Examine whether the following statements are true or false. Give an explanation.

(a) Serial correlation in the errors u leads to biased estimates and biased standard errors when the regression equation y =Bx+u is estimated by ordinary least squares.

(b) The Durbin-Watson test for serial correlation is not applicable if the errors are hetero- skedastic.

(c) The Durbin-Watson test for serial correlation is not applicable if there are lagged dependent variables as explanatory variables.

(d) An investigator estimating a demand function in levels and first differences obtained R's of 0.90 and 0.80, respectively. He chose the equation in levels because he got a higher R. This is a valid reason for choosing between the two models.

(e) Least squares techniques when applied to economic time-series data usually yield biased estimates because many economic time series are autocorrelated.

(f) The Durbin-Watson test can be used to describe whether the errors in a regression equation based on time-series data are serially independent. (g) The fact that the Durbin-Watson statistic is significant does not necessarily mean that there is serial correlation in the errors. One has to apply some other tests to come to this conclusion. (h) Consider the model y, ay-1+ Bx, +u,, where the errors are autoregressive. Even if the OLS method gives inconsistent estimates of the parameters, we can still use the equation for purposes of prediction if the evolution of x, during the prediction period is the same as in the estimation period. (i) Consider the mode! y = a + x, +u u = pu-1+e, Osp1,

e, are IN(0, 2). By regressing Ay, on Ax,, it is possible to get more efficient estimates of B than by regressing y, on x (j) The Durbin-Watson test is a useless test because it is inapplicable in almost every situation that we encounter in practice.