A Monte Carlo Study of Some Small Sample Properties of Tests for Specification Error

Abstract

Abstract Some tests for the specification errors of omitted variables, incorrect functional form, simultaneous equation problems, and heteroskedasticity previously developed by the author are further considered in this article. Monte Carlo methods are used to examine the statistical independence of the test statistics, the small sample power of the tests under various alternatives, and the effect on the tests of using ordinary least squares instead of Theil residuals.