The value of time and forecasting of flowsin freight transportation

This paper gives an overview of the current state of research and present own results in some areas regarding cost benefit analysis (CBA) for road infrastructure investments. It deals with the robustness of the estimated average value of time (VOT) currently used in Swedish infrastructure appraisals and also analyses the estimator in the case of different heteroskedastic errors and functional forms. It finallly consider forecasts of road freight flows in cases where information are scarce and may lack precision. Critical questions are how heterogeneity and non-linearities influences forecasts in the freight market. In detail it consdier the following: The impact of VOT for road freight when the origins of transports are taken into account. VOT is found to be dependent on combinations of region, transported distance, industry bransch and if transports are owned or hired. However, available data does not allow for significant values in each category. Altogether the study indicates a VOT spanning from 0 to 732 SEK, which should be compared with the average value of 80 SEK used today. The results also indicate pair-wise differences in VOT between short and long as well as hired and internal transports. The ownership condition is found to have a significant impact on VOT, although current data not gives significant differences between each of the four ownership/distance categories. It is found that further studies should focus on the hypothesis that VOT for short internal freight flows are significantly higher than the average VOT. The traditional logit model is compared with the semi-parametric weighted average density (WAD) estimator. It is found that the performance for the WAD estimator in terms of bias and mean square error is similar to the logit ML estimator for spherical errors in a latent variable specification. Methods for prediction of road freight flows are also investigated. Three traditional gravity model specifications (OLS, NLS, and Poisson regression) are compared with a neural network specification. On a data set of Norwegian inter-regional freight flows it is found that the Poisson model performs best in terms of root mean square error (RMSE) but also that the size of predicted flows is dependent on the method chosen to evaluate available estimation methods. Finally, we integrate freight flow prediction and estimation of VOT in one analysis. Logit models and neural networks with linear and non-linear profit functions are compared. The study indicates that the average VOT may decrease when prediction improves as models are given more non-linear specifications.

[1]  Lung-fei Lee,et al.  Specification error in multinomial logit models : Analysis of the omitted variable bias , 1982 .

[2]  Börje Johansson,et al.  Affinities and frictions of trade networks , 1994 .

[3]  Börje Johansson,et al.  Affinities and network attributes in Sweden's trade with Europe , 1993 .

[4]  H. Ichimura,et al.  SEMIPARAMETRIC LEAST SQUARES (SLS) AND WEIGHTED SLS ESTIMATION OF SINGLE-INDEX MODELS , 1993 .

[5]  James L. Powell,et al.  Estimation of semiparametric models , 1994 .

[6]  Terry L. Friesz,et al.  Integration of Freight Network and Computable General Equilibrium Models , 1998 .

[7]  Clifford Winston,et al.  The demand for freight transportation: models and applications , 1983 .

[8]  Ziona Austrian,et al.  Freight transportation demand: A survey of recent econometric studies , 1989 .

[9]  C. Winston,et al.  The demand for transportation: models and applications , 1998 .

[10]  I Hussain.  BENEFITS OF TRANSPORT INFRASTRUCTURE INVESTMENTS. A SPATIAL COMPUTABLE GENERAL EQUILIBRIUM APPROACH , 1996 .

[11]  Lori Tavasszy,et al.  Modelling European freight transport flows. , 1996 .

[12]  J. Berkson.  Application of the Logistic Function to Bio-Assay , 1944 .

[13]  Tony E. Smith,et al.  Gravity Models of Spatial Interaction Behavior , 1995 .

[14]  A. Anas.  Discrete choice theory, information theory and the multinomial logit and gravity models , 1983 .

[15]  L. Mattsson.  Equivalence between welfare and entropy approaches to residential location , 1984 .

[16]  Thomas M. Stoker,et al.  Optimal bandwidth choice for density-weighted averages , 1996 .

[17]  Kurt Hornik,et al.  Multilayer feedforward networks are universal approximators , 1989, Neural Networks.

[18]  Chuan-Zhong Li,et al.  Semiparametric Estimation of the Binary Choice Model for Contingent Valuation , 1996 .

[19]  T. Amemiya.  QUALITATIVE RESPONSE MODELS: A SURVEY , 1981 .

[20]  Per Johansson,et al.  Weighted Derivative Estimation of Quantal Response Models: Simulations and Applications to Choice of Truck Freight Carrier , 2000, Comput. Stat..

[21]  Joseph N. Prashker,et al.  Discrete choice analysis: Theory and application to travel demand: Moshe Ben-Akiva and Steven R. Lerman. MIT Press, Cambridge, MA, 1985. 390 pp. + xx. $32.50 , 1988 .

[22]  W. Michael Hanemann,et al.  Welfare Evaluations in Contingent Valuation Experiments with Discrete Responses , 1984 .

[23]  James L. Powell,et al.  Semiparametric Estimation of Selection Models: Some Empirical Results , 1990 .

[24]  Thomas M. Stoker,et al.  Semiparametric Estimation of Index Coefficients , 1989 .

[25]  Charles F. Manski,et al.  Nonparametric and semiparametric approaches to discrete response analysis , 1993 .

[26]  K. Haynes,et al.  Gravity and Spatial Interaction Models , 1985 .

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

[28]  Alan Wilson,et al.  A statistical theory of spatial distribution models , 1967 .

[29]  J. Horowitz.  SEMIPARAMETRIC ESTIMATION OF A WORK-TRIP MODE-CHOICE MODEL / , 1993 .

[30]  Zvi Griliches,et al.  Specification Error in Probit Models , 1985 .

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

[32]  N. F. Stewart,et al.  The Gravity Model in Transportation Analysis - Theory and Extensions , 1990 .

[33]  Joel L. Horowitz,et al.  2 Semiparametric and nonparametric estimation of quantal response models , 1993 .

[34]  Erik Bergkvist,et al.  Regional valuation of infrastructure improvements. The case of Swedish road freight , 1998 .

[35]  Mark Dougherty,et al.  A REVIEW OF NEURAL NETWORKS APPLIED TO TRANSPORT , 1995 .

[36]  F. Windmeijer,et al.  R-Squared Measures for Count Data Regression Models With Applications to Health-Care Utilization , 1996 .

[37]  Morton E. O'Kelly,et al.  Spatial Interaction Models:Formulations and Applications , 1988 .

[38]  G de Jong.  FREIGHT AND COACH VOT STUDIES , 1996 .

[39]  I. C. Ogwude.  THE VALUE OF TRANSIT TIME IN INDUSTRIAL FREIGHT TRANSPORTATION IN NIGERIA , 1993 .

[40]  D. E. Rumelhart,et al.  chapter Parallel Distributed Processing, Exploration in the Microstructure of Cognition , 1986 .

[41]  H. Mohring,et al.  MAXIMIZING, MEASURING, AND NOT DOUBLE COUNTING TRANSPORTATION-IMPROVEMENT BENEFITS: A PRIMER ON CLOSED- AND OPEN-ECONOMY COST-BENEFIT ANALYSIS. IN: URBAN TRANSPORT , 1993 .

[42]  Erik Bergkvist,et al.  Estimation of gravity models by OLS estimation, NLS estimation, Poisson, and Neural Network specifications , 1997 .

[43]  P. Ruud.  Sufficient Conditions for the Consistency of Maximum Likelihood Estimation Despite Misspecifications of Distribution in Multinomial Discrete Choice Models , 1983 .

[44]  Halbert White,et al.  Artificial neural networks: an econometric perspective ∗ , 1994 .