How Much Do Starting Values Really Matter? An Empirical Comparison of Genetic Algorithm and Traditional Approaches

This research evaluates the impact of using different starting conditions in estimating meat demand systems. Results suggest that as the econometric task becomes increasingly nonlinear, specification of starting conditions becomes increasingly important. This work demonstrates implications of failing to use the best available starting value conditions and how these implications vary with the complexity of the underlying econometric model of interest. Furthermore, this piece proposes a universal approach to be used by all applied econometric practioners to developing appropriate starting values for use in subsequent model estimation.