In: Economics
Comment on whether each of the following statements is true or false.
(a) The Durbin Watson statistic is useless because it is based on assumptions that are rarely satisfied in time series regressions.
(b) a Durbin Watson statistic below 2 shows negative serial correlation because it is a statistic based on the AR(1) coefficient of the error term.
(c) Consider a time series process that follows a random walk drift. Is the first difference of that time series stationary? Is it weekly dependent?
(d) Coefficients of an autoregressive model can be estimated in an unbiased way by OLS, as long as all right hand side variables are strictly exogenous.
(e) a time series regression on non-stationary variables cannot produce unbiased estimates.