Yesterday, David Brooks published an interesting essay in the New York Times entitled “Is Our Adults Learning?” The basic premise behind his essay can be summarized in the following quote, “Government doesn’t profit from experience because of the way it goes about testing its policy problems. It should try learning the way businesses do.”
My purpose in posting this commentary on Brooks’ article is to challenge his presumption that it’s even remotely possible for the government to conduct itself like a business. Avinash Dixit’s 1997 American Economic Review article entitled “Power of Incentives in Private versus Public Organizations” provides the framework for my critique. In this (admittedly somewhat dated) academic journal article, professor Dixit takes on an earlier version of Brooks’ ideas as exemplified in Al Gore’s 1995 tome entitled “Common sense government”. In this book, Gore argues for “reinventing” government by measuring and rewarding “results, not red tape.” However, Dixit shows that the problem with Gore’s and Brooks’ platitudes about getting the government to act more like a business is that these “theories” fundamentally ignore the nature of government bureaucracy. Dixit argues that “a distinct feature of government bureaucracies is that they must answer to multiple principals”, and he goes on to “… develop a model of a common agency to show how the interaction among many principals results in a loss of the power of incentives.” To illustrate this, Dixit notes that in the real world, a government agency may be formally answerable only to the executive, but in practice Congress, courts, media, and organized lobbies all have a say. He notes that one way to resolve this “weak incentive” problem at the federal level is to devolve political power to states or localities, where “…agencies can be so designed that each performs fewer tasks, thus reducing the externalities among the principals affected by its actions.” So basically Dixit rigorously proves with the game theory the wisdom behind federalism!
In closing, it would seem that the least likely place for “decentralizing policy experimentation as much as possible to encourage maximum variation” to be successful would be at the level of the federal government. The weak incentive problem identified by Dixit explains why government agencies are typically focused on red tape, and not results; it’s all about process and procedure, and rarely ever about results.