Why there is no science of public administration.

• Author: Fukuyama, Francis

Is public administration a science or an art? The answer to this question matters greatly to state building, because good public administration is key to effective governance. If public administration--the organization and management of individual public agencies--is a science, then presumably it can be transferred to developing countries much like knowledge of how to maintain jet aircraft or operate a factory. On the other hand, if it is more of an art, then teaching good public administration and setting up competent public agencies will be much more problematic.

This article will argue that there is no optimal form of organization, either in the private sector or among public sector agencies. The absence of globally valid rules for organizational design and management means that public administration will be more of an art than a science. This implies that most good solutions to public administration problems, while having certain common features of institutional design, will not be clear-cut "best practices" because they will have to incorporate a great deal of context-specific information.

Theorizing about organizational design has been dominated by economists in recent years, who have sought to incorporate organizations into broader microeconomic theory. This effort has yielded certain important and useful insights into public sector reform. But in the end, the behavioral assumptions on which neoclassical economics rests--in particular, the assumption that people in organizations are motivated primarily by individual self-interest--are too limited to provide understanding of key aspects of organizational behavior. This line of thought has eclipsed earlier, more sociological understandings of organizations and has obscured some important insights of that tradition.

INSTITUTIONAL ECONOMICS AND THE THEORY OF ORGANIZATIONS

For all of its richness and complexity, a huge amount of organizational theory revolves around a single, central problem: that of delegated discretion. The conundrum of organization theory is that while efficiency requires the delegation of discretion in decision-making and authority the very act of delegation creates problems of control and supervision. Observes one leading organization theorist,

Because all information cannot be moved to a central decision maker, whether a central planner in an economy or the CEO in a firm, most decision rights must be delegated to those people who have the relevant information. The cost of moving information between people creates the necessity for decentralizing some decision rights in organizations and the economy. This decentralization in turn leads to systems to mitigate the control problem that results from the fact that self-interested people (with their own self-control problems) who exercise decision rights as agents on behalf of others will not behave as perfect agents. (1) The problem of delegated discretion is currently conceptualized by economists under the rubric of "principal-agent" relationships, something that has become the overarching framework for understanding governance problems.

Berle and Means recognized long ago that the divorce of ownership from management in modern corporations created significant corporate governance problems. (2) The owners, or principals, designate managers, or agents, to look after their interests, but the agents often face individual incentives that differ sharply from those of the principals. This is a problem with all forms of hierarchical organization, and can exist at multiple levels of the hierarchy simultaneously. Jensen and Meckling introduced the concept of agency costs, which were the costs that principals had to pay to ensure that agents did their bidding. (3) These costs included the costs of monitoring agent behavior, bonding the agent and the residual losses that occurred when the agent acted in ways contrary to the interests of the firm. Jensen and Meckling assumed that it was primarily the residual risk-bearers or owners who did the disciplining, and on this basis developed a sophisticated theory of capital structure and its relationship to corporate governance. Fama argued, however, that the residual risk-bearers were not the only source of agent discipline; managers or agents monitored and disciplined each other's behavior, due to the fact that agency relationships involved repeated interactions and that there was a competitive market for managerial talent in which these evaluations would be important. (4)

Once principal-agent theory had been articulated with regard to private firms, it was a relatively simple matter to adapt the framework to explain public sector behavior. (5) In the private sector, the principals are the shareholders, corporate boards of directors are their agents, and the senior management is the agent of the board. In the public sector, the principals are the public at large. In a democracy, their first-level agents are their elected representatives; the legislators act as principals with regard to the executive branch agents delegated to carry out the policies they have legislated. Political corruption occurs when individual agents--government officials--put their own private pecuniary interests ahead of those of their principals. But agents can act contrary to principal wishes for other reasons as well, such as the desire to preserve their agencies and job security, or from ideological motivations that are at variance with those for whom they nominally work.

A great deal of the work that is now being done on improving governance is therefore the effort to better align agent incentives with those of the principals. The general approach to aligning principal and agent interests is to promote greater transparency in the activities of the agents (a nice way of describing the monitoring of their behavior), and then to hold the agents accountable for their actions through a variety of rewards and punishments. Much of the work of the public choice school is to devise constitutional and legal arrangements that minimize rent-seeking and other kinds of agency costs. Another approach--more workable in the private sector than for public agencies--is to reunite owners and managers by giving the agents stock options or other forms of equity ownership. (6)

It is certainly worthwhile to try to understand problems of corporate governance or public corruption in principal-agent terms, and to use this framework to design institutions that try to bring divergent incentives back into alignment. But there are at least three basic reasons why there can be no optimal specification of formal institutions, and thus no optimal form of organization. These considerations are particularly true for public sector agencies.

First, the goals of many organizations are unclear. Agents can carry out the will of principals only if the principals know what they want the agents to do, but this is not always the case. Goals often emerge and evolve through complicated interactions between organizational players, or are defined by the roles assigned to players in the organization. Labor can be divided functionally in a variety of ways that necessarily favor one organizational goal...