Mixed Logit Estimation of the Value of Travel Time

In this paper we use mixed logit specifications to allow parameters to vary in the population when estimating the value of time for long-distance car travel. Our main conclusion is that the estimated value of time is very sensitive to how the model is specified: we find that it is significantly lower when the coefficients are assumed to be normally distributed in the population, as compared to the traditional case when they are treated as fixed. In our most richly parameterised model, we find a median value of time of 57 SEK per hour, with the major part of the mass of the value of time distribution closely centred around the median value. The corresponding figure when the parameters are treated as fixed is 89 SEK per hour. Furthermore, our finding that the ratio of coefficients in a mixed logit specification differ significantly from the ones in a traditional logit specification is contrary to the results obtained by Brownstone & Train (1996) and Train (1997). Whether the ratios will differ or not depends on the model and the data generating process at hand.

[1]  L. Kotlikoff,et al.  Health, Children, and Elderly Living Arrangements: a Multiperiod-Multinomial Probit Model with Unobserved Heterogeneity and Autocorrelated Errors , 1990 .

[2]  L Hultkrantz,et al.  THE VALUE OF TRAVEL TIME CHANGES IN A RANDOM NON-LINEAR UTILITY MODEL , 1997 .

[3]  P. Zarembka.  Frontiers in econometrics , 1973 .

[4]  K. Train.  Recreation Demand Models with Taste Differences Over People , 1998 .

[5]  Michael Keane,et al.  A Computationally Practical Simulation Estimator for Panel Data , 1994 .

[6]  K. Train,et al.  Forecasting new product penetration with flexible substitution patterns , 1998 .

[7]  V. Hajivassiliou,et al.  Smooth unbiased multivariate probability simulators for maximum likelihood estimation of limited dependent variable models , 1993 .

[8]  Christian Gourieroux,et al.  Simulation-based econometric methods , 1996 .

[9]  Steven Stern,et al.  Simulation-based estimation , 1997 .

[10]  Paul A. Ruud,et al.  Handbook of Econometrics: Classical Estimation Methods for LDV Models Using Simulation , 1993 .

[11]  Paul A. Ruud,et al.  Simulation of multivariate normal rectangle probabilities and their derivatives theoretical and computational results , 1996 .

[12]  D. McFadden,et al.  MIXED MNL MODELS FOR DISCRETE RESPONSE , 2000 .

[13]  Kenneth Train,et al.  THE GOODS/LEISURE TRADEOFF AND DISAGGREGATE WORK TRIP MODE CHOICE MODELS. IN: URBAN TRANSPORT , 1978 .

[14]  S Algers,et al.  FURTHER RESEARCH ON THE NATIONAL SWEDISH VALUE OF TIME STUDY , 1999 .

[15]  Beth F. Ingram,et al.  Estimation By Simulation , 1987 .

[16]  Daniel McFadden,et al.  Modelling the Choice of Residential Location , 1977 .

[17]  Moshe Ben-Akiva,et al.  Estimation of travel choice models with randomly distributed values of time , 1992 .

[18]  K. Train,et al.  Mixed Logit with Repeated Choices: Households' Choices of Appliance Efficiency Level , 1998, Review of Economics and Statistics.

[19]  J. E. Glynn,et al.  Numerical Recipes: The Art of Scientific Computing , 1989 .

[20]  William H. Press,et al.  The Art of Scientific Computing Second Edition , 1998 .

[21]  D. McFadden.  Conditional logit analysis of qualitative choice behavior , 1972 .