The welfare effects of minority-protective judicial review

Abstract

Constitutional theorists usually assume that minority-protective judicial review leads to outcomes more favorable to the protected minority and less favorable to the majority. Our analysis highlights an indirect effect of judicial review that complicates this conventional wisdom. Without judicial review, pro-majority and pro-minority leaders adopt different policies. Because judicial review limits the degree to which pro-majority leaders can adopt anti-minority policies, it becomes easier for pro-minority leaders to ‘mimic’ pro-majority leaders by adopting the most anti-minority policy that the judiciary would uphold. Furthermore, if judicial invalidation of anti-minority policies is probabilistic rather than certain, pro-majority leaders may propose even more extreme anti-minority policies in order to deter pro-minority leaders from mimicking. These effects can sometimes nullify, or even reverse, the assumed relationship between minority-protective judicial review and pro-minority outcomes. When such reversal occurs, majoritarian democrats should favor minority-protective judicial review, while those concerned with protecting unpopular minorities should oppose it.

Keywords

Countermajoritarian difficulty judicial review political agency

1. Introduction

Constitutional judicial review, which empowers unelected judges to reject the decisions of elected legislatures and executives, raises what Bickel (1962) described as a ‘countermajoritarian difficulty.’ The countermajoritarian nature of judicial review has led many constitutional theorists to advocate rejection or curtailment of judicial power (Tushnet, 2000; Waldron, 2006). Others, however, have emphasized that majoritiarian policymaking may lead to an undesirable ‘tyranny of the majority’ in which certain vulnerable minority groups consistently lose out (Guinier, 1994). Many influential legal scholars see judicially enforced constitutional limits as a possible remedy for the pathologies of majoritarianism (Chemerinsky, 2004; Dworkin, 1985; Ely, 1980). Important contributions to the political economy literature suggest similar conclusions (Maskin and Tirole, 2004; Rogers and Vanberg, 2007).

All sides of the normative debate over countermajoritarian judicial review make certain positive assumptions about the effect that such review is likely to have. One such assumption is that minority-protective judicial review leads to policy outcomes that are closer to those favored by the relevant minority group and further from those favored by the majority. This assumption, which undergirds both the standard critique of judicial review as antidemocratic and the standard defense of judicial review as a bulwark against majoritarian excesses, is natural and intuitive. As this paper will show, however, it may often be wrong. Although minority-protective judicial review sometimes benefits the minority at the expense of the majority, such review can sometimes benefit both the majority and the minority, or may affect the expected welfare of one group without affecting the other group’s expected welfare at all. In some cases, seemingly countermajoritarian judicial review may even benefit the majority at the expense of the minority.¹

These results arise because judicial review, which, as we model it, imposes an outer limit on how anti-minority the final policy outcome may be, not only has the direct effect of prohibiting overly anti-minority policies, but may also have an indirect effect on leader strategies. Specifically, a judicially imposed constraint on policy outcomes alters leaders’ ability and incentive to use their policy choices to conceal or reveal their true characteristics, and thereby to affect their popularity with voters. Leaders who share the majority group’s policy preferences have an incentive to credibly signal this fact, while leaders who sympathize with the minority may have an incentive to conceal this by mimicking the expected behavior of pro-majority leaders. Minority-protective judicial review can interfere with the pro-majority leader’s ability to signal his type, thereby leading pro-minority leaders to mimic more often. To illustrate, suppose that without judicial review, a pro-majority leader would adopt a policy so unfavorable to the minority group that a pro-minority leader would not be willing to mimic it. Under these circumstances, the pro-minority leader would instead adopt the minority group’s most-preferred policy. Now suppose that judicial review imposes a constraint on the pro-majority leader’s choice, forcing him to enact a policy somewhat more favorable to the minority faction. This reduces the cost to the pro-minority leader of mimicking the pro-majority leader; if this cost becomes small enough, the pro-minority leader will pool with the pro-majority leader, adopting a policy that is less favorable to the minority group than the pro-minority leader would have adopted otherwise.

Moreover, if there is uncertainty as to whether the judiciary will invalidate policies that impinge on minority interests, then judicial review may expand the set of anti-minority policies that the pro-minority leader would be willing to propose. Because judicial review means that the pro-minority leader may not have to live with the consequences of an extreme anti-minority policy, he is more willing to propose relatively extreme policies if this would allow him to mimic the pro-majority leader. However, the pro-majority leader reacts by making an even more extreme anti-minority policy proposal, so that he can continue to distinguish himself. Although the court will sometimes invalidate the pro-majority leader’s extreme proposal, making the minority group better off than it would have been in the absence of judicial review, there will also be cases in which the court allows the pro-majority leader to implement this more extreme anti-minority policy, making the minority group worse off than it would have been in the absence of judicial review.

This paper uses a family of related formal models to illustrate the above effects and to analyze their implications. Importantly, the implications of the analysis extend beyond the domain of judicial review. Indeed, our results would apply, with appropriate modifications, to any institution that imposes a constraint on the degree to which a policy decision may impinge on some group or cause. In that sense, our paper contributes more generally to the literature on the formal analysis of checks and balances.

In Section 2 we develop a baseline model of policy choice in the absence of judicial review. In this model, the incumbent leader may either share the policy preferences of the majority group or otherwise share the preferences of an unpopular minority group. In addition to caring about policy, the leader also has an electoral incentive to appear to share the majority’s preferences. This baseline model allows us to isolate a particular set of political incentives and to assess their effect on equilibrium political outcomes. Section 3 introduces judicial review, modeled as a limit on the extent to which the policy outcome may harm the minority group’s interests. Section 3 demonstrates that judicial review, by reducing the cost to the pro-minority leader of mimicking the pro-majority leader’s policy choice, can lead pro-minority leaders to be less faithful agents of the minority group. Section 4 analyzes a variant of the model in which judicial invalidation of anti-minority policies is probabilistic rather than certain, and shows how uncertain review induces pro-majority leaders to make even more extreme anti-minority proposals. A brief conclusion follows, and an online appendix provides proofs of formal statements.

2. The baseline model

We first develop a simple model of competitive politics with voter uncertainty regarding the politicians’ preferences.² Consider an elected leader responsible for enacting some policy, represented as a point in a one-dimensional policy space, $x \in ℝ$ . There are two groups in society, a majority faction (M) and a minority faction (m). The minority’s ideal policy is y_m, while the majority’s ideal policy is y_M, where y_M > y_m. We might imagine that the minority faction is some traditionally disadvantaged ethnic or religious group. In this case, x = y_m might represent a policy of remedial affirmative action, while x = y_M might represent a principle of formal non-discrimination (‘colorblindness’) that in practice tends to favor the majority group. Intermediate polices, x ∈ (y_m, y_M), might represent more robust non-discrimination policies (including those that forbid facially neutral policies that have an adverse ‘disparate impact’ on the minority group). Values of x > y_M are more extreme policies, such as those that overtly discriminate against the disadvantaged minority. Alternatively, as in Acemoglu et al. (2013), we might think of the minority and majority factions in economic terms, for example as a wealthy elite and a working-class majority. In that case, x = y_m might represent a laissez-faire economic policy (most preferred by the wealthy minority), with higher values of x representing greater redistribution. In this case, x = y_M would be the level of redistribution preferred by the working-class majority, while values of x > y_M denote expropriative policies that go beyond what even the workers who ostensibly benefit from redistribution would favor.

The leader has policy preferences that may align either with the minority or with the majority. That is, the leader has a type, t ∈ {m, M}, where type-m leaders prefer x = y_m and type-M leaders prefer x = y_M. The probability that t = M is π ∈ (0, 1); the probability that t = m is 1 − π.³ The voters do not know t, but π is common knowledge: π can be thought of as the voters’ prior belief about the leader’s type when voters have taken into account all observable information about the leader before he selects x; such information might include the leader’s partisan affiliation, prior statements and actions, and demographic characteristics.⁴ Sometimes, the publicly observable information about the leader (including, for example, the leader’s own membership in the majority or minority group) will enable voters to make very confident inferences about the leader’s type; this would correspond to π very close to 0 or 1. In other settings, though, voters will have substantial residual uncertainty about the leader’s true sympathies. Our analysis is most relevant to these circumstances.

Since the leader and the voters care about policy, each suffers a utility loss when the chosen policy diverges from his or her ideal policy. Specifically, we assume that when policy x is implemented, an individual with ideal point y_t receives a policy payoff of v_t(x) = −(x − y_t)².

After the leader selects policy, an election is held. We assume that the leader wants to win the election, so his utility is increasing in his reelection probability, denoted r. A type-t leader’s preferences over (x, r) pairs are represented by the utility function

u_{t} (x, r) = v_{t} (x) + δ r .

The parameter δ ≥ 0 captures the weight the leader attaches to holding office; the greater this weight, the more he is willing to sacrifice his policy goals in order to win reelection.⁵

Figure 1 depicts how the leader trades off policy objectives against electoral objectives. This figure illustrates the indifference curves of the type-M leader (the gray curve) and the type-m leader (the black curve) through the policy and reelection probability pair (x′,r′). Note that these curves cross only once, with that of the type-M leader crossing from above, meaning the type-M leader needs less electoral compensation than the type-m leader in order to find a rightward shift in policy worthwhile. This single-crossing property, which is common in many canonical signaling games (e.g. Spence, 1973), ensures that our solution concept (D1 equilibrium) provides unique predictions.⁶

Figure 1.

Leaders’ trade-off between policy and reelection: illustration of single-crossing property. The indifference curve of the type-m leader through the policy and reelection probability pair (x′,r′) is depicted by the black curve, and the indifference curve of the type-M leader through the policy and reelection probability pair (x′,r′) is depicted by the gray curve.

The voters in our model form posterior beliefs about the leader’s type, given the leader’s policy choice. These beliefs, in turn, influence the incumbent’s electoral prospects. We represent the voters’ beliefs about the leader’s type with the function $μ : ℝ \to [0, 1]$ , where μ(x) is defined as the probability that the voters assign to the leader’s type being t = M when the leader chooses x. We assume (and micro-found in our online appendix) that the leader’s probability of reelection is an increasing function of the voters’ posterior that t = M. Given that the majority group is larger than the minority group, this assumption is natural if we also assume that each voter wants to elect a leader who shares her policy preferences, either because such a leader is more likely to act in the voter’s interest on other issues (which may not be directly observable), or because the voter has an innate preference for a leader who shares her values. To make the link between beliefs and reelection prospects explicit, we define a continuously increasing function F : [0,1] → [0,1], such that when the incumbent chooses x, his reelection probability is r = F(μ(x)). Thus, when voters infer that the incumbent’s type is t = M, the incumbent is reelected with maximal probability r = F(1), whereas when voters infer that the incumbent’s type is t = m, the incumbent is reelected with minimal probability r = F(0).

We solve for perfect Bayesian equilibria (PBE) that survive criterion D1. A candidate PBE consists of a strategy for the type-M leader, a strategy for the type-m leader, and a system of beliefs for the voters.⁷ In an equilibrium, the leader’s policy choice maximizes his expected utility given the voters’ beliefs, and the voters’ beliefs are derived via Bayes’ rule whenever possible. Such an equilibrium survives criterion D1 if, following any off-path policy, voters assume (with probability 1) that the leader is whichever type needs the smallest electoral inducement to make this deviation from on-path behavior worthwhile.⁸ We refer to a PBE that survives criterion D1 as a D1 equilibrium.⁹

The D1 refinement eliminates equilibria in which the type-M leader and the type-m leader pool on a common policy.¹⁰ Thus, in our baseline model, all D1 equilibria are separating and so voters are able to infer the leader’s type from his policy choice. In any such equilibrium, the type-m leader selects his ideal point, y_m, and the type-M leader selects a policy to the right of y_m. Given that the type-m leader has an electoral incentive to mislead voters about his type, the type-M leader’s policy choice must be sufficiently far to the right of y_m. This ensures that the policy cost the type-m leader would incur if he were to pool with the type-M leader would outweigh the electoral benefit. In particular, in any separating equilibrium (regardless of whether it satisfies D1), any policy x′ that belongs to the support of the type-M leader’s strategy must satisfy

u_{m} (y_{m}, F (0)) = v_{m} (y_{m}) + δ F (0) \geq v_{m} (x') + δ F (1) = u_{m} (x', F (1)) .

(1)

That is, the type-m leader’s payoff from implementing his ideal point and being identified as pro-minority must be at least that from implementing x′ and leading voters to believe he is pro-majority. Letting x⁺ denote the smallest value of x on (y_m, ∞) such that (1) is satisfied, we have that

x^{+} = y_{m} + \sqrt{δ [F (1) - F (0)]} .

In what follows, we refer to x⁺ as the separation threshold, since choosing a policy weakly greater than x⁺ is necessary and sufficient for the type-M leader to achieve separation. Consequently, when x⁺≤ y_M, the type-M leader can achieve separation by selecting his ideal point. In contrast, when x⁺> y_M, this is not possible: if the type-M leader were to select his ideal point, the type-m leader would have a strict incentive to pool. Thus, when x⁺> y_M, to achieve separation the type-M leader must propose a policy to the right of his ideal point. Note that whether x⁺ exceeds y_M depends on the weight the leader attaches to reelection (δ): the more the type-m leader cares about reelection, the more extreme the policy the type-M leader must pursue in order to achieve separation.

In addition to ensuring that equilibria are separating, D1 also eliminates all PBE in which the type-M leader incurs greater policy costs than necessary to achieve separation. Thus, in a D1 equilibrium, the type-M leader either selects his ideal point (if x⁺ ≤ y_M) or otherwise selects the separation threshold x⁺ (if x⁺ > y_M). Hence, our model admits two classes of D1 equilibria: (1) those in which both the type-m and type-M leaders select their respective ideal points, and (2) those in which the type-m leader selects his ideal point, while the type-M leader selects a policy to the right of his ideal point. We refer to equilibria in the former class as sincere separating equilibria, whereas we refer to equilibria in the latter class as extremist separating equilibria. We summarize as follows.

Proposition 1 Without judicial review, D1 equilibria exist and are separating. These equilibria have the following properties:

(Sincere separation) If x ⁺ ≤ y_M, the type-m leader selects the minority-preferred policy (y_m) and the type-M leader selects the majority-preferred policy (y_M).

(Extremist separation) If x ⁺ > y_M, the type-m leader chooses the minority-preferred policy (y_m), but the type-M leader chooses a policy that is more anti-minority than the majority group’s ideal policy (he chooses x = x⁺ > y_M).

Thus, in our baseline model with no judicial review, pro-majority leaders always perfectly distinguish themselves from pro-minority leaders, and pro-minority leaders always choose their ideal points. In the situation described in Proposition 1(a), where policy considerations weigh more heavily on incumbents (i.e. δ small), the pro-majority leader also selects his ideal point. More interesting is the case described in Proposition 1(b), where electoral considerations weigh more heavily on incumbents (i.e. δ large). In this case, the pro-majority leader adopts a policy that is too extreme even from the majority group’s perspective, such that both the majority and the minority would benefit if the pro-majority leader adopted a more moderate policy. Nevertheless, voters still ‘reward’ the pro-majority leader for adopting this sort of extreme policy by reelecting him with maximal probability.¹¹

While our main focus is on the effect of judicial review (or other minority-protective limits on the leader’s policy discretion), the result derived in Proposition 1(b) may have independent significance, for it identifies conditions under which leaders who share the majority’s preferences will go out of their way to demonstrate their antipathy to the minority group, even to the point of enacting measures that go beyond what the median voter (or the pro-majority leader himself) would actually prefer. That is, the preceding analysis suggests a possible explanation for why seemingly excessive anti-minority extremism is sometimes a successful political strategy. There are numerous examples that suggest the empirical plausibility of this dynamic. For instance, Acemoglu et al. (2013) use a model similar to ours to explain the prevalence of ‘populist’ politics, particularly in contemporary Latin America, whereby politicians implement policies well to the left of the median voter (and bad for the economy as a whole) in order to convince the electorate that they are not secretly sympathetic to the wealthy elite. Schama (1989) suggests a similar dynamic may have operated during the French Revolution, with revolutionary leaders embracing increasingly violent policies in order to signal that they were not secretly sympathetic to the royalists. This dynamic is also familiar in American politics. For example, historians have suggested that southern politicians in the mid-twentieth century staked out policy positions that were even more hostile to African-Americans than the median (white) voter would have preferred, in order to signal to these voters that the politicians were not ‘soft’ on maintaining the racial hierarchy. Perhaps the most notorious illustration of this is George Wallace’s declaration, after losing the Alabama gubernatorial election in 1958 to a Klan-backed candidate, that ‘no other son-of-a-bitch will ever out-nigger me again’ (Carter, 1995, p. 96). The dynamic also seems present in contemporary ‘culture war’-style politics, in which politicians and party activists stake out positions on high-profile issues that seem too extreme not only relative to the median voter in the electorate, but even with respect to the median voter within their own parties (Fiorina et al., 2010).

Even without this extremism effect (that is, even when both leader types implement their respective ideal points, as in Proposition 1(a)), one might be concerned about the welfare of the minority group; when anti-minority extremism of the sort predicted by Proposition 1(b) is present, one might worry about its adverse effects on the welfare of both groups. Those who share these concerns might be attracted to institutional solutions, such as judicially enforced legal constraints, that prevent leaders from enacting overly anti-minority policies. On the other hand, majoritarians might worry that such institutions would undermine basic democratic commitments. Indeed, from a majoritarian perspective, the more important problem highlighted by Proposition 1 is that pro-minority leaders implement a policy that favors the minority group, rather than catering to the median voter. Those sympathetic to that perspective might be skeptical of judicial constraints designed to protect minorities, although perhaps more mild constraints that mitigate the tendency toward ‘excessive’ anti-minority extremism (of the sort highlighted in Proposition 1(b)) might be welcome. The remainder of this paper explores whether, or under what conditions, these instincts about the effect of minority-protective judicial review are accurate.

3. The effect of judicial review

To investigate the impact of minority-protective judicial review, we modify the baseline model by introducing a judicially enforced legal limit, j, on how far to the right the leader’s policy choice may deviate from y_m, the position that most benefits the minority group (cf. Dragu et al., 2014).¹² If the leader selects an x ∈ (−∞, j], the court will uphold the policy.¹³ If the leader selects an x > j, the court will strike down the leader’s choice and implement x = j instead.¹⁴ The limit j is common knowledge.¹⁵ To reduce the number of cases considered, we take j ≥ y_m. Furthermore, since all proposals of x > j lead to a final policy of j, to simplify the exposition we assume that such polices are not proposed.¹⁶ We refer to judicial review as ‘countermajoritarian’ when j < y_M, and we refer to judicial review as ‘anti-extremist’ when j > y_M. In the former case, judicial review prevents the leader from adopting certain policies that would benefit the majority at the expense of the minority. In the latter case, judicial review precludes only extreme policies that would be disfavored by both the minority and the majority groups.

Recall that in our baseline model with no review, all D1 equilibria are fully separating. In contrast, with judicial review, pooling is now possible because the judicial limit imposes an upper bound on the policy space. However, D1 ensures that if pooling occurs, it occurs on the judicial limit (j). In such equilibria, the type-M leader selects j with certainty, while the type-m leader mixes between j and the minority-preferred policy (y_m). Separating equilibria are also possible in the model with judicial review; as in the baseline model, in any separating D1 equilibrium the type-M leader achieves separation in a manner that minimizes his policy costs from doing so.

With the above in mind, we now offer two propositions which establish how the introduction of judicial review alters the leader’s behavior, relative to the baseline case. Each proposition corresponds to one of the subcases (parts (a) and (b)) of Proposition 1. To state the first of these propositions, we define a new threshold:

x^{-} = y_{m} + \sqrt{δ [F (π) - F (0)]} .^{17}

Proposition 2 Suppose x ⁺ ≤ y_M, such that without judicial review, there would be a sincere separating equilibrium. (See part (a) of Proposition 1.) With judicial review, D1 equilibria exist. When j ≥ y_M, in any D1 equilibrium, behavior is identical to the case of no review. However, when j < y_M, judicial review induces the type-M leader to propose a more moderate policy than he would otherwise, and may also induce the type-m leader to propose a more extreme policy than he would otherwise. Specifically, behavior is as follows:

If j ∈ [x⁺, y_M), all D1 equilibria are separating, with the type-M leader selecting the judicial limit (j) and the type-m leader selecting the minority-preferred policy (y_m).

If j ∈ (x⁻, x⁺), all D1 equilibria are semi-pooling, with the type-M leader selecting j and the type-m leader playing a mixed strategy, selecting j with probability α(j) and selecting y_m with probability 1 − α(j).¹⁸Further, α(j) is decreasing in j.

If j ≤ x⁻, all D1 equilibria involve complete pooling, with both the type-M leader and the type-m leader selecting the judicial limit j.

Proposition 2(a) corresponds to the conventional wisdom: countermajoritarian judicial review does not affect the behavior of pro-minority leaders, but pro-majority leaders are constrained to implement j < y_M rather than y_M; this constraint unambiguously benefits the minority group at the expense of the majority. Parts (b) and (c) of Proposition 2 complicate the standard intuition, however.

To understand parts (b) and (c), note that since a countermajoritarian court forces the type-M leader to accept a policy outcome of j rather than y_M, the policy cost that the type-m leader incurs when he mimics the type-M leader’s choice is lower than in the case of no review. Moreover, this cost decreases as the court become more pro-minority (i.e. as j decreases). Consequently, once the judicial limit is sufficiently strict (j < x⁺), it is no longer an equilibrium for the type-m leader always to select his ideal point, because the electoral benefits of pooling with the type-M leader dominate the policy costs. Therefore, when j < x⁺, the type-m leader mimics the type-M leader sometimes (Proposition 2(b)) or always (Proposition 2(c)). Thus judicial review not only has a direct effect on policy choice (forcing the type-M leader to select j rather than y_M), but it also has an indirect effect, inducing the type-m leader to pool rather than separate. Therefore, although countermajoritarian judicial review leads to outcomes that are better for the minority (and worse for the majority) when a pro-majority leader is in office, such review sometimes leads to outcomes that are worse for the minority (and better for the majority) when a pro-minority leader is in office, compared with the no review baseline.¹⁹

How do these cross-cutting effects impact each group’s expected policy payoff?²⁰Figure 2 plots the net expected policy benefit of judicial review for both the minority group and the majority group as a function of the judicial limit j, with the top panel analyzing the welfare effects of review for a scenario in which sincere separation arises in review’s absence.²¹ The non-monotonic effects of j on minority welfare (black curve) and majority welfare (gray curve) are immediately apparent. For certain values of j, judicial review benefits the minority at the expense of the majority, as the conventional wisdom holds. This is true if judicial review constrains the type-M leader and has little effect on the behavior of a type-m leader, as in Region 5 of Figure 2.²² Likewise, if the judicial limit is sufficiently stringent (i.e. j is close to y_m), review benefits the minority at the majority’s expense even though the type-m leader sometimes pools with the type-M leader; here, the direct constraining effect of judicial review will dominate in expectation (see Region 1 of Figure 2). However, for intermediate judicial limits, judicial review can make both groups better off (Regions 2 and 4 of Figure 2), or can benefit the majority at the minority’s expense (Region 3 of Figure 2). In the latter instance, which may seem particularly counterintuitive, the costs to the minority group from type-m leaders sometimes pooling with type-M leaders outweigh whatever gains result from the fact that review forces type-M leaders to propose more moderate policies.

Figure 2.

Net expected policy benefit of certain review for the minority group (black curve) and the majority group (gray curve) as a function of the judicial limit (j). For these plots, $y_{m} = 0, y_{M} = 1, π = 0.5$ , and $F (μ (x)) = {μ (x)}^{1 / 2}$ . In the top panel, $δ = 0.8$ , and a sincere separating equilibrium arises in review’s absence. In the bottom panel, $δ = 1.6$ , and an extremist separating equilibrium arises in review’s absence. In both panels, there is a range of judicial limits for which the minority group is strictly worse off with review (Regions 3 and 9).

The next proposition considers the effects of introducing judicial review into a setting where, in the absence of review, the pro-majority leader would propose a policy more extreme than the majority group’s preferred policy in order to signal his type.

Proposition 3 Suppose x ⁺ > y_M, such that without judicial review, there would be an extremist separating equilibrium. (See part (b) of Proposition 1.) With judicial review, D1 equilibria exist. When j ≥ x⁺, in any D1 equilibrium, behavior is identical to the baseline case. However, when j < x⁺, judicial review induces the type-M leader to propose a more moderate policy than he would otherwise and leads the type-m leader to (sometimes) propose a more extreme policy than he would otherwise. Specifically, when j < x⁺, in any D1 equilibrium, there is (partial or full) pooling on the judicial limit (j): the type-M leader selects x = j, and the type-m leader selects x = j with probability α(j) and selects x = y_m with probability 1 − α(j).

The effects of judicial review in this case on each group’s expected policy payoff are illustrated in the bottom panel of Figure 2. For the minority group, as before, the direct and indirect effects of review cut in opposite directions. As such, when review is sufficiently restrictive, the minority group will benefit from review (see Regions 7 and 8 of Figure 2); when review is less restrictive, the minority group can be harmed (see Region 9 of Figure 2). For the majority group, the main difference relative to the case considered in Proposition 2 is that the direct constraining effect of review can now work to the majority’s benefit, provided that the judicial limit is not too restrictive.²³ Recall that when x⁺ > y_M and there is no review, the type-M leader signals his type by proposing x⁺, imposing policy costs on both groups. Consequently, when j ∈ (2y_M − x⁺, x⁺), the direct constraining effect of review strictly benefits the majority group, as the court forces the type-M leader to propose a policy closer to the majority’s ideal policy than he would otherwise. Therefore, when x⁺ > y_M, both the direct and indirect effects of review may redound to the majority group’s benefit. More generally, for the case considered in the bottom panel of Figure 2, anti-extremist or modestly countermajoritarian judicial review benefits the majority at the minority’s expense (Region 9); more stringent countermajoritarian review benefits both groups (Region 8); and minority-protective judicial review only has the intuitive effect of benefiting the minority at the majority’s expense when judicial review becomes sufficiently countermajoritarian (Region 7).

The main insight of the analysis thus far is that minority-protective judicial review may have two types of effects on policy outcomes. The first effect is that judicial review constrains the degree to which the policy can deviate from the minority group’s ideal outcome. The second and more novel finding is that if j<x⁺, judicial review can induce a shift from a fully separating equilibrium (in which type-m leaders select x = y_m and type-M leaders select x = max{y_M, x⁺}) to an equilibrium with either partial or complete pooling, in which even pro-minority leaders sometimes adopt policies that disfavor the minority group.²⁴ This latter effect means that the impact of ostensibly minority-protective judicial review on both minority and majority interests is more complicated than the conventional wisdom suggests, and can lead to situations in which review (in expectation) harms the minority group and benefits the majority group, as well as situations in which both groups benefit from countermajoritarian review.

While attempting to evaluate the empirical relevance of these results is well beyond the scope of this paper, and we are reluctant to use our highly stylized model to ‘explain’ any particular historical or contemporary example, it is worth observing that the basic dynamic elucidated above, particularly the less intuitive results identified in Proposition 2(b)–(c) and Proposition 3, may shed light on the phenomenon whereby legal limits on leaders’ discretion may ‘shrink’ the domain of policy disagreement, causing convergence on (the limit of) what is legal. For example, with respect to economic policy, in the absence of legal or other constraints on redistribution, politics may exhibit an overt competition between radical economic populists (the ‘pro-majority’ types) and laissez-faire libertarians (the ‘pro-minority’ types), with the former prevailing more often, but the latter prevailing at least some of the time. Constraints on the scope of wealth redistribution, such as judicially-enforced protections on property rights, may cause both factions to converge on ‘moderate’ redistributive policies, the populists because they can no longer implement anything more extreme, and the libertarians because they can now successfully mimic the populists (that is, pose as ‘moderates’) and thereby improve their chances of reelection. It is not a priori obvious whether the capitalist minority or the working-class majority would prefer the former, more polarized type of redistributive politics, or the latter setting with more consistent, moderate redistribution. Our argument is that the existence of robust and accepted legal protections for the minority may change the nature of politics in this way. This implies a comparative prediction that is, at least in principle, empirically testable.

One might likewise develop a similar hypothesis with respect to other policy issues. For example, one might predict that, in the absence of enforceable protections for the rights of marginalized racial minorities, some (pro-minority) leaders will push for aggressive affirmative action or other forms of remediation, while other (pro-majority) leaders will oppose even mild forms of anti-discrimination protection; our analysis suggests that sometimes the introduction of legally enforced protections for the minority—say, prohibitions on overt discrimination—will tend to produce convergence, helping the minority group by outlawing extreme anti-minority policies, but at the same time enabling would-be pro-minority leaders to ‘moderate’ by endorsing the same middle-of-the-road anti-discrimination policies that the pro-majority leaders have been legally compelled to accept. Again, the impact of this moderating effect on the welfare of both groups depends on other parameters: sometimes, inducing mutual moderation may make both groups better off, but under other conditions, doing so may improve the welfare of the minority at the majority’s expense, or vice versa.

One final observation is that the above discussion of welfare effects has focused on the expected outcome of the policy issue subject to judicial review. However, this focus may be misleading, because judicial review can affect welfare through another channel: by (sometimes) shifting the equilibrium from a fully separating equilibrium to a semiseparating or pooling equilibrium, judicial review may reduce the voters’ information about the incumbent’s type, leading pro-minority leaders to be reelected with higher probability and pro-majority leaders to be reelected with lower probability. A natural conjecture is that this effect tends to benefit the minority group and harm the majority group; after all, a voter might prefer a leader who shares her preferences because such a leader is likely to advance the voter’s interests on other (perhaps unobservable) issues. On this account, even when increasing the stringency of judicial review decreases the minority group’s expected policy payoff (on the observable issue in the current period), this change may still benefit the minority (at the majority’s expense) by helping pro-minority leaders remain in office, where they can work ‘behind the scenes’ on the minority’s behalf in future periods. If so, then judicial review indeed has countermajoritarian effects, but for reasons quite different from those usually emphasized.

4. Uncertain judicial review

Section 3 assumed that if the leader proposes an illegal policy, x > j, the court strikes down that policy and replaces it with x = j. In other words, the preceding analysis presumed that the judiciary consistently enforces a clear limit on government action. This assumption may make sense for some areas of law, where the minority-protective rules are sufficiently certain that no leader would ever violate them in equilibrium. In other areas, however, judicial protection of minority rights is less predictable. In this section, we investigate how the results change if judicial invalidation of anti-minority policies is uncertain.

There are many ways one might model this sort of uncertainty. Here we explore a simple framework in which, as in Section 3, there is a common-knowledge threshold, j, such that the court always upholds any x ≤ j. In contrast to Section 3, we now assume that the court will strike down a policy proposal x > j (and replace it with j) with probability q ∈ (0, 1), where q is common knowledge. That is, if the leader proposes x > j, then with probability q the final outcome is j and with probability 1 − q the final outcome is x. One can interpret q as the probability that the court will conclude that the law prohibits policies above j, or as the probability that a court willing to enforce limit j hears a challenge to the leader’s proposal.²⁵

Since enforcement of the judicial limit is no longer certain, review no longer imposes an upper bound on the policy space. Therefore, as in the baseline model, any D1 equilibrium must involve complete separation, with the type-m leader selecting his ideal point. This means that uncertain review can only affect the equilibrium behavior of the type-M leader. The next two propositions explicate this effect.

Proposition 4 Suppose x ⁺ ≤ y_M, such that without judicial review, there would be a sincere separating equilibrium. (See Proposition 1(a).) With uncertain review, D1 equilibria exist and are separating. When j ≥ y_M, in any D1 equilibrium, behavior and final policy are identical to the case of no review. However, when j < y_M, D1 equilibria are as follows.

If j ∈ [x⁺, y_M), with uncertain judicial review leader behavior is identical to that with no review, but the distribution of final policies differs: the type-M leader selects the majority-preferred policy (y_M), which is upheld with probability 1 − q and replaced by x = j with probability q, and the type-m leader selects the minority-preferred policy (y_m), which is always upheld.

Let $\bar{q} (j) = \frac{δ [F (1) - F (0)] - {(y_{m} - y_{M})}^{2}}{(j - y_{M}) (j - 2 y_{m} + y_{M})}$ and $x^{+ +} = y_{m} + \frac{\sqrt{(1 - q) (δ [F (1) - F (0)] - {q (j - y_{m})}^{2})}}{1 - q}$ .

If j < x⁺and $q \leq \bar{q} (j)$ , with uncertain judicial review leader behavior is identical to that with no review, but the distribution of final policies differs (as the type-M leader’s proposal of x = y_M is sometimes replaced with x = j).

If j < x⁺and $q > \bar{q} (j)$ , with uncertain judicial review the type-M leader proposes a more extreme policy than he would otherwise: he proposesx = x⁺⁺> y_M, which is upheld with probability 1 − q and replaced by x = j with probability q, and the type-m leader selects the minority-preferred policy, which is always upheld.

To gain intuition, begin with part (a).²⁶ Suppose the type-M leader proposes the majority-preferred policy (y_M). Since j ≥ x⁺, even if the judicial limit were always enforced, the policy cost the type-m leader would incur from pooling on the majority’s preferred policy (and having that policy replaced with policy j) would weakly exceed the electoral benefit of pooling. Uncertain review only raises the policy costs of mimicking the type-M leader’s policy choice. So when j ∈ [x⁺, y_M), the type-M leader is able to achieve separation by proposing x = y_M. Although uncertain review has no effect on the policies proposed relative to the case of no review, it does affect the distribution of final policies: when a pro-majority leader holds office, his proposal of y_M is sometimes replaced with j < y_M. So under the conditions highlighted by Proposition 4(a), the conventional wisdom holds: review benefits the minority at the majority’s expense.

Now consider Proposition 4(b). Once again, suppose the type-M leader proposes the majority-preferred policy (y_M). Because y_M ≥ x⁺, if the court always upheld x = y_M, the policy cost the type-m leader would incur from pooling on y_M would weakly exceed the electoral benefit. In contrast, because j < x⁺, if the court always struck down x = y_M and replaced it with x = j, the electoral benefit of pooling on y_M would dominate the policy cost. Consequently, when the probability of judicial enforcement is low (q small), the type-M leader can achieve separation by selecting the majority-preferred policy (y_M) (see Proposition 4(b)(i)); here, review benefits the minority at the expense of the majority, just as in part (a). However, when the probability of enforcement is high (q large), to achieve separation the type-M leader must raise the policy cost that the type-m leader incurs when he pools. This requires proposing a policy to the right of y_M (see Proposition 4(b)(ii)). In this latter case, uncertain review can induce the type-M leader to stake out a more extreme position than he would otherwise.

The next proposition characterizes the effects of uncertain review when the baseline case would produce extremist separation (i.e. the pro-majority leader chooses a policy that is more anti-minority than the majority group prefers).

Proposition 5 Suppose x ⁺ > y_M, such that without judicial review, there is an extremist separating equilibrium. (See Proposition 1(b).) With uncertain review, D1 equilibria exist and are separating. When j ≥ x⁺, behavior is identical to the baseline case. However, when j < x⁺, uncertain review induces the type-M leader to propose x = x⁺⁺, which is more extreme than the proposal he would offer in review’s absence (namely,x⁺). The court upholds the type-M leader’s proposal of x⁺⁺with probability 1 − q, and replaces it with x = j with probability q. The type-m leader selects the minority-preferred policy (y_m), which is always upheld.

The behavioral effects of uncertain review characterized by Proposition 5 seem consistent with the idea that judicial review enables leaders to adopt more extreme positions than they would otherwise, because the court may prevent the extreme policy from taking effect (Fox and Stephenson, 2011; Salzberger, 1993). However, the dynamic here differs from the conventional version of this ‘bailout effect.’ In our model, the possibility of judicial reversal means that the pro-minority leader may not have to live with the consequences of a relatively extreme anti-minority policy choice; as a result, he is willing to go further in an anti-minority direction if doing so will fool the voters into thinking he is really pro-majority. In turn, the pro-majority leader will go even further in an anti-minority direction in order to achieve separation. In equilibrium, the pro-minority leader is unwilling to go that far, even with the prospect of judicial reversal, because there is always the possibility that the judiciary will not reverse the extreme proposal. Therefore, in equilibrium the pro-minority leader will continue to propose his ideal point, but the pro-majority leader, in the shadow of the pro-minority leader’s implicit threat to mimic, will adopt a more extreme anti-minority policy. Importantly, this does not occur when judicial reversal of any x > j is guaranteed (as in Section 3), because in that case a proposal of x > j communicates no credible information to the voters. This further highlights a difference between our account and the more familiar ‘bailout’ or ‘safety net’ argument. On the conventional account, leaders are willing to make extreme proposals because they (allegedly) know that the courts will block them, but (presumptively naïve) voters will nonetheless reward the leaders for making these proposals. In our account, the fact that the court might block an extreme proposal makes such a proposal more attractive to the pro-minority leader, but the fact that the court might not block such a proposal enables the pro-majority leader to separate by becoming even more extreme, and the sophisticated (although incompletely informed) voters draw rational inferences from observed behavior.

In short, uncertain review has two main effects on the policy outcome, one direct and the other indirect. First, minority-protective judicial review invalidates some proportion of extreme anti-minority policies (the direct effect). Second, uncertain judicial review may induce the pro-majority leader to propose a more extreme anti-minority policy than he would in the absence of review. An illustration of the net impact of these effects on the expected policy payoff of the minority group and the majority group is depicted graphically in the top and bottom panels of Figure 3 (which analyze scenarios that correspond to Propositions 4 and 5, respectively).

Figure 3.

Net expected policy benefit of uncertain review for the minority group (black curve) and the majority group (gray curve) as a function of the judicial limit (j). For these plots, q = 0.6, y_m = 0, y_M = 1, π = 0.5, and F(μ(x)) = μ (x)^1/2. In the top panel, δ = 0.8, and a sincere separating equilibrium arises in review’s absence. In the bottom panel, δ = 1.6, and an extremist separating equilibrium arises in review’s absence. In both panels, uncertain review can only benefit the minority group and can only harm the majority group; that said, in the bottom panel, there exists a range of judicial limits (Region 4) for with judicial review is Pareto inefficient: for such judicial limits, the minority group does not benefit from review and the majority group is strictly worse off.

As this figure illustrates, uncertain review weakly benefits the minority group and weakly harms the majority group, just as the conventional wisdom suggests.²⁷ Since uncertain review can only benefit the minority group, there exist situations in which moving from uncertain review to certain review can make the minority group worse off. This may seem counterintuitive; it arises because certain judicial review can induce the pro-minority leader to pool with the pro-majority leader, which hurts the minority group, but with uncertain review the equilibrium is always separating, with the pro-minority leader always choosing his ideal point. Compared with the baseline case, uncertain review, if it has any effect on behavior, only affects the behavior of the pro-majority leader, inducing him to propose a more extreme (illegal) policy, thereby creating a lottery between that policy and the judicial limit j.

Three additional observations concerning the effects of uncertain review on the expected policy payoffs of the majority group and minority group are perhaps even more significant. First, in contrast to the certain review case, there are no values of j for which uncertain review is a Pareto improvement relative to no review. Second, when an extremist separating equilibrium would prevail in the baseline case, uncertain review can harm the majority group yet not benefit the minority group (Region 4 of Figure 3). In this case, then, uncertain judicial review is Pareto inferior to no review. Third, when a sincere separating equilibrium would prevail absent review (as in the top panel of Figure 3), modestly countermajoritarian judicial review helps the minority at the majority’s expense (Region 2 of Figure 3), but past a certain point, more stringently countermajoritarian review (lower j) does not lead to any further improvement in the minority group’s expected welfare, even though this greater stringency hurts the majority group (Region 1 of Figure 3).

A final observation regarding the welfare implications of uncertain judicial review: with uncertain review, since equilibrium behavior always involves complete separation, voters can always distinguish pro-majority and pro-minority leaders, as was also true in the baseline case (Section 2), but which was not true when judicial review was certain (Section 3). Thus, in contrast to certain judicial review, uncertain review does not affect welfare by altering the amount of information that voters have about the incumbent leader.

5. Conclusion

Much of the extensive normative debate in constitutional theory over judicial review presumes that minority-protective judicial review leads to countermajoritarian outcomes—that is, outcomes that favor the protected minority at the majority’s expense. While our analysis partly corroborates this conventional wisdom, we also show that minority-protective judicial review may have other effects that qualify, and in some cases undermine, this hypothesis. These effects derive from the assumption that leaders whose preferences align with the majority have an incentive to reveal this fact, while leaders whose preferences align with an unpopular minority have an incentive to conceal this fact. Thus the constraint imposed by minority-protective judicial review can induce a shift from a separating equilibrium (in which pro-majority leaders and pro-minority leaders cater to their respective constituencies) to a pooling equilibrium (in which both types of leader adopt the most anti-minority policy the court would uphold). In other circumstances, the existence of a judicial ‘safety net’ makes pro-minority leaders more inclined to propose relatively extreme anti-minority policies (which they secretly hope will be struck down) in order to mimic pro-majority leaders, and this in turn induces pro-majority leaders to propose even more extreme anti-minority policies in order to distinguish themselves.

When these indirect effects are taken into account, one can no longer rely confidently on the intuitive assumption that minority-protective judicial review leads to countermajoritarian outcomes. In some cases, minority-protective judicial review may increase the expected welfare of both the majority and the minority (as when such review induces pooling on a ‘moderate’ outcome between the majority and minority groups’ respective ideal points). In those cases, minority-protective judicial review can be ‘win–win’ on the policy issue subject to review. In other cases, making judicial review more minority-protective can reduce the expected welfare of one group without improving the welfare of the other group. Perhaps most counterintuitively, under some circumstances countermajoritarian judicial review may even benefit the majority at the expense of the minority. In such circumstances, majoritarian democrats should favor minority-protective judicial review, while those concerned with the welfare of unpopular minorities should oppose it.

Footnotes

Acknowledgements

We are grateful to Brandon Bartels, Dan Bernhardt, Rachel Brewster, Charles Cameron, Glenn Cohen, Tiberiu Dragu, Richard Fallon, Barry Friedman, Sean Gailmard, Dimitri Landa, Louis Kaplow, Michael Klarman, Jonathan Masur, Frank Michelman, Mark Ramseyer, Ben Roin, Fred Schauer, Jed Shugerman, Steve Shavell, Mark Tushnet, Adrian Vermeule, and Mark Wu, as well as to workshop and conference participants at Boston University Law School, Harvard Law School, Peking University School of Transnational Law in Shenzhen, Princeton University, Purdue University, University of Illinois, University of Miami, Washington University in St. Louis, the 2011 American Law and Economics Association annual meeting, the 2011 American Political Science Association annual meeting, the 2011 Midwest Political Science Association annual meeting, the Academia Sinica Conference on Positive Political Economy, the LSE-NYU Conference on Political Science and Political Economy, the Harvard Conference on Political Economy and Public Law, the 11th Society for Advancement of Economic Theory Conference, and the Yale Center for the Study of American Politics Summer Conference for helpful comments on earlier drafts.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

References

Acemoglu

Egorov

Sonin

(2013) A political theory of populism. Quarterly Journal of Economics128: 771–805.

Almendares

Le Bihan

(2012) Increasing leverage: judicial review as a democracy-enhancing institution. Unpublished manuscript.

Berry

Gersen

(2008) The unbundled executive. University of Chicago Law Review75: 1385–1434.

Bickel

(1962) The Least Dangerous Branch: The Supreme Court at the Bar of Politics. New Haven, CT: Yale University Press.

Carter

(1995) The Politics of Rage: George Wallace, the Origins of the New Conservatism, and the Transformation of American Politics. Baton Rouge, LA: Louisiana State University Press.

Chemerinsky

(2004) In defense of judicial review: the perils of popular constitutionalism. University of Illinois Law Review2004: 673–690.

Cho

Kreps

(1987) Signaling games and stable equilibria. Quarterly Journal of Economics102: 179–221.

Cho

Sobel

(1990) Strategic stability and uniqueness in signaling games. Journal of Economic Theory50: 381–413.

Dragu

Fan

Kuklinski

(2014) Designing checks and balances. Quarterly Journal of Political Science9: 45–86.

10.

Dworkin

(1985) A Matter of Principle. Cambridge, MA: Harvard University Press.

11.

Ely

(1980) Democracy and Distrust: A Theory of Judicial Review. Cambridge, MA: Harvard University Press.

12.

Fearon

(1999) Electoral accountability and the control of politicians: selecting good types versus sanctioning poor performance. In: Przeworski

Stokes

Manin

(eds), Democracy, Accountability, and Representation. New York, NY: Cambridge University Press.

13.

Fiorina

Abrams

Pope

(2010) Culture War? The Myth of a Polarized America, 3rd edn. New York, NY: Pearson Longman.

14.

Fleck

Hanssen

(2013) Judicial review as a constraint on tyranny of the majority. Journal of Law, Economics, and Organization29: 303–331.

15.

Fox

Stephenson

(2011) Judicial review as a response to political posturing. American Political Science Review105: 397–414.

16.

Groseclose

McCarty

(2001) The politics of blame: bargaining before an audience. American Journal of Political Science45: 100–119.

17.

Guinier

(1994) The Tyranny of the Majority. New York, NY: Free Press.

18.

Maskin

Tirole

(2004) The politician and the judge: accountability in government. American Economic Review94: 1034–1054.

19.

Morelli

Van Weelden

(2011) Re-election through division. Unpublished manuscript.

20.

Rogers

(2001) Information and judicial review: a signaling game of legislative-judicial interaction. American Journal of Political Science45: 84–99.

21.

Rogers

(2009) Casting the gimlet eye on judicial review: can judicial review be democratically debilitating? Unpublished manuscript.

22.

Rogers

Vanberg

(2007) Resurrecting Lochner: a defense of unprincipled judicial activism. Journal of Law, Economics, and Organization23: 442–468.

23.

Salzberger

(1993) A positive analysis of the doctrine of separation of powers, or: why do we have an independent judiciary?International Review of Law and Economics13: 349–379.

24.

Schama

(1989) Citizens: A Chronicle of the French Revolution. New York, NY: Alfred A. Knopf.

25.

Spence

(1973) Job market signaling. Quarterly Journal of Economics87: 355–374.

26.

Thayer

(1893) The origin and scope of the American doctrine of constitutional law. Harvard Law Review7: 129–156.

27.

Tushnet

(2000) Taking the constitution away from the courts. Princeton, NJ: Princeton University Press.

28.

Vermeule

(2006) Judging under Uncertainty: An Institutional Theory of Legal Interpretation. Cambridge, MA: Harvard University Press.

29.

Waldron

(2006) The core of the case against judicial review. Yale Law Journal115: 1346–1406.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.29 MB