Size and power of the RESET test as applied to systems of equations: A Bootstrap approach

Abstract

The size and power of various generalization of the RESET test for functional misspecification are investigated, using the "Bootsrap critical values", in systems ranging from one to ten equations. The properties of 8 versions of the test are studied using Monte Carlo methods. The results are then compared with another study of Shukur and Edgerton (2002), in which they used the asymptotic critical values instead and found that in general only one version of the tests works well regarding size properties. In our study, when applying the bootstrap critical values, we find that all the tests exhibits correct size even in large systems. The power of the test is low, however, when the number of equations grows and the correlation between the omitted variables and the RESET proxies is small. Copyright © 2004 JMASM, Inc.