MTA officials recently announced that the tunnel that carries the L train subway line between Manhattan and Brooklyn could be shut down for up to three years. Business owners in Williamsburg are already complaining about the possible effects the shutdown could have on the local economy. Can we predict these effects? We use Instagram data to quantify the effect that previous temporary L train shutdowns have had on activity in the area. We find that blocks close to the Bedford Av stop are the ones that suffer the most, with decreases up to 50% in in the number of images shared. On the other hand, blocks close to the alternative J, M, Z, or G trains see an increase in images posted between 10-40% depending on time of the day.

white

Last January MTA officials announced that the L train could be shutdown between Manhattan and Brooklyn for several years in order to repair damages caused by Hurricane Sandy. The news quickly generated backlash in the Williamsburg, Brooklyn neighborhood from local business owners that complained about the potential impact the shutdown could have on the commercial activity in the area. They state that the economic losses from not having full L train services are high stakes for their businesses. Although a lot has been said regarding the effects of the shutdown, there has been little data to support many of the claims.

The idea of this project is to quantify the effect of the L train shutdown on Williamsburg: do fewer people come to the neighborhood when the L train is shutdown? If so, how big is this effect? Luckily for us, there was no L train service between Myrtle-Wyckoff Ave station in Brooklyn and 8th Ave in Manhattan during the weekend of March 1 and 2, 2014. This allows us to compare the number of people that visited Williamsburg this weekend relative to other weekends and quantify the effect the L train shutdown has on activity in the neighborhood1. In order to capture this effect, we will rely on Instagram data. Instagram posts allow us to aggregate the number of people that are at specific parts of the city at particular periods of time. We gathered all geo-coded Instagram images taken in Williamsburg and Greenpoint between February 26 and August 3, 2014 (490,000 images in total).

The first thing that stands out is that among all weekends in our dataset, the weekend without the L train service had the fewest amount of images uploaded (6,307 images). If we contrast this to other comparable weekends, the weekend without L train service had roughly 17% fewer images posted. Intuitively, we can separate businesses in the area into two broad categories: businesses that function mostly during the daytime, which includes cafes, shops and street vendors; and those that function in the evening, like bars and nightclubs. Restaurants are harder to categorize since they tend to operate both during the day and night. The figure below shows a heatmap calculating the percentage change in Instagram images between the weekend with no L train and the average of the comparable weekends2.

Figure 1: Difference in Instagram posts during day (8am-8pm)

Weekend without L train relative to average weekend

The heatmap above shows us that the L train shutdown has heterogeneous effects among different tracts. Clearly the tracts surrounding the Bedford Av stop are the ones that suffer the largest decrease: the blocks on Bedford Avenue, north of N 7th street and into McCarren Park, see a 50% decrease in the number of images posted. On the other hand, tracts in Greenpoint, South Williamsburg and East Williamsburg see some benefits, in the order of 10%. Unsurprisingly, these tracts that see an increase in social media activity, are those that are close to either the J, M, Z or G train stops. These stations serve as convenient alternatives for getting to Williamsburg when the L train is shutdown.

Figure 2: Difference in Instagram posts during night (8pm-8am)

Weekend without L train relative to average weekend

The evening heatmap above shows a slightly different picture. We still see a large reduction in the number of images uploaded in the blocks around the Bedford Av station, with reductions up to 30%. In South Williamsburg, the benefits of the L train shutdown at night seem to benefit only a small number of blocks surrounding the J, M, Z and G stations. These gains are large, but concentrated in one specific tract. The same patterns seem to be true in Greenpoint, where the two tracts close to the G train station see large increases (40%), while the surrounding tracts have small and mostly insignificant changes. Intuitively, this figure shows that in the evenings the vicinity of the closest functioning train station is increasingly important.

What does this analysis show about the effect of the L train shutdown? First of all, the blocks closest to the Bedford Avenue L train station are the ones that suffer the largest decrease in activity. Depending on the time of the day and specific blocks, Instagram images posted in this area fall between 25-50%. Interestingly, it seems that those areas in Greenpoint and South Williamsburg that are close to the J, M, Z or G train actually see an increase in activity. These benefits are more disperse and muted during the day (at around 10%), but become extremely concentrated on the blocks surrounding the metro stations at night (between 30-40%).

To get the full picture, below is the overall effect, not differentiating between day and night. Depending on the location of their business, it seems some owners will be happy and others worried when they see this. One thing is clear though, the L train shutdown will have important consequences.

Figure 3: Difference in Instagram posts

Weekend without L train relative to average weekend

white

Note: You can contact me via email at aindaco@gmail.com. I would like to thank Lev Manovich for helpful comments and suggestions.

  1. An important caveat to note is that the proposed L train shutdown will only interrupt service between 1st Avenue station in Manhattan and Bedford Avenue station in Brooklyn, but the weekend we are studying had no service between 8th Avenue station in Manhattan and Myrtle-Wyckoff Avenue station in Brooklyn. Even though most visitors come to Williamsburg from Manhattan, under the proposed shutdown access to Williamsburg from locations further in Brooklyn will not be interrupted.↩

  2. We also leave out any tract that has less than 30 images on an average weekend, because the variation there was a bit noisy.↩