This paper discusses a tool for optimization of econometric models based on genetic algorithms. Due to the increasing popularity of semi–parametric estimators, researchers often have to optimize non–differentiable, non–smooth criterion functions that cannot be solved with conventional gradient methods. Genetic algorithms constitute an alternative opportunity for optimization. So far, however, applications are not wide–spread, as there is no software readily available yet. This paper overcomes this limitation by describing a genetic algorithm implemented in STATA that can easily be dynamically linked to any criterion function to be optimized. We describe the algorithm, provide two example applications, and provide a step–by–step guide on how to use the software.
