A Comparative-Empirical analysis of administrative courts in Mexico

Abstract

The main function of administrative courts in Mexico is to resolve disputes between administrative agencies and citizens. Mexico is a federal system with 31 states and a Federal District. Twenty-nine states and the Federal District have administrative courts of this type. Most of these courts follow the French model of reviewing administrative actions in bodies that do not form part of the regular justice system. However, almost half of the states have deviated from this model and ascribed these administrative courts to the judicial branch. How does this change in the institutional framework influence the way administrative court judges review administrative action disputes? In order to answer this question we analyzed the rulings of judges from the different types of courts empirically. The Mexican federal court structure made this experiment possible because there are both administrative courts incorporated into the judiciary and autonomous courts. We used a database of more than 4,000 cases from over twenty local administrative courts. We analyzed the influence of the branch to which the court belongs, the procedures of appointment for judges, the length of a judge’s term in office, and the protection of judges’ salaries over their actual decisions. We classified decisions into two broad categories: pro-government decisions and case dismissals. The results point toward evidence that the branch to which the court belongs, the length of a judge’s term in office and governor intervention in the appointment of judges affect judges’ decisions.

Resumen:

La función principal de la justicia contenciosa en México es resolver conflictos entre particulares y servidores públicos. México es una federación compuesta por 31 estados y el Distrito Federal. Veintinueve estados y el Distrito Federal cuentan con un tribunal que resuelve este tipo de conflictos. La mayor parte de estos tribunales se constituyeron siguiendo la tradición francesa de revisión de actos de autoridad, no incorporando a estos tribunales al sistema de justicia común. Sin embargo, casi la mitad de los estados se ha desviado de esta tendencia incorporando sus tribunales al poder judicial del Estado. ¿Cómo puede influir este cambio de diseño institucional en cómo resuelven estas disputas los jueces? Con el propósito de contestar a esta pregunta en el presente artículo analizamos empíricamente las decisiones de distintos juzgadores en cada tipo de tribunal. La estructura federal de México nos permitió realizar este experimento pues al mismo tiempo coexisten dentro del país tribunales incorporados al poder judicial y tribunales autónomos. Utilizamos una base de datos compuesta por más de 4,000 decisiones en más de veinte tribunales del país. Específicamente estudiamos la influencia del poder al que el tribunal pertenece, los procedimientos de designación de jueces, los periodos de designación y la protección de los salarios de los jueces sobre las decisiones que estos toman. Para realizar este análisis clasificamos las resoluciones en dos grandes categorías: decisiones pro-gobierno y sobreseimientos. Nuestros resultados sugieren que tanto la adscripción del tribunal como la duración del encargo e intervención del gobernador en la designación de jueces influyen en sus decisiones.

I

Introduction

For decades researchers have been questioning the pure legal prototype of courts that the architects of most of legal systems tend to assume exists.¹ Models of judicial behavior have emerged as systematic, empirical, theoretically-based attempts to explain what courts and judges do.² According to the literature on judicial behavior, courts are political complex structures that can be analyzed like other political institutions. Courts are bodies in which judicial power interacts with the executive and the legislative powers in a political context; huge organizations in which judges must administer employees and budgets; and institutions seeking to interpret rules, create law, and solve conflicts between parties.

Given this complexity, courts have to be analyzed not only from an ideal theoretical perspective, but also from an empirical one in order to obtain a real picture of what they do. Furthermore, judges have to be analyzed as agents affected by different factors, including the organization of the court; the rules applying to their jobs; their preferences, values, and political circumstances; and the interaction of the two other branches of the State.

Constitutional courts are certainly political actors, and this may be why scholars of judicial behavior have focused on them. Less attention has been given to the design of administrative courts, although they are one of the most widely used mechanisms for challenging agencies’ decisions.

The design of administrative courts is not uniform and varies over time and across countries. A divergence in the interpretation of the separation of powers doctrine prompted the appearance of two main approaches to designing administrative courts —the French model and the judicial review model. In the French model, administrative justice belongs to the executive branch, under the logic that the separation of powers requires a more restricted scope of action for the judiciary³ while the common-law interpretation places the administrative courts within the judicial branch, under the logic that any function of a truly judicial nature must be exercised by the judicial branch alone/only.⁴ Some countries use a hybrid of the two models.

In Mexico, there is significant variation between these models in the institutional design of its administrative courts. Mexico followed the French model for the solution of controversies between the State and citizens. Of Mexico’s 31 states and the Federal District, 29 jurisdictions have administrative courts that review agencies’ decisions. More than 50% of the courts are part of the local executive branch, while the rest are part of the judiciary.

Questions that arise from this divergence speak to the implications of the choice of one or other design. Does choice of design have any impact on a court’s outputs? Some scholars have shown interest in questions regarding judicial review of agency action.⁵ Empirical analysis of administrative adjudication includes studies of the reasons for creating administrative courts;⁶ as well as studies of administrative courts’ performance and their role in agencies’ performance. These studies include analyses of the performance of specialized courts in general;⁷ specialized courts in Indonesia;⁸ administrative courts in Colombia;⁹ the expansion of US administrative law and the convenience of having specialized bodies to deal with it;¹⁰ the performance of administrative courts and their role in controlling agencies;¹¹ the relationship between administrative courts and policy-making;¹² the role of administrative courts in agency performance;¹³ the impact of specialized courts in intellectual property cases;¹⁴ the role of the adversarial model in administrative tribunals’ behavior;¹⁵ the relationships between congress, executive, and judiciary;¹⁶ and the relationship between courts and agencies.¹⁷

A number of scholars have done comparative administrative law analyses on the differences between French administrative law and other administrative law systems such as the Anglo-American, German or English systems.¹⁸ But none of these studies has been able to compare actual outcomes of two variations of the French model of administrative adjudication within a single country.

The divergence in Mexico’s design makes it an interesting laboratory to study the consequences of choosing different institutional designs to create administrative courts. We will use the two main traditions of administrative adjudication as a framework to describe the Mexican system. Based on this, we will develop two models to test two hypotheses related to the design of administrative courts. Our analysis will use two datasets: an analysis of state constitutions and administrative court statutes provides the data for the first, and an analysis of more than 4,000 cases decided by 23 administrative tribunals in Mexico provides the data for the second.¹⁹

Our first hypothesis is related to the so-called “independence guarantees” for judges, such as tenure, salary protection and limitations on the executive branch in the appointment procedure. We hypothesize that judiciary courts offer more guarantees of independence for judges than those courts that are part of the executive branch. Therefore, judiciary courts are more likely to protect judges’ salaries and tenure.

Our second hypothesis examines the incentive structures for judges. We hypothesize that judges that are part of executive branch courts will decide cases differently than judges that work in a judicial branch court. We will compare pro-government decisions vs. pro-citizen decisions in both types of courts.

Although a possible approximation to evaluate judicial independence is to analyze the percentage of pro-government decisions, we believe it is very difficult to find a proxy for judicial independence. Pro-government decisions may reflect “good” administrative actions, rather than a failure to allow judges independence. Without a variable to distinguish between these factors, we will not consider administrative judges’ actual independence.

To date, discussions regarding judges’ incentives have been dominated by theoretical, rather than empirical analyses. Moreover, studies regarding Mexican courts had been focused on the federal level and on civil courts.²⁰ On the state level, there are only two empirical studies on civil courts both of which analyze civil justice issues.²¹ Unlike other contributors to the debate on the institutional design of administrative courts, our study not only relies on real data, but it also analyses such design at the subnational level. This paper sheds new light on the consequences of local legislators’ choices in creating administrative courts.

The paper is organized as follows: Section II presents an overview of the history and characteristics of administrative adjudication traditions; Section III describes the Mexican system of administrative courts; Section IV describes the data and explains our empirical model and testable hypotheses; Section V presents the findings and lastly, we present our conclusions.

II

Models of Administrative Adjudication

There are different models of administrative adjudication.²² In order to explain these models, we will first define administrative justice to better describe these models.

Although scholars have studied administrative adjudications for years,²³ the term administrative justice is recent.²⁴ Michael Adler defines administrative justice as the justice inherent to administrative decision-making.²⁵ This definition implies procedural fairness as well as substantive justice. Mashaw describes administrative justice as “the qualities of a decision process that provide arguments for the acceptability of governments’ decisions and it is referred to initial and internal decision-making”.²⁶ Other authors describe administrative justice as that concerned with the extent to which individuals affected by agencies’ decisions are treated fairly and have the ability to redress grievances in cases of a breach of fairness.²⁷ Civil law tradition administrative justice is generally associated with all administrative adjudication processes.

Regarding administrative justice functions, Buck, Kirkham and Thomson proposed a typology based on three rings that mark its functional landscape.²⁸ The inner ring, “getting it right,” refers to the initial decision-making process by public bodies, encompassing the relevant law and procedure. The middle ring, “putting it right,” refers to the whole range of redress mechanisms available to citizens who question the initial decision-making process (courts, tribunals, ombudsman or other independent complaint-handlers). The outer ring, “setting it right,” refers to the network of governance and accountability relationships surrounding both the public bodies tasked with first-instance decision-making and those responsible for providing remedies.

Following the above mentioned authors, we will use administrative justice as a broad term that encompasses the three main functions/rings and will focus on the middle ring related to the different mechanisms of redress. Hence, this paper focuses solely on the mechanisms for challenging an administrative decision, specifically in mechanisms for resolving disputes between citizens and the government that arise from decisions of officials and agencies. We will assume that the main purpose of this challenge is to determine whether or not the action of a public body is lawful.²⁹ Finally, we will focus only on those mechanisms in which decisions have to be made by a third party (different from the agency that made the initial decision).

Third parties include executive commissions (independent from the agency making the initial decision), tribunals, specialized courts and general courts.³⁰ Some scholars classify tribunals as specialized mechanisms³¹ and courts as general ones. Tribunals are sometimes referred to as court substitutes, in that they have the power to make legally enforceable decisions, but they are regarded as having the advantages over courts in terms of speed, low cost, informality and expertise.³² Other scholars³³ classify tribunals as redress mechanisms within the executive branch and courts as mechanisms within the judicial branch.³⁴ However, in the Mexican legal system, this distinction does not exist in practice. Ordinary courts within the judiciary that solve civil law cases, family cases and criminal cases are called tribunals.³⁵ At a federal level, the only body that is referred to as a court is the Supreme Court of Justice. At a state level, only a few constitutional courts are called courts. All other jurisdictional bodies within the judiciary are called tribunals. Since administrative tribunals in Mexico are designed and function as actual courts, we will use the term courts regardless of their actual name.

Regarding the purpose of administrative redress mechanisms, scholars agree with the idea that this purpose is dual: (i) individuals’ redress and (ii) the achievement of better standards of public service and administration.³⁶ To fulfill these purposes, administrative courts should decide specific cases in which one of the parties is the government, acting as the problem-solver, and working like a fire alarm system to allow courts to monitor agency performance and create incentives so that bureaucrats do not harm citizens.³⁷

Administrative courts, like every other institution, are composed of institutional tools as well as legal tools. Different models of institutional design using different institutional tools have been used over time and differ across countries. A court’s institutional characteristics are the different manners in which a court as a whole can be arranged; they include the ascription of the court (judiciary or executive branch), specialization of judges, tenure, appointment processes, salary protection, and any other independence guarantee the legal system provides. A court’s legal tools include all the rules that administrative judges may use to decide cases. These rules include procedural and substantive rules that differ from one system to the other.

1

Institutional Design of Administrative Courts

Two institutional designs characterize most administrative courts: one in which specialized judges operate within the executive branch and one in which common-law judges provide judicial review of administrative decisions.

The French model represents one of the extremes of the spectrum because, since its origins, administrative adjudication has been a function placed within France’s executive branch. French law prohibits judges from controlling executive activities.³⁸ “French tradition refused to allow courts to review administrative decisions citing the principle of separation of powers but really was being worried of any limitation to the exercise of executive power.”³⁹ The designers of the French model believed that the executive branch is best suited to decide on substantive issues in the relationship between government and citizens. During the Napoleonic period, the administrative courts evolved into the Conseil d’Etat. The Napoleonic Constitution of the Year VIII gave them the power to solve disputes that implicated administrative matters, claims against violations of economic rights and complaints from citizens deemed to have been aggrieved by any administrative authority’s arbitrary act.⁴⁰ The French model is a result of the constitutional principle that establishes “juger a l’Administration c’est encore administrer.” The model considers reviewing the acts of government part of the administrative function. Therefore, a specialized tribunal in the Conseil d’Etat, not the judiciary, revises the acts of government.

France has assigned geographical venues to a set of courts and specialized issues, such as budget supervision, to specialized courts. The Conseil d’Etat governs them all. The evolution of administrative redress mechanisms in France includes the creation of administrative tribunals to solve first-instance disputes in 1953 and second-instance disputes in 1987, but always under the authority of the Conseil d‘Etat that is part of the executive branch. Finally, judges have suggested in some recent articles that the executive branch has sufficient mechanisms to achieve independence from the executive authority.⁴¹

The common-law tradition, in contrast to the French model, arises from the principle that government and citizens should be judged by the same rules and under equal conditions. Therefore, any authority can be brought before the common courts and judged. The judiciary has the power to protect the Rule of Law and the Constitution; any dispute in the law should be brought before it. This is an appellate review model.

The US courts’ role in reviewing agency action reflects a bipolar view of administrative action.⁴² The first view stated that courts should review administrators’ actions de novo. The second view stated that no judicial review should take place, and that Congress and the agencies should analyze these cases. Therefore, relief against unlawful government action was sought in ordinary courts of first instance. An injured citizen could file for one of the prerogative writs (chief mandamus, certiorari or habeas corpus), for an injunction or for damages in tort against the offending officer. Merril also argues that judicial review reforms in states, exemplified by the Model State Administrative Procedure Act, often retain the common-law principle that administrative action is to be reviewed by ordinary trial courts. From the beginning of this century, however, the United States has frequently deviated from this model to provide for review by three-judge trial courts, by courts of appeals generally, by a single court of appeals, or by a more or less specialized tribunal.

England also follows a common-law tradition. In this tradition, the separation of powers dictates that the general regime is part of the rule of law, and public authorities have no special legal regime.⁴³ Just as in France, the creation of many specialized administrative tribunals has accompanied the evolution of administrative justice in England; however, they form part of the ordinary judicial system and depend on the Supreme Administrative Court. Just after World War II, England created an independent system of adjudication that would be entirely isolated from government intervention. This reflected the view that administrative justice is part of the judicial system.⁴⁴ India is another example of a common-law country that has recently created administrative courts.⁴⁵

As noted above, administrative tribunals following each model have proliferated. The cases of Germany,⁴⁶ Italy,⁴⁷ Spain,⁴⁸ Japan⁴⁹ and Morocco⁵⁰ exemplify this in that administrative adjudication has changed over time in all of these countries, all of which have placed it within the judiciary at some times and within the executive branch at others.

Finally, institutional design of administrative courts also includes variables related to the independence of judges. The variables affecting court independence include the process of judge’s appointments, tenure, and salary security.⁵¹ Traditionally, life tenure confers judicial independence. A number of scholars have also addressed the influence of other branches on the judiciary.⁵² For example, Congress may have control over jurisdiction, court creation, appointment, enforcement of court rulings, appropriations for the operation of the courts and the ability to impeach judges.⁵³ Other studies have argued that the judicial appointment process, lifetime appointments and prohibition to reduce judges’ salaries influence judicial independence.⁵⁴

In general, a reduced role of the executive in appointment process, longer judicial terms —up to lifetime and at least in excess of executive terms in office— and protecting judicial salaries from reduction by other branches promote judicial independence.⁵⁵ Although most scholars describing these aspects of independence refer to general courts, they also apply to administrative courts. Indeed, the particular role of administrative courts in addressing complaints against members of the executive branch makes isolation from the executive especially important.

2

Procedural and Substantive Rules Applied by Administrative Courts

Asimow and Lubbers’ classification of adjudicating models provides a starting point for the description of the procedural and substantive rules administrative courts apply to decide cases. He describes five models of adjudication, depending on the type of initial decision, reconsideration and review mechanisms.⁵⁶ The first is the adversarial hearing/combined function/limited judicial review in which the agency makes the initial decision through an administrative judge and the reconsideration phase occurs within the agency. Courts of general jurisdiction do judicial review; the review addresses the legality and reasonableness of the agency’s decision; it is prohibited for courts to re-examine the evidence and to substitute judgment on the merits of the case. The United States provides an example of this model. The second model is the inquisitorial hearing/combined function/limited judicial review. The agency makes the initial decision; a different agent makes reconsideration; courts of general jurisdiction make judicial review. The European Union uses this model. The third model is the tribunal system, in which the tribunal is separate from the prosecuting and enforcing agency, which makes the initial decision and the reconsideration decision; judicial review occurs in generalized courts with limited powers over issues of fact or discretion. The fourth model is the de novo judicial review/general jurisdiction. The agency makes the initial decision and reconsideration; general courts make judicial review and may retry. China uses this model. Finally, the fifth model is the de novo judicial review/specialized jurisdiction. The agencies make the initial decision and reconsideration; specialized courts hearing only administrative law cases review the initial decisions. France, Germany and, as we will explain in the next section, Mexico, use this model.

The French model has always used specialized jurisdiction, which in France applies not only to institutional characteristics of the courts but to the procedural and substantive rules applied to the parties. The Conseil d’Etat has a specialized procedure to invalidate an act of the administration violates the law⁵⁷ while common-law judges use the same substantive and procedural rules for every case.

This specialization is precisely what distinguishes the French model from common-law. A number of scholars have studied the implications of having a specialized tribunal rather than a generalist court.⁵⁸ There are several studies of specialized courts such as tax courts,⁵⁹ bankruptcy courts,⁶⁰ military courts,⁶¹international trade courts,⁶² drug courts,⁶³ community courts⁶⁴ and domestic violence courts,⁶⁵ among others. Several scholars sustain that specialized courts produce higher quality decisions in time and content, help to achieve legal coherence and uniformity of judicial decisions, and help to reduce regular courts’ workload.⁶⁶ On the other hand, specialization has been seen as making judges more susceptible to external control or “capture.”⁶⁷

This paper will not analyze the consequences of specialization, but other scholarship suggests that specialization effects judicial independence. Baum, for example, hypothesizes that specialized courts will review administrative decisions aggressively because specialized judges gain the confidence to take assertive positions and because the private interests that contest government actions in those courts wield considerable influence.⁶⁸ Some other scholars⁶⁹ attach the benefits of the French system of administrative adjudication to the type of case. They suggest that depending on the specific issue, generalist courts are better than specialized courts and vice versa.

III

Administrative courts in Mexico

The Mexican State is organized in the form of a federation integrated by a Federal District and 31 states. The federal system is established in the Federal Constitution and distinguishes the powers of the federation and the powers of the states.⁷⁰ The supreme power of the federation is divided into legislative, executive and judicial branches.⁷¹ Article 73 XIX-H of the Federal Constitution provides for the review of federal administrative action. The Congress has the power to create autonomous administrative courts empowered to resolve the legal controversies between the federal public administration and individuals. At a state level, Article 116 of the Federal Constitution provides for the existence of local administrative tribunals to solve disputes between citizens and local governments. Local congresses may enact legislation to regulate the administrative tribunals’ management, as well as the applicable legal procedures.⁷² It is important to point out that local congresses can decide whether to create an administrative court.

1

Institutional design of administrative courts in Mexico

Administrative adjudication in Mexico uses the French tradition of specialized jurisdiction and the specialized procedural tradition of the French model. However, some jurisdictions established their administrative courts as part of the judiciary and others established them as part of the executive branch, while granting them autonomy in their decision-making process. Different amendments and statutes captured the issue as to whether administrative courts should be part of the executive branch or the judicial branch, and these discussions generally addressed the separation of powers principle.⁷³Within the executive power, administrative courts in Mexico support their design as being in the French model. The designers of these tribunals argued that separation of powers prohibits the judiciary from controlling executive actions, and since administrative cases differ from cases between individuals in many respects, judges must have greater specialization.⁷⁴ The fact that these administrative tribunals in Mexico follow the French model should not, however, be taken to mean that they are exactly the same as French administrative courts.

Mexico’s administrative adjudication system consists of a Federal Administrative Court,⁷⁵ which is a specialized court within the executive branch,⁷⁶ and 30 state-level administrative courts. Chart 1 shows the years in which each jurisdiction created these courts.

Chat 1

Table 1 shows each court’s institutional characteristics in the period of analysis 2006-2009. Design variables include the ascription of the court, describing the branch to which a court belongs; the existence of guarantees of tenure, describing the existence of life-time appointments; salary protection, describing the existence of constitutional provisions prohibiting the reduction of judges’ salaries; and the role of governors in judges’ appointment procedures, describing the degree of governors’ participation in the appointment process.

State	Branch	Tenure	Role of executive branch in appointment process	Regulated Salaries
Aguascalientes^1	Judiciary	No tenure	Some intervention	Protection
Baja California^2	Executive	No tenure	Some intervention	Protection
Baja California Sur^3	Judiciary	No tenure	Some intervention	Protection
Campeche^4	Judiciary	Tenure	No intervention	Protection
Chiapas^5	Judiciary	No tenure	No intervention	No protection
Colima^6	Executive	Tenure	Some intervention	No protection
Distrito Federal^7	Executive	Tenure	Some intervention	Protection
Durango^8	Executive	No tenure	Some intervention	No protection
Estado de Mexico^9	Executive	No tenure	Some intervention	No protection
Guanajuato^10	Executive	No tenure	Some intervention	No protection
Guerrero^11	Executive	Tenure	Some intervention	Protection
Hidalgo^12	Judiciary	Tenure	Some intervention	Protection
Jalisco^13	Judiciary	Tenure	Some intervention	Protection
Michoacan^14	Executive	No tenure	No intervention	Protection
Morelos^15	Judiciary	Tenure	No intervention	No protection
Nayarit^16	Executive	No tenure	Some intervention	Protection
Nuevo Leon^17	Executive	No tenure	Some intervention	Protection
Oaxaca^18	Executive	No tenure	Some intervention	No protection
Queretaro^19	Executive	No tenure	No intervention	Protection
Quintana Roo^20	Judiciary	Tenure	Some intervention	No protection
San Luis Potosi^21	Executive	Tenure	Some intervention	No protection
Sinaloa^22	Executive	No tenure	Some intervention	No protection
Sonora^23	Executive	No tenure	Some intervention	Protection
Tabasco^24	Executive	No tenure	Some intervention	Protection
Tamaulipas^25	Executive	No tenure	No intervention	No protection
Tlaxcala^26	Judiciary	Tenure	No intervention	Protection
Veracruz^27	Judiciary	Tenure	Some intervention	Protection
Yucatan^28	Executive	Tenure	Some intervention	Protection
Zacatecas^29	Judiciary	Tenure	No intervention	Protection

Notes:

1

According to Article 51 of the Constitution of Aguascalientes, the Administrative Court of Aguascalientes (Tribunal de lo Contencioso Administrativo del Estado de Aguascalientes) shall be composed of one judge appointed for fifteen years, with only one term permitted. The governor proposes and congress approves such appointment.

2

According to Article 55 of the Constitution of Baja California, the Administrative Court of Baja California (Tribunal de lo Contencioso Administrativo del Estado de Baja California) shall be composed of judges appointed for six years with possibility of one more term. The governor proposes and congress approves such appointments.

3

According to Article 64.XLIV and XLV of the Constitution of Baja California Sur, the Civil-Administrative Court of Baja California’s judiciary (Sala Civil Administrativa del Tribunal Superior de Justicia del Estado de Baja California Sur) shall be composed of judges appointed for six years. The governor proposes and congress approves such appointments.

4

According to Article 82.1 of the Constitution of Campeche, the Administrative-Electoral Court of Campeche’s judiciary (Sala Administrativa Electoral del Tribunal Superior del Estado de Campeche) shall be composed of judges appointed for six years with tenure possibility after this term. The judiciary proposes and congress approves such appointments.

5

According to Article 17.c.III of the Constitution of Chiapas and Article 224 of the Judiciary Organization of the State of Chiapas, the Administrative and Electoral Court of Chiapas’ Judiciary (Tribunal de Justicia Electoral y Administrativa del Poder Judicial del Estado de Chiapas) shall be composed of judges appointed for seven years with option to be selected for one more term. Congress appoints two of the judges and the Constitutional Court of Chiapas appoints the rest.

6

According to Article 77 of the Constitution of Colima, the Administrative Court of Colima (Tribunal de lo Contencioso Administrativo del Estado de Colima) shall be composed of judges appointed for six years with tenure possibility after this term. The governor proposes and congress approves such appointments.

7

According to Articles 2 and 3 of the Statute of the Federal District Contentious Administrative Tribunal (1995), the Administrative Court of the Federal District (Tribunal de lo Contencioso Administrativo del Distrito Federal) shall be composed of judges appointed for six years with tenure possibility after this term. The governor proposes and congress approves such appointments.

8

According to Article 7 of the Constitution of Durango, the Administrative Court of Durango (Tribunal de lo Contencioso Administrativo del Estado de Durango) shall be composed of judges appointed for six years with option to be selected for one more term. The governor proposes and congress approves such appointments.

9

According to Article 87 of the Constitution of the State of Mexico, the Administrative Court of the State of Mexico (Tribunal de lo Contencioso Administrativo del Estado de México) shall be composed of judges appointed for ten years. The governor proposes and congress approves such appointments.

10

According to Article 82 of the Constitution of Guanajuato, the Administrative Court of Guanajuato (Tribunal de lo Contencioso Administrativo del Estado de Guanajuato) shall be composed of judges appointed for seven years with option to be selected for one more term. The governor proposes and congress approves such appointments.

11

According to Article 118 of the Constitution of Guerrero, the Administrative Court of Guerrero (Tribunal de lo Contencioso Administrativo del Estado de Guerrero) shall be composed of judges appointed for six years with tenure possibility after this term. The governor proposes and congress approves such appointments.

12

According to Article 97 of the Constitution of Hidalgo, the Administrative Court of Hidalgo (Tribunal Fiscal Administrativo para el Estado de Hidalgo) shall be composed of judges appointed for six years with tenure possibility after this term. The governor proposes and congress approves such appointments.

13

According to Article 65 of the Constitution of Jalisco, the Administrative Court of Jalisco (Tribunal de lo Administrativo del Poder Judicial del Estado de Jalisco) shall be composed of judges appointed for four years with tenure possibility after this term. The governor proposes and congress approves such appointments.

14

According to Article 95 of the Constitution of Michoacán, the Administrative Court of Michoacan (Tribunal de Justicia Administrativa de Michoacán de Ocampo) shall be composed of judges appointed by the Congress for five years. Two more terms are permitted.

15

According to Article 109BIS of the Constitution of Morelos, the Administrative Court of Morelos (Tribunal de lo Contencioso Administrativo del Poder Judicial del Estado de Morelos) shall be composed of judges appointed for six years with tenure possibility after this term. Congress appoints judges.

16

According to Article 47.XXXVI of the Constitution of Nayarit, the Administrative Court of Nayarit (Tribunal de Justicia Administrativa del Estado de Nayarit) shall be composed of judges appointed for six years with option to be selected for one more term. The governor proposes and congress approves such appointments.

17

According to Article 63.XLV of the Constitution of Nuevo León, the Administrative Court of Nuevo León (Tribunal de lo Contencioso Administrativo del Estado de Nuevo León) shall be composed of judges appointed for ten years with option to be selected for one more term. The governor proposes and congress approves such appointments.

18

According to Article 1 of the Law of Administrative Justice in Oaxaca (2005), the Administrative Court of Oaxaca (Tribunal de lo Contencioso Administrativo del Estado de Oaxaca) shall be composed of judges appointed for eight years with option of being selected for one more term. The governor proposes and congress approves such appointments.

19

According to Articles 72 and 73 of the Constitution of Querétaro, the Administrative Court of Querétaro (Tribunal de lo Contencioso Administrativo del Estado de Querétaro) shall be composed of judges appointed for four years with option of two more terms. Congress appoints judges.

20

According to Article 106 of the Constitution of Quintana Roo, the Administrative Court of Quintana Roo (Sala Constitucional y Administrativa del Poder Judicial de Quintana Roo) shall be composed of judges appointed for six years with one more tem permitted. The governor proposes and congress approves such appointments.

21

According to Article 124 of the Constitution of San Luis Potosí and Article 9 of the Administrative Justice Statute of San Luis Potosí, the Administrative Court of San Luis Potosí (Tribunal de lo Contencioso Administrativo del Estado de San Luis Potosí) shall be composed of judges appointed for six years with tenure possibility after this term. The governor proposes and congress approves such appointments.

22

According to Article 129 Bis of the Constitution of Sinaloa, the Administrative Court of Sinaloa (Tribunal de lo Contencioso Administrativo del Estado de Sinaloa) shall be composed of judges appointed for six years with one more term permitted. The governor proposes and congress approves such appointments.

23

According to Articles 64 XLIII Bis of the Constitution of Sonora and 3 of the Organic Statute of the Sonora Administrative Court, the Administrative Court of Sonora (Tribunal de lo Contencioso Administrativo del Estado de Sonora) shall be composed of judges appointed for six years with one more term permitted. The governor proposes and congress approves such appointments.

24

According to Article 36.XL of the Constitution of Tabasco, the Administrative Court of Tabasco (Tribunal de lo Contencioso Administrativo del Estado de Tabasco) shall be composed of judges appointed for six years with two more terms permitted. The governor proposes and congress approves such appointments.

25

According to Article 92 of the Constitution of Tamaulipas, the Administrative Court of Tamaulipas (Tribunal Fiscal del Estado de Tamaulipas) shall be composed of judges appointed for six years with tenure possibility after this term. Congress appoints judges.

26

According to Article 82 of the Constitution of Tlaxcala, the Administrative Court of Tlaxcala (Sala Electoral Administrativa del Tribunal Superior de Tlaxcala) shall be composed of judges appointed for six years. Congress appoints judges.

27

According to Article 38 C of the Law of Administrative Justice of the State of the State of Veracruz Ignacio de Llave, the Administrative Court of Veracruz (Tribunal de lo Contencioso Administrativo del Estado de Veracruz) shall be composed of judges appointed for ten years. The governor proposes and congress approves such appointments.

28

According to Article 1 of the Organic Law of the Contentious Administrative Tribunal of the State of Yucatan (1985), the Administrative Court of Yucatan (Tribunal de lo Contencioso Administrativo del Estado de Yucatán) shall be composed of a judge appointed for four years with tenure possibility after this term. The governor proposes and congress approves such appointment.

29

According to Article 112 of the Constitution of Zacatecas, the Administrative Court of Zacatecas (Tribunal de lo Contencioso Administrativo del Estado y Municipios de Zacatecas) shall be composed of a judge appointed for six years with tenure possibility after this term. The judiciary proposes and congress approves such appointment.

According to Table 1, out of a total of 29 state courts in existence in 2009, 62% were part of the executive branch and 38% were part of the judiciary.⁷⁷ Currently, Chihuahua also created an administrative court and Oaxaca and Yucatan incorporated their administrative courts into the judicial branch. Therefore, currently 53% are autonomous tribunals and 47% are part of the judicial branch.

2

Specialized Procedures of Administrative Trials in Mexico

Mexico’s administrative mechanisms of dispute resolution between the state and individuals have their own procedures and require specialized judges. This section outlines the process by which administrative courts operate. Citizens initiate the operations of the specialized administrative court when they decide to challenge an agency’s action through a nullity trial.

The main function of administrative tribunals is to determine whether the administrative agency followed the rules of decision-making as established in statutes. Judges deal primarily with procedural requirements. To perform their functions, they use specific procedures called nullity trials.

After a proceeding, which includes a hearing and the opportunity to present evidence, the tribunal offers one of three decisions:

1) Judge dismisses the case. In this case the judge does not analyze the challenged agency’s actions. This most commonly occurs in Mexico because plaintiffs violate procedural rules, such as standing rules and ripeness.

2) Judge upholds the agency’s initial decision and declares its lawfulness (pro-government decisions). After analyzing the formal requirements to sue, the judge analyzes the merits of the case. When the judge verifies that the defendant complied with administrative rules, the court upholds the agency’s action. In these cases, administrative judges must analyze every argument the plaintiff made in challenging the government’s decision.

3) Judge invalidates governments’ initial decision and declares it unlawful (against/ anti-government decisions). When a judge verifies that the defendant did not comply with administrative rules, the court strikes down the government’s action. The court may make a ruling of partial or total unlawfulness. In the first case, it orders the reversal of the agency’s act and remands it to the agency for further consideration. In the second case, the court renders a judgment and directs the agency to provide remedy (de novo review).

IV

Data, empirical model and hypotheses

1

Data

We used two datasets to analyze the differences in the rates of dismissals and pro-government decisions between courts within the judicial branch and courts within the executive branch. The first dataset includes the characteristics of each court: its year of creation, the existence of guarantees of tenure and protection of salaries, and the governor’s role in the judge appointment process. The state constitutions and the courts’ web pages provide this information.

The second dataset describes the courts’ decisions. A large-scale survey of administrative court decisions conducted by a group of Mexican researchers in the “Diagnóstico del Funcionamiento del Sistema de Impartición de Justicia en Materia Administrativa a Nivel Nacional”⁷⁸ provides this dataset of 5,400 cases decided by 23 administrative courts (22 local administrative courts and the Federal District).⁷⁹ The researchers sought to analyze the performance of administrative courts in Mexico at a state level, and therefore collected court budgets, judge’s curricula, internal organization and case specifics, such as dates, subjects, parties, quantities, decisions and appeals. The cases analyzed concluded in the years 2006, 2007 and 2008 (some courts were yet not created and therefore had no cases in 2006).⁸⁰

We disregarded 380 cases from the second dataset that did not include the variables analyzed here. We recoded the final decisions for the 5,020 remaining cases, simplifying the categories and recoding the types of cases in each of the files. Table 3 summarizes the descriptive statistics of the outcomes of this dataset.

State	Dismissals	Pro-government decisions	Against/Anti-government decisions
Aguascalientes	24%	1%	75%
Baja California	18%	26%	56%
Baja California Sur	51%	21%	28%
Campeche	38%	14%	48%
Chiapas	92%	3%	5%
Colima	4%	0%	96%
Federal District	9%	12%	80%
Durango	23%	7%	70%
State of Mexico	16%	18%	66%
Guanajuato	31%	8%	61%
Hidalgo	7%	0%	93%
Michoacan	84%	0%	16%
Morelos	56%	10%	34%
Nayarit	37%	0%	63%
Nuevo Leon	18%	27%	56%
Oaxaca	75%	2%	24%
Querétaro	29%	6%	65%
Sinaloa	26%	2%	72%
Tabasco	35%	19%	46%
Tamaulipas	64%	15%	20%
Tlaxcala	42%	15%	43%
Yucatán	43%	3%	54%
Zacatecas	21%	1%	78%
Overall Total	31%	10%	59%

2

Variables

We developed two models to predict the outcomes in the two datasets. Both models used the court’s branch, judiciary (coded 0) or executive (coded 1), as the main independent variable. Since empirical analysis of judges’ performance across different court designs raises many important issues regarding the homogenization of contexts and decisions of the compared courts, we added several control variables. The 10 variables across the models are as follows:

• 1)

  Executive nomination: The 23 courts use five types of appointments to designate judges, as found from a review of the local constitutions and the statutes of each local administrative court. The judiciary, legislative and executive branches have varying levels of responsibility for proposing and approving [or confirming] judges. For the purposes of the research, we classified all of the procedures into two categories: the ones where the executive branch nominates judges (coded 0) and the ones in which it does not (coded 1).⁸¹

• 2)

  Judges’ tenure greater than appointer tenure: Appointments made for a term length greater than the appointer’s term length were coded as 1 and 0 otherwise.

• 3)

  Protection of salaries: This variable describes whether a state constitution explicitly prohibits reducing judges’ salaries. While the Supreme Court has also forbade the reduction of salaries in decisions that apply to administrative judges and we have no empirical evidence that an administrative judge has suffered an actual salary decrease, we believe the mere mention of the guarantee may have some effect on judges’ behavior. We coded the prohibition as 1 and its absence as 0.

• 4)

  Panels: This variable refers to the number of judges required to decide a case. We classified courts into two categories —those requiring one judge to make the decision (coded 1) and those requiring more than one judge (coded 0).

• 5)

  Type of plaintiff: We divided the cases into those brought by individuals (coded 0) and those brought by companies (coded 1). This division reflects the fact that companies may be able to spend more money on their complaints than individuals, and may thereby increase their chances of winning.

• 6)

  Type of case: The two main categories of cases are administrative law issues and tax issues. Administrative issues include licensing, traffic fines, permit reversals, labor cases between the government and its employees (including police departments), expropriations and state liability, among others. The tax issues category includes property taxes and water consumption taxes. We classified cases in traffic ticket cases and non-traffic ticket cases in order to capture the real effect of the courts’ design on the rest of the cases. We coded all traffic ticket cases 1 and 0 otherwise.

• 7)

  Age of the court: We decided to control for the age of the court because the experience level of the judges may influence outcomes.

• 8)

  HDI (2008): The Human Development Index is a United Nations index that controls specific state characteristics because it measures the general wellbeing of the state. We wanted to control for general wellbeing as an external factor influencing court outcomes. We used this index because it is the only one made for each state and it incorporates various measures of economic and social variables.

• 9)

  Year the trial ended: We controlled for specific changes over time in order to avoid omitted time-variable problems.

3

Hypotheses and Empirical Models

Our empirical models and hypotheses rely on the assumption that the in ternal organization of courts and judges’ incentive structures should reflect each court’s design.

Our first hypothesis is concerned with institutional design characteristics such as tenure, appointment procedures and the protection of judges’ salaries. We predict that the branch to which an administrative court belongs will affect these characteristics. As explained, recent literature on judicial behavior relates tenure, salary protection and the appointment of judges to the actual independence of the court, which is a fundamental quality of administrative courts in which one of the parties is the state itself.

This analysis will not seek to measure the actual independence of administrative judges in Mexico, but rather to measure which models have more guarantees of judicial independence.⁸² For the purpose of our analysis we analyzed if judges’ tenure was superior to their appointers’ tenure, the limitation of the executive’s participation in judges’ nominations and the protection of judges’ salaries, all of which are forms to guarantee judicial independence. We hypothesize that courts created within the judiciary will have more guarantees of independence for judges than those courts created within the executive branch. Below is our empirical model:

We acknowledge the limitations of a regression with only 30 variables (30 states and the Federal District); however, Mexico has only 30 administrative courts.

Our second hypothesis has to do with judges’ incentive structures as a distinct influence on a judge’s decisions. The influence of institutions⁸³ on judges’ behavior has long been acknowledged. Judicial behavior literature has been focused on the choices judges make as rational individuals.⁸⁴ Developed models include how ideology,⁸⁵ aggregation schemes,⁸⁶ supra-subordination interactions⁸⁷ and precedents⁸⁸ affect judges’ choices. We hypothesize that judges in executive branch courts would decide cases differently from judges that work in a judicial branch court (pro-government decisions or dismissals). Below our empirical model:⁸⁹

We ran regressions on dismissals and pro-government decisions of 5,020 administrative trials using as our independent variables the branch to which the court belongs, the judges’ tenure greater than appointer tenure, executive nomination and the protection of salaries.

In order to add extra controls to our analysis, we ran the regression only on important cases and on tax cases. We will define important cases as those that are not traffic ticket cases. The second control has to do with tax cases. We decided to run the regression on these cases because there might be a difference in decisions associated with economic issues.

V

Findings

Model: Analysis of the Relationship between Branch and Independence Guarantees for Judges

The first hypothesis was confirmed only in the cases of judges’ tenure being greater than appointer tenure and executive nomination. Indeed, those courts that were created as part of the judiciary had a higher probability of having provisions guaranteeing that judges’ tenure would be greater than their appointers’ tenure. Additionally, courts incorporated into the judicial branch guaranteed less intervention of the executive branch in judges’ nominations. The protection of salaries was not significant; therefore, we cannot attach any effect of the branch to which a court belongs to the existence of judges’ salary protection. The following table describes our findings:

Regressor	Dependent variables with control variables
	Judges’ tenure greater than appointer tenure	Executive nomination	Salary protection
Administrative courts within the Executive Coefficient Standard Error P>|t|	-.4192327^*(.1675025)0.019	.495001^***(.1646347)0.006	.1484777(.1812336)0.420
Year of creation Coefficient Standard Error P>|t|	.0009544(.0077686)0.903	.0052484(.0076356)0.498	-.0140957(.0084055)0.106
HDI Coefficient Standard Error P>|t|	-4.704147^*(2.203285)0.042	3.047897(2.165563)0.171	-5.101426(2.3839)0.042
R_Squared	0.3386	0.3152	0.2152
N	30

t statistics in parentheses

*

p < 0.05, **p < 0.01,

***

p < 0.001

As the table illustrates, when a court is within the executive branch, the probability of having judges’ tenure greater than appointer tenure is negative (less) compared to courts that belong to the judiciary. Scholars have linked judicial tenure to judicial independence by alleviating fears about future jobs and earnings. Without tenure, judges might think of the governments as possible employers in the future, which would threaten their impartiality. Along the same line, when a court is within the executive branch, the probability of having governors as nominators of judges is greater.

Regarding our second hypothesis, we ran an OLS regression with time fixed effects as a control for the years in which decisions were made. Table 5 describes the findings:

Regressor	Dependent variables with control variables
	Dismissals	Pro-government decisions
Administrative courts within the Executive Coefficient Standard Error P>|t|	.0265423(.0172636)0.124	.049347^***(.0185217)0.008
Judges’ tenure greater than appointer tenure Coefficient Standard Error P>|t|	.321257^***(.0402684)0.000	.3142995^***(.0432036)0.000
Executive nomination Coefficient Standard Error P>|t|	.4224028^***(.0413954)0.000	.440221^***(.0444079)0.000
Salary protection Coefficient Standard Error P>|t|	-.0826914^***(.013177)0.000	-.0771635^***(.0141397)0.000
Panel decision Coefficient Standard Error P>|t|	-.076566^***(.0157672)0.000	-.0196248(.0169157)0.246
Type of plaintiff Coefficient Standard Error P>|t|	.0228991(.0167275)0.171	.0531568^***(.0179466)0.003
Traffic ticket cases Coefficient Standard Error P>|t|	-.2488434^***(.0152331)0.000	-.3256305^***(.0163435)0.000
Age Coefficient Standard Error P>|t|	-.0042822^***(.0007103)0.000	-.0045647^***(.0007621)0.000
HDI Coefficient Standard Error P>|t|	-1.64775^***(.1750752)0.000	-.7715674(.1878541)0.000
R_Squared	0.1915	0.1789
N	5020

t statistics in parentheses

*

p < 0.05, ^**p < 0.01, ^***p < 0.001

Our results varied across the two outcomes and the four independent variables we were testing.

Regarding pro-government decisions we found that judges in courts within the executive branch support governments’ decisions more often. While the coefficient is not very large, the two-model comparison does show a distinction. The “Judges’ tenure greater than appointer tenure” variable was also significant in the analysis of pro-government decisions. Judges enjoying a lesser tenure than their appointer’s more often decide cases in favor of the government compared with judges enjoying greater tenure. Protection of salaries was also significant. However, it had the opposite effect of what we had postulated. Therefore, the hypothesis suggesting that judges who enjoy explicit constitutional salary protection more often decide against the government was not confirmed. Finally, the executive nomination variable was also significant. Those judges whose nomination was made by the government decide cases in favor of the government more often.

Regarding dismissals, we did not find a significant correlation with the branch to which the courts belong. Dismissals are in a sense pro-government in that the judges dismissing cases do not invalidate the agency’s action. However, judges may dismiss cases without notifying the government of the complaint against them. In these situations, executives will not appreciate these dismissals. Thus, publicity of judges’ decisions might affect the meaning of dismissals. However, dismissals may reflect lawyer’s level of knowledge—that is, they may not bring poor-quality cases to court when the judge is part of the judiciary. This point is beyond the scope of our analysis. However, the age of the court also affects dismissals, suggesting that the younger the court, the less experienced the lawyers and the poorer the suits. “Judges’ tenure greater than appointer tenure” is also significant and confirms our hypothesis. Judges without tenure will dismiss more cases than judges with tenure. The most obvious explanation for such finding is that dismissals constitute pro-government decisions. Judges hoping for future employability within the government make more dismissals. As in the case of pro-government decisions, those judges whose nomination was made by the government decide cases more often in favor of the government, again confirming the hypothesis. Finally, the case of salary protection is counterintuitive and our hypothesis was not confirmed.

As explained, we also ran the regression with only important cases and tax cases. The following table presents our findings:

Regressor	Dependent variables with control variables
	Important cases	Tax cases
	Pro-government decisions	Pro-government decisions
Administrative courts within the Executive Coefficient Standard Error P>|t|	.0235593(.0433357)0.587	.0118864(.0633406)0.851
Judges’ tenure greater than appointer tenure Coefficient Standard Error P>|t|	.291044^*(.1274919)0.023	.471319^*(.2271253)0.038
Executive nomination Coefficient Standard Error P>|t|	.2422731(.1295834)0.062	.5288321^*(.2320203)0.023
Salary protection Coefficient Standard Error P>|t|	.1079939^***(0334019)0.001	.2047336^***(.0511714)0.000
Panel decision Coefficient Standard Error P>|t|	-.0719693(.0376943)0.056	-.0742327(.0547283)0.175
Type of plaintiff Coefficient Standard Error P>|t|	.0521001(.0320534)0.104	.0091286(.0396772)0.818
Age Coefficient Standard Error P>|t|	-.0047844^***(.001565)0.002	-.0021482(.002073)0.300
HDI Coefficient Standard Error P>|t|	-2.262612^***(.4075289)0.000	-.751942(.654552)0.251
R_Squared	0.0717	0.0529
N	1342	690

t statistics in parentheses

*

p < 0.05, ^**p < 0.01,

***

p < 0.001

The branch to which the court belongs was not significant in any of the cases. The variable “Judges’ tenure greater than appointer tenure” was also significant in the analyses of important and tax cases. Judges enjoying a lesser tenure than their appointer’s more often decide cases in favor of the government compared with judges enjoying greater tenure. The executive nomination variable was only significant in tax cases. Those judges whose nomination was made by the government decided cases more often in favor of the government. Protection of salaries was also significant. Judges not enjoying explicit constitutional protection for their salaries decide in favor of the government more often.

VI

Conclusions

Administrative justice is an inexorable companion of public administration based on the rule of law in democratic governments and implies the existence of legal remedies against decisions of administrative authorities.⁹⁰

In this paper we described the different models of administrative adjudication born within the French tradition of administrative law and within the judicial review doctrine of administrative law in order to accommodate the Mexican administrative system of justice in the spectrum of such models.

Institutional design, as well as procedural and substantive norms ruling administrative courts, has changed over time and across countries. Mexico is no exception and its system contains a number of different institutional designs. We identified administrative courts within the judiciary and administrative courts within the executive as a crucial distinction.

Our hypotheses focused on the impact of the distinction between executive and legislative branches, first on the way legislators design independence guarantees for judges and second on the way judges decide cases. The main purpose was to contribute to the discussion on institutional design of administrative courts with data.

We showed that the choice of creating an administrative court within the judiciary or within the executive branch has consequences both in the institutional arrangement of issues concerning the theoretical independence of judges and in the specific decision of cases. Regarding institutional design, we found out that the two variables that the executive branch affects were judges’ tenure greater than appointer tenure and executive nominations. When a congress decides to create a court within the judiciary, the probability of guaranteeing that judges’ tenure will be superior to the appointer’s tenure is greater than when creating the court as part of the executive branch. In the same line, when a congress decides to create a court within the judiciary, the probability of guaranteeing that the executive will not participate in the nomination of the judge is greater.

After analyzing the institutional design we studied the influence of the same variables in court decisions. To carry out this analysis we examined the whole universe of cases first, only the important cases second and only tax cases third. Although both the branch to which a court belongs and the intervention of the executive branch in the nomination of judges were significantly correlated with decisions favoring the government, the analysis of important cases was not consistent with such findings. Therefore, the only variable that was consistent throughout the three different analyses made in this paper was judges’ tenure greater than appointer tenure. This, again, is not surprising, but in analyzing our results it is important to recognize that attributes of courts within the executive branch apart from judges’ tenure may not matter.

Judges within executive branch courts made more pro-government decisions than judges in judicial branch courts. However, when analyzing judges’ behavior in important cases and in tax cases, the significance disappeared; therefore, we cannot derive a strong argument regarding the impact of the branch to which a court belongs with the decision judges make in these courts.

With these findings we want to address some effects that might shape them and some issues that require further analysis. A recurrent problem of studies of judicial cases is the selection effect of judicial cases, which arises when plaintiffs recognize the likely biases of the judge who will decide the case. Plaintiffs may invest less in bringing complaints to trial in those states where administrative courts are within the executive branch, or even refrain from bringing cases at all. If plaintiffs do not sue, then the results of the existing cases have no selection effect. In any case, the selection problem would be more of a problem when the stakes are low than when the stakes are high.

The interdependence of the control variables may also affect our results. Some of the findings with the incorporation of the control variables were mixed and we have no reasonable theory to explain our results. The characteristics of the control variables drive this problem —most depend on the existence of another. For a future analysis, control variables should refer more directly to states’ characteristics and not to the courts’ characteristics.

Our findings were not surprising, but the use of real data makes this study important. Legislatures act more readily on information based on real data. As Part II of this paper discusses, administrative courts perform two main functions: the redress of individual disputes and improvement of government agencies by monitoring their performance. Independence guarantees for judges may be more helpful in the former case and less important in the latter. It would be interesting to find out whether legislators creating administrative courts within the executive branch were more concerned with providing the executive with an effective control over its agencies rather than creating mechanisms of redress. By contrast, legislators that have created administrative courts as part of the judiciary may be more concerned with providing citizens with redress mechanisms in which one of the essential characteristics is the independence of its judges. This account seems to provide a good explanation of our findings, but deeper knowledge of the motives for the creation of these courts is needed.

Finally, since pro-government decisions create many problems as proxies for judges’ independence, this paper also gives rise to the question on how we should empirically measure judges’ independence. This is an important question and implies the test of the effectiveness of theoretical variables such as tenure or protection of salaries as guarantees for judges.

The authors would like to thank Michael Asimow, Nuno Garoupa, Jacob Gersen, Tom Ginsburg, William Hubbard and the 2010 Stanford-Harvard Junior Faculty Forum participants for their comments and suggestions for the improvement of this paper; and Jose Manuel Toral for his assistance in the research.