| Let us call et,1 = yt + 1 – ft,1 the 1-quarter-ahead forecast error that you have computed in Exercise 6. Is the expected value of the forecast errors equal to zero? If not, what do you conclude? Run the regression of the forecast errors on several lags, that is, et,1 = ?0 + ?1et,–1 +B2et–2,1 + . . . . .ut+1. Comment on the regression results. Perform an F-test for the joint hypothesis H0: ?1 = ?2 = ?k = 0 for your choice of k. What do you conclude? Is the forecast error predictable from its own past? |
