made behind closed doors. To be sure, the outputs of this opaque process will eventually be made public, when the concurrent resolution is enacted and the committees begin connecting allocations to particular programs, but the deliberation in the budget committees that produces the outputs need never be publicized. Finally, at the allocation stage there is a connection between the two senses of “transparency” identified above—transparency as intelligibility and transparency as open decisionmaking process. The budget resolution is a relatively intelligible document. It allocates money to general government functions, not to detailed programs or tax subsidies that are hard for voters, nonspecialists, and outsiders to fully understand; a decision about how government trades off guns against butter is less detailed and less obscure than the decision about how to design the various programs to deliver the guns or butter. The relative intelligibility of the budget resolution strengthens the case for opacity during the committee deliberations that produce the resolution; any reduction in good accountability is less here than in other settings. Conversely, the budget resolution, because it proceeds at a high level of generality, tends to be the majority party’s statement of its principles. This makes any public compromise particularly costly for the minority, which in turn encourages posturing. Here opacity can lower the reputational and political stakes of disagreement, helping to avert bitter strife over principles and easing negotiations at later stages of the budget process. At those later stages, the calculus changes. As legislators gain highly specific information about particular programs, as the fog of uncertainty 202

Mechanisms of Democracy

Optimizing Transparency: The Budget Process

thins out, there is increasing reason to fear that opacity will produce unchecked self-interested bargaining. Conversely, the principal cost of transparency—the poisoning of deliberation—diminishes in any event, just because there is less deliberation of any kind occurring. Bargaining over specifics comes to the fore, and transparency can help to chill or deter the most self-regarding bargains. The legislative work in the wake of the concurrent budget resolution is drafting appropriations bills to allocate money within the budget functions to various specific programs, writing revenue legislation with detailed tax subsidies, or constructing or revising actual entitlement programs. This is the prototypical arena of logrolling; the discussions in these committees tend to be less partisan and more pragmatic, at least for the vast majority of programs that are below most voters’ radar screens. Organized interests seeking transfers are vitally concerned in all aspects of appropriation, entitlement and revenue bills because they actually receive their benefits from this legislation. In contrast, the concurrent budget resolution merely defines the likely universe of resources available to fund the actual programs and tax subsidies.37 How do these prescriptions match up with the current rules? Committee meetings considering appropriations bills, tax proposals, omnibus budget reconciliation acts, and entitlement programs are where much concrete bargaining takes place, and these hearings are generally subject to strong transparency requirements. This is as it should be given the logic set out here. However, that logic suggests important changes in the framework shaping deliberation at other stages. First, committee deliberations at the macro-allocation stage culminating in a concurrent resolution should be opaque, contrary to current practice. Second, budget summits, where the ultimate deals are struck in case ordinary processes break down, are currently opaque; but they should take place in the public eye. Budget summits are entirely matters of logrolling and pragmatic compromise as the two branches, and sometimes the two parties, hammer out a bargain to keep the government running and avert a fiscal disaster.

Second Proposal: Delayed Disclosure So far we have addressed only the question whether decisionmaking that takes place at a given stage of the budget process should be transparent 37

Id. at 415–16, for a discussion of the difference in intensity of interest group activity at the concurrent resolution stage and at later stages of budgeting.

Institutional Design Writ Small

203

Chapter 

at all. A separate issue involves the timing of disclosure, assuming disclosure should occur. The second proposal is that delayed disclosure can optimize across good and bad forms of accountability. The general tradeoff between good and bad accountability arises from the existence of two audiences—general constituents and organized narrow interests—with frequently diverging objectives and with differing ability to monitor lawmakers. Legislators know that the two audiences have vastly different capabilities to monitor and punish, so transparency operates to reduce principal-agent slack between organized interests and lawmakers more than it reduces slack between voters and their elected representatives. In principle, the solution would be to keep organized interests in the dark about legislative behavior while fully revealing it to voters. Although that solution is impossible—once information is provided to voters, it is provided to everyone—it may be possible, in some cases, to deprive transfer-seeking organized interests of the information while the deals are being struck and interest group influence is most problematic, while at the same time ensuring that voters have access to information before they cast their ballots. Delayed disclosure is a small-scale design mechanism that may provide many, if not all, of the accountability benefits of transparency while mitigating its accountability costs.38 A similar technique is used in some countries’ open records laws, which preserve confidentiality of some documents during official decisionmaking processes but allow broad dissemination at a later time to enable voters to learn about the inputs relevant to a decision once it has been announced.39 It is also the technique used since 1993 by the Federal Open Market Committee of the Federal Reserve System. Minutes of Open Market Committee meetings in which policy decisions are made are released three weeks after the meeting, and full transcripts from meetings held in one year are released after a delay of five years.40 Laws embargo some of the official papers of Presidents of the 38

39 40

The idea of delayed disclosure—or to put it differently, temporary secrecy—is not new. Others who write about fiscal transparency identify the timing of disclosure as a key variable in the success of open meetings in increasing accountability without disrupting deliberation that occurs best out of the public eye. See, e.g., Stiglitz, supra note 32, at 22; David Heald, Fiscal Transparency: Concepts Measurements and UK Practice, 81 Pub. Admin. 723, 746 (2003). However, none has proposed a formal system of delayed transparency as a solution for the problem of dual audiences with divergent interests and different monitoring capabilities. See Andrea Prat, The Wrong Kind of Transparency, 95 Am. Econ. Rev. 862, 869–70 (2005). 12 C.F.R. § 261b.11 (1979).

204

Mechanisms of Democracy

Optimizing Transparency: The Budget Process

United States for later release,41 while in a less formal manner justices of the United States Supreme Court often delay release of their official papers until well after their retirement or death.42 The basic scheme might work as follows. A system of delayed disclosure would allow legislative bargaining in appropriations and taxwriting committees, within the framework of a concurrent budget resolution, to occur behind closed doors. Transcripts of deliberation would be kept, and any documents generated during the process would be retained for later release. If ongoing committee deliberations can be kept secret, then interest groups’ ability to affect the details of spending and revenue decisions and to monitor lawmakers would be reduced. Permanent secrecy is not desirable, however, because it would insulate corrupt deals from publicity and eliminate the deterrent effect of disclosure. Thus transcripts and other documents of committee deliberations must be fully disclosed after the budget has passed but before Election Day. Knowing that their discussions and deals will not be kept secret forever, lawmakers would have strong incentives to refrain from entering into bargains that could not withstand the sunlight of public disclosure. Moreover, because the output of the committee deliberations—the final mix of appropriations, entitlement spending and taxes—would be public and floor deliberations would be open, obviously corrupt and questionable deals would likely be discovered and scrutinized as the budget was being developed. The basic rationale for delayed disclosure here is to maximize the benefits of information to voters generally while minimizing the benefits of information to interest groups. This general aim in turn has two components. First, delayed disclosure prevents interest groups from bringing immediate pressure to bear on legislators and other policymakers while deliberations proceed. To be sure, legislators may anticipate retaliation from interest groups who later learn that their demands have not been met. Under delayed disclosure, however, it is hard for interest groups to make effective real-time interventions, to the extent that they are acting in the dark. Delayed disclosure gives decisionmakers breathing space and room to maneuver.

41 42

Presidential Records Act of 1978, 44 U.S.C. § 2204 (2000). See, e.g., News from the Library of Congress, Papers of Supreme Courts Justice Harry A. Blackmun Opened for Research at Library of Congress, (Mar. 4, 2004), http://www.loc.gov/ today/pr/2004/04–041.html (last visited Nov. 10, 2006).

Institutional Design Writ Small

205

Chapter 

Second, delayed disclosure can be structured to mitigate the threat of future interest-group retaliation at election time. How can electorally relevant information be channeled to voters but not to interest groups? It is unclear whether this trick can in fact be accomplished, as I emphasize below; but it is possible that it can, because of the structure and timing of the modern political campaign. The key point is that interest groups seeking to maximize their influence must act earlier in the election cycle than voters, who can wait until Election Day to pass judgment on their agents. Interest groups must target campaign funds, advertising and other resources at earlier stages of the electoral cycle; indeed, such resources are often most influential at the earliest stage, as parties are assessing the pool of potential candidates to decide which will become actual candidates, and as incumbents seek to discourage serious challengers. If interest groups must act before voters, then delayed disclosure can force interest groups to act in the dark while voters act with sufficient information. To be sure, the electoral game is repeated, but one should not overlook that the players and payoffs change over time. Interest groups will be able to use the information disclosed before Election Day in one political cycle to allocate resources and enforce threats or promises in a later cycle. But the political discount rate—the rate at which politicians discount the future—will assure that an interest group’s threat to punish the politician two elections hence will be less impressive than a threat tied to the next election. New issues will intervene in the meantime, creating new alliances; the threat may be forgotten as interests change and emotions decay over time. Delayed disclosure effectively lengthens the legislators’ term of office, but only for groups enforcing bad accountability. Such groups would plausibly prefer more frequent elections in order to maintain a tighter hold on legislators. For purposes of good accountability to voters, however, delayed disclosure does not change the effective frequency of election; voters would still receive information in time to vote competently in the next succeeding election. Moreover, there is a positive (albeit low) rate of turnover even among federal legislators, for reasons unrelated to interest group activity. The consequence is that in some fraction of cases, the legislator who angers the interest groups now will simply not be around to suffer punishment two elections hence. Legislators who know that they are in either their last or second-to-last terms of office—and these tend to be highly senior and thus influential—cannot be affected by the threat. Overall, 206

Mechanisms of Democracy

Optimizing Transparency: The Budget Process

because elections are a repeated game, delayed disclosure cannot eliminate the ex ante threat of punishment by interest groups; but inevitable turnover and the political discount rate means that delayed disclosure can at least reduce the threat, thereby improving matters. The devil is in the details, of course. The hard part is determining the right time to release the details of the budget negotiations. The delay should be long enough to reduce interest groups’ ability to pressure lawmakers as they make decisions, but the information must be publicized early enough to influence electoral outcomes. This latter requirement means that the information must be available well before the actual elections because it will influence the decisions of other viable candidates, who must decide whether to challenge incumbents who participated in unsavory or questionable deals. The information, then, must be made public not only before the primary election but also before the time candidates file to appear on the ballot. Disclosure must also occur early enough for challengers to begin to raise money necessary for a successful campaign. Such early disclosure also increases the power of interest groups because they can use the information to direct campaign resources to lawmakers who work energetically on their behalf or to the opponents of those who reneged on deals. Disclosure after the period when interest groups’ campaign contributions can make the most difference is the optimal time for publicity, but revealing information that late in the campaign might also reduce the ability of voters—and intermediaries like the press and challengers—to use the information appropriately. This is an optimization problem; the problem is to find the disclosure point that maximizes the difference between the benefits of information to voters and the benefits of information to efficiency-reducing interest groups. The available information about costs and benefits is too crude to determine exactly where the optimum is located, but a reasonable estimate is that disclosure could be delayed until a few weeks before first the primary elections for congressional seats. For many House seats, the primary is the only possible venue for competition, so continuing secrecy past this stage of the electoral cycle will deny voters a meaningful chance to act on the information. Uncertainty abounds, and this date is only a guess—it may be too late in the electoral process to allow serious challengers to emerge, and it may also be too early, in that it would allow interest groups substantial influence in the campaign for the general election, Institutional Design Writ Small

207

Chapter 

although at least such groups will have reduced sway in the selection of the two major candidates. As throughout the book, my aim here is to illustrate the conditions under which small-scale democratic mechanisms would produce desirable effects. In this setting, the relevant conditions might not hold. It might turn out, as a factual matter, that there just is no disclosure point that is both (1) sufficiently late so as to hamper the ability of interest groups to retaliate against legislators in the relevant election cycle and also (2) sufficiently early as to give would-be challengers and informational intermediaries the material they need to inform voters. But the current system, in which transfer-seeking interest groups get immediate information from budgeting committees, is an extremist solution, one that is most unlikely to be the best possible arrangement from a democratic perspective. Quite plausibly some moderate delay in disclosure would improve matters, both to give legislators breathing room and to hamper interest-group retaliation, although it is difficult to say in the abstract and with any precision how much delay would be best. Finally, a clarifying point is in order. The suggestion that some deliberations be kept opaque and some be subject to delayed disclosure does not mean that the former will be kept secret permanently. In a long enough time frame everything is disclosed, even presidential papers and highly classified documents. For present purposes, following the logic set out above, “delayed disclosure” just means disclosure that occurs after legislative deliberations but before the next following election (after the budget is adopted). “Opacity,” then, just means that the relevant deliberations are not disclosed in time for the next following election. How long opacity should be preserved is an open question. When the Federal Open Market Committee adopted the five-year delay for disclosure of meeting transcripts, some argued that the delay period was not long enough to ensure that deliberations were unaffected by the expectation of publicity, and that fear appears to be accurate. An important finding is that Committee deliberations after 1993 include fewer dissents, and appear to consist more of “canned” statements, as opposed to the genuine discussion that occurred when participants believed their views would never be disclosed.43 43

Meade & Stasavage, supra note 26, at 4–5. Although the Open Market Committee did not disclose transcripts before 1993, it kept full transcripts beginning in 1976, so it is possible to measure the effect of delayed disclosure on deliberations.

208

Mechanisms of Democracy

Optimizing Transparency: The Budget Process

Second-Best Problems: Costs and Political Constraints So far I have discussed the optimal or first-best framework for transparency in the budget process, on the assumption that such a framework could be imposed and enforced by an impartial institutional designer. Let us now turn to a range of second-best problems. Such problems arise from two related sources: institutional and political constraints on implementation, and the motivation of participants. Is a regime of the sort sketched here institutionally and politically feasible? My answer will be that there are obstacles, but none seems insuperable. The hope is that the discussion will shed light on the conditions under which transparency-adjusting mechanisms of the sort discussed above, and framework proposals generally, will prove both possible and desirable. In general, although the problems are real, I do not believe that the constraints are so tight as to rule out adoption of the small-scale changes required by the transparency framework suggested here.

The Supply of Optimal Rules The first problem arises on the supply side. Even if the framework sketched here is desirable, which actors, if any, will be motivated to adopt it? Two points are worth highlighting. First, it is valuable to identify (even approximately) the optimal structure of transparency rules. Absent that knowledge, well-motivated reformers do not even know in which direction to push; in that case the question whether it is possible to attain ideal reforms does not even arise. So the principal aim here is to sketch desirable reforms, at the level of first-best. As discussed in the Introduction, theorists can only propose; politics will determine the shape of the political constraints, but it is not systemically desirable for theorists to selfcensor based on an estimate of whether their proposals will pass through the political filter. Furthermore, with particular relevance to transparency, the analysis here should draw into serious question reformers’ typical reaction to problems of accountability: increasingly aggressive disclosure rules aimed at immediate and all-consuming transparency. Because transparency has costs and benefits—because it produces both good and bad accountability, and has ambiguous effects on the relationship between accountability and deliberation—a more nuanced approach is likely to yield better results. Second, the supply-side problem is frequently overcome within Congress. Chapter 2 examined some self-defeating proposals for framework Institutional Design Writ Small

209

Chapter 

statutes, especially ones intended to govern the war powers and emergency powers of government. But consider the range of successful framework statutes by which legislators act to dampen interest-group influence and to check the (future) self-interested motivations of other legislators, and perhaps even of themselves. The Electoral Count Act of 188744 was intended to dampen partisanship and self-interest in the context of disputed presidential elections; the Base Realignment and Closure Acts,45 first passed in 1988, reduce lawmakers’ ability to protect military bases important to the economy of their district but not vital to the nation’s defense; the Line Item Veto Act of 1996, later invalidated by the Supreme Court,46 attempted to enlist the executive as an outside enforcer who would deter budgetary logrolling; finally, aspects of the federal budget process make it more difficult for members of Congress to freely engage in distributive politics.47 As in Chapter 1, an important mechanism at work here is the veil of uncertainty. Legislators’ uncertainty about how decisions will affect their interests not only helps to determine optimal transparency rules, as discussed above, but also helps to ensure that impartial transparency rules will be enacted. At any particular time, legislators can and do agree on framework statutes that improve future decisionmaking for all concerned. In such cases agreement is possible because it is not clear, at the time of adoption, exactly whose interests will be benefited and whose hurt by the application of the framework rules in future periods. It is an open question whether such a mechanism might allow enactment of a framework statute implementing the transparency structure I have proposed. On the other hand, Chapter 2 explained that impartiality trades off against motivation and commitment problems. I criticized various framework proposals, such as John Hart Ely’s and Bruce Ackerman’s proposals for framework statutes governing warmaking and emergencies, on the grounds that (1) precisely because of their impartiality no critical mass of supporters would be motivated to adopt them; and (2) the very circumstances of their enactment (if it did occur), and their subject matter, would make them particularly vulnerable to repeal by later majorities. How do 44 45 46 47

24 Stat. 373 (1887). For the first such act, see Defense Authorization Amendments and Base Closure and Realignment Act, Pub. L. No. 100–526, 102 Stat. 2623 (1988). See Clinton v. City of New York, 524 U.S. 417 (1998). See Garrett, supra note 36 (providing examples of framework laws).

210

Mechanisms of Democracy

Optimizing Transparency: The Budget Process

the proposals offered here fare along those two dimensions—enactability and durability? The answer, I believe, is that they fare better on both dimensions, but especially on the second. The same motivation-sapping effect of impartiality works on both proposals, so enactment is a problem for both. But the smaller the costs of the change, the less motivation there must be to push the proposal over legislative hurdles. The adjustments to legislative transparency offered here are in the spirit of institutional design writ small; they are much less sweeping than, say, Ackerman’s ambitious proposal. On the score of enactment and transition costs, the comparison works in favor of the transparency framework. The difference on the second dimension is even more pronounced. The problems of commitment or time-inconsistency are far less serious for framework statutes covering, not wars and emergencies, but ordinary domestic issues such as the budget process. In such contexts, the status quo bias of the legislative process is most likely to protect the impartial enactment from repeal in later periods, if anything can, short of outright constitutional entrenchment. Moreover, process rules of the sort proposed here have fewer directly discernible effects on particular distributive outcomes than do substantive restrictions; by contrast, Ely’s proposal for a tighter War Powers Resolution is mostly substantive, while Ackerman’s proposals are a mixed bag of voting rules and substantive protections for civil liberties. Finally, note that any advantage on the dimension of durability itself converts, indirectly, into an advantage on the dimension of enactability. As Porat and Yadlin point out,48 a major reason that impartial enactments are not adopted is that political coalitions in the first period anticipate the threat of later repeal. Reduce that threat, and one disincentive to initial enactment is removed. Overall, the minimal hope is that the proposals offered here—opacity at some stages and delayed disclosure in others—might garner more sincere support from lawmakers than the current tendency toward immediate disclosure of all deliberation. It seems likely that at least some legislators have supported full disclosure only in response to episodic constituent outrage following political scandals, and would be interested in a more nuanced approach. Legislators might prefer to conduct some of their budget 48

See Ariel Porat & Omar Yadlin, Promoting Consensus in Society Through Deferred-Implementation Agreements, 56 Univ. of Toronto L. J. 151 (2006).

Institutional Design Writ Small

211

Chapter 

deliberations out of the public eye—or at least away from the immediate pressure of narrow interest groups and their lobbyists. Whether such a proposal could be sufficiently explained to constituents, given high current levels of distrust for elected representatives, is an open question. But there is no reason to think that it should be ruled entirely out of court.

Circumvention A basic question is whether transparency is ever feasible. Sometimes transparency requirements seem futile, because the requirement can be circumvented through informal arrangements. Perhaps all relevant actors desire to preserve opacity and can collude to keep real decisionmaking out of the public eye, in a way that is difficult for external enforcers to monitor and prevent. At the extreme, perhaps transparency rules are a sham, because they are systematically or at least frequently evaded. The evasion phenomenon is real, but any conclusion that transparency requirements are ineffectual would be far too cavalier. Almost any institutional rules can be circumvented, but circumvention is costly; the higher the costs, the more effective the rule. Where transparency is imposed in order to chill self-interested bargains, the fact that selfinterested bargains must be attempted through furtive whispers behind closed doors may be a sign that the transparency rules are working just as intended, rather than a symptom of failure. Where self-interested actors must act by indirection, the costs of striking bargains will rise, and fewer of them will be struck.

Leaks The circumvention problem is that actors can opt for secrecy in violation of rules prescribing transparency. The reverse problem is that of leaks, where actors can opt for transparency in violation of rules prescribing secrecy. Perhaps uncoordinated leaks, difficult to deter (especially where the leaker is a legislator or staff member protected by the Speech and Debate Clause),49 will undermine opacity, whatever the formal rules say. Recall in this connection Ferejohn’s claim that legislators and other agents have individual-level incentives to supply transparency to constituents,

49

U.S. Const. art. I, § 6. See Gravel v. United States, 408 U.S. 606 (1972) (applying Speech or Debate Clause to Senate staffer who leaked Pentagon Papers to media).

212

Mechanisms of Democracy

Optimizing Transparency: The Budget Process

even if doing so would be undesirable from the social point of view.50 Reporters and lobbyists will expend significant resources to ferret out the information before it is released, which could result in unproductive jockeying for position as officials leak valuable information to favored members of the press or lobbyists with connections and clout.51 It is much too casual, however, to assume that leaks will inevitably undermine any regime of opacity. In both the executive branch and Congress, most secrets are kept, most of the time. Leaks make news because they occur against a less salient backdrop, that of a system in which national security intelligence, privileged and confidential documents and deliberations, and other information are routinely kept secret. The problem of leaks is real, but one must give specific and concrete reasons to think that opacity in one domain will be unsustainable, when opacity in other domains is in fact consistently sustained. Recall, as an important mechanism that reduces incentives to leak, that noncredible leaks bring no benefit to the leaker. Any legislator can come out of a closed session and report to interest groups that he advocated their pet projects, but the interest groups know that the legislator knows that the interest groups have no way to verify the claim. These remarks are general; let us focus concretely on the problem of leaks in various contexts where delayed disclosure might be adopted. The main challenge for delayed disclosure, and the main variable that differs across these contexts, is the size of the group that would have access to the confidential information at the time that transparency could adversely affect deliberations. The fewer people involved in decisionmaking, the easier it is to detect and prevent leaks. Let us begin with those in which leaks are hardest to prevent and move to those where prevention is less costly. Floor debate. Debate on the floor of either House presents the extreme case where preventing leaks would be quite simply impracticable, because of the sheer number of people involved—not just lawmakers themselves, but their personal staffs and the staff of Congress. For that reason among others, I have not suggested that floor deliberations on the budget should ever be closed to the public—even during deliberation on concurrent budget resolutions and macro-budgetary policies.

50 51

See Ferejohn, supra note 3. See Stiglitz, supra note 32, at 11–12.

Institutional Design Writ Small

213

Chapter 

Congressional committees. Committees often have large memberships and substantial staffs, making it unclear whether delayed disclosure is a workable solution when applied to committee deliberations. Even if only the members of the committee and their staffs were privy to the details of deliberations, then several dozen, perhaps even more, people would have to maintain confidentiality in the face of temptations to disclose the information to interested parties. These parties can entice disclosure through implicit promises of campaign contributions and other help to lawmakers and through hints of future employment to staff. Because records of the deliberations would be maintained for future release, people involved in the negotiations might be able to credibly provide confidential information to interested parties. These considerations are hardly decisive, however. For one thing, there are legal penalties for showing nonpublic records to outside parties, such as interest groups or journalists, so the would-be leaker faces a dilemma between issuing a credible leak at the risk of prosecution, and avoiding the risk of prosecution by limiting himself to cheap talk. For another, the suggestion here is for complete opacity only at the stage of overall allocation by budget committees developing a budget resolution; at the stage of actual dickering, I merely propose that disclosure be delayed. At the former stage, the incentive to leak details to special interest groups is much reduced, because there are fewer details to leak and because it is less clear how particular interests will be affected by macro-level allocation. The very condition that makes opacity desirable—that actors work behind a partial veil of uncertainty—also reduces the benefits to be gained from leaks. At the latter stage, the incentive to leak is greater, but secrecy need only be temporarily enforced. Most generally, many congressional committees do manage to maintain secrecy or opacity for many issues. Absent some further account suggesting special reason to fear leaks in the budget setting, there is no obvious reason that a regime of committeelevel opacity in budget matters would be systematically infeasible, although occasional leaks would surely occur. Budget summits. Here leaks can rather easily be monitored and deterred; the small number of participants makes it difficult for the leaker to hide in the crowd. Historically, budget summit participants have usually managed to keep their deliberations confidential against the threat of internal leaks. Although some contemporaneous news stories surrounding budget summits include unnamed sources floating trial balloons or strategically 214

Mechanisms of Democracy

Optimizing Transparency: The Budget Process

leaking information that harms their opponents,52 the political and fiscal conditions that lead to an interbranch summit are typically serious enough that participants understand that secrecy is in their best interests. Details may become available some time after the summit as participants write memoirs or participate in interviews for analysis in books, magazine articles, or scholarly research. But leaks from these entities are unlikely to be prevalent because enforcement is much less challenging in the context of small groups. Overall, the technique of delayed disclosure should be seriously considered as a way to achieve the goals of traditional disclosure while responding to the concern that interest groups and lobbyists are the true beneficiaries of open meetings. The strategy is least likely to work when the group that must keep the secret is so large that it is difficult to detect who leaks the information, and when the leaks can be accompanied by credible proof of the information’s accuracy. Small groups such as budget summits should pose little problem; congressional committees are a larger challenge, but probably not an insuperable one.

Transparency: An Overview Finally, let us stand back from the details to make sure the larger picture is in focus. My principal aim has been to sketch an optimal framework for transparency in the federal budget process. The details of the framework are complex, but these are its outlines: less transparency should obtain at earlier stages of the budget process, when participants are deliberating over broad goals, and the risk of posturing in the glare of transparency is at its height. Conversely, more transparency should obtain at later stages, when actors are dickering over specific programs, and the risk of corrupt bargains is at its height. Where transparency is indicated, delayed disclosure should be used where feasible to maximize the benefits of transparency for voters while minimizing its benefits for interest groups. Small-scale design mechanisms of this sort are optimizing devices that, by adjusting transparency, can produce large benefits in accountability and in the quality of democratic deliberation. 52

See David Hoffman & John E. Yang, Bush Repeats “No Preconditions” for Deficit Talks, Wash. Post, May 11, 1990, at A24; David S. Broder, Optimistic Democrats Seek Issues, Wash. Post, Sept. 16, 1990, at A8; Robert Pear, G.O.P. Feud in House Stalls Budget Talks, N.Y. Times, July 22, 1997, at A15.

Institutional Design Writ Small

215

This page intentionally left blank

Chapter 7

Optimizing Deliberation: Constitutional Issues in Congress

T

he United States Congress frequently deliberates upon and decides questions of constitutional interpretation, and many of those decisions are immune from subsequent judicial review, as a matter of law or as a matter of practice. Constitutional interpretation in representative legislatures must lie near the core of any account of democratic deliberation. It is therefore crucial to ask whether and how Congress’s interpretive and deliberative capacities can be improved—understanding improvement in an uncontroversial way, relative to a broad range of democratic theories. This chapter suggests several mechanisms for improving Congress’s deliberative capacities where constitutional questions are involved. The focus is squarely on questions of institutional design. The hope is to find democratic mechanisms that enable Congress to perform its constitutional responsibilities well, whatever those responsibilities are. Given some allocation of constitutional authority between Congress, the President, the judiciary, and other constitutional actors, how can the rules

Institutional Design Writ Small

217

Chapter 

that structure congressional operations be arranged to produce the right quantity and quality of congressional deliberation on constitutional questions? Throughout, we should understand “constitutional” in a broad sense, to include not only lawyers’ questions about the meaning of particular constitutional texts, judicial precedents, or historical episodes, but also larger arguments from tradition, national commitments and values, and abstract principles of political morality. In this extended sense, constitutional deliberation encompasses much of what is important about democratic deliberation more generally. Let us begin by situating the problem against the background of relevant ideas in constitutional theory, the public choice strand of political theory, and empirical political science; by making explicit some assumptions about legislators’ behavior; and by giving more precise content to the goal of improving Congress’s deliberative capacities where constitutional issues are involved.

Legislators’ Behavior: Three Views The concern to improve Congress’s deliberative and constitutional capacities assumes that at least some legislators, some of the time, give some decisional weight to reasoned constitutional argument, in the broad sense of “constitutional” I have described. That premise has been challenged by professedly “realist” strands of political science and public choice, but the challenge rests on empirical presuppositions about legislative behavior that turn out to be untenable. The starting point is the question of what ends legislators pursue. The literature contains three distinct answers to this question. As we will see, the first two are excessively optimistic and excessively jaundiced, respectively.

Legislators Act in the Public Interest Early discussion of Congress’s constitutional performance assumed an optimistic picture of legislator-statesmen who act strictly to promote their understanding of the common good. The public-interest view is the positive counterpart to Burke’s trustee model of representation, a normative stance that sees a good representative as one who exercises independent judgment for the common weal, rather than simply acting so as to satisfy the

218

Mechanisms of Democracy

Optimizing Deliberation: Constitutional Issues in Congress

preferences of a (geographically defined) constituency. The public-interest assumption persisted for a remarkably long time in the constitutional-law literature. But naïve forms of optimism about legislators have today been displaced, in most sectors of the constitutional-law academy, by a far more skeptical account of legislative behavior.

Legislators Maximize Their Chances of Reelection, or Personal Gain More Broadly Defined The technical public-choice literature generally models legislative behavior on the explicit assumption that legislators’ sole goal is to maximize their chances of reelection. This is largely a methodological assumption, one dictated by the positivist aspiration of public choice to render testable predictions, which are unattainable without a precise maximand. As explained below, the assumption that legislators act solely to maximize their chances of reelection is clearly mistaken if it is treated as an empirical claim, but it is unobjectionable as a methodological premise within its proper domain.1 But there is also a broader, looser, and more objectionable strand in the non-formalized political science literature. This strand is harder to define; these works, sometimes called “realist,” principally share a common atmospheric that describes legislators as maximizing personal gain in a crudely venal sense. All legislative behavior, on this view, is rooted in relatively tangible forms of self-interest, such as the quests for money, fame, and power. Realists typically ignore broader motives, such as personal satisfaction from justified accomplishment or the promotion of ideological goals. Some of this work even suggests that all constitutional discourse within legislatures (and maybe generally) is a sham, a cover for selfregarding motives and tactics. Despite its hard-headed appeal, the “realist” view either represents a pre-empirical methodological commitment or else turns out to be indefensible. To the extent that it makes an empirical claim, it has been falsified outright by empirical work in mainstream political science. That work advances a third, intermediate view. 1

Thus, David Mayhew carefully explains that certain features of legislative behavior may best be explained as if legislators care only about getting reelected. See David R. Mayhew, Congress: The Electoral Connection 45–49 (1974). That the premise is methodological, not empirical, has not always been remembered by subsequent public-choice scholars.

Institutional Design Writ Small

219

Chapter 

Legislators Pursue a Complex Set of Public and Personal Goals The optimistic public-interest view and the skeptical realist view both represent implausible extremes. The best account takes a more nuanced view of legislators’ motivations. Richard Fenno’s classic study found that most legislators pursue a variety of ends simultaneously, trading goals off against one another and giving no goal overriding priority.2 Although ensuring reelection is one such goal, along with gaining colleagues’ respect, another prominent goal is legislators’ desire to promote their vision of the public interest. Reelection is, for most legislators, a necessary means to their preferred goals of influencing public policy for the better and accumulating prestige with colleagues; and it is not the case that legislators will trade everything else to maximize their chances of reelection. In general, empirical work in mainstream political science describes legislators’ diffuse ideology—legislators’ beliefs about morals, justice, good public policy, and other intangibles—as a far more powerful determinant of legislative behavior than the realist tradition acknowledges.3 The subsequent discussion follows this account by assuming that some legislators sometimes treat constitutional argument as one indicator of the “public interest” or “good public policy,” which in turn has some weight as against legislators’ other interests, goals, and aims. This empirical starting point is not necessarily inconsistent with the methodological assumption in the technical public-choice literature that legislators maximize their chances of reelection in preference to all other aims. It is crucial to understand that even if legislators are solely oriented to reelection, several mechanisms might cause legislators to give some weight to constitutional argument. First, some constituents might desire a representative who takes constitutional argument seriously, and might punish a representative who 2

3

Richard F. Fenno, Jr., Congressmen in Committees (1973). Subsequent political science work has confirmed this account. See, e.g., John W. Kingdon, Models of Legislative Voting, 39 J. Pol. 563, 569–70 (1977). See generally John W. Kingdon, Congressmen’s Voting Decisions (3d ed. 1989) (discussing legislators’ beliefs about public policy as one determinant of their voting decisions); William K. Muir, Jr., Legislature: California’s School for Politics 105–37 (1982) (asserting that the nature of politics, among other factors, causes legislators to acquire and share knowledge). James B. Kau & Paul H. Rubin, Economic and Ideological Factors in Congressional Voting: The 1980 Election, 44 Pub. Choice 385, 385 (1984). See generally Jerrold E. Schneider, Ideological Coalitions in Congress (1979).

220

Mechanisms of Democracy

Optimizing Deliberation: Constitutional Issues in Congress

appears wholly opportunistic about the Constitution. “[A] senator may gain political points by being a ‘person of principle,’ rejecting constituent demands for substantive legislation because it violates the senator’s understanding of the constitutional scheme.”4 Second, there are many legislators who enjoy slack in their agency relationship with constituents, usually because they are from safe districts and reelection is not a serious concern. Even on the public-choice premise, those legislators will shift to pursuing other aims, and one of those aims will be to implement the legislators’ conceptions of good public policy, including good constitutional law. Third, constitutional argument is given weight by the “civilizing force of hypocrisy.”5 Even a wholly self-interested legislator cannot afford to take positions in constitutional argument that are too transparently favorable to his own interests. So legislators who want to invest in credibility will have to adjust their positions to disfavor or disguise their own interests to some degree. Likewise, the pressure to maintain a reputation for consistency will, to some degree, cause even self-interested legislators to adhere to a previously established constitutional position when, in changed circumstances, that position works to the legislator’s disadvantage. For purposes of optimizing the legislative capacity for constitutional deliberation, the key question is just whether legislative discourse about the Constitution is always strategic rather than sincere. The empirical premise for what follows is that the realist account of legislators’ behavior is itself unrealistic. The claim that public-regarding discourse within legislatures, including constitutional discourse, is invariably a mask for narrowly defined self-interest verges on incoherence. That view finds it difficult to explain why legislators engage in constitutional discourse in the first place. After all, if everyone mouths constitutional formulae out of self-interest, it is unclear why anyone takes constitutional argument seriously, and thus unclear why there is any audience demand for the empty discourse. The realist account can only be salvaged either by assuming widespread myopia in the audience for constitutional discourse—assuming, 4

5

See Mark V. Tushnet, Clarence Thomas: The Constitutional Problems, 63 Geo. Wash. L. Rev. 466, 469 (1995); see also Edward L. Rubin, Beyond Public Choice: Comprehensive Rationality in the Writing and Reading of Statutes, 66 N.Y.U. L. Rev. 1, 21 (1991) (“[R]e-election maximizing itself does not preclude ideological motivations, because the articulation of an ideological position might be the best way to secure re-election.”). Jon Elster, Alchemies of the Mind: Rationality and the Emotions 341(1998).

Institutional Design Writ Small

221

Chapter 

in other words, that self-interested constitutional discourse successfully dupes other participants over the long term, the sort of assumption that realists usually reject in other contexts—or else by recourse to extremely fragile theoretical models.6 In what follows, then, I will generally ignore the skeptical position.

Improving Constitutional Deliberation in Congress The aim here is the same that I have pursued, in different circumstances, throughout the earlier chapters: to propose small-scale institutional reforms that can improve the functioning of democratic institutions. Here the hope is to suggest incremental changes in the internal design of Congress that will improve its constitutional performance, while remaining agnostic among contentious theories of democratic and constitutional deliberation, and among controversial views of substantive policy and politics. To that end, we may define an “improvement” as a design proposal that produces a net gain when assessed along three dimensions: (1) changes in the cost of constitutional deliberation and decisionmaking by the Congress; (2) changes in the cost of constitutional errors by the Congress, defining error relative to an overlapping consensus of background theories of constitutional deliberation; and (3) the costs of transition from the current design to the proposed design.

Decision Costs and Deliberative Benefits The most striking fact about Congress is its severely constricted agenda. The paramount legal status of the Constitution does not entail that deliberation over constitutional questions is the most important good that Congress supplies; constitutional decisionmaking is one activity among many. So proposals for improvement must account for the opportunity costs of constitutional deliberation. Deliberation, however, also provides institutional and process benefits. Deliberation exploits the collective character of legislatures in ways that can, in principle, improve Congress’s constitutional performance. Among the concrete benefits of deliberation are its tendencies to encourage the 6

Elster discusses models in which all speakers disguise self-interest in the language of the public interest, because each speaker fears that some listener will falsely believe that another listener will punish the first listener if the first listener fails to punish the principal speaker for failing to speak in a public-spirited fashion. See id. at 370–84.

222

Mechanisms of Democracy

Optimizing Deliberation: Constitutional Issues in Congress

revelation of private information, to expose extreme, polarized viewpoints to the moderating effect of diverse arguments, to legitimate outcomes by providing reasons to defeated parties, and to require the articulation of public-spirited justifications for legislators’ votes.7 To be sure, deliberation also suffers from pathologies, quite apart from opportunity costs: it can reduce candor, encourage posturing, trigger herd behavior, and silence dissenters.8 Yet some deliberative vices may also produce systemic deliberative benefits overall, as when extreme views and hot rhetoric force the moderate and reasonable to consider new perspective and to improve their arguments. In any event, the only alternatives to deliberation are (1) pure bargaining followed by voting or (2) voting without either preliminary deliberation or bargaining. The former is occasionally adopted in later stages of the budget process, as discussed in Chapter 6, but is never a general practice in democratic legislatures, all of which (purport to) engage in reasoned argument and not merely bargaining, on most subjects, as a preliminary to voting. Voting without either arguing or bargaining is even less common, although Rousseau urged it for citizen-voters.9 At a minimum, both approaches are procedures that no modern legislature, and few if any collective bodies generally, have ever adopted. No one seriously questions that, on net, some congressional deliberation is better than none at all, especially on constitutional questions.

Error Costs Any reference to constitutional “error” presupposes substantive criteria of right and wrong, or good and bad, in constitutional interpretation. Two such criteria dominate the legal literature on Congress’s constitutional performance. The first is that the Congress commits error whenever it deviates from the Supreme Court’s interpretation of the Constitution. The second is that Congress commits error when it deviates from the outcomes dictated by whatever particular constitutional theory the interpreter holds. Both of these criteria are unattractive. The first applies to an excessively narrow range of constitutional questions and arbitrarily privileges 7 8 9

James D. Fearon, Deliberation as Discussion, in Deliberative Democracy 44, 53–56, 63–64 (Jon Elster ed., 1998). For an overview of the pathologies of deliberation, and its benefits, see Cass R. Sunstein, Infotopia (2006). Jean-Jacques Rousseau, Of the Social Contract or Principles of Political Right: Discourse on Political Economy 48–49 (Charles M. Sherover ed., Perennial Library 1984) (1762).

Institutional Design Writ Small

223

Chapter 

judicial analysis over other modes of reasoning about constitutional questions. The second overlooks that the aim of an institutional-design project is not to entrench some highly contentious, substantive theory, but rather to suggest consensual improvements—democratic optimizers that would improve Congress’s deliberative performance as judged by any of the leading constitutional approaches. So the best criterion identifies “error” by reference to Congress’s skill at using the wide range of sources admissible under all (plausible) interpretive theories, and by reference to Congress’s ability to achieve outcomes that all (plausible) theories deem reasonable. The first criterion holds that Congress should take the Constitution to mean whatever the Supreme Court says it means. As defended by Larry Alexander and Frederick Schauer,10 this view does not assert the implausible interpretive claim that the Constitution actually means whatever the Court says. Instead, the primary criterion for good constitutional law, the argument runs, is that it should be clear and stable. Clarity and stability in turn require a single, paramount constitutional interpreter, and that interpreter should be the Court. Alexander and Schauer sensibly acknowledge that limitations on the judiciary’s political reach and logistical capacity create a broad domain of constitutional determinations by the Congress that go unreviewed by the judges. What, for example, is the Senate to do when, sitting as an impeachment jury, it must decide whether some presidential malfeasance amounts to a high crime or misdemeanor? The United States Reports do not speak to that question. So one objection to the Alexander and Schauer view is not that it is wrong, although it may be, but that it is incomplete; it covers only a fraction of the terrain. A second objection is that, even where judicial review is in the picture, the value of stability does not entail that the Court should be the supreme or paramount interpreter.11 The instability whose costs Alexander and Schauer correctly emphasize derives from the presence of multiple institutions sharing power to interpret the Constitution. Supposing that it is better to have a single interpreter, why should not Congress be that interpreter? In reply, Alexander and Schauer acknowledge that the 10 11

Larry Alexander & Frederick Schauer, On Extrajudicial Constitutional Interpretation, 110 Harv. L. Rev. 1359, 1387 (1997). Many commentators have advanced this objection; for citations, see Larry Alexander & Frederick Schauer, Defending Judicial Supremacy: A Reply, 17 Const. Comment. 455, 458 n. 12 (2000).

224

Mechanisms of Democracy

Optimizing Deliberation: Constitutional Issues in Congress

questions are empirical and outline the relevant institutional variables, including the strength of precedent and the extent of institutional status quo bias in legislatures and courts.12 It is not clear whether their claim is right or wrong, so the choice of a single exclusive interpreter would have to be made on other grounds. A major consideration against their proposal is that there is a large domain of questions, like impeachment or the determination whether a state has a republican form of government, as to which interpretive authority must for all practical purposes be entrusted exclusively to Congress. There is then a decisive asymmetry. One can imagine saying that Congress should be the final interpreter of the whole Constitution—many countries have given their national legislatures exactly that power—but it is unimaginable that the Court could have the same power, given the many domains in which the Court does not even pretend to review legislative decisions on constitutional questions.13 Other views posit that Congress errs whenever it arrives at outcomes that deviate from those indicated by some particular substantive account of constitutional interpretation. A list of the current contenders includes, in no particular order, original-intention interpretation, originalmeaning interpretation, common-law constitutionalism, process theory, law as integrity, minimalism, and pragmatism. Although some of these are better described as theories of constitutional adjudication than as interpretive theories, and are thus of attenuated relevance to constitutional deliberation by legislators, it is still true that a substantive approach must necessarily elevate one of the contested interpretive accounts over its competitors, despite intractable disagreement among proponents of these views. Substantive approaches intended to improve congressional deliberation must be addressed to a Congress whose members do not all subscribe to that account, and who would deliberate about the proposals themselves under diverse standards of constitutional evaluation held by the members. Congress is not an institution noted for facing and resolving fundamental disagreements, let alone abstract fundamental disagreements about constitutional theory. But Congress is an institution skilled at reaching specific agreements that allow all parties to preserve their 12 13

See Alexander & Schauer, Defending Judicial Supremacy, supra note 10, at 476–78. For a list of such domains, see Adrian Vermeule, Judging Under Uncertainty: An Institutional Theory of Legal Interpretation 269 (2006).

Institutional Design Writ Small

225

Chapter 

abstract commitments. The approach offered here exploits this institutional strength. Paralleling previous chapters, I will assess institutionaldesign proposals in part by their ability to improve Congress’s constitutional performance relative to an overlapping consensus about the criteria for successful constitutional argument. Admissible proposals, in other words, should identify improvements that are attractive to proponents of all views. Such consensual improvements should in principle be available (however difficult they are to identify), because competing accounts of constitutional interpretation display broad overlap along two dimensions: sources and outcomes. All of the major approaches to constitutional interpretation, for example, agree that constitutional text is relevant and admissible, agree that a broad range of potential techniques for interpreting text are barred (interpreters must assume that the text is written in English, that it has public rather than private or coded meaning, and so forth), and that certain other techniques are useful, such as the cautious use of canons of construction. The same is true for precedent, or even more so. As for outcomes, consider the Senate’s decision in 1998 not to censure President Clinton for the behavior on which he was impeached but acquitted. That decision justified a wide range of approaches to constitutional interpretation, is others have argued.14 Other easy cases abound,15 meaning cases in which text, purpose, original meaning, and political morality all converge. A twelve-year-old cannot be elected president, nor can Congress itself appoint executive officers. If all of the major interpretive approaches agree upon a core set of interpretive sources, skills, and even outcomes, then deliberative “error” can be defined relative to that overlapping consensus without taking sides on fundamental questions. Proposals can be evaluated by their tendency to improve Congress’s use of those sources and skills and to improve Congress’s ability to deliberate over hard cases, while avoiding outcomes that are condemned by all constitutional approaches.

Transition Costs The Constitution fixes some features of the design of congressional institutions: bicameralism, the length of terms in each house, and the 14 15

See, e.g., Jack Chaney, The Constitutionality of Censuring the President, 61 Ohio St. L.J. 979, 1004–12 (2000). See Frederick A. Schauer, Easy Cases, 58 S. Cal. L. Rev. 399 (1985).

226

Mechanisms of Democracy

Optimizing Deliberation: Constitutional Issues in Congress

impermissibility of state-imposed term limits are examples. It also leaves many institutional features unspecified, such as committee structure, rules governing debate that might allow or restrict the ability to filibuster or amend proposals, the role of political parties in legislative organization, and the size and organization of staff. The proposals that follow are confined to small-scale, feasible improvements, rather than large restructurings, that Congress can make without the consent of any external actor, under the expansive constitutional power of each house to “determine the Rules of its Proceedings.”16 One justification for this restriction is that controversial distributive implications will almost inevitably doom any proposed improvement; minor restructurings, on the other hand, are less likely to inflict large losses on any interested parties, and are thus more likely to gain widespread support. Another justification is that large-scale restructurings carry a greater risk of harmful unintended consequences. Finally, even if all goes as planned, the costs of a large-scale transition from one institutional arrangement to another may outweigh any gains in reduction of decision costs or error costs. In positive terms, a proposal that entails small transition costs is more likely to prove beneficial on net. And a series of such proposals can, I will suggest, cumulate to produce large beneficial changes to constitutional deliberation in Congress, and to democratic institutions generally.

Is Congress’s Constitutional Performance Optimal? Even if the institutional-design question is the right one to ask, and even if the foregoing criteria for assessing Congress’s constitutional performance are sensible, there remains the possibility that Congress’s performance is already optimal. This view would not hold that Congress never commits constitutional errors, but would rather hold that no cost-justified improvements in congressional performance are possible. This view follows from a simple account of the political incentives that affect congressional deliberation. On this account, opposing legislative coalitions will ventilate opposing constitutional arguments, thereby ensuring fully adequate deliberation. Members may raise constitutional concerns during committee deliberations or floor consideration, if they choose to do so, without adopting a special procedure to force such activity. 16

U.S. Const. art. I, § 5.

Institutional Design Writ Small

227

Chapter 

The view that Congress is optimally designed for constitutional deliberation is surely counterintuitive; it is also implausible. The final test of whether there are democratic mechanisms that can produce cost-justified improvements in congressional constitutional deliberation is, of course, the content of the proposals themselves, described shortly. But it is worth noting here that the simple account suffers from numerous conceptual and empirical difficulties. First, the simple account assumes an implicit “fire-alarm” model17 of constitutional argument in Congress: coalitions and interest groups monitor proposed bills and sound the alarm when they detect constitutionally troublesome provisions or policies. But there is no particular reason to believe, a priori, that legislators and interest groups engage in just the right amount of monitoring if they rely on post hoc fire alarms to trigger attention and discussion. To the contrary, the empirical record suggests that Congress, as a collective body that continually adjusts its own rules and procedures over time, often rejects the fire-alarm model in favor of ex ante framework legislation that structures congressional deliberation on constitutional and policy questions. For a recent and important example of framework legislation addressed specifically to constitutional questions, consider the Unfunded Mandates Reform Act of 1995 (UMRA), which requires congressional committees to specify, quantify, and describe any federal mandates that the proposed bill would impose on state, local, and tribal governments and to identify those that are unfunded by the federal government. The UMRA also allows legislators to raise a point of order during floor deliberation in order to focus debate on any unfunded mandate and to require a recorded vote to waive the objection. The UMRA is best understood as a species of collective precommitment. It represents a constitutional judgment by legislators in their collective capacity, outside the divisive context of specific proposals, that fire-alarm monitoring of the federalism questions implicated by unfunded mandates had provided insufficient consideration for constitutional values. The widespread perception of interest groups and lawmakers before passage of the UMRA was that Congress enacted unfunded mandates in some 17

Cf. Mathew D. McCubbins & Thomas Schwartz, Congressional Oversight Overlooked: Police Patrols Versus Fire Alarms, 28 Am. J. Pol. Sci. 165, 166 (1984) (using a fire-alarm model in the context of congressional oversight of the executive branch agencies).

228

Mechanisms of Democracy

Optimizing Deliberation: Constitutional Issues in Congress

cases without being aware of their existence in an omnibus bill or in ignorance of their scope and effect. It should not be surprising that legislators sometimes turn to ex ante framework legislation in order to improve congressional performance in constitutional settings. Fire-alarm monitoring will be most successful in an arena dominated by organized and sophisticated interest groups on all sides of an issue with clear and established lines of communication to lawmakers with jurisdiction (environmental policy, for example). As with unfunded mandates, however, constitutional issues arise throughout the legislative arena, and affected groups will often lack the expertise to discover, analyze, and alert legislators about substantial questions of constitutionality. Furthermore, even when some members are aware of a serious constitutional objection, the chamber’s rules may bar public deliberation on the issue. In the House of Representatives, for example, special rules usually structure consideration of legislation and sharply constrain members’ ability to raise objections or make amendments. Individual legislators could, in principle, address these deficiencies in the fire-alarm system by engaging in individual monitoring or by privately expending the political capital needed to obtain full consideration of constitutional issues. But in other deliberative contexts legislators have responded to the inadequacies of the fire-alarm model through collective action, such as the UMRA. Structural constraints explain this pattern of behavior. The basic constraint is that deliberation within the Congress constitutes a collective good. All legislators benefit when a particular legislator spends time developing information, analyzing constitutional questions, and working with specialized personal staff on constitutional issues. Yet if an individual member provides those benefits, she has taken time away from the tasks that contribute directly to her reelection. She loses time for fundraising, casework, media appearances, and obtaining particularized spending projects in her district; she will thus be at a disadvantage and receive less of the pie of limited federal resources unless all members of Congress spend a similar amount of their time on constitutional issues. If constitutional deliberation is an individually supplied good, individual legislators do not internalize all of the benefits of constitutional deliberation but do shoulder the costs. In such a system, constitutional deliberation will be underproduced. In the face of the public-good character of constitutional deliberation, Institutional Design Writ Small

229

Chapter 

all members would benefit from a system that requires lawmakers to allocate some of their scarce time to the consideration of constitutional issues, that provides collective funding for the staff required for this deliberation, and that enforces collective commitments that support deliberation. The constitutional framework proposed here, like the UMRA, seeks to solve the collective-action problem and enforce the institutional commitment to spend some time and resources on these matters. Not infinite time and resources, of course; although legislators value a process that provides the opportunity for focused and serious deliberation about difficult constitutional issues, they also have other substantive objectives and value other collective activities. Institutional-design proposals must take opportunity costs into account in calibrating the incentives for legislators’ deliberation on constitutional questions.

Deliberation-Enhancing Mechanisms Some General Principles of Design Four design principles guide the necessary tradeoffs between decision costs, error costs, and transition costs. First, members of Congress must have adequate information about constitutional issues raised by legislation. Second, members must be afforded an opportunity to raise constitutional issues and to deliberate about them fully. Third, the institutional design should encourage broad involvement from experts and interested parties. Fourth, the congressional structure for the consideration of constitutional questions should reflect a balance between improving congressional capacity to deliberate on constitutional questions, and the need to enact legislation without undue delay or extreme difficulty, and without providing opportunities for strategic action by coalitions seeking to derail bills that they oppose on other grounds.

Production and Dissemination of Information In some cases, Congress may neglect its responsibility to consider constitutional issues, either because members are not aware that a proposal has constitutional ramifications, or (as argued above) because information about and analysis of constitutional questions is a collective or public good that individual legislators will underproduce. Fortunately, models for more regularized collective treatment of such issues exist. The most influential 230

Mechanisms of Democracy

Optimizing Deliberation: Constitutional Issues in Congress

and ubiquitous procedural framework designed to produce and structure information is the congressional budget process, detailed in Chapter 6. A recent addition to the budget rules, the Unfunded Mandates Reform Act (UMRA) described above,18 requires that the Congressional Budget Office (CBO) provide authorizing committees written statements identifying federal mandates in reported bills. With respect to mandates that exceed certain thresholds, CBO must provide more detailed information about the costs to state and local governments or to the private sector. A modern committee report contains a great deal of mandatory information—some required by budget rules, some by other congressional rules—designed to address systematic gaps in information or to provide information to lawmakers who do not serve on the specialized committee and thus might overlook important aspects of policy. With respect to a similar framework for the consideration of constitutional issues, identifying bills that implicate constitutional questions must occur early so committees with jurisdiction can hold hearings and gather further information about the issue. The parliamentarian, typically a distinguished lawyer with a reputation for nonpartisanship, can determine at the time of referral to committee whether a proposal appears to raise a significant constitutional issue. The referral decision, which is published in the Congressional Record, can also contain the parliamentarian’s description of any constitutional issue. In some cases, constitutional issues will arise as the proposal is considered and amended; thus, the trigger for special procedures should occur not only at the initial referral but should also remain available throughout the committee process. In addition, the parliamentarian will specifically identify any constitutional issues implicated by the proposal that are issues the judiciary declines to review or reviews only under a rational-basis standard. Such identification will alert members of Congress to their special responsibility with respect to these bills. Initially, the parliamentarian may want to rely on a nonpartisan group of constitutional experts to provide guidelines for this process and to update the list of issues that receive very little or no judicial scrutiny.19 18

19

Pub. L. No. 104–4, 109 Stat. 48. For analysis of the Act, see Elizabeth Garrett, Enhancing the Political Safeguards of Federalism? The Unfunded Mandates Reform Act of 1995, 45 U. Kan. L. Rev. 1113 (1997). Donald G. Morgan, Congress and the Constitution: A Study of Responsibility 349 (1966).

Institutional Design Writ Small

231

Chapter 

To provide members the data required for them to make informed decisions when the bill reaches the floor, every bill will be accompanied by a constitutional impact statement. The constitutional impact statement will provide a summary of the committee’s findings on the proposal’s constitutional implications. If there is no significant constitutional issue raised by the bill, the statement will include that information. The constitutional impact statement will refer to any more comprehensive analyses (perhaps records of public hearings or analyses prepared by expert staff ), and it will contain any dissenting views. The constitutional impact statement will be written so that non-lawyers can understand the arguments, not only because many members and most constituents are not lawyers, but also because the institutional strength of Congress is not its attention to legalisms but its expertise in the policy aspects of constitutional decisions. Constitutional impact statements will be more detailed than the parliamentarian’s initial framing of the issue. The burden on drafters will not be excessive, however, because statements will be required only with respect to reported bills, a much smaller universe of legislation. Except in the case of a finding that the legislation does not implicate a significant constitutional issue, the statement will be a brief summary of the constitutional issue, together with the committee’s views and any dissenting views. If the constitutional issues raised by the proposal are ones that the judiciary declines to review, the statement will identify them as such in order to signal to legislators that their deliberation and decision on these constitutional questions are likely to be the final determinations. To respond to judicial requirements for clear statements in a number of quasi-constitutional areas, the statements will also declare whether provisions in the bill are severable, identify any retroactive provisions, and provide other clear statements required by the judiciary,20 a category that may evolve over time as the jurisprudence of interpretive rules of clear statement changes. Finally, there will be explicit statements declaring whether the legislation allows a pre-implementation challenge to its constitutionality and whether any constitutional challenge receives special or expedited judicial consideration. To ensure that a constitutional impact statement is produced for all legislation, any bill that comes to the floor without such a statement will 20

William Eskridge, Jr. & Philip Frickey, Quasi-Constitutional Law: Clear Statement Rules as Constitutional Lawmaking, 45 Vand. L. Rev. 593, 598–611 (1992).

232

Mechanisms of Democracy

Optimizing Deliberation: Constitutional Issues in Congress

be subject to a point of order, waivable only by a majority vote. Enforcement is vital. Requiring information without also providing a sanction will result in little information, as members and staff target limited resources on higher priority issues.

Expert Congressional Staff on Constitutional Issues The modern Congress increasingly relies on others to provide information and analysis necessary for decisionmaking, and it will do so in this context as well. The idea of creating a body of trained professional staff to review constitutional issues is consistent with larger institutional trends in the legislative branch. Over the last century, as Congress has become more professional and the issues it faces more complex, the number of staff has increased substantially. Moreover, Congress occasionally establishes an internal body of experts to counterbalance expertise in the other branches of government. Thus, Congress created the CBO so that the legislative branch can deal more successfully with the executive branch, which includes the Office of Management and Budget (OMB) and countless budget experts in the agencies. The President often has significant influence on constitutional and other legal matters because he is assisted by the Attorney General, the Solicitor General, their staffs, and the staff of the Office of Legal Counsel. If Congress wants to step out of the shadows of the judicial and executive branches with regard to constitutional determinations, it must establish an equivalent set of experts. There are several options for a structure to provide constitutional information and analysis to all members of Congress. First, Congress could expand the duties of existing entities and provide additional staff. For example, each house has an Office of Legal Counsel, established in the 1970s. The duties of the counsel revolve around representing the House and Senate in court and defending the constitutionality and legality of congressional enactments, subpoenas, and other legislative actions. The Offices of Legislative Counsel assist members in drafting legislation, but currently the counsel members do not formally advise lawmakers about the constitutionality of their proposals, restricting their advice mainly to drafting issues. To place this new task within the jurisdiction of either of these offices would work a fundamental change in their jobs and require significant additional staff. The duties of the American Law Division of the Congressional Research Service could be expanded so that the staff would consult more Institutional Design Writ Small

233

Chapter 

extensively and regularly with Congress on constitutional issues. Now, analysts provide testimony and written analysis when requested; the division functions as a “law office for Congress” with sixty-five attorneys, paralegals, and support staff. The option of relying on CRS staff is unlikely to appeal to lawmakers in the context of the new constitutional framework, however. CRS is perceived as an extremely neutral entity with few if any partisan connections (although individual analysts can develop a reputation for particular perspectives associated with political ideologies). Although nonpartisanship is sometimes perceived as an asset, in this context, members are apt to want some closer connection between political considerations and legal ones. Moreover, if it remains separate from any new staff organization, CRS could provide a check on the new staff, which is likely to be more partisan, much as the Government and Finance Division of CRS now provides a check on the economic analyses of the CBO, the JTC, and other congressional committees. Thus it seems best to form a new congressional office, along the lines of the CBO or the JTC: call it The Office for Constitutional Issues (OCI). This office will be headed by a Chief of Staff appointed in bipartisan fashion, as are many other congressional staff. The Chief of Staff will appoint the Office’s staff, including any deputies, and all appointments will be based solely on professional competence, without regard to political affiliation. Much like the CBO and the JTC, the staff will be a mix of lawyers and other professionals and scholars, in this case political scientists, historians, and public policy professionals with interests in constitutional law. The institutional advantage of Congress with regard to constitutional issues is its ability to blend policy considerations with technical legal arguments. Thus, it is important that members receive not only legalistic arguments, but also learn of the broader policy implications of a particular constitutional interpretation relative to others. The analyses and reports produced by the staff will be publicly available so that citizens will have access to the information that shapes the constitutional deliberation and decisionmaking of their representatives. OCI will consult with its counterparts in the executive branch, as CBO staff does with OMB and other agency officials, and as JTC staff does with Treasury and Internal Revenue Service staff. Such consultations will often be largely informal, although it may make sense to formalize some interactions so that they occur as a matter of course and so that the opinions of executive branch experts are available to all members of Congress. 234

Mechanisms of Democracy

Optimizing Deliberation: Constitutional Issues in Congress

Congress will have to balance the advantage of widespread dissemination of views with the inevitable chilling effect formal and public disclosure will have. OCI should also determine whether its staff or lawmakers will consult formally with members of the judiciary. It seems very unlikely that sitting judges will be comfortable giving their opinions about constitutional issues likely to come before them, but retired judges may well serve as a source of expertise. Finally, OCI will no doubt also rely on input and analysis by academics, many of whom will be eager to participate in the process as a way for their ideas and scholarship to shape policies. Political scientist Barbara Sinclair, commenting on this proposal, expresses the concern that “[s]imply by virtue of its charge, [OCI] would acquire a vested interest in finding constitutional problems, and highly technical and complex ones at that.”21 Sinclair is surely correct, but the observation does not amount to an objection. The proposal aims to harness precisely the institutional motivation that Sinclair identifies; some motivation of that sort is needed to overcome the collective-action or public-goods character of constitutional deliberation among legislators. Given the background of chronic underproduction of constitutional deliberation in Congress, it is not plausible to worry that a missionoriented OCI will cause Congress to spend too much time or agenda space deliberating constitutional issues. The real concern is whether Congress will do so enough, even with an OCI in place.

Committee Structure to Consider Constitutional Issues Most congressional deliberation does not occur on the floor; it is done in committees. So the relevant—and perhaps the most important—consideration is the committee framework through which such issues are analyzed. There are at least three alternative designs of committee structure to consider. First, the jurisdiction of the current Judiciary Committees could be expanded so that they would also have responsibility for considering bills identified at referral or later as implicating significant constitutional issues. The Judiciary Committees would have jurisdiction to consider the constitutional implications of the bill, to work with the OCI to prepare 21

Barbara Sinclair, Can Congress Be Trusted with the Constitution?: The Effects of Incentives and Procedures, in Congress and the Constitution 310 (Neal Devins & Keith E. Whittington eds., 2005).

Institutional Design Writ Small

235

Chapter 

the constitutional impact statement, and to make amendments to the language designed to reflect the constitutional findings. In the House, multiple referral techniques, amended in 1995, would provide the framework for the shared jurisdiction between the substantive committee and the Judiciary Committee. The substantive committee would serve as the primary committee, so that when it discharged a bill, the Judiciary Committee would have only a limited time period in which to perform its role. (Of course, in some cases, the Judiciary Committee would also be the substantive committee, and thus its deliberations would include discussions of the substance of the proposal as well as the constitutional issues raised.) The Judiciary Committee could also hold hearings and consider the bill concurrently with the substantive committee’s deliberations. Multiple referral occurs less often in the Senate, but it is not unheard of, and it could be structured much like current House procedures. The advantages of using the Judiciary Committees are obvious. These committees already have some expertise in constitutional issues, and they have reputations for relatively serious and careful consideration of legal and constitutional questions.22 The disadvantages, however, are substantial, although somewhat less apparent. First, because the committees are composed almost entirely of lawyers, their analyses tend to be legalistic and to replicate what they think judges would say on an issue. They are among the most deferential of lawmakers to courts and traditional legal reasoning.23 Second, committee assignments are largely a matter of self-selection, and the lawmakers who want to serve on the Judiciary Committees tend to be at the extremes of the ideological spectrum. Thus, the Judiciary Committees are often more polarized than other committees and less representative of the body.24 Adding significant constitutional issues affecting pending legislation to the committees’ portfolios would only exacerbate this tendency, attracting more lawyers with intense preferences on constitutional interpretation. The polarization might affect the committees’ deliberations, making compromise less possible and potentially holding up legislation. Moreover, it is likely that the substantive 22 23 24

Mark C. Miller, Congressional Committees and the Federal Courts: A Neo-Institutional Perspective, 45 W. Pol. Q. 949, 959–61 (1992). Mark C. Miller, Congress and the Constitution: A Tale of Two Committees, 3 Const. L.J. 317, 339–40 (1993). See e.g., Miller, supra note 22, at 959–60; see also Stephen F. Ross, Legislative Enforcement of Equal Protection, 72 Minn. L. Rev. 311, 358 n.190 (1987).

236

Mechanisms of Democracy

Optimizing Deliberation: Constitutional Issues in Congress

committees with primary jurisdiction over legislation would vehemently object to sharing power with the Judiciary Committees. In the budget context, framers discovered that standing committees resisted reallocations of jurisdiction to other existing committees because of turf jealousies. A second solution would be to create entirely new committees with little substantive responsibility but with significant power to coordinate the actions of other committees. This was the approach used in 1974 in the budget arena, and it may well be the best framework for the consideration of constitutional issues. By analogy, the United Kingdom Parliament maintains a Joint Committee on Human Rights that has, by most accounts, improved deliberation on rights in Parliament.25 That approach may be generalized to all constitutional questions. There are several advantages to this approach. Congress has more flexibility in designing new committees and thus can avoid some of the weaknesses of the Judiciary Committees. For example, the rules setting up the new Committees on Constitutional Matters could specify that only a certain number of members could be lawyers, and the rules could require representation from other standing committees (as the House Budget Committee does). It might make some sense to appoint a few members of the Judiciary Committees to the new committees, at least in the early years, to gain from their expertise. Because the committees would be new, and thus lack a strong tradition of seniority in committee assignments, and because they would have far-ranging jurisdiction affecting many pieces of major legislation, it is likely that party leaders would rely heavily on party loyalty in making appointments and would generally exert more control over committee decisions. Referrals to these new committees would work in the same way as the first option of using the existing Judiciary Committees. In both cases, committee members would rely on their own committee staff as well as the technical staff of the OCI. If lawmakers decide to establish new committees, they should also consider giving these new committees a role in the selection of the Chief of Staff of the OCI, with the ultimate decision vested with party leaders. This second option for committee organization has undoubted costs, many shared with the first option of giving jurisdiction to the 25

See Janet L. Hiebert, New Constitutional Ideas: Can New Parliamentary Models Resist Judicial Dominance When Interpreting Rights?, 82 Texas L. Rev. 1963, 1978 (2004).

Institutional Design Writ Small

237

Chapter 

Judiciary Committees. First, members willing to serve on the committees would likely have intense and outlying preferences just as do the current members of the Judiciary Committees. Second, situating the constitutional analysis in specialized committees, rather than in the substantive committees that have primary jurisdiction over legislation, artificially separates constitutional issues from the larger policy issues. Determinations about rights guaranteed by the Constitution are often abstract, and perhaps even meaningless, without simultaneous decisions about how resources will be directed toward vindicating those rights. Pragmatically, it is very difficult to clearly separate the two kinds of issues. Thus, a bifurcated committee structure could cause confusion, duplication, or conflicting messages from the various committees to the full Congress and the public. The third organizational option, consistent with the design of the UMRA, addresses this problem of separation. Under this model, the substantive committees would perform the constitutional analysis as they considered any bill implicating serious constitutional issues. Although members would develop less expertise on constitutional issues than would be the case in the other two models, they could rely on OCI’s technical staff, on their own committee counsel, and on lawyers on their personal staff. In addition, individual members might well become more involved in the constitutional issues because of personal preferences (much as occurs now on the generalist tax-writing committees where individual members specialize in agriculture provisions, oil and gas incentives, or other narrow areas), although this development might be unwelcome if the specialists were lawyers who received too much deference to their overly legalistic approaches. The dynamics of committee consideration would be different if the constitutional assessment were left in the substantive committees rather than placed in specialized committees. First, the substantive committees would likely be less polarized on constitutional issues than the Judiciary or similar committees. Members would have selected the substantive committees because of their interest in the policies that fall within their jurisdiction, and only secondarily (if at all) because of their views on constitutional issues. In other words, members of the Agriculture Committees may have firm positions on farm policy, but they have less intense views on constitutional issues that their bills implicate, such as federalism, the delegation doctrine, or the federal spending power. This characteristic of 238

Mechanisms of Democracy

Optimizing Deliberation: Constitutional Issues in Congress

the substantive committees’ membership might allow for more moderated, and perhaps more thoughtful, deliberation. On the other hand, it might mean that constitutional issues would be slighted, because members would not care much about them. In that case, the attention constitutional issues received would depend on the salience provided by the other proposals offered here and by interest-group agitation, the latter a sort of improved fire-alarm mechanism enhanced by the improved information that the proposals should help to produce. Second, the interest-group dynamics would be very different in the substantive committees. Substantive committees attract the attention of groups with stakes in the policies within their jurisdiction. So the members of the Agriculture Committees interact mainly with farm groups, environmental groups, consumer groups, state and local officials with responsibility for farm policy, and others with particular interests in agriculture. The committees are monitored primarily by trade publications concerned with farmers, food production, and rural policies. These groups, and their lobbyists, would invest resources in studying constitutional issues, because the new framework set out by the proposals here would provide another strategic opportunity to affect a bill’s fate, but they would not initially possess expertise in such issues. They would invest in developing expertise and producing useful information, because and insofar as doing so would affect the policies adopted by substantive committees. Their information might be particularly helpful because they could infuse their legal analysis with their knowledge of the underlying policies. In contrast, using multiple referrals and specialized committees would mean that the groups with the most sustained interaction with lawmakers on constitutional issues would be those interested in legal issues and constitutional law. Groups like the American Bar Association, the American Civil Liberties Union, the Institute for Justice, and legal academics would exert the primary influence on the deliberations of the Judiciary Committees or the Committees on Constitutional Matters. These groups would have less influence if the constitutional inquiry were done by dozens of substantive committees, because their attention would be fragmented and their resources deployed widely. The structure of interest-group activity and conflict is a crucial element in committee design. All three options present strengths and weaknesses in this respect. It would be easier for substantive committees, working with the expert OCI and getting advice from the executive Institutional Design Writ Small

239

Chapter 

branch and other outside experts, to include the law-oriented interest groups, when appropriate, than it would be for specialized committees to blend policy and constitutional law in their deliberations. OCI’s formal and informal consultation with expert groups and lawyers in the executive branch would help these interest groups to overcome the problems of fragmentation and to learn of significant constitutional issues in time to weigh in on the outcome. In a sense, the contact entity for the constitutional-law-oriented interest groups would be OCI, which would also serve as a conduit to inform the groups about decisions they want to influence.

Floor Consideration and Points of Order Procedures structuring floor deliberation seek to balance a number of competing concerns. First, certain structures, notably points of order, reduce the chances that Congress will inadvertently or intentionally ignore difficult or controversial issues. Even though most members may want to avoid such issues, and the party leaders may work to structure a bill or floor consideration to spare their members difficult votes, the availability of a point of order can allow one lawmaker (or a few) to halt proceedings, highlight the issue, and force a roll-call vote. Especially in the House, where the floor is tightly controlled by the Rules Committee, a point of order that cannot be waived in a special rule empowers individual members and reduces the chance of success of avoidance techniques. Furthermore, the point-of-order process focuses legislative attention on the constitutional issue and provides the opportunity for sustained debate for which members can be held accountable. Occasionally, constitutional issues identified by a handful of members have been brushed aside during the rush of floor debate and activity. Points of order make that more difficult. On the other hand, points of order can be used to stall or derail legislation by lawmakers who oppose the proposal but who do not care about the constitutional issue. Strategic use of the point-of-order procedure can nonetheless force sincere debate and deliberation about important issues, so the motivation behind the objection may not be relevant in all cases. But as Congress adopts new procedural frameworks, it must be aware that additional process makes enacting laws more difficult. Floor procedures must balance the need to allow the opportunity for lawmakers to focus on constitutional issues and the need to enact some legislation. 240

Mechanisms of Democracy

Optimizing Deliberation: Constitutional Issues in Congress

Currently, the Senate allows members to raise constitutional points of order;26 the House does not, although representatives have objected to legislation on the ground that it violated the prerogatives of the House under the Origination Clause of the Constitution.27 The practice in both houses is that the presiding officer does not rule on the question but instead refers it to the full chamber for a vote. The best idea is to institutionalize these points of order, drawing on experience with other congressional procedural frameworks that are enforced in similar ways. The following rules seem most plausible. First, a point of order can be raised by any lawmaker to object to considering a bill that is not accompanied by a constitutional impact statement. Such enforcement is required to ensure that the statements are systematically produced in a timely fashion. Second, a point of order can be raised against any bill with provisions designed to require early judicial determination of constitutional issues. If these provisions encourage Congress to shirk its responsibility to think carefully about constitutional issues, they should be more difficult to include in legislation. Third, a member can raise a point of order against any bill that she believes raises a significant constitutional issue. The information provided in the constitutional impact statements will help alert members to objectionable or worrisome proposals, although members may also discover issues on their own. It is important that, under this scheme, a legislator will not be limited to raising constitutional issues identified initially by the parliamentarian or discussed in the constitutional impact statement. For this process to be effective in the House of Representatives, the Rules Committee should not be allowed to waive points of order in a special rule, thereby circumventing the objection and recorded vote. To reduce the strategic use of the third point of order that raises a significant constitutional issue with regard to any bill, a member objecting to consideration of the bill must present to the presiding office a petition signed by twenty members in the House and ten members in the Senate indicating their support for the objection. This is a submajority threshold, of the type 26 27

Floyd M. Riddick & Alan S. Frumin, Riddick’s Senate Procedure 52–54 (1992). Ross, supra note 24, at 359 n.193. This conclusion was confirmed in a phone conversation with a lawyer in the House parliamentarian’s office. However, some House precedents suggest that other constitutional points of order may be raised and voted on by the House. See e.g., 15 Lewis Deschler & William Holmes Brown, Deschler-Brown Precedents of the United States House of Representatives 67–71 (1999) (stating that it is “for the House . . . to determine on the constitutionality of the bill”).

Institutional Design Writ Small

241

Chapter 

discussed in Chapter 3; it has the same accountability-forcing and deliberation-forcing virtues, and the same costs. In the constitutional context, as elsewhere, the submajority requirement is an optimizing device: it makes the point-of-order strategy more costly to those trying to use it to force changes in the bill, thereby dampening the risk of holdup by very small groups of legislators, but it still allows a small group of intensely concerned lawmakers to bring the attention of the full body to a constitutional issue. Consider the analogy of cloture, where a petition to vote to cut off debate in the Senate must be signed by sixteen Senators. The basic argument for the submajoritarian threshold is that constitutional points of order will arise more frequently than, for example, budget points of order. The latter typically pose fairly discrete issues (e.g., is an amendment revenue-neutral?) that can be avoided by bill drafters. Constitutional issues, on the other hand, can be raised in more contexts— either sincerely or strategically—and thus pose a more far-reaching procedural threat to legislation. Of course, this proposal could be modified by raising or lowering the submajority threshold, perhaps on a subjectby-subject basis. For example, a single member could be allowed to raise a point of order if the issue were one that courts are unlikely to review (as defined by the parliamentarian advised by the group of legal experts and the OCI staff ), and a group of members would be required only with respect to issues that receive robust judicial scrutiny. This modification would increase the complexity of the procedure, but it would tailor the proposal to mirror the greater concern with Congress’s deliberation of constitutional issues that are unreviewed by the courts. Sinclair says that “points of order . . . could too easily be used for purposes of delay, no matter what sorts of purported safeguards are attempted.”28 To the extent that some delay would contribute to optimal deliberation of constitutional issues, delay is a good, not a bad. Sinclair means excessive delay, somehow defined, but she does not explain why a submajority threshold or trigger cannot be adjusted upwards or downwards to prevent that, nor does she explain whether she thinks that cloture petitions are too easily filed. Her logic suggests that they are, but this will seem implausible to most observers of Congress. Finally, a bare majority vote should be all that is required to waive any of the three points of order. Some budget points of order can be waived 28

Sinclair, supra note 21, at 310.

242

Mechanisms of Democracy

Optimizing Deliberation: Constitutional Issues in Congress

in the Senate only by a vote of sixty members, and generally supermajority voting requirements afford greater protection to minority interests than the traditional majority voting rule. The suggestion for a majority voting rule in this setting is an optimizing move—the procedural rules should allow opportunities for deliberation and modification but not halt a great deal of legislative activity. The voting rule may not matter very much in the Senate, because there much legislation would also be subject to filibusters that can be broken only by a supermajority vote of sixty Senators. But lawmakers may be more willing to vote to cut off lengthy debate than to vote against a serious constitutional objection, so it is not clear that the two votes are interchangeable. A supermajority requirement in the House would be a significant new hurdle; no budget point of order in the House requires more than a majority to waive, and only a rule applying to a narrow subset of tax increases is formally enforced through a supermajority voting requirement. A separate majority vote is sufficient protection. By virtue of its separateness, a vote of this sort disaggregates the lawmaker’s stand on the constitutional issue from her final vote and eliminates or reduces her ability to explain away a troublesome position on the constitutional matter as necessary to passing an omnibus bill with numerous provisions that her constituents like. So the separate vote is an accountability-forcing device as well as a deliberation-forcing one. Not only will it improve the chances for meaningful congressional deliberation on constitutional issues, it will force all concerned to stand and be counted on the constitutional questions. These recommendations are flexible, not cast-iron. Rules can be changed and enforcement procedures calibrated over time to account for experience. For example, one could require supermajority votes in the context of constitutional issues left unreviewed by the judiciary and require only a majority for other constitutional points of order. Furthermore, there may be an advantage to supermajority requirements peculiar to the constitutional realm. Building consensus and demonstrating wide margins of support for constitutional positions may be important for the legitimacy of those determinations. Congress already has a number of formal and informal structures that result in wide margins of victory for most major legislation.29 The question is whether an additional supermajority

29

Tushnet, supra note 4, at 52.

Institutional Design Writ Small

243

Chapter 

vote would be beneficial, or whether a constitutional determination even by a bare majority of both houses of Congress, already a supermajority requirement, has more legitimacy in the eyes of the public than, for example, a five-four decision by the Supreme Court.

Deliberation and Institutional Design Whether Congress retains its current role in constitutional interpretation or whether that role is expanded, procedural and institutional devices like those described here promise to enhance congressional performance in this arena. A democratically well-functioning Congress is one that takes advantage of well-designed democratic mechanisms that structure constitutional deliberation and decisionmaking. Those mechanisms count as well-designed, in turn, if they help both to improve deliberative processes and to improve outcomes, relative to a background consensus of democratic theories.

244

Mechanisms of Democracy

Conclusion

Institutional Design as a Going Concern

I

n an illuminating metaphor, constitutional design has been compared to rebuilding a ship while at sea.1 It is rare, however, that the whole ship is thoroughly restructured while underway. More often, sailors patch here and tinker there, accepting most of the ship’s structure as a constraint while attempting small-scale improvements where feasible. In the previous chapters, I have assumed a background constitutional order that sets up the standard institutions of liberal democracy. Even within the large-scale constraints imposed by such an order, however, institutional design never ceases. There are gaps in old institutions; new institutions are continually required as circumstances change; and even well-functioning institutions have play in the joints. Institutional design, writ small, is a going concern. The gaps in the standing design leave space for improvement at the margins, but there are usually alternative improvements to be debated, and questions about what would count as an improvement anyway. 1

Jon Elster, Institutional Design in Post-Communist Societies: Rebuilding the Ship at Sea (1998).

Institutional Design Writ Small

245

Conclusion: Institutional Design as a Going Concern

We might imagine that such questions are resolved in a conclave of the sailors—in a kind of second-order assembly of democrats, convened to decide, not first-order policy questions, but how the institutions themselves can be shaped within the standing constraints. A commitment to democracy, even to the core set of democratic values I have discussed, is compatible with a broad range of views on these questions. But democrats of differing views can deliberate and decide together about what democracy permits or requires in particular contexts, even if they lack a full account of what democracy entails. To be sure, some of these democrats will rise to say that the assembly must turn, first of all, to the high-level questions of political theory. First, they say, we must decide what counts as an improvement; only then can we talk about institutions. They further suggest that the questions of theory go all the way down. The assembly of democrats will face a standard problem of infinite regress that arises in assemblies in which institutions are designed: what voting rule should govern the choices to be made in the assembly—and if there is disagreement about that question, under what voting rule should the disagreement be resolved? When the group remembers that it is already at sea, however, these questions will seem less impressive. There is no neutral standpoint to occupy until the high-level questions are resolved, no way to escape having some status quo or other. While the discussion continues, the ship must sail on, with all its leaks and flaws. The crucial question is what improvements can be made in the interim, and the sensible among the assembly will seek ways to address and resolve that question before things get too bad. In the happiest case, the assembly will consider and approve, unanimously, proposals that count as improvements on any conception of democratic theory. In a more realistic case, however, a series of proposals can be adopted by large supermajorities—perhaps differing supermajorities on different issues—whose members share enough common ground, and whose causal and empirical hunches are sufficiently similar, that they can get on with the job together. There will be protests that, after all, adoption by a supermajority of democrats begs theoretical questions. How can a proposal that is opposed by some democrats in good standing count as democratically desirable, as opposed to partisan or sectarian? Why doesn’t the voting rule in the assembly itself require unanimity? But those protests will rightly be ignored, because something or other must be done.

246

Mechanisms of Democracy

Conclusion: Institutional Design as a Going Concern

The winning proposals will typically be small-scale, feasible suggestions that respect the principle of marginalism, creating substantial gains on some margins of democratic value even if they create smaller losses on others. My aspiration has been to suggest proposals of this sort. Perhaps the particular proposals I have offered will be defeated in the assembly of democrats. But to reject all suggestions for improvement is itself a democratically controversial choice of a particular institutional design. Quite plausibly, it is the least feasible choice of all.

Institutional Design Writ Small

247

This page intentionally left blank

Acknowledgments & Sources

P

arts of the underlying material in Chapters 5, 6, and 7, and in the introduction to Part II, although heavily revised for this book, were written in collaboration with three superb co-authors: Elizabeth Garrett, whose formidable expertise in congressional institutions is apparent in both Chapters 6 and 7, Jacob Gersen, and Eric Posner. Our joint work is cited below. I thank them for their kind permission to use that work as a starting point; any errors I have introduced are my own. John Ferejohn, Daryl Levinson, Sandy Levinson, Frank Michelman and Fred Schauer provided thorough and helpful comments, while Cass Sunstein encouraged the project and helped me past the bad moments that every author experiences. Jon Elster has been a formative influence, and encouraged the project that eventually turned into Part I of this book. His work is cited throughout the notes, but my intellectual debt to him is much greater than the citations indicate. Ellen Keng and Daniel Klaff provided helpful logistical support and research assistance. Chris Collins at Oxford University Press championed the project while Amanda Orchanian steered it to completion. Most important of all, at every stage my wife, Yun Soo, and my children, Emily and Spencer, have been supportive, patient, and delightful. I am deeply grateful for their encouragement, and hope they will one day read and enjoy the indirect results of their work. I would like to acknowledge, and to thank, various journals and volumes in which earlier versions of some of the material appeared, as follows: For Chapter 1, Adrian Vermeule, Veil of Ignorance Rules in Constitutional Law, 111 Yale Law Journal 399 (2001); for Chapter 2, Adrian Vermeule, Political Constraints on Supreme Court Reform, 90 Minnesota Law Review 1154 (2006), and Adrian Vermeule, Self-Defeating Proposals: Ackerman on Emergency Powers, 75 Fordham Law Review 631 (2006) (a few

Institutional Design Writ Small

249

Acknowledgments & Sources

paragraphs from the latter paper were also adapted for the Introduction); for the introduction to Part II, Eric A. Posner and Adrian Vermeule, Interring the Nondelegation Doctrine, 69 University of Chicago Law Review 1729 (2002); for Chapter 3, Adrian Vermeule, Submajority Rules: Forcing Accountability on Majorities, 13 Journal of Political Philosophy  (2005); for Chapter 5, Jacob Gersen and Adrian Vermeule, Chevron as a Voting Rule, 106 Yale Law Journal 676; for Chapter 6, Elizabeth Garrett and Adrian Vermeule, Transparency and the U.S. Budget, in Fiscal Challenges (Elizabeth Garrett, Elizabeth Graddy and Howell Jackson, eds., Cambridge University Press, forthcoming 2007); for Chapter 7, Elizabeth Garrett and Adrian Vermeule, Institutional Design of a Thayerian Congress, 50 Duke Law Journal 1277 (2001), a version of which was reprinted in Congress and the Constitution (Neal Devins and Keith Whittington, eds.) (Duke University Press 2005). Chapter 4 is based on an article entitled “Absolute Majority Rules” that will appear in the British Journal of Political Science, Volume 37 Part 4 (October 2007). Special thanks to the Journal and to Cambridge University Press for permission, and to Linda Hart, Patrick McCartan and Prof. Albert Weale for their willingness to accommodate. The colleagues and editors who provided helpful suggestions on these efforts are too numerous to name, but I am very grateful to them.

250

Mechanisms of Democracy

Index

A Absenteeism, absolute majority voting rules, 124, 131–32, 138–39 Absolute majority voting rules, 115–42 absenteeism, 124, 131–32, 138–39 absolute voting rules, defined, 117 abstention, 132, 138–40 as negative vote, 117–18, 121, 138–39 alternatives, 126–33 proxy voting, 131–33 quorum rules and quorum-breaking, 128–30 reversal by voting, 126–28 supermajority multipliers, 130–31 asymmetries, 136 chamber majority, 131 Condorcet Jury Theorem, 134 “demi-prevarication,” 138 Democratic accountability, 133–37 generally, 84 House Rule XI, 128 insurance effect of, 122–25 legislative majorities, 133–37 majoritarian control, 122–25 May’s Theorem, 133–34 minimum majority, 116 minority action, threat of, 124 multiplicands, 116–17, 119, 123 Napoleonic minorities, 123, 125–28, 130–31, 135–36 and nonvoters, 120

Institutional Design Writ Small

overview, 118–22 probabilistic voting, 135 proxy voting, 131–33 quorum rules and quorum-breaking, 128–30 Reed rule, 129 reversal by voting, 126–28 Robert’s Rules, 118–19 rump majority, 128 simple majority rule, 117, 119–20 standard quorum rule, 119 strategic nonparticipation, 137–42 supermajority multipliers, 130–31 unanimity, 131 Abstention, voting, 132, 138–40 as negative vote, 117–18, 121, 138–39 Accountability absolute majority voting rules. See Absolute majority voting rules Chevron doctrine. See Chevron doctrine defined, 4 and delegation, 81–84 generally, 75–76 and democracy, 81–83 good vs. bad, 11 judicial decisionmaking. See Judicial decisionmaking and lawmaking, 83–84 legislative, generally, 23 optimizing, 84 promotion of, 8

251

Index submajority voting rules. See Submajority voting rules theorists, 5–6 and transparency, 180–82 accountability vs. transparency, 11 budget process, 196–200, 204 Ackerman, Bruce proposal on regulating emergency powers, 64–65, 67–70, 210–11 Administrative Procedure Act (APA), 154 Africa proxy voting, 132 Agencies “reasonable” interpretations, 172-73 submajority voting rules, 97–98 Agendas, submajority voting rules agenda-setting, 104–7 distributing power, 91–96 generally, 96–97 Aggregate norms versus individual norms judicial decisionmaking, 150–51 Aggregation Chevron doctrine, 159–66 attitudinal model of judging, 164–65 bias, 164–65 conceptual problems, 159–62 first-order interpretation, 159–60 internalization versus aggregation, 165–66 psychological burdens, 162–64 of information, and deliberation, 7 Agriculture Committees, 238–39 Alabama absolute majority voting rules, 120–22 American Bar Association, 239 American Civil Liberties Union, 239 Antiretroactivity rules, 40–41 APA. See Administrative Procedure Act (APA) Appointments, Senatorial deference, 148–49 Arguing budget process, transparency, 194–96 Asymmetries absolute majority voting rules, 136 supermajority voting rule, 169n. 34 Attitudinal model of judging, 164–65 AT&T, MCI v., 163 Attorney General, 233

252

B Bargaining budget process, transparency, 194–96 constitutional issues in Congress, improving costs, 223 Base Realignment Act, 210 Bentham, Jeremy, 5 absolute majority voting rules, 122, 127, 140, 142 “intentional surprise,” 122 strategic nonparticipation, 138 Bias Chevron doctrine, 164–65 information, restriction to reduce, 59 self-serving, 33 and veil rules, 31 Bicameralism, 58, 81 Bill of Attainder Clause, 35 and delay mechanisms, 54 generality requirement, 41, 43–44 and impartiality, 58 and prospectivity, 60–61 uncertainty and distribution of veil rules, 56, 58 Bolton, John, 140 Brazil proxy voting, 132 Breyer, Stephen, 162–64 Budget Act (1974), 201 Budget process budget summits, 214–15 circumvention, 212 congressional committees, 214 floor debate, 213 generally, 24 leaks, 212–15 political constraints, 209–15 circumvention, 212 leaks, 212–15 optimal rules, supply of, 209–12 supply-side issues, 209–10 supply-side issues, 209–10 transparency, 3, 183–215 accountability, 196–200 arguing, 194–96 bargaining, 194–96 costs, 209–15 delayed disclosure (proposal), 203–8

Mechanisms of Democracy

Index democratically harmful transfers, 193–94 federal budget process, 186–91 overall allocation, opaque (proposal), 200–203 overview, 185–86 political constraints, 209–15 proposals, 200–208 selective transparency, 184–85 tradeoffs, 192–200, 204 Burke, Edmund, 140, 218 C Calibration Chevron doctrine, 166 versus fuzziness, judicial decisionmaking, 151–52 California absolute majority voting rules, 115 Canada submajority voting rules, 99 CBO. See Congressional Budget Office (CBO) Certainty Chevron doctrine, 167–68 versus variance, judicial decisionmaking, 152–53 Chamber majority, 131 Chevron doctrine aggregation, 159–66 attitudinal model of judging, 164–65 bias, 164–65 conceptual problems, 159–62 first-order interpretation, 159–60 internalization versus, 165–66 psychological burdenrs, 162–64 background, 145–46 bias, 164–65 calibration, 166 certainty, 167–68 Condorcet Jury Theorem, 168, 170–72 costs, 168, 172–73 decision costs, 172–73 and deference regimes, 157–58 and democracy, 153–56, 175–76 and expertise, 153–56 generally, 144–45

Institutional Design Writ Small

internalization versus aggregation, 165–66 internal point of view versus voting rules, 173–74 majority rule, 158 May’s Theorem, 168–69 objectives, 168 permissive interpretations, 156–57 supermajority voting rule, 144–45, 158 supply-side, political constraints and, 174–75 voting theorems, 168–73 Chevron v. Natural Resources Defense Council. See Chevron doctrine Circumvention, budget process, 212 Clinton, Bill, 68, 69n. 26 Closed-class test, 43 Closure Act, 210 Cloture petition submajority voting rules, 86, 140 Collateral decisions submajority voting rules, 109 Commitment to equality, 16 and impartiality, 10–11 and uncertainty, 66–70 Committees on Constitutional Matters, 237, 239 Communist Party, 43 Condorcet Jury Theorem, 7 absolute majority voting rules, 134 Chevron doctrine, 168, 170–72 submajority voting rules, 107–9 Congress. See U.S. Congress Congressional Budget Office (CBO), 231, 233, 234 Congressional Record, 231 Congressional Research Service (CRS) American Law Division, 233–34 Government and Finance Division, 234 Consequentialist theories of democracy, 5, 15 Constitution. See U.S. Constitution Constitutional Convention. See U.S. Constitution Constitutional impact statements, 232 Constitutionalism and delegation, 80–81 and deliberation, 80–81

253

Index Constitutional issues, Congress, 217–44. See also Deliberation for detailed treatment Contracts clause, 61 Cox, Adam redistricting plans, proposal, 66, 71 Criminal statutes, retroactive, 57 CRS. See Congressional Research Service (CRS) Cycles, submajority voting rules, 104–7 D Decisionmaking and constitutional choice, 34–35 impartiality, 33 judicial, 23, 143–76. See also Judicial decisionmaking passive decisionmaker, 55 self-interested, 33, 36–37, 44 and uncertainty, 10 De facto norm of precedent, 46 De jure norm of precedent, 46 Delay Delayed Implementation Agreements (DIA), 66–67 disclosure. See Delayed disclosure and uncertainty, 62–66 Delayed disclosure, 8 budget process, transparency, 185, 203–8 Delayed effectiveness, 49–54 Delayed Implementation Agreements (DIA), 66–67 Delegation. See also Nondelegation doctrine and accountability, 75–76, 81–84 and constitutionalism, 80–81 and deliberation, 80–81 and democracy, 75–76, 81–83 and lawmaking, 83–84 Deliberation and aggregation, 7 and constitutionalism, 80–81 constitutional issues in Congress, 217–44 Committees on Constitutional Matters, 237 committee structure to discuss issues, 235–40

254

decision costs, improving, 222–23 definition of constitutional, 218 deliberation-enhancing mechanisms, 239–44 deliberative benefits, improving, 222–23 design principles, 230 dissemination of information, 230–33 error costs, improving, 223–26 expert staff, 233–35 “fire-alarm” model, 228–29 floor considerations, 240–44 improving deliberation, 222–27 and institutional design, 244 Judiciary Committee, 235–40 legislator’s behavior, 218–22 optimal nature of performance, 227–30 points of order, 240–44 production of information, 230–33 transition costs, improving, 226–27 as controversial ideal, 6–7 defined, 5 and delegation, 80–81 good vs. bad, 11–12 and institutional design, 244 legislator’s behavior, 218–22 and personal goals, 219–22 public interest, acting in, 218–22 reelection, maximizing chances of, 219 promotion of, 8, 24, 217–44 and transparency, 8, 180–81 deliberation vs. transparency, 11 Democracy and delegation, generally, 75–76 Democratic accountability absolute majority voting rules, 133–37 Democratic theories, 15–16 Democratic values, conflict, 14 Design, institutional. See Institutional design DIA. See Delayed Implementation Agreements (DIA) Discharge rule, submajority voting rules, 86 Disclosure, delayed. See Delayed disclosure

Mechanisms of Democracy

Index Due Process Clause, 39n. 23, 40 Durability generality, distinction, 47 veil rules, 45–49 E Eastern Europe constitution-making, 2 Effectiveness, delayed, 49–54 Electoral Count Act of 1887, 210 Elitism, 1 Ely, John Hart War Powers Resolution, proposal for strengthened version, 65, 210–11 Emergency powers regulation of, proposed, 64–65, 67–70, 210–11 Emoluments Clause, 35, 50, 52–54, 56 English Parliament submajority voting rules, 97 Equality, commitment to, 16 Error constitutional issues in Congress, 223–26 Executive branch impartiality, 58 uncertainty and distribution of veil rules, 56–57 Ex Post Facto Clause, 35, 54 and generality requirement, 42 prospectivity rules, 39, 60–61 uncertainty and distribution of veil rules, 56 F Factional oppression, 34 Federal budget process transparency, 186–91 current rules, 188–91 open meeting rules, 189 senses of, 187–88 strong secrecy, 188 weak secrecy, 188 Federal Communications Commission, 171 Federal Reserve System Federal Open Market Committee, 204, 208

Institutional Design Writ Small

Felons, 39 “Fire-alarm” model, 228–29 First-order interpretation judicial decisionmaking, 159–60 Floor considerations constitutional issues in Congress, 240–44 Floor debate budget process, 213 Framers, U.S. Constitution. See U.S. Constitution France proxy voting, 132 French National Assembly (1790s), 180 Fuzziness, calibration versus judicial decisionmaking, 151–52 G Gang of Fourteen, 149 Generality durability, distinction, 47 and prospectivity, 42–43 and veil rules, 41–45 Generalizations. See Generality H Hamilton, Alexander, 61–62 House of Representatives. See U.S. House of Representatives Howard-Barkley Bill, 103 Humean theory of political psychology, 59 Hypocrisy, criticizing force of, 194, 221 I Impartiality and Bill of Attainder Clause, 58 and consequentialist theories of democracy, 5 defined, 4 first-order, 27 and motivation, 62, 70 promoting, 28 promotion of, 8 second-order, 27 and self-interest, freedom from, 27 and uncertainty, 55–56, 66–70 and self-defeating proposals, 70 vs. commitment, 10–11

255

Index Impartiality (cont.) vs. information, 10 vs. motivation, 10 what constitutes, 34 Incentive incompatibility, 20 India (Parliament of India) absolute majority voting rules, 116 Information and aggregation of information, 7 dissemination of, constitutional issues in Congress, 230–33 and mechanism design, 9 neutrality, tradeoff, 59 production of, constitutional issues in Congress, 230–33 submajority voting rules, 96–97 vs. impartiality, 10 Institutional design, 1–29 challenges to, 10–13 and conflict, 10–13 consequentialist theories of democracy, 5, 15 democratic values, 4–8 as going concern, 245–47 incentive incompatibility, 20–21 and information, 8–10 aggregation of information, 7 institutional variables and values, 17 limiting conditions, mechanisms and, 17 margin, design at, 3 mechanism design, 8–10 mechanisms of democracy defined, 4 place of, 13–16 self-defeating proposals, 20–21 supply and demand, 18–20 theory-dependent, 12, 14–15 and tradeoffs, 10–13 writ small, 2–3 Institutional problems, judicial decisionmaking, 146–49 appointments, Senatorial deference, 148–49 Gang of Fourteen, 149 legislatures, deference to, 147 precedent, 148 Rule of Four, 148 “Intentional surprise,” 122 Interest groups, 198

256

analysis, 39–40 transfer-seeking, 199 Internalization versus aggregation Chevron doctrine, 165–66 Internalizing norms judicial decisionmaking, 150–51 Internal point of view versus voting rules, 173–74 International Emergency Economic Powers Act, 68 Iraq absolute majority voting rules, 116 Irreversibility thesis submajority voting rules, 104 Italian Parliament submajority voting rules, 97 J Japan submajority voting rules, 99 Johnson, Lyndon, 139 Joint Committee on Human rights (United Kingdom), 3, 20, 237 Journal clause submajority voting rules, 86, 89, 97, 100, 106, 112 JTC, 234 Judges decisionmaking. See Judicial decisionmaking tenure of justices, 63–64 and veil effects, 57 Judicial decisionmaking, 23, 143–76. See also Chevron doctrine accountability, 153–56, 175–76 agency interpretations, weighted in favor of, 158 aggregate norms versus individual norms, 150–51 aggregation, 159–66 attitudinal model of judging, 164–65 calibration, 166 versus fuzziness, 151–52 certainty, 167–68 versus variance, 152–53 Condorcet Jury Theorem, 168, 170–72 costs, 168, 172–73 decision costs, 172–73

Mechanisms of Democracy

Index deference regimes, 157–58 agency interpretations, weighted in favor of, 158 de novo judicial interpretation with majority rule, 157–58 majority rule, 157–58 supermajority rule, 158 and democracy, 153–56, 175–76 de novo judicial interpretation with majority rule, 157–58 and expertise, 153–56 first-order interpretation, 159–60 generally, 144 institutional problems, 146–49 appointments, Senatorial deference, 148–49 Gang of Fourteen, 149 legislatures, deference to, 147 precedent, 148 Rule of Four, 148 inter-judge variation, 167 internalizing norms, 150–51 internal point of view versus voting rules, 173–74 legal doctrine, development of, 146–49 appointments, Senatorial deference, 148–49 Gang of Fourteen, 149 legislatures, deference to, 147 precedent, 148 Rule of Four, 148 majority rule, 157–58 May’s Theorem, 168–69 objectives, 168 permissive interpretations, 156–57 Rule of Four, 152 Rule of Three, 152 supermajority rule, 158 and status quo, 169–70 supply-side, political constraints and, 174–75 tradeoffs, 149–53 aggregate norms versus individual norms, 150–51 calibration versus fuzziness, 151–52 certainty versus variance, 152–53 internalizing norms, 150–51 Rule of Four, 152

Institutional Design Writ Small

Rule of Three, 152 voting theorems, 168–73 Judiciary Committee, 235–40 Jury Theorem. See Condorcet Jury Theorem Justices’ tenure, 63–64 K Kosovo conflict, 68, 69n. 26 L Lawmaking and accountability, 83–84 and delegation, 83–84 Leaks, budget process, 212–15 Legal doctrine, development of, 146–49 appointments, Senatorial deference, 148–49 Gang of Fourteen, 149 legislatures, deference to, 147 precedent, 148 Rule of Four, 148 Legislative majorities absolute majority voting rules, 133–37 Legislator’s behavior constitutional issues in Congress, 218–22 and personal goals, 219–22 public interest, acting in, 218–22 reelection, maximizing chances of, 219 Legislatures constitutional deliberation, 24 legal doctrine, development of, 147 Line Item Veto Act of 1996, 210 Logrolling, 195n. 24 M Madison, James Constitutional Convention, 195n. 25 Emoluments Clause, 53–54 on self-interested decisionmaking, 28, 36–37, 58, 60–61 Twenty-Seventh Amendment, 51–52 Majoritarian control, absolute majority voting rules, 122–25 Manipulation, submajority voting rules, 104–7

257

Index Mason, George, 45, 100, 126 May’s Theorem, 133–34, 168–69 MCI v. AT&T, 163 Mead decision, 153, 155 Minnesota absolute majority voting rules, 115 Minoritarian holdup, submajority voting rules, 99–101 Morris, Gouverneur, 46, 100 Motivation and impartiality, 62, 70 and uncertainty, 62–66, 70 vs. impartiality, 10 Multi-member courts, supermajority voting rule, 7–8 Multiplicands, absolute majority voting rules, 116–17, 119, 123 N Napoleonic minorities, 123, 125–28, 130–31, 135–36 National Emergencies Act, 68, 69 Natural Resources Defense Council, Chevron v. See Chevron doctrine Near relations and submajority voting rules, 87–89 Neutrality information, tradeoff, 59 submajority voting rules, 90 Nondelegation doctrine generally, 23, 75, 143 standard view, 76–80 Nonvoters and absolute majority voting rules, 120 Norms aggregate norms versus individual norms, 150–51 O Office for Constitutional Issues (OCI), 3, 20, 234–40, 242 Office of Legal Counsel, 233 Office of Management and Budget (OMB), 233 Offices of Legislative Counsel, 233 Other-regarded partiality, 27

258

P Paine, Tom, 103–4 “Parchment barrier,” 36–37 Passivity, 55 Philadelphia Constitutional Convention, 126, 180, 195 Philippines absolute majority voting rules, 124, 141 Pluralism, 94–95 Points of order constitutional issues in Congress, 240–44 Political community of citizens, 5 Political constraints budget process, 209–15 circumvention, 212 leaks, 212–15 optimal rules, supply of, 209–12 supply-side issues, 209–10 Chevron doctrine, 174–75 submajority voting rules, 111–13 Popular enlightenment, 104 Posner, Richard, 152 Precedent legal doctrine, development of, 148 President of the United States emergency powers, 69 Probabilistic voting, 135 Prospectivity, 37–41, 60–61 and generality, 42–43 Proxy voting, 131–33 Public-choice theory, 82–83 Publicity, submajority voting rules, 96–97 Q Quorum rules and quorum-breaking absolute majority voting rules, 128–30 standard quorum rule, 119 R Railway labor disputes, 103 Rawls, John, 32–33 Redistricting plans, proposal, 66, 71 Reed rule, 129 Reform constraints on, 18–20 and delay, 63

Mechanisms of Democracy

Index proposals, 18 and supply and demand, 18–20 tenure of Justices’, 63–64 Reversal by voting absolute majority voting rules, 126–28 Reversibility submajority voting rules, 101–4 Robert’s Rules, 118–19 Roll-call voting, 240 submajority voting rules, 99–100, 109 Rule of Four institutional problems, judicial decisionmaking, 148 judicial decisionmaking, 152 legal doctrine, development of, 148 submajority voting rules, 86, 89–91, 112 intervening factors requirement, 102 Rule of Six, 109 Rule of Three, 152 Rules Committee, 240–42 Rump majority, 128 S Saxbe fix, 54 Scalia, Antonin, 163 Schumpeter, Joseph, 1, 5–6 Self-censorship, 19 Self-dealing, 34 costs of, 58 uncertainty and distribution of veil rules, 56 Self-defeating proposals institutional design, 20–21 and uncertainty, 70 Self-interest, 28 and decisionmaking, 33, 36–37, 44 freedom from, 27 and partiality, 27 and uncertainty, 35, 55 Senate. See U.S. Senate Seven Member Rule submajority voting rules, 86, 96, 112 Solicitor General, 233 Speech and Debate Clause, 212 Staff and constitutional issues in Congress, 233–35 Standard quorum rule, voting, 119 Stare decisis doctrine, 46

Institutional Design Writ Small

Status quo, submajority voting rules, 87, 109–11 Chevron doctrine, 169–70 Story, Justice, 52 Submajority voting rules, 85–113 and accountability, 85–113, 89–91. See also lines throughout this topic and agency, 97–98 agenda-setting, 104–7 distributing power, 91–96 agendas, generally, 96–97 Canada, 99 cloture petition, 86, 140 collateral decisions, 109 Condorcet Jury Theorem, 107–9 cycles, 104–7 discharge rule, 86 English Parliament, 97 generally, 84 House Rule XI, 86, 96, 112 House Rule XXVII, 88 information, 96–97 institutional rules and norms, 103 intervening factors requirement, 102 irreversibility thesis, 104 Italian Parliament, 97 Japan, 99 Journal clause, 86, 89, 97, 100, 106, 112 and majority motivations, 111–13 manipulation, 104–7 minoritarian holdup, 99–101 and near relations, 87–89 neutrality, 90 pluralism, 94–95 and political constraints, 111–13 predefined minority, 88 problems with, 99–111 agenda-setting, 104–7 Condorcet Jury Theorem, 107–9 cycles, 104–7 manipulation, 104–7 minoritarian holdup, 99–101 reversibility, 101–4 status quo point, selection of, 109–11 publicity, 96–97 reversibility, 87, 101–4 roll-call voting, 99–100, 109

259

Index Submajority voting rules (cont.) Rule of Four, 86, 89–91, 112 intervening factors requirement, 102 Rule of Six, 109 Senate Rule XXII, 86, 100–101 Seven Member Rule, 86, 96, 112 and status quo, 87, 109–11 supply-side issues, 111–13 transparency, 97–98 Supermajority multipliers, 130–31 Supermajority voting rule, 7–8 asymmetrical rules, 169n. 34 Chevron doctrine, 144–45, 158 and status quo, 169–70 internal point of view versus voting rules, 173–74 Supply and demand institutional design, 18–20 Supply-side issues budget process, 209–10 Chevron doctrine, 174–75 submajority voting rules, 111–13 Supreme Court. See U.S. Supreme Court T Takings clause, 40 Taxation, 82–83 Tenure, Justices’, 63–64 Texas absolute majority voting rules, 130 Tradeoffs budget process, transparency, 192–200 accountability, 196–200, 204 arguing, 194–96 bargaining, 194–96 democratically harmful transfers, 193–94 and institutional design, 10–13 gains produced, 14 judicial decisionmaking, 149–53 aggregate norms versus individual norms, 150–51 calibration versus fuzziness, 151–52 certainty versus variance, 152–53 internalizing norms, 150–51 Rule of Four, 152 Rule of Three, 152

260

Transparency and accountability, 180–82 budget process, 196–200, 204 budget process, 3, 183–215 accountability, 196–200, 204 arguing, 194–96 bargaining, 194–96 costs, 209–15 delayed disclosure (proposal), 203–8 democratically harmful transfers, 193–94 federal budget process, 186–91 overall allocation, opaque (proposal), 200–203 overview, 185–86 political constraints, 209–15 proposals, 200–208 selective transparency, 184–85 tradeoffs, 192–200, 204 defined, 4–5 and deliberation, 8, 180–81 federal budget process, 186–91 current rules, 188–91 open meeting rules, 189 senses of, 187–88 strong secrecy, 188 weak secrecy, 188 legal rules mandating, 16 necessity of, 6 overview, 215 promotion of, 24 selective transparency, 24, 184–85 strong secrecy, 188 submajority voting rules, 97–98 vs. accountability, 11 vs. deliberation, 11 weak secrecy, 188 Turnover, legislators, 206–7 Twenty-Seventh Amendment, 50–52, 54, 56 U UMRA. See Unfunded Mandates Reform Act of 1995 (UMRA) Unanimity, voting, 131 Uncertainty and commitment, 66–70 and decisionmaking, 35

Mechanisms of Democracy

Index and delay, 62–66 and distribution of veil rules, 56–62 executive branch, 56–57 skewed distributions, 57–58 and impartiality, 55–56, 66–70 and self-defeating proposals, 70 limits, 29, 55–71 and commitment, 66–70 and delay, 62–66 and distribution of veil rules, 56–62 and impartiality, 66–70 and motivation, 62–66, 70 and motivation, 62–66, 70 and self-defeating proposals, 70 and self-interest, 55 and veil rules, 32 Unfunded Mandates Reform Act of 1995 (UMRA), 228–31, 238 United Kingdom Joint Committee on Human rights, 237 rights, 3 Wednesbury case, 144 United States. See specific topic U.S. Congress accountability, 76–80 congressional committees, budget process, 214 deliberation, constitutional issues, 217–44. See also Deliberation for detailed treatment Electoral College, 77 federal budget process. See Federal budget process First Congress, 51 median member, 78 Office of Constitutional Issues (proposed), 3, 20 submajority voting rules, 106–7 supervision of agencies, 82 veil rules, 28 U.S. Constitution. See also Philadelphia Constitutional Convention amendments, 46. See also specific amendment Article V, 35, 46, 50 clauses. See specific clause Constitutional Convention, 32, 48 submajority voting rules, 97–98, 100

Institutional Design Writ Small

distribution of veil rules, 58 Framers, 180 Humean theory of political psychology, 59 submajority voting rules, 98 veil rules, 21, 29, 59 view of executive, 62 Origination Clause, 241 prospectivity, 38 veil-generating rules, 21 veil rules limits on distribution, 59, 60 optimal use of, 28 U.S. House of Representatives absolute majority voting rules, 124, 128–29 agenda-setting power, distributing, 94 discharge rule, 86 House Rule XI absolute majority voting rules, 128 submajority voting rules, 86, 96, 112 House Rule XXVII submajority voting rules, 88 and Rules Committee, 241–42 submajority voting rules. See Submajority voting rules U.S. Senate federal budget process. See Federal budget process Senate Rule XXII, 86, 100–101 submajority voting rules. See Submajority voting rules U.S. Supreme Court closed-class test, 43 nondelegation doctrine, 76 and prospectivity, 61 V Variance, certainty versus judicial decisionmaking, 152–53 Veil of ignorance arguments, 45 Veil rules, 31–54 and bias, 31 and constitutional choice, 34–35 defined, 31 delayed effectiveness, 49–54 distribution of veil rules and uncertainty, 56–62

261

Index Veil rules (cont.) durability, 45–49 generality, 41–45 generally, 22 limits, 22, 28–29 prospectivity, 37–41 self-interested decisionmaking, 36–37, 44 and uncertainty, 32 distribution of veil rules, 56–62 executive branch, 56–57 skewed distributions, 57–58 veil mechanisms, 37–54 veil of ignorance arguments, 45 Voting absolute majority voting rules. See Absolute majority voting rules

262

alternative rules, 3 Chevron doctrine. See Chevron doctrine democratic values, promotion of, 9 majority rule, 9 submajority voting rules. See Submajority voting rules supermajority voting rules. See Supermajority voting rule W War Powers Resolution, 69 strengthened version, proposal, 65, 210–11 weakness of, 68 Wednesbury case, 144 Wilson, James, 98

Mechanisms of Democracy