Tests of Inference for Dummy Variables in Regressions with Logarithmic Transformed Dependent Variables

The interpretation of dummy variables in regressions where the dependent variable is subject to a log transformation has been of continuing interest in economics. However, in the main, these earlier papers do not deal with the inferential aspects of the parameters estimated. In this paper we compare the inference implied by the hypotheses tested on the linear parameter estimated in the model and the tests applied to the proportional change that this parameter implies. An important element in this analysis is the asymmetry introduced by the log transformation. Suggestions are made for the appropriate test procedure in this case. Examples are presented from some common econometric applications of this model in the estimation of hedonic price models and wage equations.