Desktop version

Home arrow Economics arrow Commitment and cooperation on high courts : a cross-country examination of institutional constraints on judges


The most common-sense implication of the discretion to set panel size and composition is that differences in panels can lead to differences in results. A party whose appeal was heard by five judges of a nine-person court may worry that the result would have been different if the whole court or if a different set of five judges had decided the appeal. A court may have rules to overcome this concern. For example, at the federal court of appeal level in the united States a party can petition for an appeal to be heard or reheard en banc. The uS federal government recently made use of this procedure when it lost a challenge to President Obama’s healthcare act in the DC Court of Appeal in the case of Halbig v. Burwell, in which two (Republican-appointed) judges of a three-judge panel voted to overturn the law on the same day that three judges on the Fourth Circuit unanimously upheld the same law in a separate case. The DC Court of Appeal vacated the first judgment and scheduled a new hearing en banc without giving any reasons for the decision.[1] Similarly, it may be that fear of differing panels yielding different results is behind the rule in Australia that a full court is required for constitutional cases.

However, the panel selection power may not influence outcomes. If the judges on a court are similar in their personal or political views, the chief justice or president may have less scope for influencing the outcome through panel selection than a highly divided court. If all judges on the bench have the same background and the same policy views, the chief justice would not be able to use this tactic to alter outcomes at all. Moreover, panel selection may not influence outcomes overall if the panels are set randomly or if the chief justice specifically sets panels to ensure that the composition of the panel reflects the composition of the court as a whole. However, even if overall outcomes are not different from what the full bench would decide, panel selection may influence individual cases. As result, we have the following hypothesis about panel selection:

H2: Choice of panel composition by a chief justice will influence the outcomes

of appeals.

There is some evidence that the ability to choose the panel makes a difference as to how a court decides. On the uK House of Lords, for example, a third of possible alternate bench compositions would have produced a majority of justices favoring a different outcome in Bancoult (no 2),[2] an important administrative law and human rights case.[3] Similarly, which justices heard the appeal was very influential to the outcome of Charter claims before the Supreme Court of Canada between 1983 and 1989.[4] Justices had very different approaches to the Charter, and Charter claims had a much better chance of being accepted when the majority of the deciding panel were justices who were more receptive to Charter claims.[5] Panel selection was also important in mandatory criminal jurisdiction cases before the Israeli Supreme Court. Different combinations of justices varied widely in how they decided these cases.[6] The panel most favorable to defendants decided 33.3 percent of its cases in their favor, whereas the panel least favorable to defendants only decided 2.8 percent of its cases in their favor.[7]

On the other hand, strategic panel assignments by South African chief justices on the South African Appellate Division during the apartheid era may have had only limited effects on the outcomes of cases.[8] Different chief justices did not appear to cause different results through their panel selection.[9] For example, despite Chief Justice Rabie’s attempts to appoint more Afrikaansspeaking judges and fewer liberal judges to civil rights cases, Rabie’s tenure as chief justice was not statistically correlated with more conservative outcomes in this area of law.[9]

A chief justice or president may not influence outcomes overall through panel selection where panels are set to achieve the same result as if the full bench heard the appeal. If there is no ideological diversity on the bench the chief justice will not even be able to influence individual cases. even assuming a chief justice wishes to influence the outcomes of appeals through manipulating the composition of panels, she might not succeed, even if there is some ideological diversity across the judges on the court. It depends on the court’s minimum bench size and the total number of judges. In order for her to influence the outcome, there must be enough judges who would support her preferred outcome for her to form a majority of the minimum allowable bench size. For instance, in Canada where the minimum bench size is five justices, the chief justice could appoint herself and two others who support her preferred result to a five-judge panel. If the Canadian Supreme Court decides a case on a 7-0, or 9-0 basis, no other combination of judges could have produced a majority in favor of the opposite outcome on a different panel.[11] A 6-3 decision, on the other hand, could have been decided differently if the three dissenters on the actual decision had been placed on a panel of five judges with any two of the other judges on the majority.

One measure of the extent to which a chief justice can alter outcomes then is the proportion of cases that could potentially have been decided differently with a differently constituted panel. We term this measure the Gaming Prospect Ratio (GPR)—the potential for a chief justice or president to “game” the panel selection by using panel selection rules to achieve a preferred outcome. We construct the GPR by dividing the number of decisions that could have been decided differently with a different panel by the total number of cases heard by the court.[12] More ideologically divisive courts will exhibit higher GPRs, because in such courts, more cases are likely to be decided by a close split, which opens the door for potential gaming.

Figure 4-2 sets out the GPR for various high courts—the percentage of cases that could potentially have been gamed through panel selection. Note that for India and the UK Houses of Lords, the GPRs are 100 percent—that is, all the cases could potentially have been decided differently with a different set of justices on the bench at the time. Because of the size of the court and the sizes of panels, all cases could be heard by a completely different set of justices, which means that if there is any ideological diversity (and the case would turn on ideological voting by the justices), the chief justice could rearrange the panels to influence any decision. Even in countries that ostensibly are in the mixed

Gaming Prospect Ratios

Figure 4-2 Gaming Prospect Ratios (GPR) for various countries for 1970 to the early 2000s. A country’s GPR is the number of decisions that could have been decided differently with a different panel, divided by the total number of cases heard by the court (the figure converts this rate to a percentage). Because of the small panels and large bench, India and the UK have 100 percent GPR—all the cases could potentially come out differently with a different panel. Canada has the lowest GPR. The US GPR is hypothetical, assuming the US Supreme Court heard cases in panels of five, seven, or nine judges.

category such as India, the chief justice has essentially unconstrained powers to set panels; with 31 judges on the Court, the chief justice has a wide range of panels of two, three, or five from which to choose.

The other three countries have lower GPRs. Canada has the lowest of these GPRs, with about 20 percent of cases potentially coming out differently with differently constituted panels. Australia has a slightly higher GPR of 30 percent. Although the United States hears appeals en banc (with a resulting GPR of 0), we also looked at what the GPR would be in a hypothetical scenario where it heard cases in panels of five, seven, or nine judges as in Canada. It would have the highest GPR at about 34 percent, meaning that given the justices on the bench about a third of the cases would be decided differently. The ratio would be even higher if the court could use panels of two or three judges as in India.

The GPR then is a function not only of the rules directly relating to panel selection but broader institutional rules relating to the size of the court, the appointments process (how ideologically split the court is), and the jurisdiction of the court (which may determine how contentious the cases before the court are, with higher GPRs for more contentious dockets). It will also likely depend on norms about consensus and dissents. However, interestingly, the GPRs have been relatively stable in recent years both in Canada and in the hypothetical uS hearing in panels. Figure 4-3 shows the differences in the GPR at the Supreme Court of Canada and the uS Supreme Court over time. The GPR for the uS Supreme Court ranged from a low of 34 percent during the Warren Court to a high of 41 percent during the Burger Court. It has, however, stayed fairly consistent for the last two chief justices (that is, since 1986). The GPR for the Canadian Supreme Court was much lower, generally ranging from 12 to 15 percent since 1973. Again, the GPR of the Supreme Court of Canada has been fairly stable since the mid-1980s (after Canada adopted the Charter of Rights and Freedoms).

Although the GPR shows the proportion of cases that theoretically could have been decided differently, many of these in fact may not have been decided

Gaming Prospect Ratio

Figure 4-3 Gaming Prospect Ratio (GPR) for Canada and the United States over time (percent). The GPR for both countries has been fairly constant over the past few decades, with the Canadian rate about a third of the hypothetical uS rate.

differently. The most obvious example is the US Supreme Court, which decides all cases en banc, and hence does not provide opportunities for gaming through panel selection. In addition, a chief justice may choose to forgo the opportunity to game the outcome of appeals by choosing to hear the appeal with a full court instead of a smaller panel.

To what extent is a chief justice or president willing to take up this potential to alter outcomes? To look at this, we use a measure we call the Gaming Resistance Ratio (GRR). It is the proportion of cases that could have been decided differently by a smaller panel but were instead heard by a larger panel. For example, in Canada panels of nine where the votes were 6-3 could have been decided differently as a panel of five, and where the votes were 5-4 could have been decided differently as panels of five or seven. As seen in Figure 4-4, the GRR of the uS Supreme Court is approximately 100 percent, because the potential gaming possibilities are essentially eliminated by the fact that the court decides all cases en banc. Conversely, the GRR for India, South Africa, and the uK were essentially zero as they did not hear decisions in panels large enough to alter the probability of gaming.

Gaming Resistance Ratios

Figure 4-4 Gaming Resistance Ratios (GRR) for various countries for 1970 to the early 2000s. A country’s GRR is the number of cases that could have been decided differently by a smaller panel but were heard by a larger panel. For Canada and Australia, about 30 percent of cases could have come out differently if the Court had used smaller panels. Because of the small panels, India and the UK have a GRR of 0. The United States has an almost 100 percent GRR because it never hears decisions in smaller panels.

The GRR in Canada and in Australia, on the other hand, are more informative. The GRR is slightly higher for the Supreme Court of Canada than for the Australian High Court (28 percent versus 26 percent). For each court, the chief justice resisted the opportunity to potentially influence the outcome in about a third of the cases where he or she had the opportunity to do so. However, as with the GPR, the GRR changes over time, but the changes were much more extreme. In Canada, the gaming resistance ratio varied from a low of 15 percent under Chief Justice Dickson to a high of 69 percent under Chief Justice McLachlin, raising the possibility that different chief justices may have had different tastes for influencing outcomes through panel selection. Chief Justice McLachlin, for example, appears to have largely refrained from using the power to game outcomes through using smaller panels. Alternatively, she may have either faced a set of cases for which it was more difficult to predict the resulting votes or, even with a similar set of cases, been less able to predict how the justices on her court will vote than other chief justices.

Therefore, chief justices in high courts outside of the united States have scope within which to try to influence the outcomes of appeals via panel selection. For some countries such as India, there are almost unlimited opportunities for gaming the system through panel selection, though it is difficult to tell from the evidence whether these opportunities were taken. For other countries such as Canada and Australia, the nature of the court provides the chief justice with potential for influencing outcomes through panel selection but she sometimes resists, pointing perhaps at differences in norms concerning the legitimacy of influencing outcomes through panel selection.

  • [1] Order Granting Appellant’s Pet. for Jacqueline Halbig et al. v. Sylvia Mathews Burwell,No.14-5018 (DC Cir., Sept. 4, 2014), 2014 WL.
  • [2] R (Bancoult) v Secretary of State for Foreign and Commonwealth Affairs [2008] UKHL 61,[2009] 1 AC 453.
  • [3] T.T. Arvind & Lindsay Stirton, “Lawyers and the Legal Model: Judicial Ideology, JudicialProfessionalism and Institutional Strategy among the Law Lords” (2012) at 30, online:
  • [4] Andrew Heard, “The Charter in the Supreme Court of Canada: The Importance of WhichJudges Hear an Appeal” (1992) 24:2 Canadian Journal of Political Science 289 at 305 [Heard].
  • [5] Heard, supra note 56 at 305.
  • [6] Theodore Eisenberg, Talia Fisher & Issie Rosen-Zvi, “Does the Judge Matter? ExploitingRandom Assignment on a Court of Last Resort to Assess Judge and Case Selection Effects”(2012) 9:2 Journal of Empirical Legal Studies 276 [Eisenberg et al., “Does the Judge Matter”].
  • [7] Eisenberg et al., “Does the Judge Matter,” supra note 58 at 276.
  • [8] Stacia Haynie, “Judicial Decision-Making and the Use of Panels in the South AfricanAppellate Division, 1950-1990” (2002) 29:2 South African Journal of Political Studies 147 at 154.
  • [9] Hausegger & Haynie, “Use of Panels,” supra note 60 at 155.
  • [10] Hausegger & Haynie, “Use of Panels,” supra note 60 at 155.
  • [11] This assumes the lack of panel effects, which will be discussed in Chapter 5.
  • [12] Again assuming no panel effects.
< Prev   CONTENTS   Source   Next >

Related topics