Microfoundations for macroeconomics

One of the major research programs over the past two decades has been the search for microfoundations of macroeconomic theory, particularly of the New Keynesian variety. (A sampling of this work is in Mankiw and Romer (1991); for recent reviews of this research program see the articles by Colander (1992) and van Ees and Garretsen (1992) in this volume.) The hallmark of this activity has been to ascertain what economic behavior patterns operating at the level of the household and firm could be responsible for certain observed or hypothesized relationships between aggregate macroeconomic variables as GDP, inflation, real wages, productivity, and unemployment. An impressive array of competing behavioral hypotheses have been modeled formally and plausibly argued on choice-theoretic grounds. The empirical track record is somewhat less compelling. Most of these models have not been tested with aggregate data; for those that have, clear victories have been few and many puzzles remain. At this point in time it seems that more foundations than houses have been built, and a coherent, progressive direction to this research program is not at all obvious. There may be good reasons for this.

It has been known for some time (e. g., beginning with Leontief (1947), Gorman (1953), and Theil (1954), and later Eisenberg (1961) and Green (1964) that the logical requirements of consistent linear aggregation are so restrictive on functional forms that choice-theoretic microfoundations at the level of the individual agent have few implications for the behavior of large-scale aggregates unless one is prepared to make a number of auxiliary assumptions.