Introduction
As MBTA ridership continues to recover from its drop in 2020, certain areas of the region are recovering at different rates. This blog post examines spatial patterns in the recovery of ridership across the MBTA’s bus and rapid transit systems. We measured ridership recovery by calculating ridership in the fall of 2021 as a percent of ridership in the fall of 2019, and ridership was measured as the number of boardings at bus stops and rapid transit (subway) stations. Viewing ridership recovery at the stop level tells a more nuanced story than viewing ridership trends across the system as a whole or at the level of individual subway lines or bus routes. This analysis supports previous non-spatial analyses which suggest that bus riders have largely driven ridership recovery. The spatial analysis presented here adds to the previous analyses by revealing areas that have seen particularly high levels of recovery, which include southern and western portions of Boston, as well as Chelsea, Everett, Revere, and Lynn.
Data
The analysis used data from two of the MBTA’s main passenger data sources: Automated Passenger Counters (APCs) and the Automated Fare Collection (AFC) system. Composite day data from the APCs provided the number of boardings at each bus stop for typical service days. Data from the AFC system supplied the number of boardings at subway stations, which are mostly derived from CharlieCard taps and CharlieTicket validations at station fare gates. Both the subway and bus boardings were separated into three day types, weekday, Saturday and Sunday, as each of those day types have different levels of service. We used ridership data from the fall rating (September through December) to compare the two years because, with schools in session, the fall can be considered the typical T-riding experience. Some limitations of the data are that the AFC data does not include boardings on the T’s streetcar services on the Green Line and the Red Line and that boardings during holidays were removed from the analysis.
Methodology
The first step of the methodology included joining the 2019 and 2021 boarding data to each GTFS bus stop in ESRI GIS software. The analysis included ridership data for over 6,000 stops across the region, which presented a visualization challenge. You can imagine the difficulty of trying to discern visual patterns among thousands of points on a map. To smooth variation and make spatial patterns more visually legible, the stops were aggregated to a grid of identically sized hexagons, each of which would take roughly five minutes to travel directly from the center to an edge.
Consequently, the maps depict changes in ridership based on the sum of boardings at any bus or subway station within each hexagon. After boardings were summed for each year in each hexagon, the number of boardings in 2021 was divided by the number of boardings in 2019. The final step was to remove any hexagon whose 2019 total boardings fell below the 40th percentile in that year for each day type. Those hexes were removed because a modest increase in boardings in 2021 from very few boardings in 2019 created substantial percent changes. At a lower scale of analysis, e.g., looking at changes in ridership on a single bus route, that would be important information. However, at the scale of the greater Boston area, those kinds of fluctuations are insignificant.
The Geography of Ridership Recovery
Changes in weekday ridership show the clearest spatial patterns. Figure 1 shows the ridership recovery for weekday service. The areas of highest recovery (darker colors) do not consistently coincide with subway stations (small white dots), with the exception of the Blue Line. Instead, areas near key bus routes in Chelsea, Revere, Everett, and the neighborhoods from Dorchester to Uphams Corner show clusters of high recovery. Additional clusters of recovery occur around Oak Square and Allston as well as in Lynn.
Figure 1: Ridership recovery for Weekday service, comparing Fall 2021 to Fall 2019
Figures 2 and 3 show the recovery for weekend service. While less clustering is apparent, the pattern of higher ridership occurring near bus routes as opposed subway lines holds. For Saturday service, some clustering is also evident in Oak Square, Allston, and the towns northeast of Boston. One notable difference between the weekday and weekend recovery is that the magnitude of recovery is much higher on the weekends, as seen in the classifications in the map legends.
Recovery Patterns in Context
As noted in the introduction, previous analysis has shown that bus ridership retained a higher share of its pre-pandemic ridership levels than the subway lines, with the exception of the Blue Line. This analysis augments the non-spatial investigations by revealing the varied landscape of ridership recovery. Each of the above maps show areas in the system where boardings have met or surpassed the numbers seen in 2019, a phenomenon not detectable at the level of the route or line. A possible explanation for the preponderance of bus riders in the higher recovery areas could be that bus riders are more likely to have jobs that require in-person work, as opposed to teleworking. A next step in this analysis could be to layer in demographic and income data to identify which socioeconomic factors correlate with individuals’ decisions to resume riding the T.
