Bootstrap Tests for Regression Models

Modern computer systems are now so powerful that they can be used to carry out simulation-based statistical investigations without involving delays or the need to access high levels of equipment. When carrying out econometric analyses, the routine use of computer-based methods offers a valuable alternative to the standard approach in which approximations are based upon what happens as the sample size grows without limit. Applied work has to be based upon a finite number of observations. Computationally-intensive techniques and, in particular, bootstrap methods provide ways to improve the finite-sample performance of well-known tests. Bootstrap tests can also be employed when conventional theory does not lead to a test statistic, which can be compared with critical values from some standard distribution. This book uses the familiar linear regression model as a framework for introducing simulation-based tests to applied workers, students and others who carry out empirical econometric analyses.