Prelude to Macroeconomics
构建一个严格遵循理性原则但包含不完全信息的模型，产生类似凯恩斯主义的总产出和就业行为，并证明政府支出能增加总产出和福利。
Two schools of macroeconomic thought compete today. The Keynesian school attempts to analyze each sector of the economy using the usual tools of optimizing models, but produces general equilibrium descriptions of a macroeconomy which are rarely Pareto efficient. For this reason, economists question the internal consistency of Keynesian models. In contrast, the neoclassical or .rational expectations school maintains consistency with the principles of perfect competition and flexible prices. In essence, the neoclassicists' macroeconomy behaves as an Arrow-Debreu general equilibrium. Once this is understood, we realize that the economy is Pareto efficient, though this does not preclude occasional ex post bad draws. Government policy can be expected to be either neutral or damaging. (In fairness, I am describing polar cases of the Keynesian and neoclassical view.) Nearly all economists are extraordinarily prejudiced in favor of models exhibiting rational behavior (as am I) and the last decade has seen an almost complete intellectual victory for the neoclassical school. Complete victory has been elusive for a single reason. In apparent ignorance of the intellectual arguments of the neoclassical school, the economy persists in behaving pretty much as the modern Keynesian models predict. As premier examples, neither the Great Depression nor the recent massive recession was (in my opinion) a Pareto-efficient equilibrium. The model I present below rigorously adheres to the rule that agents should follow rational principles. In this paper that rule means that agents equate marginal rates of substitution to relative prices. At the same time, I insert a single piece of imperfect information which prevents the formation of a complete Arrow-Debreu general equilibrium. The model predicts qualitative behavior of GNP and employment which is analogous to Keynesian predictions. Government spending is shown to increase GNP and economic welfare. The substantive results of the paper appear in the next three sections. Section I presents the role of imperfect information in the labor market and then goes on to solve for general equilibrium in the absence of government intervention. Section II examines the Keynesian-like behavior of this equilibrium. Section III examines the impact on GNP, aggregate labor supply, and welfare of balanced-budget government spending. The model produces four major results. 1) Aggregate spending and labor supply decisions are not simply the sum of individual decisions. The model identifies the logical error that Paul Samuelson has labeled the fallacy of composition. 2) Say's Law fails. A unit increase in aggregate supply produces a less than unit increase in demand for output. 3) An increase in government spending increases GNP and reduces unemployment. 4) An increase in government spending can generate a Pareto improvement in individual utility. The last section of the paper discusses some of the ways this model differs from the way we usually think about the economy. While the paper develops a particular model of aggregate demand, its real goal is to demonstrate a general principle: once the right set of mathematics is put together, it is easy to produce a model of the economy which is at once rational and Keynesian. In the specific setting I present, all the results fol*Department of Economics, University of Washington, Seattle, WA 98195. The first version of this paper was written while visiting at the Graduate School of Business, Stanford University. The final version was written while a member of the Finance Department, University of Pennsylvania. Shelly Lundberg deserves many thanks for extensive aid and partial absolution from any remaining errors. Stanley Fischer, Mark Flannery, Robert Solow, a number of other friends and colleagues, and two anonymous referees all contributed valuable constructive criticism for which I am most grateful.