Welfare Estimation Using Aggregate and Individual-Observations Models: A Comparison Using Monte Carlo Techniques

Due to the weak behavioral foundations of aggregate demand models, zonal travel cost models have been largely abandoned in favor of models based on individual observations. However, sample selection difficulties in individual-observation models often require the use of distribution-sensitive limited-dependent variables estimators. In this paper I use Monte-Carlo simulations to investigate whether the bias from aggregation is worse than possible bias from these narrowly specified estimators. Somewhat surprisingly, the results indicate that zonal models often outperform the individual-observation models, especially when using an aggregate model that incorporates intrazonal variance of the explanatory variables.