It is not, however, straightforward to see this, because the RBC conjecture is advanced jointly with a claim that models should be assessed using a novel strategy. We must therefore evaluate the conjecture and the assessment strategy simultaneously.
Since the publication of Kydland and Prescott's "Time to Build and Aggregate Fluctuations" (1982 ), the paradigm RBC model, a large and active group of new classical macroeconomists have elaborated and developed the RBC model. As important as these developments are to the RBC program, none of them fundamentally affects the critical points that we will make. Our assessment will, therefore, focus on the original Kydland and Prescott model and its successor models in a direct line. We will also refer frequently to the programmatic statements and methodological reflections of Kydland, Prescott and Lucas, the most articulate defenders of the aims and methods of equilibrium business cycle models.
The RBC model does not present a descriptively realistic account of the economic process, but a highly stylized or idealized account. This is a common feature of many economic models, but RBC practitioners are bold in their conjecture that such models nevertheless provide useful quantifications of the actual economy. While idealizations are inevitable in modeling exercises, they do limit the scope of the virtues one can claim for a model.
In particular, the RBC program is part of the larger new classical macroeconomic research program. Proponents of these models often promote them as models that provide satisfactory microfoundations for macroeconomics in a way that Keynesian models conspicuously fail to do (e.g., Lucas and Sargent, 1979). The claim for providing microfoundations is largely based on the fact that new classical models in general, and RBC models in particular, model the representative agent as solving a single dynamic optimization problem on behalf of all the consumers, workers, and firms in the economy. However, the claim that representative agent models are innately superior to other sorts of models is unfounded. There is no a priori reason to accord RBC models a presumption of accuracy because they look like they are based on microeconomics. Rather, there are several reasons to be theoretically skeptical of such models.
Most familiar to economists is the problem of the fallacy of composition, which Samuelson's (1948) introductory economics text prominently addresses. It is difficult to deny that what is true for an individual may not be true for a group, yet, representative agent models explicitly embody the fallacy of composition. The central conceptual achievement of political economy was to analogize from the concerns of Robinson Crusoe-alone in the world-to those of groups of people meeting each other in markets. The complexities of economics from Adam Smith's invisible hand to Arrow and Debreu's general equilibrium model and beyond have largely been generated from the difficulties of coordinating the behavior of millions of individuals. Some economists have found the source of business cycles precisely in such coordination problems. By completely eliminating even the possibility of problems relating to coordination, representative ag