To apply MLE we need error distribution assumption

Always

Only for small samples

Only if we want to match OLS coefficients

Never

2) Is it true that OLS loses its attractive properties such BLUE when applied to AR(1) model?

Yes

No, because under some conditions such as stationarity and weak exogeneity OLS will still be BLUE

No, because we can put the lagged Y into the same stacked matrix X and proceed with OLS

No, because in time series disturbance of today never depend on observations of yesterday

3) MLE of error variance of a linear regression model is unbiased for finite samples

Agree

Disagree

4) Wald test can be used for hypotheses testing

only in large samples

both in large and small samples

only in small samples

only asymptotically, i.e. when the sample size in infinite

5) Feasible GLS can only be used when the covariance matrix of errors is known

Yes

No

6) Adding irrelevant variables introduces bias into the estimated coefficients

Yes

No