A Simulation Experiment into the Numerical Properties of General Equilibrium Models of Factor Market Distortions

Abstract

HOW big are static efficiency losses from market distortions? How do these loss estimates vary with the size of distortions? How are the effects of distortions compounded (or offset) when more than one distortion occurs? How do efficiency loss estimates vary with the parameters of a model for given distortions? How are efficiency loss estimates affected by the level of aggregation within a model? These questions all require quantitative answers which have, until recently, been difficult to obtain due to the absence of adequate computational devices for analysing them. This paper reports some numerical results which have a bearing on the questions posed above. Fixed point computational techniques are applied to some simple general equilibrium models often used in economics but whose numerical properties are less well known. The techniques which are used derive from some recent proposals by Shoven and Whalley (1973) for procedures for the computation of competitive equilibria in the presence of taxation. These, in turn, build on the fundamental work in this field due to Scarf (1973). The main focus is the two industry general equilibrium literature on factor market distortions that has developed from the work of Harberger (1966) and Johnson (1966) (see the survey article by Magee (1973)). In this paper a process of repeated solution for competitive equilibria in the presence of factor market distortions is followed for different parametric values of coefficients and distortions. Results are reported for a number of cases of interest, and a final section attempts to synthesize results into some broad conclusions on numerical properties of these models. Even though these formulations are themselves oversimplified representations of economic processes, it seems useful to investigate possible rules of thumb for simple models which can be appealed to in more complex policy situations.