Suppose we have the standard multiple regression problem of y = XB +e and B = (XTX)-1XTy. (a) Assuming E(c) = 0 and Cov(e) = 021, verify formula that Var ( 3) = 02 ( XT X ) -1 (b) Assuming E(c) = 0 and Cov(e) = E where E is an arbitrary covariance matrix, derive the appropriate formula for Cov(B). (c) Suppose E(ci) = / for all i = 1, 2, ..., n. Determine expected values of Bo, B1, . .., Bp and verify that B1,. .., Bp are all still unbiased, but that So is biased