Introduction

Business cycles, or fluctuations of aggregate economic activities, have long attracted economists' attention. A glance at Haberler (1964) reveals that all kinds of theories had already been advanced by the end of the 1950s.

Some theories consider monetary factors, such as changes in money supply, the fundamental factor causing business cycles whereas others consider real factors, such as technical progress, fundamental. Some theories emphasize the role played by exogenous shocks. The famous theory of Jevons (1884), for example, singled out sun spots as the ultimate factor, and traced the chain of causation from sun spots to weather conditions, weather conditions to harvests, and finally from harvests to general business conditions. In contrast, other theorists purged exogenous factors, and constructed models in which endogenous cycles are generated. Goodwin's (1951) model of the nonlinear accelerator is a primary example of such models, Kaldor (1940) is another. More recently, Grandmont (1985) offers another example in which persistent deterministic cycles appear in a purely endogenous fashion without any exogenous shocks. Endogenous cycles produced in these models all rest on the particular nonlinearity of behavior of the representative agent.

Earlier, following the lead of Frisch (1933), Samuelson (1939) demonstrated that the second-order ordinary linear difference equation based on the multiplier and accelerator could generate cycles. Introducing the “ceiling” and the “floor” into such a model, Hicks (1950) ingeniously got around “explosions,” and showed that the model could account for a dazzling variety of cycles.