Hessian and approximated Hessian matrices in maximum likelihood estimation: a Monte Carlo study

Full information maximum likelihood estimation of econometric models, linear and nonlinear in variables, is performed by means of two gradient algorithms, using either the Hessian matrix or a computationally simpler approximation. In the first part of the paper, the behavior of the two methods in getting the optimum is investigated with Monte Carlo experimentation on some models of small and medium size. In the second part of the paper, the behavior of the two matrices in producing estimates of the asymptotic covariance matrix of coefficients is analyzed and, again. experimented with Monte Carlo on the same models. Some systematic differences are evidenced.