Further Results on Interpreting Coefficients in Regressions with a Logarithmic Dependent Variable

Abstract

Abstract Estimators are presented for quantifying the proportional rate of change in the continuous variable Y from a regression in which the dependent variable is the logarithm of Y, and the data generation process includes explanatory variables of interest that may be binary (dummy), continuous, or logarithmic. Estimators from earlier works that examine the binary explanatory variable are special cases of the results presented here. The additional estimators provided here will be useful to practitioners who must convert coefficients estimated from regression models specified with a logarithmic dependent variable into proportional rates of change.