A Two-Way Generalisation of the Chow Test Under Dependent Errors

The Chow test is not robust under dependency of errors. The presence of dependency of errors will affect level of significance as well as power of the test, especially when the sizes of the samples are small. This paper attempts to construct a test procedure where not only the equality between sets of coefficients in two linear regressions are compared, but also, in case they are not equal, detailed informations about the inequality of the sets are provided. Also, all these are accomplished for not just only two linear regressions but for all possible pairs of linear regressions out of any number of given linear regressions, resolving at the same time the problem of dependency in the errors, thus generalizing the Chow test in two directions under dependent errors. The procedure is then illustrated through comparison of growth rates of population of India for different decades, using NSSO data.