Question 4 (20p). a) Explain intuitively how and why autocorrelation matters. b) How would you see if there is autocorrelation? How would you test for autocorrelation with a regression? Explain the steps clearly. c) Could you estimate the coefficients and/or standard errors differently, taking autocorrelation into account? How? d) Are there any concerns in correcting for autocorrelation if it is not present? e) Assume that there are omitted variables that are uncorrelated with your regressors but that change slowly over time as an AR(1) process. How can adding the lagged dependent variable handle this problem. Derive and explain.