Dissent Aversion and Sequential Voting in the Brazilian Supreme Court

• DOI: http://doi.org/10.1111/jels.12236
• Published date: 01 December 2019
• Author: Felipe Mendonça Lopes
• Date: 01 December 2019

Journal of Empirical Legal Studies

Volume 16, Issue 4, 933–954, December 2019

Dissent Aversion and Sequential Voting

in the Brazilian Supreme Court

Felipe de Mendonc¸a Lopes*

The literature on judicial behavior indicates that dissent is costly to the individual judge,

for it means an increased workload and the incurrence of reputational costs. This observa-

tion leads to the concept of dissent aversion, which is the notion that judges may refrain

from dissenting, even though they disagree with the court’s position, in order to avoid

those costs. In this article, I explore the fact that voting at the Brazilian Supreme Court is

sequential, and that this sequence varies in almost every single case, to generate a quasi-

experiment that allows for a clearer identification of dissent aversion than is found in

other contexts. The main idea is that after a majority has been formed, the justices who

vote in sequence know that their votes cannot change the outcome of the case. Hence,

they may deviate from their preferred votes and join the majority to avoid the costs of dis-

senting. Using a unique dataset with vote-level data of all abstract review cases heard by

the full Court from 1990 until 2015, this study finds strong evidence of dissent aversion in

the Brazilian Supreme Court. Judges who vote after the pivotal judge are significantly less

likely to dissent.

I. Introduction

Dissent is costly. Judges who choose to openly disagree with the court’s ruling have to

bear the costs of doing so. Those costs can be due to effort—the intellectual effort and

the time spent writing a separate opinion. They can be reputational—judges dislike being

criticized and a judge who frequently dissents may become less well-liked at court and

find it harder to persuade fellow judges to join his opinions in future cases. Dissents can

also be costly to the court, not just to the dissenting judge. By signaling that there is not

a single legal answer, dissents can be detrimental to the court’s legitimacy and under-

mine the ability to enforce its decisions.

*Address correspondence to Felipe de Mendonc¸a Lopes, Senior Consultant at LCA Consultores, Rua Cardeal

Arcoverde 2450, S~

ao Paulo 05408-003; e-mail: felipe_lopes25@hotmail.com.

The views expressed are those of the author and do not necessarily represent those of his current or past

employers. This article is based on a chapter of the author’s Ph.D. thesis at the S~

ao Paulo School of Economics,

Fundac¸ ~ao Getulio Vargas (EESP-FGV). The author is grateful to the participants of the Conference on Empirical

Legal Studies at University of Michigan Law School (especially discussant Veronica Santarosa), and Rodrigo

R. Soares, Paulo F. de Azevedo, Nuno Garoupa, Bruno Ferman, Vladimir Ponczek, and Bernardo Mueller for their

important suggestions. The author also thanks the editor and the two anonymous referees.

933

The observation of the costs of dissent has led to what Judge Richard A. Posner

calls dissent aversion, which is the idea that judges may not dissent, even if they disagree

with the court’s position, in order to avoid such costs. As Posner himself put it, most

judges do not like to dissent since “not only is it a bother and frays collegiality, and usu-

ally has no effect on the law, but it also tends to magnify the significance of the majority

position” (Posner 2008:32).

The empirical literature that investigates judge’s incentives toward dissent indicates

that, in some contexts, dissent aversion is high to the point of concealing the ideological

cleavages that exist among judges. This is especially true in courts of the civil-law tradi-

tion. As Garoupa et al. (2012) show in the Spanish context, where judges face strong

pressures toward consensus, a high unanimity rate may mask the underlying political ten-

sions and disagreements within the court.

Conversely, in the U.S. Supreme Court, where the cost of dissent is comparatively

low, as justices have a light workload, enjoy the support of several law clerks, and benefit

from an environment where dissent has come to be expected, dissent is rampant

1

(in Epstein et al. (2013), 57.4 percent of cases have at least one dissenting vote). In this

Court, however, judges have an added benefit of dissent. As precedents are more likely to

be reexamined (due to low workload), justices face the prospect of influencing future

decisions through their written dissents.

In this study, I explore the voting procedures of the Brazilian Supreme Court to

generate a quasi-experiment that allows for a clearer identification of dissent aversion

than is found in other contexts. Here, voting order is key to identification. The idea is

that once a majority has been reached, that is, after six out of eleven justices have voted

in a particular direction (typically, either to uphold or to strike down the law under

review), the outcome of the case is settled. The judges who vote subsequently know that

their votes cannot change the outcome, no matter how strongly they may disagree with

it. Consequently, if they are dissent averse, they may deviate from their preferred vote

and join the majority.

In the Brazilian Supreme Court, the voting sequence changes in nearly every case,

hence making the identity of the judges who vote after the formation of a majority

equally variable. Among the main sources of voting order variation are Rapporteur assign-

ment, which is random

2

within each class of petitions that are presented to the Court,

the rotating position of Chief Justice, and changes in the Court’s composition, which are

relatively frequent since retirement is mandatory at age 75.

3

Underlying this identification strategy is the notion that the formation of a majority

reveals valuable information to judges. Up until the pivotal vote is cast, there is

1

This was not always the case. See, for instance, Hendershot et al. (2013) on the demise of the U.S. Supreme

Court’s “norm of consensus.”

2

There is an automated system that allocates cases to judges. The draw is random within each class of petitions,

ensuring that there is an even workload among justices.

3

Until 2015, mandatory retirement used to be at age 70.

934 Mendonc¸a Lopes