On nonnested regression through modified regressor design

This paper examines the consequences of improper use of Monte Carlo (MC) simulation design in order to compare some nonnested testing procedures within the framework of Dubbelman et al's (1978) work. It also investigates the aspect of the MC experiment of Godfrey and Pesaran's (1983) study. In doing so, a small MC experiment with modified parameters in the generating mechanism for the regressors in the null model is carried out. However, the results obtained in this study confirm that the JA² test can be regarded as a serious contender within the array of inferential tools in nonnested setting.