The State in a Market Economy
In framing a government to be administered by men over men, the great difficulty lies in this: you must first enable the government to control the governed; and in the next place oblige it to control itself.
James Madison, Federalist 51
The transition from command to market economy entails not only economic but also political reforms. As the allocation of resources becomes decentralized, the inevitable question concerns the role of the state in the economy. This is why reform of the state appears on the agenda of all countries transforming their economic systems. The goal of this reform is to build institutions that would empower governments to do what they should while impeding them from doing what they should not. And this means that reform must be not only administrative, but also broadly political.
What one thinks about the proper role of the state depends on the model of the economy, as well as of the state itself. One question is "What should the state do?"; the other is "What kind of state will do all and only what it should?" Hence, this chapter begins with a brief recapitulation of debates about the proper role of the state in the economy, and only then addresses the issue of state reform. The next section briefly reviews the history of controversies concerning the relation between the state and the economy; the discussion includes consequences of the economic theory of incomplete markets and imperfect information for our understanding of this relation. The following section examines three classes of principal-agent relations: between govern-
An earlier version of this paper was originally prepared for the Seminar on "State Reform in Latin America and the Caribbean," organized by the Ministério da AdministraÇ o Federal e Reforma do Estado, Brasilia, May 16-17, 1996. I appreciate comments by Luiz Carlos Bresser Pereira, Zhiyuan Cui, Garriela Montinola, Susan Stokes, and Elisabeth Wood.
ment and private economic agents (regulation), between politicians and bureaucrats (oversight), and between citizens and government (accountability). I conclude that the quality of state performance depends on the institutional design of all these mechanisms, and that well-designed institutions will allow and induce governments to intervene in the economy in a way that is superior to what occurs in a noninterventionist state.
THE STATE AND THE ECONOMY: CONTRASTING PERSPECTIVES
To understand the rationale for state reform, we need to look back at the history of debates about the proper role of government in the economy. These debates have gone around in circles, with arguments about market failures being countered by claims about regulatory failures. As one reviews the history of these controversies, they appear almost as a boxing match, with the state and the market alternately on the ropes.
In a standard neoclassical model of the economy, there are markets for everything, now and for the future; everybody knows everything, and they know the same things; and there are no public goods, no externalities, no transaction costs, and no increasing returns. Since under these assumptions the market generates the first best allocation of resources, state intervention, in any form or fashion, is but a transfer of income; in turn, transfers of income, by causing rates of return to diverge from the competitive allocation, reduce incentives and misinform about opportunities. This conclusion follows directly from this model of the economy: since the state has nothing to contribute, anything it does is pernicious. The market wins round 1.
Yet the very fact that this model must be characterized at least in part negatively—by the absence of public goods, externalities, transaction costs, and monopolies—indicates an immediate problem. In the presence of these "failures," markets no longer allocate efficiently. This was the observation underlying the doctrine of state intervention enshrined in the 1959 Bad Godesberg Programme of the S.P.D.: "markets whenever possible, the state when necessary." The general prescription that emerged from this observation was that markets should be left alone to do what they do well—allocate private goods in those cases where the private rate of return does not deviate from the social rate—while the state should provide public goods, facilitate transactions, correct externalities, and regulate monopolies due to increasing returns. Round 2 goes to the state.
Neoliberals attacked this view in several ways: (1) by arguing that in the absence of transaction costs, the market can deal efficiently with market imperfections under a suitable reassignment of property rights (Coase, 1960); (2) by pointing out that the notion of market imperfections, including public goods, is unclear, and no theory specifies them ex ante (Stigler, 1975); and (3)
by remarking that even if the market fails to act efficiently, there is no guarantee that the state would do any better (Stigler, 1975; Wolf, 1979). Neoliberals maintain that prescriptions for state intervention are based on a naive model of an omniscient and benevolent state. They claim that the reason the state intervenes is the same as the reason for any other economic action: someone's private self-interest. Hence while the state is necessary for an economy to function, it can and does damage the economy. Here lies the fundamental dilemma of economic liberalism: "The economist recognizes that government can do some things better than the free market can do but he has no reason to believe that democratic processes will keep government from exceeding the limits of optimal intervention" (Posner, 1987:21). Indeed, analyses of the downfall of Keynesianism presented in the mid-1970s, whether from the left (Habermas, 1975), the center (Skidelsky, 1977), or the right (Stigler, 1975) were almost identical: the state became powerful, and for this reason it offered an attractive target for rent seeking by private interests (Buchanan et al., 1980; Tollison, 1982). As a result, the state was permeated by special interests, private logic prevailed, and the internal cohesion of state interventions disintegrated. Thus round 3 ends with the state on the ropes.
The goal of "constitutional" economics became to disable state interventions, particularly those that discriminate among private projects, respond to current economic conditions, or directly transfer incomes. Thus, for example, in Posner's (1987:28) view, "a government strong enough to maintain law and order, but too weak to launch and implement ambitious schemes of economic regulation or to engage in extensive redistribution, is probably the optimal government for economic growth." The neoliberal approach for limiting the state includes (1) reducing the size of public administration, (2) reducing the size of the public sector, (3) insulating the state from private pressures, (4) relying on rules rather than allowing discretionary decisions, and (5) delegating decisions subject to dynamic inconsistency to independent bodies that have no incentives to yield to political pressure. Public administration should be reduced because the state is ''bloated," and the productivity of public services is allegedly lower than that of the private sector. 1 The public sector should be privatized because governments are supposed to be more responsive to political pressures from public than from private firms. The state should be "insulated" from political pressures so it will not fall prey to rent seeking by private interests. Economic policy should be governed by rules, such as the gold rule or the balanced-budget amendment in the United States, that would eliminate discretion and thus overcome the suboptimality due to dynamic
inconsistencies (Kydland and Prescott, 1977). Finally, an alternative to rules is to delegate important policy decisions, particularly in the monetary realm, to institutions that are independent from political pressures and thus have no incentives to yield to dynamic inconsistencies (Cukierman, 1992).
Yet the view that markets are efficient even in the absence of "traditional" failures now appears dead, or at least moribund. The inefficiencies originating from the absence of some markets and from imperfect (more accurately, endogenous) information2 are both more profound and more devastating than the imperfections that blemish the neoclassical market. In a recent summary, Stiglitz (1994:13) put it bluntly: "The standard neo-classical model—the formal articulation of Adam Smith's invisible hand, the contention that market economies will ensure economic efficiency—provides little guidance for the choice of economic systems, since once information imperfections (and the fact that markets are incomplete) are brought into the analysis, as surely they must be, there is no presumption that markets are efficient." When some markets are missing, as they inevitably are, and information is endogenous, as it inescapably is, markets need not clear in equilibrium, prices do not uniquely summarize opportunity costs and can even misinform, externalities result from most individual actions, information is often asymmetric, market power is ubiquitous, and "rents'' abound. These are no longer "imperfections": there is nothing out there to be blemished, no unique "market," but many possible institutional arrangements, each with different consequences.
Moreover, some forms of state intervention are inevitable (Cui, 1992). The economy can function only if the state insures investors (limited liability), firms (bankruptcy), and depositors (two-tier banking system). But this kind of state involvement inevitably induces a soft budget constraint. The state cannot simultaneously insure private agents and not pay their claims, even if the latter result from negligence induced by the insurance ("moral hazard"). If markets are incomplete and information imperfect, moral hazard and adverse selection render first-best allocations unattainable.
Even the most ardent neoliberals admit that governments should provide law and order, safeguard property rights, enforce contracts, and defend from external threats. The economics of incomplete markets and imperfect information allows room for a much greater role for the state. The neoclassical complacency about markets is untenable: markets simply do not allocate efficiently. Even if governments have only the same information as the private economy, some government interventions would unambiguously increase welfare.
Thus the state has a positive role to play. Yet round 4 ends at best in a
draw. All we know now is that there are important things the state could do. But the consequences of the neoliberal punch still linger: will the state do what it should and not do what it should not?
Once we understand that markets are inevitably incomplete and economic agents have access to differing information, we discover that there is no such thing as "the" market, only differently organized economic systems. The very language of "the market" subject to interventions by "the state" is misleading. The problem we face is not that of "the market" versus "the state,'' but of specific institutions that would induce individual actors—whether economic agents, politicians, or bureaucrats—to behave in a collectively beneficial manner.
Suppose your car has been making strange noises. You go to a mechanic, explain the problem, leave the car, and wait for the result. One day later, the car is ready, the mechanic tells you that the shock absorbers needed changing and that it took 5 hours, and you pay and drive out of the garage, the noise gone. You chose the mechanic, and you can reward him by going back to him if you are satisfied with the outcome or punish him by going somewhere else if you are not. But there are all kinds of things the mechanic knows that you do not: whether he wanted to do the best job possible or as little as he could get away with, whether the car required a major repair or just a minor adjustment, whether he in fact did the job in 5 hours or only 1. You are the "principal," and he is the "agent." You hire him to act in your best interest, but you know that he has interests of his own. You can reward or punish him, but you will have to decide which to do with imperfect information, since he knows things you do not and does things you do not see. What can you do to induce him to work for you as well as he can?
When some markets are missing and when particular individuals have access to differing information, relations between classes of actors are those of principals and agents, linked by explicit or implicit contracts. Agents have some information that principals do not directly observe: they know their own motivation, they have a privileged knowledge of their capacities, and they may have a chance to observe some things principals cannot. They also undertake some actions that are, at least partly, hidden from the principal. The generic problem facing the principal is thus how to induce the agent to act in the principal's interest while meeting the "participation" constraint—providing the agent with income (or utility) above the next best opportunity—and the "incentive compatibility" constraint—allowing the agent to act in self-interest. You must pay the mechanic enough for him to want you to come back, and you must make him know that you will come back only if he has done a good job.
The "economy" is a network of multifarious and differentiated relations
between particular classes of principals and agents: managers and employees, owners and managers, and investors and entrepreneurs, but also citizens and politicians, and politicians and bureaucrats. The performance of firms, of governments, and of the economy as a whole depends on the design of institutions that regulate these relations. What matters is whether employees have incentives to maximize effort, whether managers have incentives to maximize profits, whether entrepreneurs have incentives to take only good risks, whether politicians have incentives to promote public welfare, and whether bureaucrats have incentives to implement goals set by politicians.
Institutions organize all these relations—those that are purely "economic," such as between employers and employees, owners and managers, or investors and entrepreneurs; those that are purely "political," such as between citizens and governments or politicians and bureaucrats, as well as those that structure state "intervention''; and those between governments and private economic agents. If the economy is to operate well, all these principal-agent relations must be structured appropriately.
At the cost of being schematic, let us consider only three classes of such relations: (1) between the government (politicians and bureaucrats) and private economic agents, (2) between elected politicians and appointed bureaucrats, and (3) between citizens and government.
The performance of an economic system depends on the design of all these relations. Private agents must benefit by behaving in the public interest and must suffer when they do not, and so must bureaucrats and politicians.
Government and Economic Agents (Regulation)
The role of the state is unique since the state sets the incentive structures for private agents by exercising its legally qualified coercive power: mandating and prohibiting some actions by law or changing relative prices through the fiscal system.
Let us take an example. Suppose I buy car-theft insurance. I drive to my destination and have a choice of parking a few blocks away from where I am going, in a place where the car is unlikely to be stolen, or parking just in front, in a place where the car is more likely to be stolen. Given that I am insured, I take the risk and park in the more dangerous place. Now the state comes in: it taxes me and uses the tax revenue to place a policeman in the dangerous place. As a result, car theft is less likely, the insurance company loses less money, and my premium goes down, more than compensating for the increased tax. The state is inextricably present in my relation with the insurer: although our relation is strictly "private," it is shaped by the state. The state permeates the entire economy; it is a constitutive factor of private relations. Problems of institutional
design cannot be avoided by throwing the state out of the economy. They must be confronted as such.
Yet government intervention in the economy—what is referred to as "regulation" in the United States—is not simple even on paper, not to speak of practice. The generic problem is the following (Baron, 1995). The regulated firm has information about some of its conditions, such as its cost of production or the demand for its outputs, which is superior to the information available to the government (the "regulator," understood in broad terms as elected politicians or appointed bureaucrats). Moreover, this firm undertakes some actions the regulator cannot observe directly, but can only either infer from outcomes or monitor at a cost. The regulator has a legal authority to set prices or rules. Once the regulation is issued, the firm decides whether to produce and how much. The regulator's problem is then to set up the best trade-off between the rents of the firm and the surplus of consumers.
Given hidden information and hidden actions, first-best regulation is not possible. The firm will always extract some rents. Thus regulation can be optimal only subject to the information available to the regulator; at most it is "second-best optimal regulation" (Baron, 1995:14).
Moreover, since any kind of state economic intervention has distributive consequences, different groups affected by regulation—firms, industries, employees, consumers, or public action lobbies—have incentives to seek regulation that benefits them and resist regulation that hurts them. The regulators can in turn benefit privately by supplying the interventions sought by the private actors. These private gains may consist not only of being (re)elected, but also of getting rich while in office or once out of office. As a result, regulation may induce clientilistic ties between the regulators and the regulated groups. To this extent, regulation is "endogenous"; that is, it is supplied in response to demand by the groups potentially affected by it.
As an example, consider a situation simplified from Laffont and Tirole (1994:Ch. 16): There are two periods. In period 1, a firm that is a natural monopoly has either high or low costs with some probabilities. A firm with high costs can invest to lower them. This investment is socially beneficial. A good intervention—one that maximizes consumer surplus—is then one in which the government subsidizes investment if the firm has high costs in period 1, and the government does not pay for investment otherwise. A bad intervention is one in which the government fails to subsidize investment by a firm with high costs, or subsidizes a firm with low costs and splits the rents with the firm.3
The institutional problem is then twofold: (1) how to enable the government to intervene in a good way and (2) how to induce it to act well. Only the first question is considered here; the second is discussed below.
To be able to intervene in a good way, the government must have some information about the costs faced by a given firm, and must be either legally capable of setting prices for the regulated firm (so that the cost of investment will be borne by consumers) or financially capable of subsidizing the firm out of tax revenues. Yet this is not sufficient. The reason is that even if the investment is subsidized, either by consumers or directly by the state, the firm will not undertake it unless there is a reasonable certainty that the profit due to the investment will not be confiscated once the costs are sunk. Suppose the firm expects that the government may change, and the new government will tax away its increased profits. Then the firm will not invest even if it receives the subsidy, and knowing that the firm will not invest, the optimal intervention for the period 1 government will be not to subsidize investment, even thought it is socially beneficial. 4 Hence, to be able to intervene well, the government must be committed to not confiscating the firm's profits during the second period.
The problem of commitment stems from the moral hazard of the principal. Even though the government would want the firm to invest, once it does so, the government will want to tax away its profits. Hence, agents cannot be sure that they will be rewarded if they behave well. This problem is present in many principal-agent relations, including purely private ones. But it is inherent in political relations. The ultimate source of political sovereignty—exercised by the democratic process—rests with "the people," in its eighteenth-century singular. And this implies that no government can fully precommit all future governments. An absolute guarantee of property rights is not possible. True, property rights can be, to various degrees, protected by the constitution. But constitutions cannot specify everything and must leave leeway for legislative discretion, as well as judicial interpretation. Moreover, even though the process may be difficult, constitutions can be changed: vide the nationalization of the Chilean copper industry by a constitutional amendment in 1970. Hence, property rights are inherently insecure.5
The result is a paradox of regulation: in those countries where it is easy to pass regulatory legislation, it is also easy to change it, so commitment is not credible; in those countries where it is difficult to change laws, so that commitments are credible, laws are also difficult to pass. Spiller (1995) shows the difficulty of making credible commitments in different institutional contexts. In different countries commitments are enforced by either (1) judicial review of decisions of regulatory bodies (prevalent in the United States, where 80 percent of the decisions of the Environmental Protection Agency are contested
in courts); (2) highly detailed legislation (e.g., Chile's 1980 electricity regulation); or (3) contracts between the government and the firm, enforceable under contract law (e.g., Bolivian COBEE since 1912). But here the paradox appears. If a political system generates majorities and party discipline, even detailed legislation can be overturned when legislative majorities change. In turn, where the political system generates a highly divided party system, such legislation is difficult to overturn once it has been adopted, but it is also difficult to adopt.
Yet even if the cost of this insecurity may be that resources are underutilized,6 commitment is not always optimal. For the danger is that a particular government will make a bad commitment, one that serves its own interests or those of its private allies, rather than those of the nation.
To return to our example, note that a commitment is socially beneficial only if the government intervened well during the first period, that is, if it subsidized a firm with high costs. If the government gave a subsidy to a firm that had low costs to begin with, the firm will have gained rents at the cost of the public, and if all future governments are committed to not increasing the tax on the firm, the new government will not be able to recoup these rents. As Laffont and Tirole (1994:620) observe, "The cost of commitment is that the government may identify with the firm and bind the nation to a bad outcome over the long run."
There are thus good and bad commitments.7 Imagine the following situation, due to Calmfors and Horn (1985). Early in its term, the government announces that if unions push wages up and create unemployment, it will not accommodate by expanding public employment. But come election time, the government will want to win and will employ. Thus, the initial announcement is not credible, unions push wages up, the government accommodates, and the outcome is suboptimal. The government must precommit itself, by rules or delegation, to not increasing public employment on the eve of the elections. This is a good commitment. But suppose the government does not precommit, unions push wages up, and the election time arrives. Now the government wants to expand public employment. But the unions anticipate that once reelected, the government will fire the new public employees. Hence, the government commits itself to not doing so, say, by passing a law of "immovability" of public employees. This is a bad commitment. If governments bind themselves in response to pressures from special interests, their commitments will not be optimal. Hence, a central institutional issue of state reform is how
to enable governments to make good commitments while preventing them from making bad ones.
I have used throughout the example of government regulation of a monopoly. Yet the same considerations apply to other forms of economic intervention. They also apply to "social" regulation, for example of health, safety, the environment, and employment (Baron, 1995). State intervention can be superior to nonintervention when governments have some information about private agents, when they have legal or fiscal instruments to regulate, and when the institutional framework allows credible commitments.
Yet none of these conditions guarantees that governments will intervene in the public interest. The very capacity of the state to intervene makes it an attractive target for influence by private interests, and the very ability to commit opens the possibility of collusion. Hence, there are reasons to expect that the quality of state intervention in the economy depends on the internal organization of the state—in particular, on the relations between politicians and bureaucrats—and on the design of the democratic institutions that determine whether citizens can control politicians.
Politicians and Bureaucrats (Oversight)
In a democracy, the authority of the state to regulate the life of the society coercively is derived from elections. Yet many of the functions of the state and all of the services the state supplies to citizens are delegated by the elected representatives to someone else, specifically to the public bureaucracy. Delegation is inevitable. As Kiewiet and McCubbins (1991:3) observe, "desired outcomes can be achieved only by delegating authority to others."
Delegation raises the standard principal-agent problems. Since it is impossible to write legislation that would fully specify the actions of agents under all contingencies, the executive and administrative agencies are left with a significant degree of discretion.8 But the objectives of bureaucrats need not be the same as those of citizens or of the elected politicians who represent them. Bureaucrats may want to maximize their autonomy or the security of their employment, render clientilistic favors to friends and allies, shirk their duties in office, aggrandize their budgets, (Niskanen, 1971), or simply get rich—all at the expense of the public. Again, they have private information concerning the benefits and costs of their actions, and they undertake actions that cannot be observed directly, but only inferred from outcomes or monitored at a cost. Hence delegation must inevitably give rise to agency costs.
Indeed, given the discretion bureaucrats must enjoy, the question is how to avoid a regime of "policy without law," as Lowi (1979:92) describes the U.S. political system.
Some of the agency problems inherent in managing the public bureaucracy are not different from those confronted by private organizations (on these, see Hammond and Miller, 1985; Holmstrom, 1982; Groves, 1985; Miller, 1992:128-158). Yet public bureaucracies differ in some important ways from private ones. One difference stems from the difficulty of setting criteria by which not only individual agents, but also teams of them, could be evaluated in the public sector. Private firms often perform multiple tasks, but to the extent to which they face a market constraint, their performance can be measured by financial criteria. Yet public bureaucracies are faced with multiple targets that are not simple to set9 and cannot be reduced to a single dimension. Another difference between private firms and public bureaucracies is that the latter are more frequently monopolies, which in turn implies that there are no comparative yardsticks by which their performance can be evaluated. As Tirole (1994) observes, the performance of management at Ford can be compared with that at General Motors, but this manner of measuring performance is not available when public agencies constitute monopolies.
Faced with these difficulties, public bureaucracies are prone to rely on conformity with rules rather than on incentives. This managerial style is called "police patrol" by McCubbins and Schwartz (1984). It consists of an a priori control of processes, as opposed to an a posteriori control of results. Tirole (1994:14) observes that "the central feature of a bureaucracy is that its members are not trusted to make use of information that affects members other than themselves, and that decisions are therefore based on rigid rules." The principal establishes rules, such as "work from 9 to 5,'' "do not use the telephone for private conversations," or "do not spend more than 20 minutes on a client," as well as reporting requirements. Agents are judged by whether they are observed to conform to such rules and by what they report. Needless to say, this is not a very effective mode of control: not only is it costly (the principals bear the costs of monitoring and of the time agents spend on filing reports), but it does not establish any direct relation between incentives and performance. Nevertheless, this is the way most public bureaucracies operate, and perhaps it is for good reason: if monitoring individual effort and eliciting private information are prohibitively expensive, reliance on rules may be the third-best.
There are a number of measures that can be taken to alleviate these agency problems.
Designing contracts. Even with the difficulties involved in monitoring individual efforts of team members, the principal can create incentives for the agents by (1) setting wage levels sufficiently high to attract agents of high quality, who have higher opportunity costs; (2) offering sufficient career advancement (which entails wage differentials); and (3) establishing monitoring systems that will make job loss likely in the case of bad performance (see Haggard, 1995).
Screening and selection. Recruitment to the public sector must be sensitive to costly signals, such as education, that indicate the potential performance of the agents.
Institutional checks. Kiewiet and McCubbins (1991:33) point out that "agents are often in a position to do more harm to the principal than to simply withdraw effort: embezzlement, insider trading, official corruption, abuse of authority, and coups d'etat are all testaments to this fact. Whenever an agent can take actions that might seriously jeopardize the principal's interests, the principal needs to thwart the agent's ability to pursue such courses of action unilaterally." The solution offered by Kiewiet and McCubbins is "institutional checks [that] require that when authority has been delegated to an agent, there is at least one other agent with the authority to veto or to block the actions of that agent."
Creating multiple principals or multiple agents with dissonant objectives. Tirole (1994) observes that most governments are divided in such a way that it is not the task of any particular position or agency to maximize general welfare, yet their interaction is supposed to generate this effect. One example is the division between "spending ministries," which are supposed to promote substantive goals, and the finance ministry, which is supposed to control spending.
Creating competition. Such competition can be either between state agencies and private counterparts (say, in delivering mail) or among state agencies in the exclusive sector. While there are some costs due to duplication of effort and perhaps economies of scale, competition facilitates the measurement of performance and, if combined with correct incentives, enhances performance.
Decentralization. This is a complex and controversial topic. Arguments in favor of decentralization typically rely on observing that local provision of public services enhances the accountability of the government by bringing it closer to the people it serves. Arguments against decentralization assert that it diminishes the capacity of the government to reduce regional income dispari-
ties, requires higher administrative capacity (Haggard, 1995), and can induce a soft budget constraint in which the less efficient jurisdictions will be more highly subsidized by the central government.10 Moreover, as Prud'homme (1995:204) observes, "the decentralization of taxes and expenditures works against the decentralization of activities and is likely to lead to a concentration of growth in a few urban locations."
Finally, public bureaucracies are different from private firms in one fundamental way that opens the possibility of more effective monitoring. To explain what is entailed, we need to step back.
A private firm has owners and a bureaucracy, which is an agent of the owners. It delivers products to the public, which does or does not buy the goods and services offered by the firm at a price. By revealing its demand for the firm's products, the public sanctions the firm; it generates profit or losses for the owners. Hence the principals, the owners, get free information from the public about the performance of the firm. The owners read the "bottom line," perhaps compare it with the performance of similar firms, and reward or punish the bureaucracy.
We have already seen one contrast between a private and a public agency—that the latter does not have a single "bottom line." Another contrast, to be discussed below, is that politicians may or may not want to sanction the bureaucracy. But most important, citizens' evaluations of the performance of public agencies do not have any direct sanctioning power. State services are produced and delivered by a bureaucracy, whose members are appointed by politicians. Citizens' control over the bureaucracy can be only indirect, since democratic institutions contain no mechanisms that would allow citizens to sanction directly the legal actions of bureaucrats. Citizens can at most consider the performance of the bureaucracy when they sanction elected politicians. As Dunn and Uhr (1993:2) observe, we do not even seem to know how to think about principal-agent relations involved in controlling bureaucrats: "It is by no means clear what place executive officials are meant to play as representatives of the people. Are they agents of the government or of the people?" Hence, while the state bureaucracy is supposed to deliver services to citizens, it is accountable to politicians (or to other bodies appointed by politicians, such as courts or administrative oversight agencies).
Yet precisely because the state bureaucracy delivers services to citizens, they are best informed about its performance. Moreover, if politicians care about citizens' welfare, the citizens have the same interests as the politicians, who are the principals, rather than the bureaucrats, who are the agents. The
principals can thus rely on the information provided by the affected parties; this is, in the terminology of McCubbins and Schwartz (1984), "fire alarm" oversight. This form of oversight has two advantages: (1) it allows the principals to gather information at a lower cost than that of "police patrol" oversight, and (2) it provides better information, particularly about the most egregious violations by the agents. Even though the legal authority rests with the elected politicians, fire alarm oversight is a mechanism for the accountability of the bureaucracy to the citizens.
Fire alarm oversight requires institutional arrangements that facilitate citizens' monitoring of the bureaucracy, transmission of information, and sanctioning of violations. As Haggard (1995:41-42) argues, "The ultimate check on government must come through institutionalized forms of participation. This may either be 'corporatist,' such as building in NGO [nongovernmental organization] participation in areas in which they have expertise, or 'legislative,' such as adopting forms of local governance in which citizen participation is maximized."
To summarize the conclusions of the two previous sections, government intervention can be effective if the regulatory institutions are well designed, and politicians can better control bureaucrats if they solicit the cooperation of citizens. But the question that still remains open is whether politicians will want to intervene well and to control the bureaucracy.
Citizens and Governments (Accountability)
The problem of citizens is to induce politicians to enhance the citizens' welfare rather than to pursue their own objectives in collusion with the bureaucracy or with private interests.
In many political systems, including democratic ones, bureaucracies appear to have autonomy from any control, completely insulated from public scrutiny. Moe (1990) offers a suggestive explanation for this pattern. Note first that in a democracy, bureaucrats can never be certain which political forces will control the government in the future, and they have reason to fear that a future government will be less favorable to their interests than the current one. Hence to protect themselves from the moral hazard of the principal—the possibility that their good behavior will not be rewarded by a future government—they seek to free themselves from any political control. In turn, the current government may fear that if it loses, the political forces that come into office will want to use the bureaucracy for their own advantage. Hence when the current government fears losing office, it has incentives to insulate the bureaucracy from political control, even at the cost of sacrificing its own influence over the bureaucracy. As a result, politicians and bureaucrats collude to make the bureaucracy autonomous, which implies that the bureaucracy will
not be well designed to further social objectives and that bureaucrats will have no incentives to promote them.
Moreover, the principal-agent relation between citizens and elected politicians is a very special one, without parallel in the private world. Because sovereignty rests with citizens, they are the principals with regard to politicians they elect. But because the state is a centralized and coercive mechanism, it is the agents who decide which rules the principals must obey and who coerce them to comply. As Moe (1990:232) put it, "while citizens are nominally the superiors in this hierarchy, it is the legislators who actually hold public office and have the right to make law. Their role, as agents, is to exercise public authority, backed by the police powers of the state, in telling principals what to do."
Why then would politicians respond to citizens rather than collude with bureaucrats or some special groups to which they are beholden? The standard answer to this question is that under democracy, government can be controlled by citizens because it is elected.
In particular,11 if the incumbents want to be reelected, whether because they value tenure in office per se or see it as an opportunity to advance private interests, citizens can induce government to represent their best interests by conditioning their voting decisions on the government's performance. Government is "accountable" if citizens can discern whether it is acting in their best interest and sanction it appropriately, so that those incumbents who act in the best interest of citizens win reelection, and those who do not do so lose. Accountability, in turn, induces representation through government's anticipation of the retrospective judgments of citizens. If political institutions meet politicians' participation ("self-selection") constraint (i.e., make it at least minimally attractive for people who have other opportunities to want to be reelected12) and the incentive compatibility constraint (i.e., make it in the interest of politicians to do what citizens would want them to do), then, anticipating the retrospective judgments of voters, governments will choose policies they believe will be positively evaluated by citizens by the time of the next election (Downs, 1957; Fiorina, 1981; Manin, 1995).
Yet asymmetric information between governments and voters makes representation difficult to enforce. Citizens typically do not observe some conditions of which politicians are aware. Such conditions may include the negoti-
ating posture of foreign governments or international financial institutions (whether they are accommodating or adamant) or the level of demand among the major importers of the country's exports. Citizens typically do not know as well as governments the impact of policies on outcomes, for example the impact of minimum wages on unemployment or of taxes on labor supply. With such incomplete information, accountability is not perfect: when citizens set their demands too high, politicians have no chance to be reelected and may dedicate themselves to the pursuit of short-term rents; when citizens set their demands too low, politicians can extract excess rents and still be reelected (Ferejohn, 1986).
While first-best representation is not feasible, institutional design matters. The behavior of the government is sensitive to the information available to citizens. Hence, we must ask again what institutional arrangements would alleviate these informational problems.
The first concerns the opposition. Citizens have two agents, not one: the incumbent government, which chooses policies, and the opposition, which wants to become the government. The opposition is an agent of citizens since it wants to win office, and to do so must anticipate the retrospective judgments voters will make about the incumbents at election time. Anticipating these judgments, the opposition has incentives to monitor the government and inform citizens (truthfully or not) about the bad performance of the incumbents. It can win elections if it persuades voters that the incumbent government has not been representative. Even if citizens initially care only about outcomes rather than the policies that generated them, the opposition can induce voters to care about policies if it succeeds in persuading them that different policies would have led to better outcomes (Arnold, 1993). And if opposition parties inform citizens about the misdeeds of the government or just about the sources of incumbents' money, they lower the cost of information to voters.13
Second, the mechanisms of accountability are not only "vertical"—of elected politicians to voters—but also "horizontal"—of different branches of the government to each other (O'Donnell, 1991). Elections are inevitably a plebiscitarian device: however informed voters are, their choice is only intermittently to ratify or reject decisions made by competing and cooperating teams of their representatives (Bobbio, 1989:116). A deliberative and open legislative process forces representatives to justify publicly the courses of action they advocate and to pool information they have. The legislative process is the occasion for spelling out the technical relations between policies
and outcomes in concrete terms and in some detail. It not only forces the executive to justify and defend its actions to other organs of the government, but also informs citizens.14 Decree powers bypass this process and deprive citizens of the opportunity to learn about the quality of policies. By depriving the legislature of its deliberative function and citizens of information about the relative merits of alternative policies, decrees reduce the effectiveness of accountability mechanisms. Indeed, rule by decree should create a presumption that the executive is hiding from citizens, as well as legislators, some reasons for choosing particular policies.
Finally, the quality and amount of information available to citizens can be enhanced by institutions independent of other organs of the government. Such institutions may include (1) an independent board to ensure transparency of campaign contributions, with its own investigative powers; (2) an independent auditing branch of the state, an auditor general (World Bank, 1994:32), in the vein of the Chilean controlaría; (3) an independent source of statistical information about the state of the economy; and (4) a privileged place for the opposition in overseeing the publicly owned media. These would be, to use the language of an Australian commission, "accountability agencies" (Dunn and Uhr, 1993).
Let us conclude with an example. It is widely reported that worldwide, only about 40 percent of the funds allocated to targeted food subsidies ends up as food in the mouths of the poor; the rest disappears along the way. Faced with this fact, we can react in two ways. One is to say that since such programs are inefficient, they should not be undertaken. The other is to reform the delivery systems: to organize incentives in such a way that only those who really need the subsidies will apply for them; that government agencies will be able to recognize the needy; that citizens will be able to inform politicians as to whether those who need, and only those who need, subsidies are receiving them; and that politicians will fear being thrown out of office if most of the money ends up in the wrong pockets. I have argued that this latter solution is feasible and preferable. This is not to claim that a first-best allocation is possible; the best food delivery programs spend about 20 percent on delivery, and this cost is unavoidable. Moreover, some inefficiency is unavoidable since governments, as well as citizens, are constrained
by information and transaction costs. The task of state reform is to make the government operate as well as possible under these constraints.
The reform of the state should be conceived in terms of institutional mechanisms by which governments can control the behavior of private economic agents, and citizens can control governments. The question of whether a neoliberal state is superior to an interventionist one cannot be resolved in general, since the quality of state intervention depends on the specific institutional design. But the neoliberal state is at best a benchmark against which to measure the quality of state intervention: given that market allocations are not efficient, disabling the state is not a reasonable goal for state reform.
State intervention can be superior to nonintervention when the institutional design allows governments to intervene in the economy, enables politicians to control bureaucrats, and enables citizens to control governments. At the risk of repetition, it bears emphasis that all are necessary. Governments must be able to distinguish when their interventions will increase social rates of return and must have instruments of effective intervention. But governments themselves must have incentives to intervene well and must be subject to sanctions when they do not act in the public interest. The elected politicians must want and be able to control bureaucracies, which are not subject to direct popular sanctions. Citizens must be able to discern good from bad governments and to sanction them appropriately, so that those incumbents who act well will win reelection, and those who do not will lose.
These conditions are stringent, and they can never be fully satisfied. But institutional design matters. The fact is that during the past 200 years, we have thought little about the institutional design of democracy. Since the great explosion of institutional thinking when the present democratic institutions were invented—and they were invented—there has been almost no institutional creativity. Except for the never-implemented provisions for workers' comanagement in the Weimar Constitution, the discovery of proportional representation in the 1860s and of mass parties in the 1890s were the last major political inventions. All democracies that have emerged since the end of the eighteenth century, including the most recent ones, have merely combined in different ways, often piecemeal, the preexisting institutions. Hence, there is a great deal of room for institutional creativity.
Arnold, D. 1993 Can inattentive citizens control their elected representatives? In Congress Reconsidered, 5th ed, L.C. Dodd and B.I. Oppenheimer, eds. Washington, DC: CQ Press.
Baron, D.T. 1995 The economics and politics of regulation: Perspectives, agenda, and approaches. Pp. 10-63 in Modern Political Economy, J.S. Banks and E.A. Hanushek, eds. Cambridge, England: Cambridge University Press.
Barro, R.J. 1973 The control of politicians: An economic model. Public Choice 14:19-42.
1990 Government spending in a simple model of endogenous growth. Journal of Political Economy 98:S 103-S 126.
Bobbio, N. 1989 Democracy and Dictatorship. Minneapolis: University of Minnesota Press.
Buchanan, J.M., R. Tollison, and G. Tullock, eds. 1980 Toward a Theory of the Rent-Seeking Society. College Station: Texas A&M University Press.
Calmfors, L., and H. Horn 1985 Classical unemployment, accommodation policies and the adjustment of real wages. The Scandinavian Journal of Economics 87:234-261.
Cheibub, J.A., and A. Przeworski 1997 Government spending and economic growth under democracy and dictatorship. In Understanding Democracy: Economic and Political Perspectives, A. Breton, P. Salmon, and R. Wintrobe, eds. Cambridge, England: Cambridge University Press.
Coase, R.H. 1960 The problem of social cost. The Journal of Law and Economics 3:1-44.
Cui, Z. 1992 Incomplete Markets and Constitutional Democracy. Manuscript. University of Chicago.
Cukierman, A. 1992 Central Bank Strategy, Credibility, and Independence. Cambridge, MA: MIT Press.
Downs, A. 1957 An Economic Theory of Democracy. New York: Harper and Row.
Dunn, D.D., and J. Uhr 1993 Accountability and Responsibility in Modern Democratic Governments. Paper presented at the Annual Meeting of the American Political Science Association, September 2-5, Washington, DC.
Elster, J. 1995a The impact of constitutions on economic performance. Pp. 209-226 in Proceedings of the World Bank Annual Conference on Development Economics. With comments by Adam Przeworski and Pranab Bardhan.
1995b Ulysses Revisited. Manuscript. Columbia University, New York.
Ferejohn, J.A. 1986 Incumbent performance and electoral control. Public Choice 50:5-25.
Findlay, R. 1990 The new political economy: Its explanatory power for the LDCs. Economics and Politics 2:193-221.
Fiorina, M.P. 1981 Retrospective Voting in American National Elections. New Haven, CT: Yale University Press.
Groves, T. 1985 The impossibility of incentive-compatible and efficient full cost allocation schemes. Pp. 95-100 in Cost Allocation: Methods, Principles, Applications, Peyton Young, ed. Amsterdam: Elsevier.
Habermas, J. 1975 Legitimation Crisis. Boston: Beacon Press.
Haggard, S. 1995 The Reform of the State in Latin America. Paper presented at the World Bank's Annual Conference on Development in Latin America, June 12-13, Rio de Janeiro.
Hammond, T., and G. Miller 1985 A social choice perspective on authority and expertise in bureaucracy. American Journal of Political Science 29:1-28.
Holmstrom, B. 1982 Moral hazard in teams. Bell Journal of Economics 13:324-340.
Kiewiet, D.R, and M.D. McCubbins 1991 The Logic of Delegation: Congressional Parties and the Appropriation Process. Chicago: University of Chicago Press.
Kydland, F.E., and E.C. Prescott 1977 Rules rather than discretion: The inconsistency of optimal plans. Journal of Political Economy 85:473-491.
Laffont, J., and J. Tirole 1988 The dynamics of incentive contracts. Econometrica 56:1153-1175.
1994 A Theory of Incentives in Procurement and Regulation. Cambridge, MA: MIT Press.
Lowi, T.J. 1979 The End of Liberalism: The Second Republic of the United States. New York: W.W. Norton.
McCubbins, M., and T. Schwartz 1984 Congressional oversight overlooked: Police patrols versus fire alarms. American Journal of Political Science 28:165-179.
Manin, B. 1995 Principes du Gouvernement Représentatif. Paris: Calmann-Lévy.
Manin, B., A. Przeworski, and S.C. Stokes 1997 Democracy, Representation, and Accountability. New York: Cambridge University Press.
Miller, G.J. 1992 Managerial Dilemmas: The Political Economy of Hierarchy. Cambridge, England: Cambridge University Press.
Moe, T..M. 1990 Political institutions: The neglected side of the story. Journal of Law, Economics, and Organization 6:213-253.
Niskanen, W.A. 1971 Bureaucracy and Representative Government. Chicago: University of Chicago Press.
O'Donnell, G. 1991 Delegative Democracy? East-South System Transformations Working Paper No. 21, University of Chicago.
Persson, T., G. Roland, and G. Tabelini 1996 Separation of Powers and Accountability: Towards a Formal Approach to Comparative Politics. Discussion Paper No. 1475. London: Centre for Economic Policy Research.
Posner, R.A. 1987 The constitution as an economic document. The George Washington Law Review 56:4-38.
Prud'homme, R. 1995 The dangers of decentralization. The World Bank Research Observer 10:201-220. With comments by Charles E. McLure, Jr., and David O. Sewell.
Przeworski, A., and F. Limongi 1993 Political regimes and economic growth. Journal of Economic Perspectives 7:51-69.
Ram, R. 1986 Government size and economic growth: A new framework and some evidence from cross-section and time-series data. American Economic Review 76:191-203, with comments by Jack L.Carr and by V.V. Bhanoji Rao, American Economic Review 79(1989):267-284.
Shihata, I. F.I. 1995 Judicial reform in developing countries and the role of the World Bank. Pp. 219-233 in Judicial Reform in Latin America and the Caribbean. Proceedings of a World Bank Conference, M. Rowat, W.H. Malik, and M. Dakolias, eds. World Bank Technical Paper No. 280. Washington, DC: The World Bank.
Skidelsky, R., ed. 1977 The End of the Keynesian Era: Essays on the Disintegration of the Keynesian Political Economy. New York: Holmes and Meier.
Spiller, P.T. 1995 Regulatory commitments and utilities' privatization: Implications for future comparative research. Pp. 63-79 in Modern Political Economy, J.S. Banks and E.A. Hanushek, eds. Cambridge, England: Cambridge University Press..
Stigler, G. 1975 The Citizen and the State. Essays on Regulation. Chicago: University of Chicago Press.
Stiglitz, J.E. 1994 Whither Socialism? Cambridge, MA: MIT Press.
Tirole, J. 1994 The internal organization of government. Oxford Economic Papers 46:1-29.
Tollison, R.D. 1982 Rent seeking: A survey. Kyklos 35:575-602.
Weder, B. 1995 Legal systems and economic performance: The empirical evidence. Pp. 21-26 in Judicial Reform in Latin America and the Caribbean. Proceedings of a World Bank Conference, M. Rowat, W.H. Malik, and M. Dakolias, eds. World Bank Technical Paper No. 280. Washington, DC: The World Bank.
Wolf, C., Jr. 1979 A theory of nonmarket failure: Framework for implementation analysis. Journal of Law and Economics 22:107-139.
World Bank 1994 Governance: The World Bank's Experience. Washington, D.C.: The World Bank.