Testing Superexogeneity and Invariance in Regression Models.

This paper introduces tests of superexogeneity and invariance. Under the null hypothesis, the conditional model exhibit s parameter constancy while under the alternative shifts in the proces s of the independent variables induces shifts in the conditional model. The test is sensitive to particular types of parameter nonconstancy, especially with changing variances and covariances. The authors rela te the test to rational expectations models and the Lucas critique. An empirical example of money demand finds prices and interest rates superexogenous in a conditional model, but when the inflation specification changes, superexogeneity fails although standard specification tests do not