Behind the scenes with MBTA data.

In this article we investigate the drop in riders from FY 2008 to FY 2015.

A few weeks ago, we described in this post how passenger “taps” are recorded by the MBTA’s automated fare collection (AFC) system and are adjusted to arrive at the number of unlinked trips the MBTA reports to the National Transit Database (NTD). In this post, we will show how the number of unlinked trips can vary relative to the number of AFC interactions. 

Unlinked trips over time

While developing the charts for the “State of the Green Line” presentation to the Fiscal and Management Control Board, we compared the reported ridership numbers for the Green Line from Fiscal Year 2008 (when we first instituted the CharlieCard system) to Fiscal Year 2015. This view over time showed a loss of over 50,000 average weekday trips from FY08 to FY15. Although some variation in ridership year-to-year is expected, we do not expect a 20% loss of weekday trips in normal variation.


Average Weekday Unlinked Trips on the Green Line

Other sources of data (like customer reports of continued overcrowding on the Green Line) make us doubt that the drop in unlinked trips is due to a sustained drop in riders. As explained previously, unlinked trips count each segment of a linked trip. Fewer unlinked trips do not necessarily mean fewer riders are using the MBTA; it could also mean that people are making fewer transfers.  However, any time we see a large drop in unlinked trips, we investigate further to make sure passengers are not leaving the system.

Why the drop in unlinked trips?

There are two obvious possible reasons for the loss in unlinked trips in FY15: the closure of Government Station and the winter of 2015.

Government Center closed for renovations in March 2014, three quarters of the way through Fiscal Year 2014. In the unlinked trip calculations for the period when the station was closed we “lost” both the station entries at Government Center and the behind-the-gate transfers in that station. Since Government Center is one of the most used stations in the system, and is the only Green-Blue transfer connection, we expect a huge drop in Green Line unlinked trips from FY14 to FY15 that continues into FY16.

The closure of Government Center changed people’s boarding and transferring patterns. So while the number of unlinked trips dropped, the closure likely had a much smaller impact on the number of linked trips, or actual passengers in the system. Some people might have replaced a short Green Line trip with a walk or bike trip, reducing overall ridership. Others likely walked farther to a different downtown station and eliminated a Green Line transfer, reducing unlinked trips, but not linked trips. We will monitor station boardings now that Government Center has reopened to check if they return to pre-closing levels.

The winter of 2015 (also in FY 2015) had an impact on ridership and data collection on the entire system. The MBTA operated three fewer days in FY15 due to system closures, operated through five additional days of lower ridership due to school closures, and ran fewer vehicles due to mechanical difficulties. This decreased ridership temporarily and we are continuing to monitor for long-term ridership impacts. Preliminary data from February 2016 indicates an increase of approximately 20% average weekday ridership systemwide over FY15, a rebound to more typical service usage.

Average Weekday Unlinked Trips with Additional Notes

Fare collection counts over time

In order to help determine if we are losing unlinked trips or actual riders, we looked at the fare collection system counts for the Green Line over the same timeframe.

The interactions with the AFC system, especially on the Green Line, do not reflect the number of trips that people actually take the Green Line. For example, if you tap in at Coolidge Corner and transfer to the Red Line to go to Harvard Square, then your return trip, where you tap back in at Harvard Square, does not register that you also took the Green Line back. Since many Green Line trips involve transfers, the number of AFC tap-ins is much lower than the unlinked trips we report to the NTD.

In addition, there is a large non-interaction factor on the Green Line partially due to people boarding at rear doors at surface stations during rush hour. We correct for this factor when reporting unlinked trips, because these are riders on our system even if they don’t interact with the farebox and aren’t counted in this method.

Assuming generally consistent behavior over time (e.g. roughly the same proportion of riders transferring to other lines, and roughly the same proportion of riders boarding at the rear door, children, and use of flash passes), the AFC counts serve as a good check for change in actual number of customers over time.

Green Line Average Weekday Unlinked Trips with Farebox Taps

Over the same time period (FY08 to FY15), the number of people tapping a CharlieCard, swiping a CharlieTicket, or paying in cash on a Green Line vehicle or at a Green Line station hasn’t varied much except for the drop from FY13 to FY15 that coincides with the closure of Government Center. In this measure, we lost the station entries at Government Center as people walked to other stations to enter the system. While some people walked to other Green Line stations, many people who used to enter at Government Center instead walked to Orange or Blue Line stations; as a result, we “lost” these people from the AFC counts on the Green Line.

Because transfer and non-interaction factors change yearly, the number of unlinked trips is more volatile than the number of AFC interactions. There will be (sometimes dramatic) yearly fluctuations in unlinked trips as a result of this volatility. However, even with these fluctuations, unlinked trips are a better reflection of customer experiences (riders who don’t interact with the farebox or who transfer behind the faregates are still on the train with you).

AFC counts serve as a check in explaining dramatic changes in unlinked trips. Any time that there is a large drop or rise in unlinked trips, we check to see if it’s due to how unlinked trips are calculated or due to a change in actual underlying AFC numbers. This helps us attribute changes to the correct cause.