Microeconomic foundations of cyclical irregularities or ‘chaos’

Abstract

Phenomena described variously as pseudorandom, irregularly-cyclical, or chaotic can appear in comparatively-simple dynamic aggregate models. This paper takes a microeconomic perspective and examines conditions in the small which appear to underlie irregular behavior in the large. The point of departure is Day's (1982a) nonlinear version of the neoclassical aggregate growth model which exhibits both conventional and irregular dynamics. The microeconomic approach, which employs one-dimensional cellular automata to simulate departures from equilibrium investment paths, yields information on the likely frequency of chaotic phenomena and on their expectational properties.In particu;ar, the automata-theoretic approach generates a rigorous and easily-applied classification scheme for dynamic systems and the data produced by them. It is shown that plausible variations in business behavior within a deterministic microeconomy support chaotic macrodynamics as well as two simpler types of behavior: uniformity over time and regular periodicity. A surprising result is the finding that the same economy can also produce a type of behavior that is qualitatively richer than chaos and far more interesting in its expectational and economic implications. The last phenomenon can manifest itself in aggregate data compiled from the micro-model as an apparent profound structural or regime change. It turns out that only seemingly minor alterations in the parameters of a model microeconomy account for which of these four qualitative macro behaviors will appear.