Complementary Constraints: Separation of Powers, Rational Voting, and Constitutional Design

Nzelibe, Jide O., Stephenson, Matthew C., Harvard Law Review

TABLE OF CONTENTS

I. SEPARATION OF POWERS AND ELECTORAL

    ACCOUNTABILITY: AN OVERVIEW                         624

II. HOW SEPARATION OF POWERS AFFECTS

    RATIONAL RETROSPECTIVE VOTING                       627

    A. The Baseline Case: A Single Elected Agent        627

    B. The Impact of Separation of Powers on Electoral  631
       Strategies and Political Behavior

III. IMPLICATIONS OF THE ANALYSIS                       636

    A. Presidential Empire Building                     637

    B. Responsibility Shifting                          639

    C. Gridlock                                         643

    D. Voter Welfare                                    645

IV. LIMITATIONS OF THE ARGUMENT                         647

    A. Rational Retrospective Voting                    647

    B. Effective Electoral Discipline                   650

    C. Clarity of Responsibility                        652

CONCLUSION                                              653

The voter, the political theorist V.O. Key once observed, is the "rational god of vengeance and of reward." (1) The extent to which voters can influence policy choices by rewarding or punishing politicians has long been a central concern for both political science and legal scholarship. Few political systems, however, rely solely on voter discipline to constrain elected policymakers. Instead, most advanced democracies rely on a combination of both electoral discipline and some form of internal separation of powers to reduce political "agency slack" (the deviation between the behavior of political agents and what the voter-principals would prefer). Electoral accountability ameliorates agency slack by punishing poorly performing incumbents and rewarding successful ones. (2) Separation of powers ameliorates agency slack by reducing the possibility that a biased or parochial interest group will be able to use the power of the state for its own ends. (3)

While the academic literature contains numerous sophisticated treatments of both elections and separation of powers, the literature has paid surprisingly little attention to how these different forms of institutional control interact. (4) The possibility of such interaction raises a number of important questions. To what extent does the separation of powers influence how voters wield their electoral power to reward or punish incumbent politicians? Is the separation of powers principally a substitute for electoral discipline, restraining biased government actors when voters are not able to do so effectively? Or might certain separation of powers regimes complement electoral accountability, making electoral discipline more effective than it would be otherwise? The answers to these questions bear on a more significant question: What institutional configuration best serves voter welfare? Are voters better off when one institution has exclusive control over a policy decision? Are they better off when two or more institutions must act together, as when the President must get congressional authorization or Congress must secure presidential approval? Or would rational voters prefer a system in which the different branches of government have the option of acting jointly or unilaterally?

While this Article does not aspire to provide comprehensive answers to these questions, it contributes to our understanding of these issues by analyzing how separation of powers and electoral accountability interact in a particular political environment. We consider a setting in which rational retrospective voters can observe both ex post policy outcomes and the decisionmaking process--including which institutions participated and the positions that they took--but voters can observe neither politicians' true preferences nor whether the decision taken was the correct one from an ex ante perspective. Moreover, we assume that while voters cannot observe whether a politician is biased, voters can infer (from the politician's party affiliation or other information) the likely direction of any such bias. (5) In this environment, we show that rational voters will adjust their punishment and reward strategies to compensate for the expected bias of the political decisionmakers but will do so in markedly different ways under different separation of powers regimes. This interaction between the institutional regime and rational voters' electoral strategies has positive implications for political behavior as well as normative implications for optimal institutional design.

A sketch of our argument is as follows. Consider as a baseline case a situation in which all authority over some policy decision is concentrated in a single elected official, such as the President. The President may be biased, but the voters are not sure whether he is. The voters are also unsure about which policy would best serve their interests. In this case, a rational voter would prefer an asymmetric reward and punishment strategy: When the President takes an action that is consistent with his possible biases, voters will reward him less generously if the policy succeeds and punish him more harshly if it fails. (For example, voters who suspect that the President might be excessively enthusiastic about military interventions abroad might punish the President more harshly if an intervention fails and reward the President less generously if it succeeds.) In contrast, if the President's action goes against his likely biases, voters will reward success more generously and punish failure less harshly. In this way, voters offset (imperfectly) the biases to which the President may fall prey.

This asymmetric strategy, however, has significant costs. These costs are due to the voters' uncertainty about both the correct policy choice and the true magnitude of the President's bias. If the incumbent President is more biased in favor of a particular policy than the voters expected, the voters' electoral strategy will not adequately deter unwarranted action (resulting in "false positives"). On the other hand, if the President turns out to be less biased than expected, the voters' electoral strategy will deter action even in some cases where that action would have been in the voters' interest ("false negatives").

An institutional separation of powers allows for a more refined strategy, reducing the probability of false positives and false negatives. The reason is that when voters can use electoral rewards and punishments to manipulate the incentives of two actors rather than one, they do not need to impose such large asymmetries in the rewards for success and the punishments for failure. So, for example, when independent branches make policy jointly--as when the President seeks and receives congressional authorization for a military intervention abroad--rational retrospective voters can employ a more nuanced strategy, with less of an asymmetry in the magnitude of political credit and blame, because Congress will screen out some of the President's undesirable policies. This dynamic, upon which we elaborate in Part II, has a number of implications for both positive and normative constitutional theory. Four in particular stand out:

First, our analysis casts doubt on the oft-repeated claim that presidents will usually oppose sharing power with Congress. On the contrary, presidents will often forego the opportunity to act unilaterally because doing so reduces their political risk.

Second, our analysis provides a rationalist account for forms of political behavior that are sometimes thought to derive from voter confusion or irrationality, such as the apparent ability of politicians to "shift responsibility" for unpopular or risky actions by involving other political actors in the decision. Whereas the more conventional responsibility-shifting explanations imply that the separation of powers tends to undermine electoral accountability, our alternative hypothesis suggests that behavioral patterns that superficially resemble responsibility shifting may actually indicate that the separation of powers is facilitating effective electoral control.

Third, contrary to conventional wisdom, we show that separation of powers does not necessarily increase "gridlock" or "status quo bias"--or at least that claims to this effect may be exaggerated. In a system in which one actor (say, the President) has authority to act unilaterally, the President has great formal flexibility to choose whatever policy he wants, but rational voters will punish the President's policy failures quite harshly in order to offset the President's possible bias. Adding another veto player--for example, by requiring congressional approval of presidential initiatives--has two effects. The first-order effect is to reduce the President's freedom of action because there are some policies an unconstrained President would prefer to enact that Congress would oppose. The second-order effect derives from rational voters' awareness of the first-order effect: because voters know that Congress will block some fraction of undesirable presidential initiatives, voters do not need to punish the President for policy failure as harshly in order to compensate for his possible bias. This second effect--the relaxation of the electoral deterrent to action--tends to increase the President's propensity to initiate policy change. While there is no guarantee that this second-order effect will fully offset the first-order effect of adding a veto player, such a result is possible. Thus, claims that the separation of powers necessarily reduces the quantity or frequency of policy change may need to be reconsidered.

Fourth, and perhaps most importantly, our analysis suggests that voters may be better off under a system of separated powers in which the agenda-setting political actor (for example, the President) has the option to seek the approval of another branch (for example, Congress) than under a regime in which such joint approval is required. We also contend that a system of separated powers (whether mandatory or optional) is better for voter welfare than one in which all political authority in some policy domain is concentrated in one branch of government. The intuition for this result is that voters are better off when they can employ more nuanced electoral strategies. When one actor (say, the President) has unilateral authority in some policy domain, the voters have only one lever at their disposal: their electoral support for the actor with authority over the policy decision. By contrast, under a mandatory separation of powers regime (for example, when both the President and Congress must agree to enact a new policy), the voters have two levers at their disposal: their electoral support for each of the two agents. In a regime where one branch (for example, the President) may seek the permission of another branch (for example, Congress), but may also act unilaterally, the voters can assign the primary decisionmaker different levels of electoral support depending not only on the ultimate outcome of the policy, but also on whether that branch acted unilaterally or with authorization from the other branch. Thus, an optional separated powers system gives the voters three levers with which to calibrate their electoral strategies: electoral support for the President in the case of unilateral action, electoral support for the President in the case of joint action, and electoral support for Congress. (6)

While our analysis focuses on the benefits to voters of increasing the number of ways a policy choice might be made, we recognize that this additional complexity also has costs. As institutional complexity increases, voters are likely to face greater difficulty in attributing responsibility for policy decisions. Thus, despite our central normative claim, we do not think that increasing the level of institutional complexity or power diffusion will always benefit voters. To put this point another way, adding more actors to the policymaking process has two effects on the efficacy of external electoral accountability, and these two effects cut in opposite directions. On one hand, the participation of a greater number of elected agents in a policy decision enables voters to use more refined electoral strategies, so long as voters can assess not only the ultimate policy outcome, but also the process that led to that outcome. On the other hand, the greater the complexity of the process and the larger the number of actors involved, the greater the informational burden on voters. If this burden becomes too great, voters might not be able to correctly attribute responsibility for different decisions, and this may undermine electoral discipline as a constraint on incumbents.

This latter "clarity-of-responsibility" effect has received more attention than the former effect, (7) which is the subject of our analysis. While we acknowledge the clarity-of-responsibility problem--indeed, it is the principal reason why our analysis would not support the creation of an arbitrarily large number of government branches--we conjecture that this effect is not overwhelming in a simple separation of powers system with only two or three institutional actors. We suggest that, at least for high-salience policy issues, the benefits to voters of being able to employ a more refined reward and punishment strategy exceed the informational costs associated with the added institutional complexity.

This Article proceeds as follows. Part I briefly reviews the existing literature on how separation of powers interacts with electoral accountability. Part II elaborates our basic theoretical argument, and Part III outlines the argument's main positive and normative implications. Part IV discusses some of the key assumptions and limitations of the analysis. A brief conclusion follows.

I. SEPARATION OF POWERS AND ELECTORAL ACCOUNTABILITY: AN OVERVIEW

Political leaders are supposed to use the power of the state to promote the welfare of the citizenry, yet leaders may be tempted to pursue their own agenda or to serve parochial interests at the expense of public welfare. Political systems may address this agency problem through two methods. The first is competitive elections. Elections can reduce agency slack in government in two ways. First, voters can try to select "good types." (8) That is, voters may vote prospectively in order to put into office officials who are competent and who share the voters' values. Second, voters can try to sanction poor performance. (9) That is, voters may vote retrospectively, rewarding incumbents for good policy outcomes and punishing them for bad ones, in order to create stronger incentives for incumbent politicians to advance the voters' interests.

A second strategy for reducing political agency slack focuses on institutional separation of powers. The classic Madisonian justification for the separation of powers is that ambition will counteract ambition, with the institutional and political rivalries between the branches of government serving to check the excesses or abuses of any one branch. (10) There are numerous modern variants on this line of argument, but the essential idea is that the "principals" (the citizens) can mitigate the political agency problem by employing multiple agents whose potential biases are at least partially uncorrelated. (11)

Separation of powers is sometimes characterized as a partial substitute for electoral accountability. That is, some contributions to constitutional theory (in both the legal and the political economy fields) suggest that robust electoral accountability obviates the need for an entrenched separation of powers, but reliance on separation of powers becomes necessary when electoral discipline is ineffective. (12) Thus, James Madison argued in the Federalist Papers that although elections would be the "primary control" on government, separation of powers was an important "auxiliary precaution" in light of the imperfection of electoral control. (13)

While there are undoubtedly cases in which electoral accountability and separation of powers act as substitutes in this way, the separation of powers constraints and electoral constraints may also complement one another. Some strands of the literature have recognized this possibility. The existing claims regarding how the separation of powers regime might affect the efficacy of electoral discipline can be divided loosely into two lines of argument, one more pessimistic about the impact of separation of powers on electoral accountability, and the other more optimistic.

The more pessimistic view emphasizes the so-called "clarity-of-responsibility" problem. (14) According to this view, diffusion of authority among multiple government actors makes it difficult for voters to figure out whom to blame or reward, thereby weakening electoral incentives across the board. As a result, politicians may be excessively reluctant to undertake socially desirable policies because they will receive only a fraction of the credit. Leading constitutional scholars and comparative political scientists, at least as far back as Woodrow Wilson, have invoked a version of this argument to claim that the American-style separation of powers hinders effective governance. (15) The clarity-of-responsibility problem may also make politicians more likely to take actions that benefit parochial interests at the expense of a majority of voters. For example, Congress might delegate controversial or unpopular decisions to executive agencies, while continuing to influence these decisions behind the scenes, so as to deliver rents to special interest groups while blaming the agency for unpopular policies. (16) Another alleged example of this sort of blame-shifting …

Questia, a part of Gale, Cengage Learning. www.questia.com
Publication information: Article title: Complementary Constraints: Separation of Powers, Rational Voting, and Constitutional Design. Contributors: Nzelibe, Jide O. - Author, Stephenson, Matthew C. - Author. Journal title: Harvard Law Review. Volume: 123. Issue: 3 Publication date: January 2010. Page number: 617+. © 2007 Harvard Law Review Association. COPYRIGHT 2010 Gale Group.

This material is protected by copyright and, with the exception of fair use, may not be further copied, distributed or transmitted in any form or by any means.