Differences in the Travel Time Value in a Binomial Logit Model with Different Functional Forms and Estimation Process?

Abstract

This paper presents an analysis of which functional form of Travel Time Value is appropriate to the estimation with binomial logit random factors model. In order to do this, it will make an estimation methods overview, describing some structured functional forms as Tuong and Hensher (1985) for Becker, DeSerpa and a DeSerpa with a Taylor expansion. The estimation methods used are the ones with heteroscedasticity presented in Train (2003) with maximum likelihood estimation and Bayesian simulated procedures. We make a comparison between the binomial logit model and their changes with random factors models. The study indicates advantages and disadvantages of estimation methods and functional forms adopted. An analysis of combining the estimation method with functional form suggests that the Becker and DeSerpa model estimated by MVSL is a good combination to measure TTV and others economic parameter as wiliness-to-pay and elasticicity and to predict events.