8.1.1 Definition of JI

When analyzing JI, researchers must decide whether to define JI in terms of the autonomy that judges have from other branches of government and/or in terms of the ability of the judiciary to influence policy outcomes (that is, have their decisions implemented). Most scholars define JI strictly in terms of autonomy (e.g., Cox Reference Cox1996, Kornhauser Reference Kornhauser, Burbank and Friedman2002, Howard and Carey Reference Howard and Carey2004, La Porta et al. Reference La Porta, Lopez-de-Silanes, Pop-Eleches and Shleifer2004, Ríos-Figueroa Reference Ríos-Figueroa2007, Helmke and Rosenbluth Reference Helmke and Rosenbluth2009, Keith et al. Reference Keith, Tate and Poe2009, Gibler and Randazzo Reference Gibler and Randazzo2011)Footnote ⁷ or as a combination of both autonomy and influence (Ferejohn and Kramer Reference Ferejohn and Kramer2002, Feld and Voigt Reference Feld and Voigt2003). I will follow the general rule and focus on autonomy. Given that I restrict my empirical analysis to democratic countries, the implementation of judicial decisions can, in principle, be assumed.

8.1.2 Inconsistency within and across Measures of JI

A researcher’s measure of JI can be problematic if it lacks internal consistency, which occurs when the chosen proxies do not match the definition of JI. Glaeser et al. (Reference Glaeser, La Porta, Lopez-de-Silanes and Shleifer2004) and La Porta et al. (Reference La Porta, Lopez-de-Silanes, Pop-Eleches and Shleifer2004), for example, define JI as the ability of judges to enforce laws without interference (an autonomy-based measure) but measure JI in terms of whether judicial decisions are a source of law.

Internal consistency can also be a problem when researchers use measures of other related concepts as a proxy for judicial independence. For example, researchers such as Dove (Reference Dove2016) cite Barro (Reference Barro1997) as providing evidence that JI leads to economic growth, but Barro shows the effect of the central bank rather than judicial independence on growth. Similarly, Linzer and Staton (Reference Linzer and Staton2015) include the Contract Intensive Money (CIM) score from Clague et al. (Reference Clague, Keefer, Knack and Olson1999) in their composite measure of JI despite the CIM score reflecting the proportion of money in a given polity that is held in banking institutions.

This can also lead to inconsistency across measures since researchers are not all measuring the same concept. For example, Hayo and Voigt (Reference Hayo and Voigt2010) measure whether a constitution has an explicit statement of judicial independence, while Melton and Ginsburg (Reference Melton and Ginsburg2014) are skeptical as to whether this will have an effect. Hayo and Voigt (Reference Hayo and Voigt2007) argue that judges who only serve one term are more independent, while Ríos-Figueroa and Staton (Reference Ríos-Figueroa and Staton2014) argue that it is only important that the judge’s term is longer than those who elected or appointed them. Melton and Ginsburg (Reference Melton and Ginsburg2014) argue that judges with lifetime terms are more independent. Other concepts such as the number of judges, selection procedure, removal procedure, salary insulation, changes to rules, and dependence on other branches are less debated but vary in terms of whether they are included in the measure altogether (Melton and Ginsburg Reference Melton and Ginsburg2014).

As a result, independent measures of JI may be uncorrelated even when trying to capture the same dimension of independence. For example, Haggard et al. (Reference Haggard, MacIntyre and Tiede2008) show that the correlation between the La Porta et al. (Reference La Porta, Lopez-de-Silanes, Pop-Eleches and Shleifer2004) measure of JI and the measure used by the Word Economic Forum is only 0.15, even though both measures are attempting to capture the autonomy dimension of judicial independence.

8.1.3 Unclear Distinction between De Jure and De Facto JI

Both problems are also exacerbated by the strategies researchers use to distinguish between de jure and de facto JI. Tables 8.1 and 8.2 summarize existing measures of de facto and de jure JI in the literature, respectively, and Table 8.3 breaks down which indicators of de jure JI are used by each indicator. Starting with Feld and Voigt (Reference Feld and Voigt2003), many researchers consider the two to be separate concepts, with de jure JI referring to the amount of JI guaranteed in legal texts and de facto JI to the amount of JI that exists in practice.

Despite the intent to distinguish between legally guaranteed and actually occurring independence, every measure of de facto JI mixes elements from both de jure and de facto JI. Measures based on expert surveys (e.g., Feld and Voigt Reference Feld and Voigt2003, Howard and Carey Reference Howard and Carey2004, Cingranelli and Richards Reference Cingranelli and Richards2008) cannot ensure that the surveyed experts separate the influence of institutional guarantees in their assessment of the independence of a country’s judiciary, and measures using procedural checklists include structural factors that should be associated more closely with the de jure concept.Footnote ⁸

It is also not clear what the de jure concept is measuring. Most measures amount to aggregating a checklist of rules and procedures (e.g., Feld and Voigt Reference Feld and Voigt2003, Hayo and Voigt Reference Hayo and Voigt2014). But, as Ríos-Figueroa and Staton (Reference Ríos-Figueroa and Staton2014) argue, whether researchers recognize it or not, by using the de jure concept they are implicitly trying to capture the incentives that the written guarantees of independence provide the actors, rather than simply the semantic content of the written guarantees of JI. It is not clear a priori which rules should be included in a given measure of de jure JI, and, as a result, different measures of de jure JI (Feld and Voigt Reference Feld and Voigt2003, La Porta et al. Reference La Porta, Lopez-de-Silanes, Pop-Eleches and Shleifer2004, Keith et al. Reference Keith, Tate and Poe2009) are only weakly correlated with one another.

8.1.4 Composite Measures as a Corrective?

Recognizing the multiplicity of different measures of JI and the uneven coverage of these measures across both countries and time, some scholars have created composite measures of JI. Hayo and Voigt (Reference Hayo and Voigt2014, Reference Hayo and Voigt2016) generate a time-series cross-sectional measure of de jure JI for 100 countries between 1950 and 2005 using factor analysis to extract the shared information from a variety of structural indicators of JI; Linzer and Staton (Reference Linzer and Staton2015) use a Bayesian Item-Response model to create a composite measure of de facto JI from 1948 to 2012 across over 200 countries by pooling information from a variety of existing de jure and de facto measures of JI (Feld and Voigt Reference Feld and Voigt2003, Howard and Carey Reference Howard and Carey2004, Cingranelli and Richards Reference Cingranelli and Richards2008, Keith et al. Reference Keith, Tate and Poe2009, economic and investment indexes from Clague et al. Reference Clague, Keefer, Knack and Olson1999, the Global Competitiveness Report, the International Country Risk Guide, the XCONST index from Polity IV).

It is unclear how to interpret Linzer and Staton’s composite measure or what to do when other researchers use the Linzer and Staton measure of de facto JI to test the relationship between de jure and de facto JI given that the latent variable modeled by Linzer and Staton contains information from both de jure and de facto measures (Hayo and Voigt Reference Hayo and Voigt2019).

8.1.5 Implications for Existing Findings

In sum, we should be skeptical of the validity of existing measures of JI and be cautious in accepting the lack of cohesive empirical findings from the literature at face value. Recall a central confusion from the literature: Feld and Voigt (Reference Feld and Voigt2003) and Hayo and Voigt (Reference Hayo and Voigt2007, Reference Hayo and Voigt2016, Reference Hayo and Voigt2019) claim that de facto and not de jure JI is correlated with economic growth, but that there is a weak correlation between de facto and de jure JI, whereas Howard and Carey (Reference Howard and Carey2004) claim that de facto JI is correlated with political rights. La Porta et al. (Reference La Porta, Lopez-de-Silanes, Pop-Eleches and Shleifer2004) and Keith et al. (Reference Keith, Tate and Poe2009) claim that a de jure measure of JI is correlated with political and economic freedom and respect for human rights, respectively, and Melton and Ginsburg (Reference Melton and Ginsburg2014) argue that certain combinations of indicators of de jure JI are correlated with de facto JI.

The inconsistency of these results is unsurprising, considering that the de jure measures from each of these projects are very weakly correlated (Ríos-Figueroa and Staton Reference Ríos-Figueroa and Staton2014). Regarding Howard and Carey’s (Reference Howard and Carey2004) measure of de facto JI, based on US state department reports, their coding criteria mix de jure and de facto concepts. Feld and Voigt’s (Reference Feld and Voigt2003) de facto measure mixes in de jure concepts, and Melton and Ginsburg (Reference Melton and Ginsburg2014) and Hayo and Voigt (Reference Hayo and Voigt2019) use the composite measure of de jure and de facto indicators from Linzer and Staton (Reference Linzer and Staton2015).

Where do we go from here? In their review of measures of judicial independence, Ríos-Figueroa and Staton (Reference Ríos-Figueroa and Staton2014) argue,

Similarly, Melton and Ginsburg (Reference Cheibub, Elkins and Ginsburg2014) argue that the relationship between individual components of de jure JI (such as selection procedure, judicial salary, judicial tenure, etc.) and de facto independence should be considered separately and theoretically justified.

I concur that theory is needed to map the relationship between specific rules and judicial independence and that our instinct with respect to the measurement of rules should be to disaggregate rather than aggregate. In addition, in order to properly measure the effects of specific rules (rather than aggregated indices), we first need to theorize about how the presence or absence of a rule affects specific judicial behaviors instead of aggregated indices of de facto JI based primarily on expert surveys.

To this end, in the next section, I return to the concept of judicial discretion and use it to motivate a theory of JI as the interaction between judicial discretion and the preferences of the judiciary.

8.2.1 Constitutional Discretion Is Proportional to Constitutional Rigidity

Cooter and Ginsburg (Reference Cooter and Ginsburg1996) argue that judicial discretion, defined as the extent to which the judiciary can use statutory interpretation,Footnote ⁹ increases as legislative override of judicial decisions becomes more difficult. Their logic is straightforward: when judges’ preferences over policy diverge from those of the government, judges may wish to move policy closer to their ideal point by way of statutory interpretation. However, because the legislative veto players can, in most cases, come together to override the judiciary, the ability of the judiciary to interpret laws will be limited when the executive and legislative branches are aligned. Conversely, when the legislative veto players conflict, the judiciary may interpret laws to the extent that at least one of the legislative vetoes prefers the interpreted policy to the original.Footnote ¹⁰

A generalization of the Cooter and Ginsburg argument is that as the number of veto players increases, so too does the discretion of the judiciary. Let us assume that a political system has three veto players (e.g., three parties in a coalition government or three political institutions in a presidential system).Footnote ¹¹ Figure 8.1 presents the ideal points (preferences) in a two-dimensional space. Assume that the horizontal axis represents the left-right continuum and that the vertical axis represents the environment. If each of these actors prefers points closer to their own preference over points further away, then they cannot change any policy located inside Triangle 123. Indeed, for any point inside this triangle, any movement of the status quo to the north will be objected to by Veto Player 3; any movement to the south will be objected to by Veto Player 1, and any movement to the east or west will be objected to by either 2 or 3. Therefore, a legislative change from point L1 to L2 is impossible because it will be objected to by Players 1 and 2, who will find the final outcome further away from their preferences. Similarly, a change from L2 to L1 will find Veto Player 3 objecting.Footnote ¹²

Figure 8.1 Legislative core: the court can make any statutory interpretation inside it

This analysis can be used in order to explain judicial discretion since any decision inside the triangle cannot be overruled by the political system. If the judiciary in the corresponding country prefers L1 or L2, it can interpret the law accordingly without any fear of being overruled. In addition, the courts could modify their opinion (a delicate stare decisis case) from L1 to L2 without any interference from the political system. However, if it prefers points J or K, it will have to select points J’ and K’ in order to avoid a legislative decision overruling its interpretation. So, if the statutory interpretations are within the political core (Triangle 123), no reaction by the political system is possible. Therefore, the size of the legislative core is an appropriate proxy for the discretion of the judiciary with respect to regular legislation.

What would happen if the basis of the judicial decision is the constitution (constitutional interpretation) and not any particular law (statutory interpretation)? Then, instead of the legislative core of the political system, we would have to base the analysis on the constitutional core. In most countries, it is more difficult to modify the constitution than the legislative status quo.Footnote ¹³

Figure 8.2 gives a visual representation of this situation. I have added one more veto player than what was in Figure 8.1. The quadrilateral 1234 represents the constitutional core (the Veto Player 4 is also required for a constitutional revision). As a result, changes to the constitution inside Quadrilateral 1234 are impossible, and any constitutional interpretation inside this area becomes possible. The reader can verify from Figure 8.2 that while a judicial decision J would be overruled (no matter whether it was on statutory or constitutional grounds), a decision K would be overruled on statutory grounds but would be valid on constitutional grounds. So, in our hypothetical example, if the court had based its decision on the constitution, a legislative overrule would have been irrelevant, and a constitutional amendment would have been impossible. So, the larger the difference between the constitutional and the judicial core (the shaded area in Figure 8.2), the more empowered the judges are to make constitutional interpretations (as opposed to statutory ones).

Figure 8.2 Constitutional core larger than legislative core: any constitutional interpretation within the constitutional core stands

Assuming the constitutional and supreme courts do not want to be overruled, they will exercise discretion proportionally to the size of the constitutional core. It follows that when considering discretion with respect to constitutional matters, it is appropriate to use the size of the constitutional core as a measure of constitutional discretion. For ordinary courts (or, more accurately, for statutory interpretations of any court), the determinant factor will be legislative overrule (i.e., the size of the legislative core of a country), while for constitutional decisions, the decisive factor will be the size of the constitutional core.

8.2.2 The Importance of Jurisdiction

The literature often focuses on higher courts, so there is relatively no work on lower courts (Burbank and Friedman Reference Burbank, Friedman, Burbank and Friedman2002). This is because it is argued that the incentive structures of lower courts are different. Local courts may be independent of local governments, but they may also be highly reliant on senior judges (in particular for promotions) (Ramseyer and Rasmusen Reference Ramseyer and Rasmusen2003). Lower court judges want to be promoted, whereas supreme court or constitutional court judges do not, which means that lower court judges will be more beholden to superior judges (Salzberger and Fenn Reference Salzberger and Fenn1999). In addition, it is thought that lower courts are more constrained than the supreme court (Burbank and Friedman Reference Burbank, Friedman, Burbank and Friedman2002).

There are generally two different models for constitutional courts, generalized into the “American” and “European” Systems (Jacob et al. Reference Jacob, Blankenburg, Kritzer, Provine and Sanders1996). On the one hand, these models are different because they have different appointment mechanisms and different terms. For certain European courts, the appointment of judges is seen as nonpolitical – the process, usually through a constitutional tribunal, is often criticized for being too secret, unlike that of the appointment of Supreme Court justices in the US (Ferreres Comella Reference Ferreres Comella2009). We will test this expectation in Section 8.4. Unlike many ordinary judges, constitutional court judges often have term lengths and limits. This is because while the rulings of ordinary judges can be overruled by higher judges, constitutional judges are the final and only say in the matter (Ferreres Comella Reference Ferreres Comella2009). There are conflicting expectations about constitutional courts in the literature. Epstein et al. (Reference Epstein, Knight and Shvetsova2001) compare the characteristics of constitutional and supreme courts. They summarize the literature as follows:

Table 8.4 presents a list of countries with constitutional courts (which will be included as a dummy variable in my empirical analysis).

The main difference between constitutional and ordinary courts is that constitutional courts provide the final say in constitutional matters (Finck Reference Finck1997). At their most basic level, “constitutional courts have the power of judicial review and invalidation of unconstitutional statutes and statutory provisions” (Garlicki Reference Garlicki2007: 67). In most cases, constitutional courts are added later on after the judicial system of a country is well established.

The distinction between constitutional and supreme courts started with Kelsen arguing the differences between the European and American models. Within systems that have a constitutional court, only “a single court (usually called a ‘constitutional court’) can exercise judicial review; other courts are typically barred from so doing” (Epstein et al. Reference Epstein, Knight and Shvetsova2001: 121). Because of this, constitutional courts are thought of as more centralized or concentrated and specialized (for example, in Germany), whereas supreme courts are seen as more decentralized or diffuse and general (Finck Reference Finck1997, Horowitz Reference Horowitz2006). This means that in systems with only an ordinary court, it can rule on whether an act is unconstitutional, but it can only do this “a posteriori,” or after an act has occurred (Epstein et al. Reference Epstein, Knight and Shvetsova2001). Here, centralized means a clear delineation in terms of how a case will reach a constitutional court. Ordinary courts deal with all legislation – if a country has a constitutional court and an ordinary court, the ordinary court will decide whether it is a constitutional matter. In practice, there is much more overlap between the constitutional court and the highest ordinary court (in some cases, a supreme court), which means that there can and will be tensions between the two (Garlicki Reference Garlicki2007). For instance, in Germany, where the constitutional court has a vast amount of power, there are no clear and understood boundaries of the court’s jurisdictions in practice (Garlicki Reference Garlicki2007).

8.2.3 Judicial Preferences

To distinguish between cases where a court genuinely aligns with the government from those where the court aligns with the government under pressure, the positions of the executive, the legislature, the judiciary, and others must also be known (Cameron Reference Cameron, Burbank and Friedman2002). This approach can place a court into one of four categories. The first three, enumerated by Vanberg (Reference Vanberg2001), are (1) a friendly court where it shares the same preferences for the policy as the legislature, (2) a submissive court where it agrees that the policy is unconstitutional but will only do so when it knows that the legislature will abide by the court’s decision, and (3) an assertive court where it will vote that a policy is unconstitutional regardless of the legislature’s actions. Another possibility is (4) an authoritative court, which would force the executive to respect its preferences.

However, these distinctions require knowing the positions of the court in order to distinguish between a friendly, a submissive, an assertive, and an authoritative court. Therefore, in cases where the preferences of government actors are, in fact, measurable, it is possible to infer the positions of the judiciary from the rules governing judicial appointments and the positions of those who appoint. For instance, in countries where judges are appointed by legislators, the position of the legislatures can be a reliable proxy for the position of the judges. This is the case for a country like France where the “Conseil Constitutionnel members – being political appointees – are actually incapable of independent action and behave … necessarily as partisans and not as judges” (Stone Sweet Reference Stone Sweet2007: 73). Based on this, either the constitutional court legislates in the same manner as the parliament or it does not legislate at all, given that it is a product of appointees from the parliament (Stone Sweet Reference Stone Sweet2007). This is similar to the case of the US, where, although not a constitutional court, the decisions of the Supreme Court can be predicted based on the position of the median justice in the majority coalition (Carrubba et al. 2012). This is because, in both cases, the nomination process of the judiciary is fairly transparent, making it known who appointed the judge, which makes the appointee a suitable proxy for the position of the judge. The nomination and appointment mechanisms can range from just a single actor, such as the head of state, to multiple actors, such as the head of state, one or multiple chambers of the legislature, and even approval from the judiciary. How many actors – as well as which actors – there are will determine how visible the process is as well as how applicable a proxy method of appointment may be.

8.2.4 JI as the Conjunction of Constitutional Rigidity and Judicial Preferences

When we can measure both the preferences of the government and the judiciary, we should expect the relationship between judicial preferences, discretion, and independence to be conditional: High discretion will be a necessary but not sufficient condition for JI. Indeed, if a court has no discretion, it will not be independent, but if it has high discretion, it still may not be independent. If the justices are appointed by a political actor, the appointees are likely to have identical preferences with their principal, and the court will not be considered independent.

However, since the preferences of members of the judiciary are unknown, we should expect that the relationship between the two variables will be heteroskedastic (just like the relationship between constitutional rigidity and amendment frequency; see Chapter 6).

Figure 8.3 makes this point visually. High constitutional rigidity generates high judicial discretion; however, this high discretion may or may not be used by the judiciary and, consequently, under these conditions judicial independence may be high or low. On the other hand, low constitutional rigidity leads necessarily to low judicial independence.

Figure 8.3 Low constitutional rigidity is a necessary condition for low judicial independence

8.2.5 Constitutional Strikes as a Measure of JI

Figure 8.4 presents an abstraction of the set of outcomes that the majority of the constitutional court (from now on, the judiciary) prefers over the status quo (not presented in the figure). I call this set the win-set of the status quo, or W(SQ), which is the intersection of the majority of preferences and restrictions different judges impose on a piece of legislation to consider it compatible with the constitution. These preferences may include elements of political judgment (Gabel et al. Reference Gabel, Carrubba, Helmke, Martin, Staton and Ward2024),Footnote ¹⁴ or they may include absolute principles (like respect for human life), proportionality between such principles (human rights and freedom of expression),Footnote ¹⁵ or any other rule of textual interpretation one considers to be in play. The shaded area presents the intersection of the majority of such judicial preferences and constraints. In this simplified game, if the government makes a decision outside of W(SQ), it will be overruled by the judiciary on constitutional grounds. Therefore, under the conditions described above, the government will make a decision inside W(SQ). This model produces no judiciary strikes because they are anticipated by the government, and the proposed solutions are not objectionable by the judiciary. The only way that there would be judiciary overrule of government actions is if the government has a dominant strategy to provoke the judiciary and be overruled by it (or at least not care about it). Such a situation could happen if the government is involved in a Nested Game (Tsebelis Reference Tsebelis1991) where it cares about the payoffs provided in another arena (e.g., electoral) and not about the survival of its own legislation. An example of this would be a conservative government legislating against abortion rights in order to appeal to its supporters regardless of the fact that it will be overruled by the judiciary. Although such cases are possible, they cannot be the predominant explanation of disagreement between governments and the judiciary at a comparative level. In fact, they require the introduction of an additional actor (the public) and a special interaction between the government and this actor to determine (and explain) the actions of the government.

Figure 8.4 The win-set of the status quo subject to constraints on judicial decision-making

Figure 8.5 replicates the previous story with one difference that increases the realism of the model: What if the government does not have exact knowledge of W(SQ)? The lighter gray shaded area indicates the government’s uncertainty over the judiciary’s win-set of the status quo. Uncertainty stems from the fact that the government may not know the exact constitutional consequences of a particular policy decision and/or the preferences of the judiciary (just like researchers). As a result, these zones of uncertainty may be very wide indeed. In this scenario, the government may make a decision G in the zone of uncertainty that it intends to be approved by the judiciary but is instead struck down. What is the inference if we observe the judiciary striking down a government decision? First, we are in Figure 8.5 instead of Figure 8.4 (that is, the government is operating under incomplete information), and second, the judiciary has preferences different than the government and has affirmed these preferences. What is the inference if we observe no disagreement? There are several possibilities: first, it could be that the government was able to anticipate the preferences of the judiciary and made a proposal acceptable to it; second, it is possible that the judiciary does not have any significant differences from the government; and third, it may be that the judiciary is afraid to contradict the government. In other words, in the first case, the judiciary prevails; in the third, the government; and in the second, there is an identity of preferences. As a result, in the case of judicial approval, we can make no inference.

Figure 8.5 The win-set of the status quo subject to constraints on judicial decision-making, with uncertainty

However, a lack of any disapproval of existing law should be counted as a disagreement between the government and the judiciary. We should identify the cases where the judiciary makes decisions conflicting with the decisions of the current government. Constitutionally overruling old laws is not an indication of JI, but it could be an indication that the laws have become obsolete or, even more perversely, that constitutional courts strike down old laws or provisions to suit the preferences of the current executive.Footnote ¹⁶ Therefore, I operationalize the JI variable as the percentage of cases that a constitutional court strikes down the decisions of the current government as unconstitutional over the total number of cases that the court votes them either up or down.

In sum, when the judiciary is not independent of the government, we will not see strikes against the sitting government’s legislation; when the judiciary is independent of the government, we may or may not see strikes, depending on whether the preferences of the judiciary and the government align.

Some researchers have used the frequency of court overrule as a measure of judicial independence, although this is only the case for individual countries and is not always in terms of overruling against the sitting government. Their findings align with my expectation of a positive relationship between the size of the constitutional core and JI. For example, Santoni and Zucchini (Reference Santoni and Zucchini2004) examine the Italian Constitutional Court from 1956 to 1992 and show that the frequency of disputes over the constitutionality of laws is increasing with the size of the constitutional core (defined as the number of parties needed to agree on a constitutional revision), though they do not restrict their analysis to only conflicts over legislation from a sitting government.Footnote ¹⁷ Similarly, Ríos-Figueroa (Reference Ríos-Figueroa2007) analyzes all judicial decisions by the Mexican Supreme Court from 1994 to 2002 and shows that the judiciary is more likely to strike down legislation from the PRI when the fragmentation of the political system is high. Sánchez et al. (Reference Sánchez, Magaloni, Magar, Helmke and Ríos-Figueroa2011) include all Mexican Supreme Court rulings until 2007 and show that after the PRI lost the presidency in 2000 (and the political system became more fragmented), the Mexican Supreme Court became more likely to strike down laws of the sitting government via constitutional review.Footnote ¹⁸

In order to replicate these analyses at the cross-national level, we need the indicator of constitutional rigidity developed in Chapter 6 along with decisions of constitutional courts in different countries rejecting laws from the government in power. The expectation is a heteroskedastic relationship between judicial discretion and the number of judicial decisions on the unconstitutionality of current government laws.