City Desk

Bikeshare Users More Employed than Rest of D.C., Poorer than Other Commuters

Capital Bikeshare members are more likely to be employed than the average D.C. resident, and they save an average of $819 a year by using the service. But it's not all good news: they're also a little poorer than the average commuter in Washington.

The information comes from a new survey of Bikeshare users. While around 70 percent of adults in the D.C. area are employed, 90 percent of Bikeshare users have job. They're also younger, whiter, better educated, and more likely to be men than the average commuter.

40 percent of members reported that they drive fewer miles, with Bikeshare estimating that its members drive 5 million fewer miles a year than they would without the service.

Who's not benefitting from Bikeshare? The 20 percent of respondents who said joining the service didn't make them bike more. For them, Bikeshare has become a bigger analogue of the unreturned Netflix envelope.

Photo by Flickr user James D. Schwartz, attribution-NoDerivs 2.0 Generic (CC BY-ND 2.0)


Blog Widget by LinkWithin


  1. #1

    "5 million miles less a year"

    In the above phrase, "less" should be replaced with "fewer." Because miles can be counted, "fewer" is correct and "less" is incorrect.

    A better alternative would be to re-write the thought as "5 million fewer miles a year." As it is currently written, the phrase is actually subtracting a year from five million miles, which does not make sense.

  2. #2

    @ Chris:

    That's correct, and we've made the change. Thanks for pointing it out.

  3. #3

    "But it's not all good news: they're also a little poorer than the average commuter in Washington."

    Younger == "a little poorer". This is completely unsurprising.

  4. #4

    With their $819 savings, someone could purchase a pretty nice bike (and lock and chain, of course)--why rent?

  5. #5

    Most probably already own bikes. Why rent? It's essentially free, comes with parking, and you don't have to worry about your bike being stolen or vandalized while you're at work. And somebody else handles maintenance and repairs. So you save your own bike for trips where performance matters, minimizing wear and tear, and use CaBi for mundane errands and your commute.

    Which, of course, would suggest that the miles not driven stat is unrealistic. Odds are, most of the CaBi trips would be on foot, on public transit, on privately-owned bikes, or not made if CaBi bikes weren't available.

  6. #6

    It appears that CaBi calculated the "miles not driven" based on the actual results of the survey...the respondents may have over-reported their reduction in miles driven or the miles driven may have dropped for some other reason, of course, but that's actual data, not swapping a tally of miles biked for miles driven.

    I use CaBi because I only bike one-way most days/trips. I have been in numerous situations where the bike commuter in the group is scrambling to figure out what to do with their bike when the weather suddenly turns nasty at close of business or they get a last-minute happy hour invite but can't ride home after drinking. It's very inexpensive and saves me a little money. I don't own a personal bike...beyond wanting to ride a lot of one-ways, I don't have any place to safely store a bike at my home.

  7. #7

    So how do you get back from the one-way trips? And what did you do before CaBi?

  8. #8

    Walk, Metro, or occasionally get a ride from a friend, exactly how I went both ways before.

  9. #9

    *Except that now I don't have to wait 20+ minutes for a bus or train (so long as there's a bike available, I can just hop on and go). Or make a rail transfer and still make a walk that takes longer than the bike ride from a further-away, one-seat-ride Metro stop. Or suffer through massive Metro crowding in the core stations (summer evenings are AWFUL between Dupont and Union Station). And it's fun. :)

  10. #10

    So no miles not driven in your case. Which, from what I've seen, is pretty typical (and why I question the validity of the survey research and the massive extrapolation being made from it).

    Fun, convenience, exercise, etc. seem like great reasons to use CaBi. But not such compelling reasons for public subsidies.

  11. #11

    Re Anon:
    The calculation of mileage reduction is not extrapolated over all trips or all users. It is self reported by members. Their individual estimation of how many miles less they drive since joining Capital Bikeshare. If a member reports that their driving reduction is zero, then that is what is calculated for that person. Averaged across all survey respondents, the average mileage reduction is 523 miles per member. Calculation can be found on page 56 of the full report:

  12. #12

    Metro rail in the downtown core is pretty much so at max capacity at rush hour. By getting me, and others, off that ride, you make some more room on those trains. More people can use public transit, but more importantly, the trains break less when they're not stuffed past capacity. I think that deserves a public subsidy, since the alternative is more money spent on fixing trains/providing additional capacity (very expensive at the margin).

    Many people reported taking trips they otherwise wouldn't have. Frequently, these trips were to dining or retail establishments. More dining & shopping = more tax revenue = the system helping to pay for itself.

    5 million miles not driven in ONE YEAR means a whole lot less road maintenance and emissions. That also deserves a public subsidy, again, as it is cheaper than the alternative. And while we're on this subject, reducing emissions reduces public and private expenditures on healthcare, improving the economy and the government's fiscal condition. And so does increasing exercise.

    That's quite a few public benefits for a pretty small amount of money. And that's not even mentioning that CaBi is now covering its operating expenses with user fees. Government money has long been given to kick-start programs. While CaBi may need continuing subsidies for capital expenses, the amount of subsidy needed in the long run is only a fraction of the annual subsidies so far. CaBi is a bargain when compared to more Metro trains, more Metro buses, more Metro vehicle maintenance, more Metro infrastructure maintenance, more Metro rail infrastructure, more car lanes, more road maintenance, less business for retail establishments, and increased expenditures on healthcare.

  13. #13

    According to the survey, only 13% of the respondents said they would have driven/ridden (7%) or taken a cab (6%) on their last CaBi trip had they not had access to CaBi. And riders averaged 8.1 trips a month. We're also told that CaBi has about 18,000 members. At that rate of substitution & usage, you'd have to assume that the average CaBi trip (that would otherwise have involved driving) was almost 22 miles long in order to find the 5 million automobile miles/year saved figure credible.

    You'd also have to assume that a survey sample based on self-selection is representative of the membership as a whole when, most likely, such a sample will disproportionately include the most frequent users and significantly under-represent tourists and people who joined CaBi but rarely, if ever, use it.

    The 5 million miles claim is bogus and any decent researcher should have known that.

  14. #14

    re the tax revenue gains -- not at all clear. Presumably, given destinations and distances, it means that CaBi users went to restaurants further away from home but generally still in DC. No reason to believe they ate out or shopped more -- just that they patronized different stores/restaurants than they otherwise would have.

    Metro ridership is also trickier than you suggest. Presumably some substantial chunk of the $800+ savings/user/year is lost revenue for Metro. And decreased ridership only saves dollars if it's significant enough at times of peak demand to prevent the need for additional cars/trains. Seems unlikely given the numbers.

  15. #15

    So here's how my math works out for the car/bike exchange (I will show my work):

    53% of CaBi members surveyed responded that they own a car, or, generalizing that result, 9540 members. Assuming that these members are "average" CaBi users, 8.1 trips/month x 9540 users x 12 months/year = 927,288 annual CaBi trips by users who also own cars. 5,000,000 miles/927,288 trips = 5.4 miles/trip average. Seems totally reasonable to me. (BTW, the 18,000 members are ANNUAL members, CaBi has almost 200K "total members," including tourists who joined for a day and are no longer members...all this data is available, for free and in a few clicks, on the dashboard at

    As an economist, I'm also good at drawing conclusions from various points of evidence. Look at any day's ridership patterns, and note the huge spikes in usage around rush hours. Combine that with the fact that the strong majority of trips replace Metro bus/rail trips, the conclusion is that many CaBi trips remove riders from Metro at peak hours, where the marginal cost of extra riders is high and the additional revenue benefit is low.

    The survey clearly asked users if they took a trip they otherwise wouldn't have because CaBi was available. Now, I suppose that respondents could think "hey, I went out to dinner in Georgetown instead of Foggy Bottom because I rode the bike, so, yeah, I took a trip I wouldn't have," but my take-away from that question was that I did something I otherwise wouldn't have, i.e., go out to dinner in the first place.

    Maybe you should try reading the survey (check your math twice, as well) instead of grasping at straws in the comment section of the WCP.

  16. #16

    Sorry, the strong *plurality* of trips replace public transit. By far and away the most replacement, but not a majority.

  17. #17

    The stats for miles saved were also very, very clearly the reduction in driving AS REPORTED BY THE RESPONDENTS. There was no extrapolation involved to reach that 523 miles saved, just total miles saved(gained)/number of respondents. Now, could some of that reduction be attributed to other behaviors than using CaBi? Sure. If you're feeling healthier, and enjoying using your car less, maybe you walk somewhere you might have otherwise driven. Or maybe you take the train since you really liked not having to worry about parking when you rode to that destination the last time, but you're not feeling up to the ride. To say, however, that it's "totally bogus" to simply say "how much did you drive before? how about now?" and report the results is, well, totally bogus.

  18. #18

    You've just assumed that every trip any person with access to a car takes on CaBi would have been a solo driving trip or solo taxi trip but for CaBi. AS REPORTED BY THE RESPONDENTS, that's not true. The vast majority of the time, they'd have walked or taken public transit instead and ometimes they just wouldn't have made the trip.

    That's not even true of the car owners, only 38% of whom reported that they reduced their annual VMT after joining CaBi. And as the study itself reports (and as you can see from looking at the survey instrument) "the survey did not ask respondents how much Capital Bikeshare had influenced respondents' reductions in vehicle mileage." When you get most people reporting no reduction in VMT and a few reporting major reductions (14% = over 1,000 miles) and you don't ask questions about causality -- i.e. did you move from the burbs to the city in the past year? change jobs? quit work and start grad school? decide to use public transit more when gas prices soared?), then it makes no sense to extrapolate in the way you (and the study's author) have done.

    Do you know how large the subsample was that reported VMT for pre- and post-CaBi?

  19. #19

    Oh dear FSM. Please go read the study. The average drop in VMT by personal vehicle was indeed reported by the respondents (as I said reported miles saved(gained)/number of car-owning respondents). The generalization was reported average drop in VMT to all vehicle-owning CaBi members to come up with the TOTAL number, not on the side of the average drop in VMT. That is a reasonable assumption...or do you also think that all other "sample" polls, where the sample is large and representative, are totally bogus? Was it all due to CaBi riding? No, I realize that my assumptions were wrong. But CaBi never represented it was all because of riding a bike. They simply said that CaBi members reduced their average VMT in a personal vehicle by 523 miles. That is a reliable fact. Regardless of whether they made the new trips by bus, train, walking, personal bike, or CaBi, they still saved those personal vehicle miles on the road.

    There's also a bit of "e" on the flip side of the coin. I sold my car just before joining CaBi, so had I been party to this survey, my drop in VMT (over 2K miles) would not have been counted. While it would be wrong to include my total drop in VMT, I do substitue transit, CaBi, walking, and shared rides with friends for solo car share, so I am also taking SOV VMT off the roads, even though I no longer own a car. There's no way to design a study to account for all of these nuances without getting stupid complicated, and the fact was properly communicated. For vehicle-owning CaBi members, the VMT in a personal vehicle dropped an average of 523 miles each, for an estimated total of 5M fewer VMT year-over-year. Given a bit of e on each side (people who don't own cars who substitute other modes over shared SOVs - I know very few people who don't own a personal vehicle and don't occasionally rent cars, and they're usually piggybacking on me to share a Zipcar - and people who do own cars who don't follow the trend seen in the study), it's a well-constructed estimate.

    The sample was just under 5K users, 53% of whom owned a personal vehicle both before and after joining CaBi. Given a total ANNUAL membership of just over 18K, that's a more-than-healthy sample size, probably big enough to account for selection bias. If I were doing a random survey, my sample size for that population would not have been that large. I mean, election polling regularly samples a few hundred people to generalize the preferences of MILLIONS of people, with a MOE of 3 or 4%. 5K out of 18K is a great sample.

  20. #20

    Even if only 38% reduced their driving, that 38% reduced it enough to create an average drop on VMT of 523, in other words. YMMV, as with any study, but overall, the average car-owning user reduced their driving by 523 miles/year.

  21. #21

    No, the size of the sample doesn't overcome self-selection bias. If two-thirds of the people who bought monthly or annual memberships in CaBi rarely use the service and don't answer the survey as a result, then the fact that 31% answered doesn't make them representative of the whole. And the particular questions being discussed are ones where fewer than half of the people who *did* take the survey provided answers.

    Look, if you had a randomized and reliable sample and if you had a way of attributing causality for the decrease in VMT and if you had a complete data set for that particular question, then the extrapolation would make some sense.

    But you don't have any of those things and you've studiously ignored the most relevant data the survey gave you about mode choice/substitution. 87% of the survey respondents indicated that they would NOT have replaced their bike ride with a drive/taxi ride had they not had access to CaBi. (BTW, this is consistent with figures from survey research last year on Montreal's bikeshare program -- I think their number was 90%). Even if you want to attribute all of the foregone personal car trips to the car-owning members (so the 7% overall becomes 14% for this group), assume these trips were are all solo and assume car-owners were no less likely than non-owners to take cabs (6%), you'd still have a situation in which 80% of the time, car-owners would *not* have replaced CaBi trips with car trips. Yet the math you've presented here, replaces that 80% with 0% -- i.e. you assume that every CaBi trip taken by a car-owner would have been a driving trip. Which is how you got the presumed CaBi trip length down from > 20 miles (big red flag) to a little over 5 miles. Basically, you have to ignore your own data in order to lend even superficial plausibility to your conclusion.

    Even with the kinds of controls this study lacks and even with a better survey instrument, self-reporting isn't always reliable. Not all errors associated with self-reporting are detectable, but you know there's a problem when people tell you things that can't simultaneously be true and/or they tell you things that don't jibe with objective data. But we don't even have to get to that level of analysis to see what's wrong here. It's a logical fallacy so textbook it has its own Latin name -- post hoc ergo propter hoc.

    No one ever responded to this survey in a way that attributed a particular number of miles not driven to CaBi. People were asked to estimate monthly miles driven the year before they joined CaBi and the year after but not asked anything about what changed in their lives over the course of that year. When you get the vast majority of people saying little or no change in VMT and a few extreme outliers (e.g. 6% claiming decreases of 2500+ miles a year which seems highly unlikely as a CaBi effect), it doesn't make sense to assume causality, average, and extrapolate.

    I'm perfectly willing to believe that last year 14% of the people who signed up for CaBi were people who, for example, didn't live in DC or Arlington the previous year and whose lives elsewhere involved more driving. And that this pattern will continue. But that's not evidence that CaBi significantly decreases VMT. It's evidence that some people whose VMT is significantly decreasing (for whatever reason) will buy a CaBi membership.

Comments Shown. Turn Comments Off.