Suppose that, for n observations y1,,yn, we have specified a multiple linear regression model of the form yi=0+1xi1+2xi2+ei where the ei are uncorrelated errors, with E(eiX)=0 and Var(eiX)=2. Consider the situation where we are interested not in the values of the coefficients 1 and 2 separately, but instead in their difference =12. a) Show that the estimator ^=^1^2 is an umbiased estimator of . That is, show that E(^X)= b) Derive an expression for the variance of this estimator, Var(), in terms of the relevant entries of the variance-covariance matrix Var(), where is the OLS estimator of =(0,1,2)T..