This is a Guest Post by Stu Donovan
Imagine you are a policy-maker in a city, such as Auckland, which faces growing travel demands and persistent socioeconomic inequality: How might you apply your transport budget to tackle both problems at once?
In this recent paper published in the journal “Transport Policy”, my co-author and I consider this question. From a policy perspective, major transport investments would ideally be efficient—in that their benefits exceed their costs— and equitable—in that they disproportionately favour the less well-off.
To answer the second question, several studies propose accessibility-related “justice tests”. Our research extends this literature in two ways: First, we test whether a project is “just” using the correlation between the change in accessibility and socioeconomic outcomes. Second, we apply our test to a case study of Auckland’s “City Rail Link” (CRL).
Data does indeed show that deprivation is concentrated in certain parts of Auckland, as illustrated below (NB: Higher percentile indicates higher deprivation).
Figure 1: New Zealand Deprivation percentile scores (NZDep).
In our paper, we model the effects of the CRL using Auckland’s “General Transit Feed Specification” (GTFS feed), as shown below. A GTFS feed consists of standardised text files containing information on the PT network, such as stop locations, route alignments, and schedules. GTFS feeds are now available for PT networks in more than 500 cities worldwide, providing a rich and standardised way to represent PT networks.
Figure 2: Auckland’s GTFS feed. Bus, rail, and ferry routes in green, red, and blue respectively; black circles denote stops and stations. (Auckland Transport).
We edited Auckland’s GTFS feed to model the effects of the CRL. We complement the GTFS feed with pedestrian networks sourced from Open Street Maps (OSM), from which we calculate walking times to access PT services. Finally, we developed a routing tool in Python to calculate PT travel-times and accessibility across Auckland, accounting for time spent walking, waiting, and in-vehicle.
Using this tool, we can then calculate the change in accessibility to jobs across Auckland when assuming a 45-minute maximum travel-time, for example. We find the CRL increases accessibility in the west, east, and south of Auckland, which broadly aligns with the areas of high deprivation illustrated above.
One might expect, then, a positive correlation between accessibility and socioeconomic deprivation, and so it proves: In almost every scenario that we test, we find a positive correlation coefficient between the change in accessibility caused by the CRL and prevailing socioeconomic deprivation. We conclude the CRL is a relatively “just” project.
To me, this was a somewhat fascinating outcome. In the debate about the CRL, some claimed the benefits would fall to the city centre. While it is true the CRL is situated in the city centre, we find that the bulk of the benefits fall to Auckland’s most deprived suburbs. In this case, what is good for the city is good for suburbs.
Figure 3: Change in accessibility from CRL (45-minute travel-time).
We are the first to admit that our proposed test is not ideal.
Using the correlation coefficient as a criterion for justice tests may be unreliable, for example, in the presence of outliers and/or non-linear associations. In this specific case study, scatter plots of the underlying data points indicate neither of these two statistical issues were present. Other methodological issues, such as dynamic effects where households change location in response to changes in accessibility, are also omitted.
Some may also quibble whether a single justice test can help understand the complex and diverse effects of PT investments. We hedge our bets on this front: On the one hand, the literature identifies a gap between policy and practice in relation to justice tests. In this context, a single, simple, and stable justice test may be better than nothing.
On the other hand, we do not believe there exists one justice test to “rule them all”; careful application and interpretation is, as always, needed. And potential transport investments that do not pass our proposed test may nonetheless be worthy of funding. Determining that a project is just, using this test or any measure, does not imply that it is worthwhile; the latter is a distinct, equally important, question.
Acknowledgements: This paper was co-authored with Saeid Adli.
Disclosure: The author works as a transport consultant for this firm