FAST DOUBLE BOOTSTRAP TESTS OF NONNESTED LINEAR REGRESSION MODELS

ABSTRACT

It has been shown in previous work that bootstrapping the J test for nonnested linear regression models dramatically improves its finite-sample performance. We provide evidence that a more sophisticated bootstrap procedure, which we call the fast double bootstrap, produces a very substantial further improvement in cases where the ordinary bootstrap does not work as well as it might. This FDB procedure is only about twice as expensive as the usual single bootstrap.

Keywords:

• Nonnested test
• Bootstrap test
• Jtest
• JEL Classification:C12, C15, C20

ACKNOWLEDGMENTS

This research was supported, in part, by grants from the Social Sciences and Humanities Research Council of Canada. We are grateful to two anonymous referees for comments on an earlier version.

Notes

¹Davidson and MacKinnon[6] contains limited simulation results for FDB tests on the mean of a lognormal distribution and tests for omitted variables in a probit model. It does not discuss the application of the FDB to nonnested tests. The FDB is not discussed at all in Davidson and MacKinnon.[5]