Files
Abstract
We consider the true size of the Chow Test for the structural stability of a regression model when the disturbances are autocorrelated. We show that there may be considerable size distortion in the case of either AR(1) or MA(1) errors.