The bias in estimating accessibility inequalities using gravity-based metrics

Abstract

Accessibility measures have been extensively used to explore the outcomes of the spatial distribution of transport, jobs, and population groups in cities. Despite its wide use, identifying the population groups that most benefit from accessibility is not straightforward and different metrics might result in different conclusions. The present work aims to analyze the potential bias of using gravity-based measures based on revealed mobilities to identify job accessibility inequalities. By looking at two large and very different regions, the municipality of São Paulo (SP) and the Greater London Area (GLA), we argue that distance decay functions built from current trip behaviors should be carefully used in evaluations of accessibility inequalities because it may underestimate disparities between socio-occupational groups and also result in a misleading interpretation of impedance factors. Two distinct approaches were implemented to support those claims. We first estimate group-specific distance decay functions, considering only travel time. Secondly, we consider both travel time and travel cost relative to income to estimate zone-specific and city-specific distance decay functions for each one of the study areas. The population of both cases studies was stratified according to the NS-SEC standard to select the highest and the lowest socio-occupational groups and to explore job accessibility inequalities. It was found that higher-level and lower-level socio-occupational groups of SP and GLA present striking differences in terms of travel times and relative travel costs, with SP being more unequal. By applying the distance decay function of the lowest level socio-occupational group to the calculations of the job accessibility of the highest level group, and by adding travel cost to the analysis, we highlight inconsistencies between gravity-based accessibility calculations and theory, as trips taken by different groups can be mistakenly associated with willingness to travel. From a policy perspective, our findings emphasize that accessibility inequalities in large urban centers, especially in the Global South, can be underestimated if revealed mobilities are considered to represent the willingness to travel and by not taking into account the relative cost of travel.