__Question__

Load the data file __gdp1.dta__, which contains quarterly observations from 1949.Q1 to 2002.Q4 on the following variables: con (per capita consumption), and gdp (per capita GDP).

[1] Run the following two regressions with OLS: (i) log(con) on log(gdp) (ii) log(con) on L(1/2).log(con) and L(0/2).log(gdp). Based on the regression results comment on if the residuals are auto-correlated (based on the Breuch-Godfrey LM test, ).

[2] Examine if each of the following series (i) log(con) (ii) log(gdp) contains a unit root using the “dfgls” test. Interpret the results. Hint: Beware of “trend”!

[3] Examine if log(con) and log(gdp) are cointegrated. Hint: Plot the regression residual series and comment if it is mean-reverting, the use the Engel-Granger approach to test for cointegration (i.e. test if the regression residual series contains a unit root, ).

[4] Examine Granger causality between D.log(con) and D.log(gdp). Specify . Comment on the results.