MTA officials recently announced that the L train will be shutdown for 18 months between Bedford Avenue and 8th Avenue. Given that roughly 300 thousand people use the L train on a daily basis, this looming shutdown could have a huge effect on these areas. I propose studying the effect the shutdown will have on Citibike ridership to Williamsburg and how that could potentially affect the age distribution of people visiting the neighborhood. I will use temporary L train shutdowns in the past few years to extract patterns that help predict the effect of this newly announced shutdown.

white

In late July, MTA officials announced that the L train will stop running between Brooklyn and Manhattan for 18 months starting in January 2019. Several reports have come out analyzing possible effects on business, rent, urbanism, and even lifestyle. I propose studying what effect the L train shutdown will have on Citibike ridership by analyzing bike trips to the Williamsburg area during previous temporary L train shutdowns. In a previous project, I studied the number of Instagram posts in Williamsburg when the L train shutdown in 2014 for one weekend. This analysis would broaden this study and add an additional layer I had not yet studied: how alternative transportation methods, such as Citibike, are affected by the shutdown.

For this project I downloaded and analyzed all 9,937,969 Citibike rides taken in NYC during all of 2015. I found that the L train service between Bedford Avenue and 8th Avenue were shutdown for six weekends in 2015, specifically between the weekends of April 11-12 and May 16-17. This gives us the possibility to compare Citibike trips during these weekends in the Williamsburg area and compare to other weekends when the L train was in full service. This way we can study the possible effects that the L train shutdown in 2018 might cause on Citibike rides in the area. Although we are aware that this comparison is not exact, given that the shutdown we are analyzing was only in effect during weekends and for only 6 weeks, compared to the complete shutdown for 18 months that will take place in 2018, we believe this analysis can provide us with some useful insight that can prove useful to Citibike officials as well as business owners in the affected areas.

The first thing we do is plot the total number of Citibike trips taken to bike stations in Williamsburg and Greenpoint during all weekends in 2015 (Figure 1). The data seems to indicate that there was a sudden increase in Citibike rides to Williamsburg stations the weekends that the L train was shutdown. In fact, the three most popular weekends for Citibike dropoffs in Williamsburg over 2015 were all during weekends when the L train was shutdown.

white

Figure 1: Total number of Citibike trips to stations in Williamsburg

Trips are aggregated by weekend. Red plots indicate weekends when L train was shutdown.

white

The issue with Figure 1 is that we do not have a counterfactual to compare to. Other studies have already shown that weather has a large impact on Citibike ridership: given that April is the beginning of Spring in New York, it could be the case that the jump we see in bike rides to Williamsburg is simply driven by better weather conditions and nothing to do with the L train shutdown. In order to control for this, in Figure 2 we calculate the proportion of rides in the entire Citibike system that end in a station located in Williamsburg across time. This will allow us to parse out the ‘weather effect’ from the increase in bike rides to Williamsburg in April and May (that would have occurred regardless of the L train service), and get a more isolated idea of the effect of the L train shutdown.

white

Figure 2: Proportion of Citibike trips that end in Williamsburg

Trips are aggregated by weekend. Red plots indicate weekends when L train was shutdown.

white

Given that the weather is roughly the same throughout all of NYC, any large differences in the proportion of rides ending in Williamsburg stations is likely due to some other factor1. Given that the highest ratios are precisely when the L train was shutdown, it is another indicator that when the L train is not in service there is an increase in number of Citibike rides that go to Williamsburg.

I believe this is interesting and relevant for several stakeholders. On one side, this is important for businesses in the area. For them, it is crucial to understand how many fewer people go to Williamsburg when the L train is shutdown. But also, this analysis will allow us to see if and how the demographics change. For people coming from Manhattan to Williamsburg on bike, crossing the Williamsburg bridge is no easy task. So is it possible that given that more people choose to bike to Williamsburg when the train is down, that the overall distribution of age in Williamsburg is somewhat younger when the L train is down. Of course, in order to do this precisely, we would also need to consider taxi ride trips into the area. This can be done using publicly available taxi data and is something I have in mind for my complete project. My analysis shows that the average age of people going to Williamsburg on Citibikes is lower than that of people going to other destinations in NYC (34 vs. 38 years of age). Also, the distribution of males and females that bike to Williamsburg is somewhat different form the overall average: while the total average is 73% male, this drops to 69% male on trips to Williamsburg.

And this is also relevant for Citibike officials. If the L train shutdown significantly increases ridership to Williamsburg, they might need to adjust the distribution of bikes across stations–or maybe even consider opening more stations in Williamsburg.

white

Figure 3: Hourly number of trip to Williamsburg stations with and without L train

Trips are aggregated by weekend. Red line indicates weekends when L train was shutdown.

white

As we can see in Figure 3, the amount of Citibike trips to Williamsburg on weekends when the L train is shutdown is roughly the same as average in the early hours of the morning, but begins to significantly pickup after 10am. This increased activity lasts up until 9pm, when activity is similar to that of a weekend with normal L train service. This is interesting and useful for Citibike officials who might need to plan a reshuffling process according to this pattern when the L train does officially shutdown in January 2018. It would also be interesting to see the pattern of people starting their Citibike ride in Williamsburg, as this may also show a different pattern than usual. This too would be useful for Citibike officials when thinking about a proper bike reshuffling strategy. I would also be interested in seeing where the majority of these trips to Williamsburg originate: maybe the use of Citibike by users that live close to the G, J, M or Z train (all of which have stops in Williamsburg) are not affected by the L train service, so this too is important for Citibike officials when thinking about a reshuffling pattern.

For my project I would like to expand on this idea by studying in more detail the types of users that do use Citibike to visit Williamsburg when the L train is shut down: their age and gender, where do they live and possible cross this data with Census data to have a more clear picture of the socio-economic characteristics of these users. Also, I would like to develop an econometric model to try to quantify the effect the L train shutdown has on Citibike rides to Williamsburg. Initially, I am thinking the setup lends itself quite well for a differences-in-differences model, were I could use similar stations far from the L train stops, with similar daily activity as my ‘control’ group. Once this is done, I could also cross these results with the analysis I have already done on Instagram posts in Williamsburg when the L train is shutdown, in order to have a more complete understanding of the effect the L train shutdown might have on the neighborhood.

  1. Although it could be the case that weather affects differently different parts of NYC. For example, if subscribers are looking to make longer trips when the weather is nice, hence many people from Manhattan carry on longer trips to Brooklyn were they have to cross the bridge. This would mea a disproportionate increase in the number of trips to Williamsburg caused by weather (even though the weather is the same throughout NYC).↩