Estimating Simultaneous Equations Models by a Simulation Technique.

For simultaneous equations models, estimates from ordinary least squares (OLS) methods are biased and even inconsistent and those from two-stage least squares (2SLS) methods are, though consistent, still inadequate because of finite sample biases. A new simulation technique developed here produces better estimates by compensating for the simultaneous bias in those conventional estimation methods. In the simulation (SIM) estimation technique, such biases are directly compensated in situ through a synthetic use of simulations. SIM is demonstrated to outperform the existing techniques through a series of Monte Carlo experiments. Klein's macroeconomic model is used to further illustrate the practical application of SIM. The results are compared with those from OLS and 2SLS to elaborate the attractive performance of SIM. Citation Copyright 1995 by Kluwer Academic Publishers.