Assessing systematic sources of variation in public transport elasticities: Some comparative warnings

Abstract

There is an extensive and continually growing body of empirical evidence on the sensitivity of potential and actual users of public transport to fare and service levels. The sources of the evidence are disparate in terms of methods, data collection strategy, data paradigms, trip purpose, location, time period, and attribute definition. In this paper, we draw on a data set we have been compiling since 2003 that contains over 1100 elasticity items associated with prices and services of public transport, and car modes. The focus herein is on direct elasticities associated with public transport choice and demand, and the systematic sources of influence on the variations in the mean estimates for fares, in-vehicle time, and headway obtained from 319 studies. The major influences on variations in mean estimates of public transport elasticities are the time of day (peak, all day vs. off-peak), the data paradigm (especially combined SP/RP vs. revealed preference (RP)), whether an average fare or class of tickets is included, the unit of analysis (trips vs. vkm), specific trip purposes, country, and specific-mode (i.e., bus and train) in contrast to the generic class of public transport.

Introduction

The worlds of research and practice rely extensively on published estimates of price and service elasticities to develop predictions of switching behaviour towards or away from public transport. For at least 50 years we have seen an accumulation of empirical estimates of direct and cross fare and service elasticities reported in the literature. There are some classic reviews such as Goodwin (1992) and Oum et al. (1992)¹, which have synthesised many of the better studies undertaken prior to 1990.

Kremers et al. (2002), Nijkamp and Pepping (1998) and Holmgren (2007) have undertaken meta analyses on samples of public transport elasticities to identify systematic sources of variation. They find, in particular, that differences in elasticities can, in part, be explained by the functional form of the model, whether the estimates reflect the short run or long run, the nature of the data structure (e.g., cross-section and time series), location (such as country or city size), and whether data are aggregated or disaggregated.

A particularly interesting aspect of the studies is the evidence of differences obtained using revealed preference (RP) data and that reported in stand alone stated preference (SP) or combined RP/SP data when individual choice or (aggregated) share models are being used. One of the problems we have with the comparison is that the majority of SP studies do not appear to calibrate their models to reproduce base modal shares² and hence the comparison is likely to reflect as much the failure to calibrate the model constants rather than any possible systematic under- or over-estimate of mean elasticity. This is an important point, which we will return to in the empirical analysis, since it is easy for skeptics of SP or SP/RP applications to argue that they under- or over-estimate compared to models estimated with RP data (although who says that it is the correct reference anyway?).

Unlike the meta analysis studies cited above, which studied each class of elasticity separately, we have pooled the data for fare, in-vehicle time and headway direct elasticities, to give us a sample of 319 observations. In some of the previous meta-studies (e.g., Kremers et al., 2002), price elasticities from different modes of transport are pooled, but it is the pooling of price and service elasticities that is new in this study. The advantage of our approach is that we have a significant sample size to add confidence to inference, in contrast to the earlier studies where sample sizes were often as low as 12 data points. Holmgren’s sample sizes varied from 17 to 81. In addition, the focus is on establishing sources of systematic variations in a broad class of direct elasticities, and so the assessment of candidate sources across three key attributes of public transport seems appropriate. To control for possible biases attributable to a sub-class (e.g., fares), we introduce dummy variables for each sub-class, normalizing on one sub-class for identification.

The key purpose of this paper is to suggest points of assistance for policy makers in the decision as to what extent, existing knowledge on behavioural response as captured through direct elasticities, might be used in another context, and what are some lessons we can learn from a meta analysis in guiding the definition of elasticity outputs obtained from new primary data.

The paper is organized as follows. We describe the data set in the next section and the approach we had adopted to establish potential sources of systematic variation. This is followed by the empirical evidence on three key direct elasticities of public transport, namely fares, in-vehicle time, headway. The paper concludes with comments on the evidence and offers three very specific warning signals when selecting elasticities from secondary sources for use in particular contexts, and when designing new studies that collect primary data.