Simple, old fashioned Keynesian models of cyclical fluctuations provid
e a legitimate alternative to neoclassical real business cycle models.
Not only can these models mimic many of the characteristics of actual
business cycles, the ad hoc behavioral rules on which they are based
are vastly simpler and easier to calculate than the often insoluble ma
ximization problems faced by representative agents in neoclassical mod
els. In addition, these simple rules often imply a negligible loss of
utility compared to that achieved when decision rules are derived as s
olutions to an intertemporal optimization problem. As a result, even t
hough Keynesian models ate apparently based on suboptimal behavior by
individual agents, one cannot assume that these models are dominated b
y neoclassical alternatives.