This paper discusses a tool for optimization of econometric models based on genetic algorithms. First, we briefly describe the concept of this optimization technique. Then, we explain the design of a specifically developed algorithm and apply it to a difficult econometric problem, the semiparametric estimation of a censored regression model. We carry out some Monte Carlo simulations and compare the genetic algorithm with another technique, the iterative linear programming algorithm, to run the censored least absolute deviation estimator. It turns out that both algorithms lead to similar results in this case, but that the proposed method is computationally more stable than its competitor.
Genetic Algorithm Semiparametrics Monte Carlo Simulation