Answer Happy

2. Like the Chow test, the Wald test for model stability can be generalized to the case of more than two subsamples. Suppose that your sample can be divided into three subsamples, and let Bj, a K x 1 vector, be the coefficient vector for subsample j. In addition, let nj be the number of observations in subsample j. Derive the Wald statistic for testing the null hypothesis that the coefficient vectors B; are the same for all three subsamples, under the assumption that the subsamples are independent. What information would you need to compute this test statistic? Hint: First, write down the unrestricted and restricted models and the restrictions on the coefficients of the unrestricted model. Then, starting with the general formula for the Wald statistic (equation (2) of the notes), derive the specific form of the statistic for this test. (30 marks)