Abstract:
Regression model error assumptions are essential to estimator properties. Least squares model parameters are consistent and efficient when the underlying error terms are normally distributed but yield inefficient estimators when errors are not normally distributed. Partially adaptive and M-estimation are alternatives to least squares when regression model errors are not normally distributed. Vertically Integrated firms in the oil and gas industry is one industrial sector where error mis-specification is consequential. Equity returns are a common area where returns are not normally distributed, and inappropriate error distribution specification has substantive effect when estimating capital costs. Vertically Integrated Major equity returns and accompanying regression model error terms are not normally distributed, and this study considers error returns for Integrated oil and gas producers. Vertically Integrated firm returns and their regression model error are not normally distributed, and alternative estimators to least squares have desirable properties.