ESTIMATION IN MULTIMARKET DISEQUILIBRIUM MODELS

A two-market disequilibrium model with spillovers is formulated. The parameters are estimated in sampling experiments by (1) ignoring spillovers, (2) treating them as if they were exogenous, (3) maximizing the likelihood function that is appropriate in the presence of spillovers. It is found that the cost in terms of root mean square error of using the incorrect estimator is serious. However, the computational time needed to maximize the likelihood function with spillovers rises sharply with the number of markets in the model due to the multiple integrals involved in evaluating the likelihood function. © 1979.