Consider the VAR models with one and two lags in Exercise 4.

(a) Estimate the characteristic roots and vectors for the relevant matrices discussed in Section 14.8. Apply tests for unit roots and tests for cointegration.

(b) If there is a cointegrating regression, estimate it from the characteristic vectors and also from the static regression as suggested by Granger and Engle.

(c) Are the results different for the VAR models with one and two lags? Are they different from those from the static regressions? What do you conclude from these results?

(d) Repeat parts

(a) to

(c) with the seasonally adjusted data (residuals from the regression on seasonal dummies).