Testing for joint significance in nonstationary ordered choice model

- We investigate the joint significance test for nonstationary ordered choice model.
- The convergence rate of maximum likelihood estimator of explanatory variable is T.
- Wald statistic is asymptotically Chi-squared for nonstationary logit or probit model.

This paper studies the test of joint significance for the ordered choice model with multiple explanatory variables following integrated processes. The results show that for the widely used logit and probit models, when the true parameter vector of explanatory variables is zero, the classical Wald statistic is still useful and has a Chi-squared limiting distribution.