ECON 710 Final: ECON 710 UW Madison Final Exam 2003a
Document Summary
Answer key: the parameter is the m 1 , so the model is overidenti ed. The e cient estimator is e cient i + ei where zi = q0xi, and then we know that the. We can rewrite the equation as yi = z0 two-step e cient gmm estimator takes the form. = q0x 0x 1x 0xq 1 q0x 0x 1x 0y where xix0 i e2 i. X 0y: the e cient estimator for is ls, = (x 0x) 1 x 0y and that for is = 1/ 2. The natural test statistic for hypotheses on is the t-ratio tn = /s where i! x 0x 1 s = qh 0. The percentile-t interval is constructed using the quantiles of t n , the bootstrap version of tn. We simulate samples of size n by sampling iid (with replacement) from the observed sample {yi, xi : i = 1, , n}.
