Resources An online introduction to Gauss. Gauss 5.0 Quick Start Guide Gauss 5.0 User's Guide Gauss 5.0 Language Reference Example: Poisson The Gauss program The Gauss data file Example: Probit The Gauss program The Gauss data file Example: GMM The Gauss program The Gauss data file The Gauss program: 4 moments The Gauss output: 4 moments Extra Documentation MAXLIK documentation sqpSolve documentation

The GMM "Bug" The GMM estimation I originally constructed contained two offsetting "bugs" which resulted in the solver providing the correct estimation values, only one of which ended up in the code provided to class. The code as originally written reformulated jwj with each iteration of the optimization subroutine. Formulating the inverse outside the loop and letting it remain static throughout the optimization produces answers very close to those presented in Green 4th edition. However, in Green 3rd and 5th editions, the same example generated with the same data produces a very different result. As you may have surmised by now, the result in the 3rd and 5th editions is almost identical to the result produced by reconstituting jwj within the optimization loop. So it appears that Green himself has performed the optimization both ways and presented both regression results in various editions of the textbook. In addition to the discussion that was requested in the original assignment, please discuss the difference between the two methods and the results they provide. For reference, the differences (in diff format) between the original and updated files are as follows: 15c15 < OUTPUT FILE = ./gmm.out RESET; --- > OUTPUT FILE = ./gmmv2.out RESET; 40a41,44 > j = DIAGRV(ZEROS(2,2),lambda|lambda^2); > > jwj = inv(j*w*j); > 86c90 < LOCAL p, lambda, x1, x2, x3, x4, mu1, mu2, mu3, mu4, mu, jwj, ret; --- > LOCAL p, lambda, x1, x2, x3, x4, mu1, mu2, mu3, mu4, mu, ret; 98c102 < jwj = inv(j*w*j); --- > @ jwj = inv(j*w*j); @