Using the consumption-personal disposable income data for the U.S. used in this chapter:
(a) Compute the sample autocorrelation function for personal disposable income (Yt) for m = 13 lags. Plot the sample correlogram. Repeat for the ?rst-differenced series (ΔYt).
(b) Using a Ljung-Box QLB statistic, test that Ho; ρs = 0 for s = 1, . . . , 13.
(c) Run the Dickey-Fuller regression given in (14.6) and test for the existence of a unit root in personal disposable income (Yt).
(d) Run the augmented Dickey-Fuller regression in (14.7) adding one lag, two lags and three lags of ΔYt to the right hand side of the regression.
(e) De?ne Y-t = ΔYt and run ΔY-t on Y-t-1 and a constant. Test that the ?rst-differenced series of personal disposable income is stationary. What do you conclude? Is Yt an I(1) process?
(f) Replicate the regression in (14.22) and verify the Engle-Granger (1987) test for cointegration.
(g) Test for homoskedasticity assuming an ARCH(2) model for the disturbances of (14.13).
(h) Repeat parts (a) through (g) using logC and logY . Are there any changes in the above results?