Instituto Cultural Minerva

Institute of Brazilian Issues

The George Washington University

Author: Suely Dib de Sousa e Silva

Washington, DC

Introduction

The purpose of this paper is to analyze the purpose and evolution of certain rules and institutions that govern the civil service system. The main focus is on the link between these rules and the performance of the bureaucracy. In particular, the fact that several countries have opted to organize their bureaucracies with high proportions of career civil servants is examined. This fact seems paradoxical given that providing tenure to bureaucrats is widely seen as a cause of inefficient performance. There is a prevailing sense that job tenure has removed the incentives for high productivity and accountability. The bureaucratic public administration is perceived as not being able to guarantee fast, neither cheap, nor high quality services. On the contrary: Public bureaucracy as it is today, is seen as slow, expensive and not well oriented to serve the citizens. If this picture is true, why is job tenure for most civil servants maintained? Why isn’t this system changed?

In the next section, it will be shown that the current civil service model arose as a reaction to patronage and clientelism, which were prevalent in many countries and which have even lower results in terms of efficiency. Section 2 will trace the historical evolution of the civil service towards a system of meritocracy and stability.

It will be shown that this new system brought problems of its own. Employment tenure turned out to be a factor that mitigated the incentives for employees to put more effort into their jobs, since they could not be fired based on poor performance. In many countries (including, amongst others, Brazil and the United States of America), there were periodic attempts at reforming the civil service system throughout this century, in order to make the bureaucracy more efficient. However, these attempts have not been successful.

This paper intends to focus on these questions, arguing that one of the main reasons for the maintenance of this model is the competition between the Executive and Legislature to control the agencies.

Section 2 - From Patronage to Meritocracy. The Current System

According to Melo (1996), presidential control over bureaucracy can be divided in three historic phases. The first was patronage control over access to the public careers. Secondly, there was a professionalization of the bureaucracies, with stability, autonomy and insulation from the political process. This can be seen as equivalent to the process of construction of the "Weberian" bureaucracies. It is in this context of independent bureaucracies that the principal-agent problem first appears in the relationship among elected officials and the bureaucracy. The third phase, beginning in the Seventies, introduced standards instead of routines, with flexible and entrepreneurialists characteristics, in order to, among other things, demand more accountability from the bureaucracies, reducing their insulation.

Indeed, there has been major institutional change over the last 100 years, from a patronage system for federal employees to the current civil service system. When the bureaucratic public administration replaced patronage, it represented an immense innovation against corruption and nepotism. The main European countries adopted it by the end of the last century; by the United States of America at the beginning of this century; and by Brazil in 1936. Job tenure for federal civil servants is a characteristic of bureaucratic administrations. It was a way of protecting the employees and even the government against the negative effects of patronage. In this section first the evolution of the Civil Service System in the USA is examined before tracing its development in Brazil.

2.1. The Evolution of the Civil Service System in the USA

In the U.S.A., up until the early part of the twentieth century, most federal employees were hired on the basis of patronage. There was no job tenure and employees were routinely replaced after elections whenever their political benefactors were defeated. The shift away from patronage to a merit system began with the Pendleton Act of 1883, although the change was neither homogeneous nor fast.

The Pendleton Act assigned specific positions for inclusion into a merit system called the classified service. The civil service was classified into two components: classified (merit) and unclassified (patronage) employees (Johnson & Libecap, 1994). This system reduced the influence of the local party officials over federal workers. The change was gradual, however, taking over two decades for more than 50% of the total federal civilian labor force to be under the merit system.

The Pendleton Act, however, did not grant tenure to the employees, even though it did state that they could not be dismissed for refusing to perform partisan services. Nevertheless, the fact that employees could still be dismissed or declassified indicated that there remained room for further rules to avoid political interference. The first move in that direction came in 1897 when President McKinley issued an Executive Order which stated that dismissal from the classified services could only be made with written notification and the right to respond. President Roosevelt strengthened this Order in 1902. Subsequently, other acts in 1920 (Civil Service Retirement Act), 1923 and 1949 (Classification Acts) and 1978 (Civil Service Reorganization Act) reiterated the principle.

According to Johnson and Libecap (1994, 91-119), the adoption of a merit system was an attempt by federal politicians to obtain more effective control over the actions and the performance of the federal labor force. Also, pressure on politicians by interest groups demanding efficiency gave them an incentive to seek arrangements to deliver better public services to those constituencies.

The patronage system was completely scrapped by the Hatch Act of 1939 and 1940, which prohibited the direct involvement of federal employees in political campaigns.

In 1987 it was estimated there were approximately 5000 political appointees within the federal work force, amounting to less than 1% of total civilian employment. Nowadays, about 80% of federal civil servants have guaranteed stability by federal law. The federal civilian white-collar employees are hired on the basis of merit, promoted within the framework of civil service rules, and are not to be explicitly involved in political campaigns. Once beyond a probationary period, federal employees essentially have job tenure.

2.2. The Evolution of the Civil Service System in Brazil

In the history of job stability in the public sector of Brazil, the rules alternated several times. Sometimes, job tenure was guaranteed for all of the public servants, in other moments it was assured only for a few officials, or nobody.

The Federal Constitution of 1824, for the first time, clearly established the fundaments of a meritocratic system. In 1827, during the Imperial period, when Dom Pedro I was the Emperor of Brazil, some careers, such as justice officials and academics amongst others, had guaranteed tenure. The purpose of this system was to avoid the replacement of these servants each time a new government came to power. In 1868 another national decree increased the number of careers graced with tenure. Despite this decree a high percentage of employees continued to be removed whenever there were changes in government.

Finally in 1888, a further decree was passed establishing job tenure for all public servants. According to the 1888 Decree, civil servants could only be fired after a lengthy administrative process or as a direct consequence of serious misconduct. The next change in the relevant legislation came in 1896. The government then revoked all past decrees and proclaimed itself able to transfer or fire any civil servant indiscriminately.

The legislative pattern was to change in 1934 when for the first time in the history of Brazil, a Federal Constitution assured job tenure for public officials. The 1934 Constitution also established competitive examinations for new applications into some careers, assuring for all those approved in such exams the right to belong to the civil servant career. Still, the Constitution kept the possibility of appointing officials for the high careers, a tradition that is still partially maintained today. During the mandate of President Getulio Vargas, in the 1930s, the first concrete administrative reform occurred in Brazil. The main point was to introduce the bureaucratic model, as proposed by Max Weber, in order to avoid the expansion of patronage practice.

The Federal Constitution of 1946 maintained job stability for all public civil servants admitted by public examination. Officials who had been working for more than five years, even if not admitted by public examination, were also granted job tenure.

Incongruous to the trend, and solely as a means to punish adversaries of the military regime, in 1964, a federal act (AI-5) suspended job tenure for all officials for six months. The Federal Constitution of 1967 continued to guarantee job tenure for the federal public servants. The privilege of tenure was, however, revoked in 1968. It was only when the military government ended in 1985, those public servants regained job stability.

In 1988, the Congress voted a new Federal Constitution. It guaranteed job tenure for public civil servants that had been admitted by public examination, after a two-year probation period. The privilege was also given to employees who had been in their jobs for more than five years.

The Federal Congress just voted a reform for the public sector in Brazil, which will be explained in Section 6.

2.3. The trade-off between political insulation and efficiency

The merit system was adopted because it represented a better alternative to patronage in the control of corruption and nepotism. However, this new system brought problems of its own. The precepts of efficiency, in which this system was based on, turned out to be unreal (Pereira, 1996). When the small liberal state of the nineteenth century turned into the social economic state of the twentieth century, it could be observed that the bureaucratic administration was not guaranteeing fast, cheap, nor high quality services.

The existence of job tenure for federal employees is a characteristic of bureaucratic administrations. If on the one hand it may prevent the perverse effects of a patronage system, on the other hand it does imply in costs of its own. Its adoption prevents easy changes in the size of the work force to respond to necessities faced by the State. Also, the protection granted to federal employees may turn out to be a factor hindering employees from fully committing themselves to their jobs, as they are aware of the low likelihood of dismissal. The implementation of an effective system of incentives and punishments is also prevented. The merit system was very useful to finish with patronage, and served well while the Liberal State only supplied limited public services. However, its shortcomings became apparent as the role of the State grew. (Pereira, 1996).

It is true that tenured bureaucracies do have an advantage in challenging other interests. This is so for the obvious reason that, unlike elected officials, it allows bureaucrats to be spared from the immediate threat of being removed from office by disgruntled constituencies. Furthermore, the formal structure and organizational goals of bureaucratic agencies may give them both the autonomy and the capacity to pursue courses of action that are adverse to any groups. Among students of public administration there remains the notion that effective administration requires protecting bureaucrats from the intervention of politicians, so they can perform their duties in a neutral, technical and professional manner. But the rules that were adopted to shield federal career employees from political manipulation, at the same time insulated the bureaucracy from more legitimate political control, contributing directly to voter complaints about responsiveness and productivity. The increase in the autonomy of the bureaucracy should not be confused with bureaucratic insulation from political pressure. In a democratic government, the high career bureaucracy is necessarily involved with the political process, as part of the process.

Section 3 - Principal-Agent model

The principal-agent model applies across a number of situations in which a principal contracts with an agent to perform some tasks on its behalf, i.e. when there is a delegation of power.

The model can be applied to situations such as voters and politicians, owners and workers of a company, regulatory agency and regulated firm, shareholders and managers, landowners and tenants, etc. In executing the task the agent chooses a course of action which will produce a set of consequences: an outcome. It is important to note that this outcome will affect the welfare of both principal and agent. Moreover, since the principal is not on a par with the agent, it is likely that situations will arise in which their goals significantly differ.

The goal of the shareholders of a firm, for example, is that the firm makes a large profit and the price of their shares increase in value. The goal of the manager of the firm, who is hired by the shareholders, may differ significantly from the goal of the shareholders. He/she may want to maximize his/her own income and/or his/her power. Generally, the objectives of shareholders in any given business will not match those of the managers, even though they will not be completely opposed to each other. Lacking the time (relatively high opportunity cost), knowledge, and skill to run the company directly the shareholders need hire and delegate managers to do this for them. The fact that their goals differ and that monitoring is costly means that a standard contract between the two may not be sufficient to make the relationship prosper. The principal-agent model predicts that in such a situation a special contract will arise that aligns the incentives and allows it to do so. In this example share options accomplish this task.

Key issues in a principal-agent problem are delegation of power, monitoring costs and asymmetric information. The agent has more information than the principal does about the details of individual tasks assigned, as well as the actions, abilities and preferences. The standard problem in a principal-agent model arises because monitoring is costly and incentive contracts may not insure compliance with the principal’s objectives. Agents will often take advantage of the high costs involved in monitoring characteristics and performances and in enforcing contracts. Awareness of the high costs involved in monitoring will in this case act as an incentive for agents to engage in shirking or opportunistic behavior. If the principal can monitor the effort of the agent at low cost, then the problem is less likely to arise. The existence of a contract, be it formal or informal, covering the specific tasks required should be sufficient in most cases. However, in many situations low cost monitoring is simply not possible. Standard monitoring solutions for the principal-agent problem are costly and hence inadequate as a solution to the problem of institutions with tight inflexible budgets as the politicians' agency.

Alternatives need be found in order to align the objectives of both principal and agent. In some cases the parties may produce a contract which satisfies all the necessary incentives of both sides. The existence of such a contract would make it unnecessary to monitor agents' efforts as closely. In order to achieve such an agreement it is usually necessary to create incentives that makes the agent act in a way which conforms to the principal’s interests. The owners in a firm, for example, can offer the workers some participation in the profits. Therefore, both sides now share similar interests, which although are not necessarily identical are nonetheless significantly closer than before the incentives.

The principal-agent model can explain why relationships amongst people are sometimes structured in ways that at first sight may appear inefficient. The way a relationship is set up may be a response to some form of asymmetric information or high monitoring cost. Had a solution not been found, then it is unlikely the two parties would choose to enter the relationship, with the result that a mutually beneficial opportunity would be missed. The principal-agent model thus provides a framework for analyzing many every day relationships where very particular forms of contracting evolve to solve the information problems and bring the parties’ interests into reasonable alignment. Below it will be argued that the persistent question of job tenure in the civil service, despite its perverse incentives, is an outcome of the principal-agent problem mentioned above.

Principal-agent problems are at the center of democratic life. At first instance they occur between the voters and the elected officials, where the voters are principals and politicians are agents. The people elect politicians, but the elected officials have to delegate the implementation of their programs to relevant agencies. A central problem in politics is to ensure that those who are given political power will faithfully use that power in the interests of the citizens.

Problems also occur between the elected officials and the bureaucracy to which the implementation of chosen policies is delegated. In this relationship, the elected officials act as principal, with bureaucracy as agent. It is on this last type of principal-agent problem that this paper focuses. More specifically, this paper looks at a principal-agent problem with two principals (Congress and the President) and one agent (bureaucracy).

The policy process and the implementation stages in public policies are full of such relations where the principal monitors the efforts of different agents in living up to the terms of agreement made between them.

It is not easy for politicians to control bureaucracy, nor is it easy for private firms once they grow past a certain size. It is however difficult to compare the situation that is being studied with private firms because of the complexity in defining and measuring the results achieved by the public bureaucracy. For this reason, monitoring the performance of the public bureaucracy is difficult. In some cases, new structures may be needed to mitigate principal-agents problem.

For Mace (1992) the structure and design of administrative agencies can be explained as the end result of the incorporation of mechanisms for controlling the principal-agent problem between the political actors delegating authority to administrative agencies and the bureaucrats within those agencies. Since legislators delegate legislative power to administrative agencies which they themselves create (although some agencies are created by executive decree), the chances of possible administrative-agency deviation from the original intent, by the manipulation of the structure and design of those agencies, can be reduced.

Elected officials are often concerned that administrative agencies will act in ways contrary to their interests. The question of political control of agencies is a straightforward principal-agent problem: elected officials who create administrative agencies must worry about future shirking by the bureaucrats within the agency (McCubbins, Noll, and Weingast, 1987; Mitnick). In this case shirking can be defined as the administrative agency taking any action that might harm the elected officials. This may include simply not doing their job correctly, and irritating the officials’ constituency, or taking actions that have a direct perverse effect on the officials, such as benefiting other politicians constituencies. As with other sorts of principals in complex agency relationships, politicians not only find it costly to monitor the behavior of the agencies they create and fund; they find it equally costly to inform themselves of any relevant issues and determine which outcome among many may be the best. As such, standard monitoring solutions to the principal-agent problem, which include political oversight in the form of hearings and budget reviews, may prove to be a costly and hence imperfect solution to the politicians' agency cost problem (McCubbins, Noll, and Weingast, 1987: 248-53). (Macey, 1992: 93-110).

In section 5 it will be argued that the prevalence of the high level of tenured civil servants arise partly as a solution to a principal-agent problem for the control of the bureaucracy by the Congress and the President. The tensions between the two principals competing for control over the bureaucracy has important implications for understanding the civil service system and the special principal-agent problems that exist within the federal government.

Section 4 – The Relationship between Congress, President and the Bureaucracy

In a democratic system the politicians delegate the implementation of public policy to the bureaucracy. This fact brings the question of how the elected politicians can control the bureaucracy, since the policies implemented by the bureaucracy sometimes do not coincide with the policies defined by the politicians (President and Congress).

The relationship between elected politicians and the bureaucracy contains the elements of a principal agent problem.

The U.S.A. Constitution is not clear about the authority over the federal bureaucracy. The civil service system operates in an environment where there is no clear structure of political control over federal bureaucracy. These conditions complicate the principal-agent problem. Both the Congress and the President influence the agencies in many different ways. Agencies are created as parts of the legislative or the executive branches of government and are formally responsible to the Congress or the President respectively, but in reality, the lines of authority are blurred. The President has control over appointments, budget and administrative procedures. However Congress also has many ways to impose controls on the agencies and potentially get the policies they want. Neither the President nor the Congress has well-defined property rights over the federal bureaucracy.

There is great controversy in the literature as to who, if anyone, controls the bureaucracy (Hammond and Knott, 1996), on how the Congress and the President are able to influence the bureaucracy and also why the bureaucracy, in many instances, may be able to act autonomously. The objectives of Congress and the President are usually not the same. Therefore, you have two principals. In this situation controlling the agent is even harder. Indeed, not only must a principal contend with the fact that the agent has its own objectives, but also that there is another principal trying to influence the agent in a possibly opposing manner.

4.1. Control by the Congress

Coexisting with the literatures on bureaucratic autonomy and congressional dominance has been a literature arguing that the Congress controls the bureaucracy. It does so by imposing controls on the agency after it takes an action or by influencing agency choices before action is taken. For Wilson (1989), Congress retains enormous influence over the bureaucracy. The American Constitution makes the President and Congress rivals for control of the American administrative system. In comparison, in Great Britain the prime minister and Parliament are not rivals because the House of Commons has little power to supervise, investigate, intervene in, or even obtain answers from the bureaucracy. The very factors that make agencies more susceptible to presidential influence also make them more susceptible to congressional control.

McCubbins, Noll and Weingast (1987, 1989) argue that administrative procedures are designed by politicians to confine agent discretion and to force their decisions to be consistent with the enabling legislation, constraining the evolution of future policies so as to maintain the interests of the enacting coalition.

For Macey (1992), since legislators delegate legislative power to administrative agencies that they themselves create, they can reduce the chances of possible administrative-agency deviation from the original intent by manipulating the structure and design of those agencies.

4.2. Control by the President

As Wilson explained (1989: 258-259) the notion that the President, as the chief executive officer of the United States, should administer and hold accountable to him executive branch agencies is a relatively recent one. Before 1921 bureaus and departments submitted their own bills directly to Congress. It was changes that occurred mainly during Roosevelt administration that made the President the effective head of the government.

The literature that argues that Presidents have the major impact on bureaucratic policy-making describes the President as having a wide variety of tools for controlling the bureaucracy, like the control over appointments; control of the agencies budget submission; and influence over administrative procedures (Hammond and Knott, 1996). About this last item, a growing body of literature documents how Presidents have created and managed this administrative presidency.

4.3. Autonomy of the bureaucracy

The literature on who controls the bureaucracy also presents the argument that the bureaucracy is autonomous, since it is difficult to regulate all of its actions in all situations. The bureaucracy can have its own goals, not coinciding with the interests of the principals. Hammond and Knott (1996) suggested that three broad classes of reasons have been advanced for why an agency might become an autonomous policy-making body:

indifference by the President and Congress toward the agency's policy domain. It is sometimes argued that the politicians indifference stems from the fact that agency policies are of little interest to their constituents, so trying to change these policies has too little electoral payoff to make the effort worthwhile;

information asymmetries between the agency and those who would control it. Two different kinds of asymmetries can be distinguished. There may be hidden information, which refers to the fact that agencies may control information or have technical expertise which the President and Congress lack. There may also be hidden actions, referring to the fact that agencies can take actions that are not easily observable. Each kind of information asymmetry enables an agency to pursue its own goals; and

conflicts among the multiple principals who have authority over the agency. The argument is that the agency can take advantage of a presidential-congressional impasse to pursue its own policies.

Competition over the bureaucracy has important implications for understanding the civil service system and the principal-agent problem. Within this competitive environment, the roles played by political appointees and career civil servants are quite different.

The federal civilian labor force (both is the U.S.A. and Brazil) has in its structure both political appointees and career civil servants. Both American and Brazilian systems rely on the use of political appointees to head most federal agencies. There are, basically, three categories of federal civilian employees: presidential appointees, who are responsible for setting administration policies, accountable to the President, who can also remove them, like agency head level; high career officials, who are also part of agency management and policy administration, but not responsible directly, and are protected by civil service tenure guarantees; and rank-and-file career employees, who perform agency functions and have day-to-day contact with constituents, having, also, tenure protection. Each of these groups works under different constraints with separate motives, and any discussion of the control of the bureaucracy must consider the implications of these distinctions (Johnson & Libecap 1994, 155).

The purpose of this section was to show that in fact the relationship between the bureaucracy, the President and Congress can be usefully examined within a principal-agent framework, or more precisely, a multiple principle agency framework. All the elements of such a relationship are present; (i) delegation of a task; (ii) asymmetric information; (iii) high monitoring costs; and non-aligned incentives. According to the standard solution to a principal-agent problem we would therefore expect there to arise some special way of organizing the relationship so as to overcome the basic problem. The next section will argue this is one of the reasons for the persistence of tenure in the civil service, despite its shortcomings.

The purpose of this section is not to determine whether Congress, the President or the agency is more influential. It is just to show that the relationships between them are characterized by some conditions that suggest that they will seek to establish special ways of organizing their interaction.

This section examines theories that argue that aspects of the civil service rules are a response to competition between Congress and the Executive over control of the bureaucracy. In particular this section will focus on theories that explain how tenure promotes control. The theories examined are those of Libecap and Johnson (1994) and Spiller and Urbiztondo (1994).

5.1. Libecap and Johnson (1994) argue that the change from a patronage to a merit system for the federal civil service was an attempt by federal politicians to obtain more effective control over the actions and performance of the federal labor force. The change was due to the fact that, in the late nineteenth century it became increasingly costly for both the President and members of Congress to effectively control and benefit from the patronage workers as had been the case until then. Shifting economic and political conditions at that time meant that the President and certain members of the Congress had incentives to replace the system for staffing federal positions. Whereas before the patronage system provided political benefits to politicians, it gradually came to be that these returns were lower than the costs the system imposed on them. There was considerable pressure from business groups on politicians to provide government services more effectively than was possible under patronage. Indeed, the merit system that begun in 1883 addressed the control problems that the President and members of Congress had faced with respect to local politicians in the allocation and use of federal workers.

Within this context, another question emerged: whether the President or the Congress was to have direct control over the new merit employees. Incomplete property rights in the legislation allowed both the President and the Congress to control certain aspect of the agencies. Thus both sides had to deal not only with the problem of getting the bureaucracy to act according to their interests, but also to avoid the bureaucracy being influenced by the other principal. Libecap and Johnson see the creation and gradual extension of the merit system and the addition of tenure protection as an attempt of the President and of the Congress to reduce each other’s influence on federal employees. Were it the case that only one side had control over the bureaucracy, that side would not want tenure, since that would reduce its influence. But given the existence of the other competing principal, each side is willing to give up some control as long as this also removes the other side’s control, so in order to mitigate the costs of political competition over the bureaucracy, the President and members of the Congress have agreed to civil service rules that inhibit their influence over career federal employees. Since neither side wanted to run the risk of allowing the other to control the bureaucracy, the merit system with tenure protection, is preferred. Although, the rules that had been adopted to shield federal career employees from political manipulation, however, have insulated the bureaucracy from more legitimate political control, contributing directly to voter complaints about responsiveness and productivity.

5.2. Spiller and Urbiztondo (1994) see the choice of using a higher proportion of political appointees rather than career civil servants as a result of competition between the President and Congress to control the bureaucracy.

Both of these principals have ways of influencing the bureaucracy (as showed in the previous section), and they usually also have different interests. Therefore, it is natural that each side will try to influence the bureaucracy to act in such a way that will serve its own interests. Because the other principal is also trying to influence the agent, each will try to make it harder for the other to exert influence. The idea presented in Spiller and Urbiztondo's paper is that the structure of the civil service in political appointees or career civil servants is one way of doing this.

The main idea behind their paper is that the President is a short-lived principal compared to Congress. In the United States this has been true since the beginning of the century, when the average time in office for congressmen and senators became significantly longer than that of the President. From this, Spiller and Urbiztondo conclude that Congress will prefer a stable bureaucracy because it will have a greater influence on the it, since a long term relationship can be established. The President, on the other hand, will be in office for a shorter period, and therefore, will not be able to set-up means of influencing the bureaucracy that depend on repeated interaction. The President will, therefore, generally prefer political appointees, which he has greater control over.

If, however, the President and the Congress have similar preferences, then there is less need for political appointees, since there is less competition to control the bureaucracy. The authors note that parliamentary systems are such that the executive power is formed from the legislative, so preferences are similar. This fits in with the observed fact that most parliamentary systems have less political appointees than presidential systems.

Another point is that, because the President prefers political appointees, in countries where there is separation of powers and the legislative bargaining power is small compared to the executive, we would expect there to be more political appointees. This helps to explain differences between countries like the United States and Mexico, where in the case of the latter the executive power is relatively stronger and there is a higher proportion of political appointees.

They also find evidence in favor of their argument from the fact that different cities in the United States have different forms of government and civil service structures. Cities where the mayor is elected indirectly are closer to a parliamentary system and tend to have less political appointees than cities where the mayor is elected directly by voters, which resemble Presidential systems.

The basic idea of the paper is, in conclusion, that the choice of civil service structure, as relates to the proportion of political appointees to career civil servants, is determined by the attempt of the President and the congress to control the bureaucracy, and avoid the other from doing the same. As they note "…the U.S. Congress was willing to give up the power to choose or replace public employees, among other reasons, because it was useful to restrict the Presidents power" (Spiller and Urbiztondo, pg. 467).

5.3. For both theories, the structure of the civil service system with job security and merit-based hiring and promotion had a functional reason for existing. Simply put, it was a direct result of each of the principals (President and the Congress) trying to reduce the influence of the other in the bureaucracy. Thus the fact that a system which has some apparent problems is adhered to over time, despite the clear knowledge of the problems it causes, can be understood. As was stated by Dixit:

"(m)any apparently inefficient outcomes can in fact be understood as consequences of constraints imposed by various transaction costs, or as credible attempts to cope with them. For example, the weakness of incentives in bureaucracies can be seen as a consequence of their being open to influence of multiple principals…"

Section 6 – Civil Service in Brazil

In Brazil there is considerable competition between the President and the Congress, and it is reasonable to assume that many of the same forces are at play here too. While the political institutions in Brazil are similar, in many ways, to those in the U.S.A, there are also many important differences. Thus those theories may or may not be relevant for Brazil.

By the Brazilian Federal Constitution, neither the President nor the Congress has well-defined property rights over the federal bureaucracy.

Article 48 of the Federal Constitution of Brazil stipulates that it is incumbent upon Congress to provide for all the matters within the jurisdiction of the Federal Union, and especially on: the pluriannual plan, budgetary directives and annual budget (II); the administrative and judiciary organization (IX); the creation, transformation and extinction of public offices, positions and functions (X); creation, structuring and responsibilities the Ministries and government administrative agencies (XI).

Article 49 determines that it is exclusively incumbent upon Congress to stay normative acts of the Executive Branch which exceed the regulamentary authority or the limits of legislative delegation (V); to establish for each fiscal year the wages of the President and Vice-President of the Federative Republic, and of Ministers of State (VIII); to each year examine the accounts rendered by the President of the Republic and evaluate the reports on the execution of government plans (IX); to supervise and control the acts of the Executive Branch, including those of the indirect administration (X); to choose two-thirds of the members of the Federal Audit Court (XIII).

Article 52 establishes the competencies of the Federal Senate, which includes the giving of its prior approval on the selection of: the President and Directors of the Central Bank, the Justices of the Federal Audit Court that had been chosen by the President; holders of other offices as determined by law.

Article 70 determines that the control of the accounts, finances, budgets, operations and property of the Federative Republic of Brazil and of the direct and indirect administration entities as to lawfulness, legitimacy, economic application of subsidies and waiver of revenues shall be exercised by the Congress, by means of external control, with the assistance of the Federal Audit Court (Art. 71). The Court has also to examine the lawfulness of acts of hiring personal, as well as the approval of civil and military retirement and pension (III).

Article 84 establishes it is incumbent exclusively upon the President of the Republic: to appoint and dismiss the Ministers of State (I); to exercise, with the assistance of the Ministers of State, the higher management of the federal administration (II); to provide for the organization and operation of the federal administration, in accordance with the law (VI); to appoint, after approval by the Senate, the Justices of the Federal Supreme Court, the Attorney-General of the Republic, the President and the Directors of the Central Bank and other public officials, when required by law (XIV); to appoint the Justices of the Federal Audit Court (XV); to submit to Congress the proposals of the pluriannual plan, the budget and directives bill of law (XXIII); to create and extinguish federal government offices, in accordance with the law.

The understanding of Johnson and Libecap (1994) can also be applied in Brazil. They argue that the confusion over who controls the bureaucracy that exists in the American Constitution reduces the incentives of politicians to engage in long-term, meaningful reform. The absence of a clear principal who would benefit from reform of the bureaucracy seriously complicates any efforts to change the system.

The progress of modernization of the public administration in Brazil represents an attempt of substituting patronage for meritocracy. This process can be characterized as discontinuous. According to Bresser Pereira (1996), who is also the current Minister of Federal Administration and Reform of the State of Brazil, patronage had been a constant in the federal bureaucracy of Brazil. His work argues that the crises of the bureaucratic model in Brazil began in the military regime. It was not only because of the incapability to limit the patronage, that has always been present in the federal bureaucracy, but also because, instead of consolidating a professional bureaucracy, through the definition of careers, public examination for high officials etc, it was preferred as the easy way of recruiting high officers through state firms. This fact reduced the possibility of the creation of a strong and professional bureaucracy. The Federal Constitution of 1988 increased the problem. It represented a shift from one extreme point to another. The Constitution of 1988 stipulated very rigid rules for the bureaucracy. The consequences of the surviving of patronage and of the rigid system, sometimes perversely mixed, represented the high costs and low quality of the public administration in Brazil. The Minister of Administration of Brazil supports a shift to a modern management administration for the public sector, since the "weberian" bureaucratic model is no longer efficient.

Bresser Pereira defends this model of public administration management, whose implementation started in the 1980's in Great Britain, New Zealand and Australia, and in the 90's in the United States, when it started to be noticed because of the publishing of "Reinventing Government", and with the adoption of the National Performance Review, by the President Clinton administration. In Brazil the process begun with the approval, in 1995, of the Director plan of the Reform of the State (Plano Diretor da Reforma do Estado).

As mentioned in Section 2, a reform for the public sector in Brazil was just approved by the Congress, in which is contained the end of job tenure for all public civil servants. Public servants, in this new situation, may be dismissed when they are not considered efficient. Moreover, employees may be dismissed whenever the government's total wage expenditure surpasses 60% of the net revenue. The new rules are applicable not only to the Federal Government, but also to States and Municipalities.

According to this constitutional amendment approved by the Congress, job tenure is maintained only for officials in careers deemed as typical of the public service. As such, officials whose duties do not have correspondent career in the private sector - police, internal revenue officials, diplomats etc. - have guaranteed job tenure. Still, even such employees may be dismissed if they commit a serious administrative fault.

All remaining officials will be regulated by a regime that is being called the "flexible job tenure". Theoretically, public employees still enjoy stability. However, dismissals are allowed for inefficiency or for occasions when it is necessary to decrease public expenditure.

For rank-and-file career employees complete tenure guarantees make little political sense for the President and the Congress. More remote from policy determination and management, lower-level workers are less susceptible to (and less valuable for) political manipulation than are those at more senior policy-administration levels. Further, it is in the interests of politicians to have federal employees perform effectively in the delivery of services to constituents.

In the view of the World Bank , for most of the Latin American countries, it is essential to improve the incentive structures of civil service agencies. This includes improving the financial remuneration and other reward systems of public-sector workers and using "performance contracts", through which public-sector workers and agencies can be held accountable for the realization of goals. In some cases, new structures may be needed to mitigate principal-agent problems. It is important to ensure some congruence in the incentives and behaviors of bureaucrats and politicians. Reforms may include the creation of institutional checks and balances and the introduction of mechanisms to increase contestability.

For the current Minister of Administration of Brazil, Bresser Pereira, in an abstract form, it is possible to view the reform of the state through a principal-agent perspective, where the reform is a way of creating incentives and punishments for the preferences of the voters to be realized by the State. According to this model, in its most simplified form, the voters are the principals and the elected politicians their agents; these, in turn, are the principals of the bureaucracy and the civil servants. But most of the literature takes a rational choice perspective that assumes that politicians are only motivated by rent-seeking and by the desire for reelection. The fact that public interest, a third possible motivation, is excluded, reduces the explanatory power of these models. When the analysis limits the motivations of public administrators to rent-seeking and reelections, and ignores that of public interest, the behavior of a large number of administrators and the meaning of the reforms that inspire the "new public administration" - the Managerial Public Administration - become incomprehensible.

One view in opposition to this is that politicians do in fact pursue public interest, however they only do so when this does not harm their other motivations. That is, a politician has his/her own view of what is in the public interest, but is constrained from pursuing this by the need to remain in power. Since the interest of his/her constituency will generally not coincide with the public interest, a politician who pursues it anyway will tend to be removed in the next election. Thus politicians will only do so when their constituency will not be affected or will not notice. One would thus expect this to happen more frequently when the next election is far away than when it is near. Note also that the level of political participation of the population also affect how constrained politicians are by public opinion.

Section 7 – Conclusion

In this paper I have shown that the structure of the civil service system, in countries like Brazil and the US, based on job security and merit-based hiring and promotion, has a functional reason for existing. It was argued that this structure arose as a direct result of the President and the Congress trying to reduce the influence of the other in the bureaucracy. This was done by showing that the nature of the relationship between these parties and the bureaucracy had all of the conditions that characterize a principal-agent problem, that is, delegation, asymmetric information and high monitoring costs. By putting the relationship in terms of a principal-agent model it became possible to understand why an apparently inefficient way of structuring the civil service was maintained over time. The general principal-agent model predicts that in order to overcome the information problems that are present in a relationship, the parties will try to find a way to structure the contract between them so as to align their interests. In the case of the bureaucracy, the President and Congress, the contact that was arrived at was one based on high proportions of tenured workers.

The fact that this may be a very inefficient way of overcoming the problem does not contradict the hypothesis. After all, society is full of examples of objectives that are consistently pursued through apparently inefficient means, such as the U.S. sugar program which uses a costly way of redistributing income to some special interest groups when a direct transfer would be much cheaper. In the case of the sugar program, it turns out that a system of tariffs and quotas are probably the only way to do the redistribution in a politically acceptable way. As for the case of the bureaucracy and job tenure it is argued here that the inefficiency has been purposefully created and maintained, for it plays an important role. This idea is well expressed in the following quote by Williamson :

"…political property rights are especially insecure in democratic regimes where programs that are put in place by one generation of politicians are subsequently subject to reversal when incumbent politicians are voted out of office. If efficiently designed bureaus and programs are more responsive but easier to reverse than are those which are encumbered, then politicians face an intertemporal tradeoff: inefficiency (nonresponsiveness; inertia) that has been crafted into the design of bureaus is the price of assuring that original purposes are not quickly reversed. To be sure, the originators would prefer to have programs that are both efficient and responsive now and are resistant to redirection by successors who would use them for other purposes. Not having that choice, however, inefficiency – both now and later – is intentionally introduced as the means by which to achieve program persistence."

In the case of Brazil it is very likely that job tenure has been maintained, at least in part, for the reasons explained above. Future research on this subject could study if the structure arose as a direct result of the President and the Congress trying to reduce the influence of each other in the bureaucracy. To do this it would be necessary to examine, in greater detail than this paper has done, the institutional rules that govern the relationship between the parties.

Although the political institutions both in Brazil and in the U.S.A. are quite similar, there are also important differences. Future research could study if in Brazil the President has more influence over the federal bureaucracy, as seems to be the case, given the current institutional rules. If that hypothesis is correct, the President may be less concerned about the influence of the Congress over the bureaucracy, and maybe that is an explanation for the administrative reform that just occurred in Brazil, with the partial end of job stability for the federal employees.

Barbosa, Lívia. 1996. "Meritocracia à Brasileira: o que é desempenho no Brasil?" Ano 47, V 120, N 3, Revista do Serviço Público 59-102.

Burki, Shahid Javed and Guillermo E. Perry. 1997. The long march: a reform agenda for Latin America and the Caribbean in the next decade. The World Bank. Washington D.C.

Dixit, Avinash 1996, The Making of Economic Policy: A Transaction-Cost Politics Perspective, Cambridge Mass: MIT Press.

Foss, Pal. 1995. Economic approaches to organizations and institutions: an introduction. Brookfield: Dartmouth Publishing Company

Hammond, Thomas H., and Jack H. Knott. 1996. "Who Controls the Bureacracy?: Presidential Power, Congressional Dominance, Legal Constraints, and Bureaucratic Autonomy in a model of Multi-Institutional Policy-Making," 12 Journal of Law, Economics, & Organization 119-166.

Johnson, R. N. and Gary D. Libecap. 1994. The Federal Civil Service System and the Problem of Bureaucracy. Chicago. The University of Chicago Press.

Johnson, R. N. and Gary D. Libecap. 1994. "Patronage to Merit and Control of the Federal Government Labor Force," 31 Explorations in Economic History 91-119.

Macey, Jonathan R. 1992. "Organizational Design and Political Control of Administrative Agencies," 8 Journal of Law, Economics, & Organization 93-110.

McCubbins, Mathew D., Roger G. Noll, and Barry R. Weingast. 1987. "Administrative Procedures as Instruments of Political Control," 3 Journal of Law, Economics, & Organization 243-77.

McCubbins, Mathew D., Roger G. Noll, and Barry R. Weingast. 1989. "Structure and process, politics and policy: Administrative arrangements and the political control of agencies," Virginia Law Review 75, no. 2, 432-82

Melo, M. André. 1996. "Governance e reforma do Estado: o paradigma agente x principal," Ano 47, V 120, N 1, Revista do Serviço Público 67-81.

Mucciaroni, Gary. 1995. Reversals of Fortune: Public Policy and Private Interests. Washington, DC. The Brookings Institution.

Nassuno, Marianne. 1997. "Organização dos usuários, participação na gestão e controle das organizações sociais," Ano 48, N 1, Revista do Serviço Público 27-41.

Pereira, L. C. Bresser. 1997. A Reforma do Estado nos Anos 90: Lógica e Mecanismos de Controle. Brasília. Ministério da Administração Federal e Reforma do Estado. Cadernos MARE da reforma do estado.

Pereira, L. C. Bresser. 1996. "Da administração pública burocrática à gerencial," Ano 47, V 120, N 1, Revista do Serviço Público 7-39.

Spiller, Pablo T. and Santiago Urbiztondo. 1994. "Political appointees vs. career civil servants: A multiple principals theory of political bureaucracies," 10 European Journal of political Economy 465-497.

Spiller, Pablo T. and John Ferejohn. "The Economics and Politics of Administrative Law and Procedures: An Introduction," V 8, N 1 Journal of Law, Economics, & Organization 1-7.

Williamson, O.E., 1996. "The Politics and Economics of Redistribution and Inefficiency". In The Mechanisms of Gvernance, Oxford, UK: Oxford University Press