We consider the following regression model: YIX=(X1,...,X5)=1+ X1+ exp(X2)E, where X.- '12? mo, 1), i = 1, . . . ,5. are independent of 5 ~ N(0, 1). In this regression, we are interested in E(Y|X) alone. Then should the ordinary least square estimator be useful in this situation? If yes, explain why. If no, explain way, too