What Influences Ridership on Public Transit?
And some suggestions for transit operators such as Calgary Transit
When one of your Ride On writers was a little skid, he could not wait to own his own car for reasons provided in the first issue of this newsletter. And for many years after this goal was achieved, he looooved driving on the open road, especially on trips down Highway 401 between Ottawa and Acton, Ontario (a little town located between Guelph and Toronto) late at night, listening to music on the Discman connected to the car’s cigarette lighter. The world was a wide-open place ripe for exploration!
But over time, the joys of driving were diminished by increasing gasoline prices (rising above $1 per litre in the fall of 2005), skyrocketing insurance rates (roughly doubling immediately upon moving to Alberta from Ontario, and doing so again after his first year in Alberta despite a flawless driving record), and trying with mixed results to remain patient with other drivers (don’t get him started). Finally, both writers of Ride On decided that public transit was sufficient for their needs, with the occassional car rental for trips outside Calgary.
But not everyone is like us, so before we can go further on our quest for the optimal public transit system, we need to determine what factors influence the “typical” person to take transit. Therefore, the next section will begin our focus on a recent paper by Boisjoly et al. (2018) from the peer-reviewed journal Transportation Research Part A. Along the way, we will apply the lessons learned from this article to the transit system we currently understand most intimately: Calgary Transit.
Literature Review
Boisjoly et al. (2018) provide a thorough literature review in which they identify two main approaches to studying factors that might influence ridership levels. The first approach is to focus on the individual rider: the “micro” level. In these studies, researchers want to learn how aspects of the individual, such as socio-demographics and personal preferences, can affect transit usage, or how these individuals respond to changes in parameters such as income or the “built environment”. Sometimes, these researchers segment markets to determine key sectors of the population where transit uptake may be more responsive, such as students, recent immigrants, larger family sizes, and the unemployed.
On the other hand, researchers might take a macro approach to understand how larger regions as a whole respond to changes in internal factors (e.g., agency expenditures and provisions), or external factors (e.g., unemployment rates, gasoline prices, or GDP per capita). The study by Boisjoly et al. (2018) leans toward a macro approach in which they delineate between internal and external factors, where internal factors relate entirely to decisions, policies and conditions determined by the transit agencies or municipal governments providing subsidies.
The authors note that there is some debate over whether internal or external factors dominate ridership decisions. Some research shows that internal factors such as service improvements and fare reductions are able to increase ridership, even when ridership is falling for other transit agencies due to expected external factors. Other research, however finds that external factors such as metropolitan population and area, economic vitality and low levels of car access are responsible for most variation in transit ridership, although fares and service levels do have (smaller) impacts.
So which factors have been identified in the literature as significant influences on ridership? In terms of external factors, it has been found that increases in population and employment rates have consistently been associated with higher ridership levels. While one might also expect rising gasoline prices to lead to higher ridership levels, evidence from U.S. studies have been mixed, with results ranging from no significant relationship to a strongly significant relationship. However, research does find strong positive correlations between gasoline prices and public transit ridership levels in Canada, Australia and Europe (including Germany specifically).
As for internal factors, researchers cited by Boisjoly et al. (2018) have found significant relationships between fare reductions and ridership increases, as one might expect by the Law of Demand. Similarly, researchers have found that rising service levels are significantly associated with higher ridership levels, although studies differ in their measurements of service levels. But regardless of measurement, the general consensus is that increased service levels are positively associated with overall ridership.
Similar conclusions are made in other research where ridership growth is primarily associated with increased service frequency, increased amount of service in general, and increases in network coverage. Improvements in service quality have also been viewed as important for increasing transit ridership levels, specifically reliability and dependability of the system. Unfortunately, different agencies measure reliability differently, so it cannot be used in empirical studies.
While there are many studies which examine the link between the above factors and transit ridership, Boisjoly et al. (2018) argue that their study is important primarily for three reasons:
It is a longitudinal analysis, in which they use annual data over a 14-year period (2002-15). They find that very few previous studies were either cross-sectional or longitudinal, and those few studies used data prior to the late-2000s when transit ridership rates reversed from upward trends to downward ones.
It assesses a mix of modes (bus vs. rail) and how differentiating modes relates to ridership. Furthermore, they include new variables related to mobility transformations in cities, including bike-sharing and ride sourcing (e.g., Uber).
Methodological: the authors use a multilevel approach to control for clustering of data within agencies.
Inclusion of Canadian cities in the analysis, particularly Toronto, Montreal, and Vancouver, whereas previous studies did not include Canadian cities at all.
Data
Boisjoly et al. (2018) chose to include data for metropolitan areas with at least 1.5 million people in 2015, and which operate at least two modes of transport (bus, streetcar, light rail, and/or heavy rail). When a city had more than one transit agency, the authors included the one with largest fleet size. In the end, they used data for 25 metropolitan areas in Canada and the U.S. from 2002-15, with the note that transit agencies in the study did not necessarily cover their entire metropolitan areas.
As for specific variables used in their analyses, they included ridership (defined as the number of unlinked passenger trips), vehicle revenue kilometres (VRK, defined as the number of kilometres traveled by vehicles in revenue service), rail VRK, bus VRK, average transit fares, Census Metropolitan Area (CMA) population, CMA geographic area, the percentage of households without a car, unemployment rate, GDP per capita, average retail gasoline price, and the total length of highways within the CMA. The authors also included dummy variables to measure the presence of a private bus operator, the presence of ride-sourcing services, and the presence of bike-sharing.
Results
Boisjoly et al. (2018) identify several very important results. First, they find that a 10% increase in VRK is associated with an 8.27% increase in ridership, all else held constant — by far the largest contributor to ridership. Conversely, a 10% rise in fares is linked to a 2.19% decrease in ridership. These results are quite interesting because while the exact numbers will obviously not apply to any specific transit system, qualitatively they suggest that potential transit users value a more extensive transit system more than they do lower fares — although fares are still an important determinant of ridership. This makes sense intuitively (at least for your Ride On writers) because we personally have not had a problem with the cost of Calgary Transit passes, but we have experienced frustration with the time and effort necessary to travel somewhere that is not in the downtown core or otherwise within five kilometres of our home.
The second-largest contributor to ridership in this study is car ownership: the greater the number of households without a car, the more transit trips are taken. This makes sense because without a car, transit is a much more attractive option. However, this poses a problem for transit operators in Alberta which is known for having a strong car culture. However, the authors cite research showing that car ownership falls as car-sharing becomes more prominent, which leads us to the next result of this study.
Specifically, while not statistically significant, external factors such as the presence of ride-sourcing services (e.g., Uber) and bike sharing are associated with higher transit ridership, contradicting expert hypotheses. The authors argue these results seem to suggest that these services are complementary to transit — for example, using transit to get somewhere and then ride-sharing for the return trip, similar to the relationship between cabs and public transit. Again, your Ride On writers see this argument as intuitively logical, since we are quite willing to take public transit during the day to a destination, but we have a much greater incentive to use an alternative to get home at the end of the night when we are tired and the quality of transit services for our purposes has diminished greatly (if not entirely shut down for the night).
Another explanation provided by the authors in terms of bike-sharing is that these services tend to be designed with transit in mind, such as placing docking stations near transit stations and linking bus passes to bike passes. Therefore, people could take a bike to and from transit stations, but use public transit otherwise.
Moving on, Boisjoly et al. (2018) find that gasoline prices are positively (and significantly) related to ridership, which makes sense economically since it means the opportunity costs of driving have risen. However, the magnitude of this relationship is smaller than the corresponding magnitudes for ride-sourcing and bike-sharing, implying it is more important for local governments and transit agencies to promote these complementary services than it is to raise gasoline prices. Again, this makes economic sense since demand for gasoline is likely quite inelastic in the short run, meaning higher gasoline prices lead to very little reductions in gasoline purchases.
As for the population of the CMA, it has a significant and positive relationship with transit ridership: Boisjoly et al. (2018) find that a 10% increase in population coincides with a 3.39% increase in ridership, all else held equal. Inversely, a 10% rise in the geographic size of the CMA is associated with a 2.80% decrease in ridership. Both of these results imply greater population density (less urban sprawl) is associated with greater transit ridership — possibly because the farther one has to travel on public transit, the longer it will take and therefore the less likely they will take it.
The authors also find that the bus VRK regression coefficient is five times greater than the rail VRK coefficient. Again, we cannot assume that the exact numbers will hold anywhere, but qualitatively this implies a greater association between bus operations and ridership. Therefore, an increase in rail operations is not sufficient to sustain an increase in ridership, and that is important for cities like Calgary and Edmonton that are expanding their LRT systems — while it does not mean that Calgary’s “Green Line” (for example) is a bad idea, it does suggest that it is not the be-all-end-all for increasing ridership on public transit. Buses are still a very important variable in the equation.
On that note, according to Boisjoly et al. (2018), the results suggest decreases in bus operations in recent years are associated with falls in ridership. According to the authors, possible reasons for this finding are as follows:
Unlike cities in Europe, rail service in the U.S. is typically very focused on specific areas or needs, serving mainly central areas (downtown) and suburban commuting trips. So most people depend on buses for their daily lives. In most cases, the bus system partly designed to complement rail network. Thus, a decline in bus operations might be associated with a decline in service frequency and coverage of bus feeder services, reducing the quality of integrated network.
It might be easier to increase service coverage with bus services, thus attracting new users. In contrast, increases in rail service are more easily achieved with increased frequency, which might not yield so many new riders.
The authors point out that some people argue rail service is better than bus service, but studies have shown that high performance bus service with similar frequencies and service attributes as rail (e.g., exclusive right-of-way) have similar ridership attraction. With adequate investments in increasing service frequency and improving the efficiency of bus operations, such as transit priority and dedicated bus right-of-ways, bus service can effectively contribute to transit ridership.
Overall Takeaways for Calgary Transit
Based on the above results, if Calgary Transit and Edmonton Transit Service (ETS), for example, want to encourage more transit use, they need to encourage people with strong preferences for their cars to stop driving themselves around so often; for that to happen, the results found by Boisjoly et al. (2018) suggest they build more extensive transit systems and (possibly, but maybe not as importantly) lower fares. As argued above, it seems potential transit riders might far prefer better service to lower fares, so it is possible that transit operators could cover the costs of better service with higher fares, and the positive effects of the former would outweigh the negative effects of the latter.
Furthermore, they should perhaps devote relatively more attention to buses than trains, although that does not imply trains should be ignored entirely — so don’t come at us, Green Line supporters! :-)
Thank-you for reading to the end of this newsletter, and we look forward to writing for you again next week. Please let us know your experiences with public transit in the comment section below, and also let us know what you want to see in future issues of this newsletter. Finally, please consider a paid subscription for this newsletter, as it helps us to do more and better research for it.
ETS has dropped many neighbourhood routes in favour of express routes along main corridors. They argue that dial a bus services are available but I have heard that these don't work well. They take a lot more time to access and are costly. In my own case I have found information hard to come by on the new system. In order to figure out which buses will take you where you want to go you have to know the number given to bus stops. This discourages those who don't regularly use public transit. Others are discouraged by safety issues common to both Edmonton and Calgary. On a positive note the pay as you go cards are convenient though I still don't understand why the ticket dispensers can't be designed to take charge cards and debit.