The title of this post is taken from a 1994 article by John DiIulio (Journal of Public Administration Research and Theory 4(3), 1994: 277-320), which criticized the principal-agent framework being used by economists to understand organizational behavior, governance, and political corruption.
Under principal-agent, organizations (whether public or private sector) are hierarchical structures in which principals have the authority to give instructions to agents. Corruption and other dysfunctions occur, by this view, because the agents are also self-interested actors whose interests are not necessarily the same as that of the principals. Fixing a problem like corruption therefore involves restructuring incentives to better align the interests of the agents with those of the principals. Granting managers stock options in their own companies was originally seen as a way of implementing this solution. Closely monitoring the behavior of agents and rewarding or punishing them for measured performance was another, but this often involves substantial agency costs. The general emphasis in many recent initiatives at governance reform involving increasing transparency and accountability all fall within a principal-agent framework. (Transparency and accountability are just nice words for monitoring and punishing.) The International Budget Project’s Open Budget Survey and the World Bank’s Expenditure Tracking Surveys are all efforts along these lines.
I think that principal-agent is useful as a starting point for analysis of government effectiveness, and I will make reference to it in subsequent posts. No one can argue in general with the need for greater transparency and accountability on the part of governments (though whether all forms of transparency and accountability add to government quality will be the subject of a later post). But P-A also has a number of important limitations that sometimes blind its users to the real sources of dysfunction in public agencies.
DiIulio pointed to one important limitation, which was that in many high-functioning organizations, good performance was not the result of bureaucrats following economic incentives, but rather the result of norms, values, and social habits. Human beings are not simply rational individuals making self-interested calculations, but social animals conforming to rules set by their environments. If the agency creates a strong organizational culture around certain clear goals, and workers are strongly bonded to one another, they will stay late in the office and resist lucrative offers from the private sector in order to perform their jobs as best they can.
Economists puzzle about how to solve the problem of hidden or joint output, where it is difficult if not impossible to measure the contribution of a worker to a common goal. If you can’t measure individual output, you can’t create the proper economic incentives for performance. Strongly embedded social norms solve this problem by having the agents in effect monitor themselves. They don’t shirk or cheat, not out of fear of being caught, but because they live according to certain internalized principles. Hence “principled agents.”
Internalized norms are in fact a component of what we label professionalism. We do not try to precisely monitor the output of doctors, architects, professors, or software engineers and pay them by a piece rate schedule because it is not possible to measure the quality of the complex services they produce. We have to rely to a large extent on internalized standards of behavior, and the fact that they take pride in what they do.
Blindness to moral or normative motivation is only one of the limitations of principal-agent, however. Another that is critical in understanding problems of the public sector is the problem of multiple principals. The principal-agent framework assumes that there is a single principal, and that that principal issues clear instructions to the agent. In modern government, however, there are many different principals who issue often contradictory instructions to their agents. A police department, for example, may be under the gun from the mayor to get aggregate crime rates down, but it also faces substantial political pressure from minority groups to avoid racial profiling or police brutality. A public utility faces demands for cost recovery and efficient service delivery from commercial users, but countervailing pressures for universal, cross-subsidized service to poor consumers and their political representatives. The setting of these goals is not under the control of the agency, and the agents cannot satisfy all of its principals simultaneously. The fact that many public utilities in the developing world lose lots of money and nonetheless continue to fail to deliver reliable service is a direct product of the multiple principals to which they have to answer.
The existence of multiple principals is, by the way, one important reason that the public sector often performs much more poorly than the private sector. In a private business, the principal (the owner) issues a relatively simple, single mandate to maximize profits. Performance of the mandated action can be precisely measured in terms of the bottom line. The public sector, by contrast, is subject to mandates from competing political principals each of which have a degree of democratic legitimacy and therefore authority over the agent. Public officials have to wade through this thicket in a way that their private sector counterparts do not.
A further limitation of the principal-agent framework is the fact that it assumes the principals know what they’re doing. But in many real-world organizations, expertise and therefore authority lie with the agents rather than the principals: it is the former who are much more influential in setting goals and laying out strategy for the organization than the clueless bosses. Herbert Simon in his book Public Administration pointed out many years ago that authority in most organizations flows bi-directionally, downwards and upwards. Indeed, in some government organizations the original legislative mandate is so vague and unclear that there is in effect no principal in the first place.
The agents often know more than the principals for a variety of reasons. It is they who are the engineers, economists, agronomists, and medical professionals with the technical expertise to do their jobs. And they are almost always closer to sources of local knowledge. One of the most important contributions of Friedrich A. Hayek was his 1945 article “The Use of Knowledge in Society” (American Economic Review 35(4): 519-30), in which he pointed out that the vast majority of information in a modern economy was local in nature. He argued that market economies work better than centrally planned ones because the decentralized price-setting that characterizes them is done by actors who are closer to the sources of local knowledge than bureaucrats sitting in centralized ministries. This macro observation can be applied on a micro organizational level as well: organizations need to decentralize and delegate authority because it is the lower levels of the organization that have direct access to the local knowledge that is necessary to get the job done.
These two factors, expertise and local knowledge, then explain the need for bureaucratic autonomy, both in private and public sector organizations. Even though the formal organization chart says the principal is the boss, it is the agent who often is better suited to making decisions. All well-functioning organizations thus have to solve the problem of delegated discretion, in which agents are allowed sufficient autonomy while still remaining under the broad control of the principal.
In the next post we’ll see how this works with respect to military chains of command.
