Measurement error and functional form: implications for welfare estimates

Abstract

This paper describes a series of Monte Carlo simulations which investigate the accuracy of final estimates generated from the travel cost model (TCM). The results are of interest to all applied researchers who: (a) cannot select functional form on theoretical grounds; and (b) wish to use Ordinary Least Squares (OLS) estimates for further calculations. This is because the simulations indicate that certain types of measurement error decrease the probability that researchers will be able to determine the correct functional form of the estimating equation on the basis of goodness of fit statistics. While this does not necessarily mean that slope (partial coefficient) estimates will be biased, it can cause substantial and unpredictable bias in welfare estimates.