Generalised Wald type tests of nonlinear restrictions
Abstract
This paper proposes a generalised Wald type tests to test the hypothesis of the nonlinear restrictions. We circumvent the problem of singularity of the covariance matrix associated with the usual Wald test by proposing a generalised inverse procedure, and an alternative simple procedure which can be approximated by a suitable chi-square distribution. New threshold values are derived to estimate the rank of the covariance matrix. We also propose an operational procedure to test the hypothesis of the Granger non-causality in cointegrated systems.