Tests for Linear Regression Models

Abstract

The linear regression model is often used to study economic relationships and is familiar from standard intermediate and introductory courses at the level of, for example, Greene (2008), Gujarati (2003) and Wooldridge (2006). In such courses, considerable emphasis is usually placed on the important topic of testing hypotheses about the values of the parameters of the model. The text-book tests for regression models are developed using very strong auxiliary assumptions that simplify teaching but are of limited relevance in practical situations. As a consequence, applied workers often have to replace procedures that are exactly valid in finite samples under strong assumptions by tests that are based on weaker assumptions but are only asymptotically valid.