Make economics policy relevant: depose the omniscient benevolent dictator.

• Author: Holcombe, Randall G.
• Position: Essay

Economic policy analysts typically specify a model that describes an optimal outcome--for example, an efficient allocation of resources, an optimal distribution of income, an optimal growth path, or an optimal macroeconomic outcome of stability with full employment and low inflation. Because economic models can become complex when they account for many things, all models employ simplifying assumptions to focus on one particular issue. Models that analyze policies to promote economic growth, for example, differ from models that describe the optimal distribution of income or the optimal policy for internalizing an externality (see Holcombe 1989). Nevertheless, the methodology is the same: develop a model that incorporates the particular issue, then show in the model the optimal result and identify what keeps the economy from producing the optimum. The recommended policy is one that should move the economy from a nonoptimal position to the optimal one by removing impediments to the attainment of an optimal outcome, by changing the incentive structure so that market participants trade to a Pareto optimum, or, where markets pose more difficult problems, by imposing regulations, government mandates, or government production to allocate resources optimally.

When the invisible hand of the market fails, economic policy recommendations typically do not go beyond advising that the visible hand of government move the economy to the optimal outcome, without any detailed discussion of whether government action can achieve this outcome or whether public-sector actors have an incentive to implement the optimal outcome. Government is modeled as if it is an omniscient benevolent dictator. To make economics policy relevant, however, this omniscient benevolent dictator must be deposed. Government in contemporary Western countries is neither omniscient nor benevolent nor dictatorial.

The Planner's Problem

The omniscient benevolent dictator is often represented in economic analysis as the planner's problem. The analyst derives the conditions required for an optimal allocation of resources, and the planner's problem is to create those conditions. The policy problem is couched entirely in comparative-static terms, identifying the optimal outcome and the status quo, with policy recommendations intended to change the situation from the latter to the former. The comparative-statics methodology for evaluating economic policy measures goes back at least to A. C. Pigou (1920, chap. 1), who lays out a methodology for evaluating policy measures by comparing economic welfare with and without the implementation of the policy measure. Frank Ramsey (1928) explicitly lays out a planner's problem for finding the rate of saving that maximizes consumption over successive generations. In response to Ludwig von Mises's ([1922] 1951) claim that central planners cannot allocate resources rationally, Oskar Lange and Fred Taylor (1938) argue that central planners can allocate resources at least as efficiently as the market. (1) Abba Lerner (1944) delivers a policy analysis based on the planner's problem that identifies conditions for the optimal allocation of resources and then advises that the government implement these conditions.

We see, then, that the omniscient benevolent dictator who solves the planner's problem has been a part of economic analysis for at least three-quarters of a century, but it gained increased stature with the Arrow and Debreu (1954) proof of the uniqueness and stability of competitive equilibrium. Using competitive general equilibrium as a benchmark, one can develop a formal framework to depict cases in which the market fails to reach that optimum. In a frequently cited article, Takashi Negishi shows that the "existence of an equilibrium is equivalent to the existence of a maximum point of this special welfbxe function" (1960, 92). Francis Bator (1957) shows the conditions for welfare maximization and demonstrates (Bator 1958) conditions under which markets fail to reach this maximum. It is a short step indeed to show the conditions for an optimum and to demonstrate a reason why the market fails to get there.

The well-known public-goods articles by Paul Samuelson (1954, 1955) offer a good example. Samuelson titles his first article "The Pure Theory of Public Expenditure," indicating that his analysis of a possible market failure in the production of public goods is in fact not a theory, but the theory of public expenditure, even though the article contains no analysis of how government would succeed in producing public goods where the market would fail. (2) The only way Samuelson's public--good theory can be a theory of government expenditure is if the government is an omniscient benevolent dictator. This is the planner's problem: how to get from the imperfect allocation of resources in the real world, as depicted in a mathematical model, to the outcome that has been demonstrated mathematically to be optimal.

Milton Friedman, commenting on Lerner's book The Economics of Control (1944), observes: "Most of the book is devoted to the formal analysis of the conditions for an optimum. The institutional problems are largely neglected and, where introduced, treated by assertion rather than [by] analysis.... [N]ot only the title and the introduction but even a first reading somehow generate the expectation and the illusion that the book contains a concrete program for economic reform.... Much of what at first sounds like a concrete proposal, particularly about the general structure of society, turns out to be simply an admonition to the state that it behave correctly and intelligently" (1947, 405). Friedman's critique of Lerner holds more generally for the work that has followed, which tends to show the mathematical conditions required for an optimal allocation of resources but does not explain how a real-world government can make the economy satisfy these conditions. Because government is not an omniscient benevolent dictator, in many cases government failures will be at least as significant as market failures.

To make economic analysis policy relevant, the omniscient benevolent dictator must be deposed. Government is neither omniscient nor benevolent, and in contemporary Western countries it is not a dictator.

Not Omniscient

In many cases, the mathematical formulation of a market failure shows that the information required to correct that market failure is not available to policymakers. In a classic example from welfare economics, Pigou (1920) demonstrates that a corrective tax on an externality can bring about the internalization of external costs. As William Baumol (1972) shows, for the tax to be optimal, it must be precisely equal to the external cost and vary with it so that if the external cost rises, the tax must rise by an equal amount, and if the external cost falls, the tax must fall by an equal amount. However, the amount of the external cost cannot be observed, so unless government is omniscient, it cannot calculate the optimal tax, and therefore it cannot eliminate the market failure. It may be...