2. Consider the following regression: Yi =0 1Xi ui We believe that X may be measured with error, and the error term is defined as ei = Xi - X i*, where X* is the true value of X, but is unobserved, E[e]=0 and E[Y|X*,X]=E[Y|X*]. a) If we assume Classical Errors in Variables (CEV), that is if we assume Cov(X*,e)=0, show that Cov(X,e)0. (4) b) Under CEV our composite error term (the actual error in our regression) is vi=ui - 1ei. Since CEV implies Cov(X,e)0, then Cov(X,v)0. Explain what this implies about the (un)biasedness of our OLS estimator 1 of the regression above. (2) c) Let Xi= the number of marijuana joints that person i reported to have smoked in the past month. Explain why it is or is not reasonable to assume CEV for this variable X. (2)