A correction for local biasedness of the Wald and null Wald tests

This paper proposes a procedure to overcome the problem of small-sample local biasedness of the Wald test of a single restriction. The new test procedure involves finding two critical values, in order to obtain the desired size and local unbiasedness. Simulation results indicate that the Wald test corrected in this way still shares the non-monotonic power behaviour of the original test reported in the literature. We extend the new test procedure to the null Wald test recently developed to resolve power non-monotonicity, and obtain good power properties.