Behind the scenes with MBTA data.

Conclusion that ridership behaviors may have changed regarding gov't centers closing, but will eventually return.

Government Center Station was closed for just over two years, from March 2014-March 2016, for extensive renovations. During this time, regular users of the station had to find other ways to reach their destinations. We were curious what these people did while the station was closed, so we dug into the data to see what clues it could give us.

The most readily available data are validations from the MBTA’s AFC database. These record every “tap” at a faregate for a specific station. We pulled these data by day, by station for Government Center and the stations nearby (that people might reasonably divert to): Bowdoin, Downtown Crossing, Haymarket, Park Street, and State. A more detailed look, taking into account transfer patterns behind the gates, could be done using the MBTA’s ODX data.

Caveats with these data are that they only show station entries, and only show card and ticket interactions with the gate. We generally assume entries are close to equal with exits on the course of a day (especially a weekday), as people tend to take round trips. Some entries are missed from people not interacting with the gate (for example, children being allowed through by an agent), but we believe these to be a low amount, and they should be randomly distributed between the stations we are looking at.

Station entries around Government Center in Spring 2014

This chart shows the 7-day moving average daily entries at each of the above stations for the period before and after the initial closing of Government Center. Obviously, entries at Government Center itself dropped from around 7500-8000 to zero, and station entries increased at the other stations, most significantly at Park Street Station and State Station. Some of the increase at these stations may be due to overall seasonal variation. At these downtown stations, entries generally tend to increase during the summer months as tourists visit the attractions near these stations. To get an idea of these, let’s take a look at the same time period in 2015:

Station entries around Government Center, Spring 2015

In looking at this data, unfortunately, we run into another caveat: the winter of 2015, where the system was closed for several days and entries were low when the system was running. But by looking at the period from March onward, when travel returned to normal, we see a similar increase in entries at these downtown stations as the weather warmed.

Let’s look at entries as Government Center returned to service in Spring 2016:

Station entries around Government Center Spring 2016

Some interesting trends emerge here. We see the surrounding stations, especially State, reduce in entries after Government Center returned to service. Government Center itself gains back entries, but does not reach the numbers we saw before it closed, remaining in the low 5000s as compared to 7500+ before. Downtown Crossing also has consistently more entries than Park Street, which wasn’t the case before. How do we explain these observations?

  1. Drop in entries at Government Center: Government Center Station has not moved, and the makeup of the surrounding area is largely the same now as it was in 2014. The overall number of entries at these six stations are not especially different. It seems plausible that passengers whose behavior changed during its closing have realized that other stations actually work better for their trip. For example, perhaps passengers didn’t know about the entrance to State Station between the State and Court St intersection and City Hall, but began to use it when Government Center was closed, and realized that this was a more direct trip for them.
  2. Change in entries at Downtown Crossing and Park Street: Downtown Crossing’s daily entries gained on Park Street over the time studied, and overtook Park Street’s in the fall of 2015. The City of Boston has focused greatly on Downtown Crossing in recent years, and multiple new buildings have been built, including the Millennium Tower development and the opening of the Roche Brothers grocery store at the former Filene’s building. It seems likely that these developments have contributed to increased entries at Downtown Crossing’s many entrances.

This table shows entries at the above stations on an average weekday during the time period. “Pre-closing/opening” is January 1 through the date of closing or opening, and “post-closing/opening” is the date of closing / opening through June 24 of the applicable year. Note that the “before” period is generally lower ridership than the “after,” so we should expect an increase separate from any effects of Government Center opening or closing:

  Avg. Weekday Pre-Closing 2014 Avg. Weekday Post-Closing 2014 Difference Avg. Weekday Pre-Opening 2016 Avg. Weekday Post-Opening 2016 Difference
Gov't Center 9,214 0 -9,214 0 6,041 +6,041
State 13,191 16,845 +3,654 16,853 15,204 -1,649
Park St 17,642 24,277 +6,634 20,592 19,817 -775
Haymarket 10,553 13,887 +3,334 12,546 12,249 -298
Downtown Crossing 21,642 23,264 +1,622 25,493 25,603 +110
Bowdoin 1,898 2,518 +620 2,613 2,273 -339
Total Near Gov't Center 74,140 80,790 +6,650 78,117 81,187 +3,090

Finally, here are two charts of total entries at the above stations. These show daily entries, and the yellow lines shows a 7-day moving average. As you can see, there doesn’t appear to be a noticeable change in total entries when Government Center closed or when it reopened. This confirms what the above charts show: passengers continued to take the T, but shifted their travel to other nearby stations.

Daily entries total near Gov't Center 2014

Total daily entries near Government Center 2016

We will continue to monitor station entries and other data at these stations to see if entries return to pre-closure distributions. But based on the trends of total entries at the stations near to Government Center, we are not concerned about a long-term drop in ridership from Government Center closing. It appears that some riders may have permanently changed their behavior due to the closing, but they did not leave the system entirely.