It is commonly held wisdom that as the price of gasoline rises—or the price of oil as its corollary which I have shown is highly correlated to the price of gasoline—ridership on mass transit increases. That’s what CNN reports. I do not like to subscribe to commonly held wisdom because it is so often wrong. Anecdotal evidence only counts for so much. I like hard facts and statistics.
Using the same series of gas price information used to determine the correlation between oil prices, demand, and relative U.S. dollar strength I compared them with a set of data related to ridership on all forms of mass transit. The data on mass transit comes from the American Public Transportation Associations (APTA). The data is available here.
The dataset I chose was the total of unlinked transit trips. This dataset combines heavy commuter rail with light rail and bus systems as well as less widespread systems like ferries. Originally plotted, the transport data was not indicative of anything because it was jagged month to month:
The dataset also bounced within a fairly narrow range. However, a twelve month average looked different:
The scale is a little different, but the 12 month rolling average shows a peak and then a decline. What does that trend correlate with? How about gas prices:
The correlation between the monthly average for gas prices and the 12 month rolling average is quite weak. It calculates to ~(0.174), which connotes weak if any correlation. The correlation between the monthly average for gas prices and the monthly unlinked transit trips is higher—~.351—but it is weak nonetheless.
So, gas prices and transit trips do not correlate strongly. Why? My theory is that mass transit ridership is impacted by gas prices, but its driving factor is people getting to and from work. Look at the curve in the 12 month rolling average of unlinked transit trips. The decline corresponds to the deepening of the recession that began in 2008. See what happens when you plot it against the monthly unemployment rate:
Nothing to see here, right? Wrong. The datasets are actually negatively correlated. The math works out to ~(0.694). That is to say that as unemployment goes up—more people out of work—then transit ridership goes down—fewer people going to work. The converse is also true. The unemployment statistics were taken from the Bureau of Labor Statistics.
I do not think any of this is definitive, but it suggests a more complex relationship than merely higher gas prices driving behavior. Ironically, the president of the APTA Michael Melaniphy said “As people get jobs and go back to work, they get on mass transit more. And then when people look at gas prices, they really get on transit more.” The converse of the statement is true. As gas prices rise people get on mass transit more, but as they get back to work they really start to ride mass transit.
Another aspect I am struggling with how to model is the capacity of mass transit relative to ridership and gas prices. The capacity of mass transit, outside of adding busses, is not something that can be scaled with any rapidity. There is a long lead time in building rail lines or adding trains. The need for additional capacity is probably reduced by the time that new capacity is installed meaning that only the farthest looking of civic planners actually put systems in place to handle future demand growth.