The transferability of logit-based share elasticities for evaluating congestion pricing

The sensitivity of travel demand to changes in the generalized cost of travel can be tailored, through disaggregate mode choice models, to reflect specific trip conditions. The logit mode split function can provide share elasticities that vary not only with tripmaker characteristics, but also with travel conditions encountered along the individual route. Frequently however, average estimates of elasticities are used to predict aggregate or market wide changes in mode share which may result from congestion surcharges. Grouping individual elasticities into market averages is mandated by lack of time and money, or insufficient detail in the travel data. To the extent that logit-based elasticities reflect wide variations across tripmaker and trip characteristics, group averaging can introduce significant distortions in obtaining an accurate market response to travel cost modifications. In the absence of transferability, a reliable evaluation of congestion pricing strategies would be difficult to obtain. The main objective of this paper is to compare a cross-sectional sample of point elasticities obtained from a binary logit function against an overall average and to assess the nature of any variability in terms of selected contextual travel variables.