13.6: Principal-Agent Problems

Concerns about the bureaucracy becoming too powerful and out of control might seem incongruous with complaints about its wastefulness and inefficiency. However, both critiques stem from the same fundamental feature of bureaucracies: they are rife with principal-agent problems. A principal-agent problem occurs when someone (a principal) asks someone else with different motivations (an agent) to do something. This difference in motivations can cause the agent to perform the assigned task in a way other than how the principal would prefer it to be done.

Any situation in which someone acts on behalf of someone else can become a principal-agent problem. When you take your car to a mechanic, order food at a restaurant, or ask your friend to dogsit for you, you become a principal by enlisting an agent. You want the mechanic to work quickly and cheaply, the cook to wash his hands and give you generous portions, and your friend to be attentive to and careful with your dog. The mechanic may prefer to work at a leisurely pace and overcharge you, the cook to save time by not washing his hands and money by not filling your plate, and your friend to provide the bare minimum in pet upkeep because of how busy or lazy she is. In each case, the agent wouldn’t perform the task exactly how you would like it to be performed because your motivations differ.

Not every principal-agent relationship becomes a principal-agent problem. If the principal and agent do not differ substantially in their motivations, the principal may be able to trust the agent to act as the principal himself or herself would have acted.

Fortunately, principals are not helpless when confronted with agents who do not share their motivations. First, they can monitor agents’ performance. This can take the form of student evaluations of professors, CCTV cameras watching a store’s cash register to make sure the cashier is not stealing from the till, or a mother inspecting her child’s room for cleanliness. Second, they can reward or punish agents on the basis of their performance. This could mean a special parking place reserved for the “Employee of the Month,” a generous tip for a timely pizza delivery driver, or the impeachment of a president for high crimes and misdemeanors.

Neither monitoring nor reward or punishment is by itself sufficient to solve a principal-agent problem. If agents don’t expect to be rewarded or punished for their performance, it won’t matter how closely they are being monitored. Likewise, performance-based rewards or punishments will be ineffective if no one is keeping an eye on the agents to determine whether they deserve to be rewarded or punished.

When Congress and the president delegate rulemaking and law enforcement to the bureaucracy, they become the principals and the bureaucrats become the agents. Monitoring bureaucratic performance is possible but challenging. Congress can conduct oversight and the president can demand reports, but it’s hard to keep track of what millions of bureaucrats are doing, especially for elected officials who are non-experts in the relevant policy areas and busy with other duties. Reward and punishment is also complicated by protections like the Pendleton Act which insulate bureaucrats from political pressure. They might still be fired for gross incompetence or corruption, but mediocrity is unlikely both to draw attention and to cost them their jobs.