Capital Bikeshare rack location, data analysis, 3/8/2019.

On this page, I describe the issue facing Vienna, and then profile the typical Capital Bikeshare user and analyze data on Capital Bikeshare use in Tyson’s Corner and Reston.  See a separate page for a quick summary of dockless bike alternatives.

Edited 3/9/2019 to correct minor errors in the data analysis.  Gray pins in the map below indicate racks with less than one bike trip per day, on average, in 2018.  Edited 3/10/2019 to add an analysis of growth trends for newly-opened Bikeshare stations.

This page started out as a little addendum to my most recent discussion of Capital Bikeshare, based on the Transportation Safety Commission meeting of 2/26/2019.  Then I decided to see whether I could find any systematic way to identify promising locations for Bikeshare racks in Vienna.

But in the end, after looking at a profile of Bikeshare users, and analyzing trip-level detailed data on Bikeshare use in Tyson’s and Reston, my conclusion is that Capital Bikeshare racks in Vienna will probably be an outstandingly poor value, in terms of tax dollars spent per bike trip.  Close to, but not quite, a total waste of money.  And that’s because most of the racks will likely be used for less than one bike trip per day, as is currently the case in Tysons and Reston.

They key finding from the data analysis the extremely low level of Bikeshare use in neighboring jurisdictions.  In Tysons, three-quarters of the Bikeshare racks are used for less than one trip per day, on average.  In Reston, 69% of the racks are used for less than one trip per day.

Best guess, on any given weekday, the entire Tyson’s Bikeshare network serves about six people.  If we characterize the “use rate” as average daily bike trips per bike-parking “slot”, the use rate of the Tyson’s Bikeshare racks is just 6% of the all-metro-area average Bikeshare use rate.  And that low use rate, combined with high capital and operating costs, makes this an expensive way to transport a few people.

I should emphasize that such low use was not expected when the decision was made to pay for Bikeshare in these localities.  In other words, this is not business-as-usual for Bikeshare.  The 2018 Reston use rate is less than 20% of the level projected for the second year of operation in the Reston bikeshare feasibility study (.pdf).

Moreover, there should be no expectation that, after more than two years in operation, there will somehow be explosive growth in the future.  As shown in the data analysis below, based on historical Bikeshare data for new stations, we should expect the 2021 ridership at these Reston and Tysons stations to be just 21 percent higher than the 2018 ridership.

The upshot is that Bikeshare sounds good, but in reality, Bikeshare performs poorly in spread-out suburban settings such as Tysons and Reston.  (To believe otherwise, you would have to expect phenomenal growth in Bikeshare use at a  time when Bikeshare use in Arlington appears to be plateauing due to competition by “dockless” alternatives (.pdf). Several members of the Transportation Safety Commission had already questioned the viability of Bikeshare in our area.  The analysis below just adds some hard numbers to that intuition.

If Vienna proceeds with installing Capital Bikeshare racks, the data suggest prioritizing work-related sites (such as a large employer) over leisure-related sites (such as a library).  Bikeshare users in Reston and Tysons (and for Bikeshare as a whole) overwhelming use the system for commuting.  Leisure-type uses are a distant second in terms of generating Bikeshare bike trips.  For example, in 2018, the Reston regional library Bikeshare rack accounted for just 48 bike trips.  (See below for exact definition of “trip”.)

 


What is Capital Bikeshare and what is the issue for Vienna?

Capital Bikeshare is a government-owned, privately-operated bike sharing system where you can rent a bike at one of their bike racks by swiping a credit card.   It is designed mainly for short (30 minute or less) trips, from a bike rack, to some other bike rack.  Most users buy an annual membership, and with that, all trips under 30 minutes are free (or, more properly, prepaid).

Capital Bikeshare appears to be coming to Vienna.  Paid for by Fairfax County, using money from the I-66 tolls.  Vienna town government has to figure out where to put the bike racks.

Bikeshare rack location was discussed briefly at the most recent Transportation Safety Commission meeting.  That discussion struck me as odd, because it sounded like a group of very bike-savvy people trying to make … an educated guess, more or less, as to where those racks should be located.

Why is that odd?  Capital Bikeshare has about 500 bike racks in operation.   By the time a business had made the same investment decision 500 times, you’d kind of expect to see some sort of  optimized (or at least standardized) decision-making process in place.  Instead, this looks like Capital Bikeshare leaves that up to the locality, which then has to make a more-or-less seat-of-the-pants decision.

That doesn’t seem like a good way to go about this.  And what caught my attention was the combination of these four elements:

  1. Capital Bikeshare is owned by our local governments, as described here).
  2. Capital Bikeshare is operated by a private entity — Motivate.
  3. These racks are expensive (as in $50K each, or so, as documented at the bottom of this page.)
  4. The choice of rack locations seems like a make-or-break set of decisions for the success of Capital Bikeshare in Vienna.

Let me just copy in some information on costs here, from one of my prior posts.  Read up on the recent history of Capital Bikeshare, and pay some attention to the cost numbers.  Per Wikipedia: “In May 2011, it cost $41,500 to install a station with six docks and $49,300 each for larger stations with 14 docks. Each bicycle cost about $1,000, and the annual operating cost per bike was $1,860.”

A 2017 study of this issue by Falls Church (.pdf) largely validates those cost numbers.  They estimated an average cost of $44,000 per station to set up 10 Capital Bikeshare stations within Falls Church.  A study for the Merrifield (Mosaic District) area (.pdf) came up with a figure of about $40,000 per stationThe feasibility study for Reston put the costs modestly higher.

To put this as plainly as possible:  The costs are substantial, one entity is paying for it (Fairfax County, for Vienna), a different entity is making the location decisions (Town of Vienna), and it is not clear that the actual day-to-day operator of the system (Motivate) has any skin in the game other than greater revenue from an expanded system.

That sort of economically-muddled situation raises an obvious question.  Is there no standard or optimized way to locate racks because that can’t be done?  Or is there no standard process because the entire scope of the decision (costs and benefits) is split across three different entities?

The apparent lack of an optimal way to find Bikeshare locations seems particularly unfortunate in this case.  In my opinion, and in the opinions of several members of the Transportation Safety Commission, Capital Bikeshare faces an uphill battle here in the spread-out suburbs.  You can see my extended discussion at the bottom of this page.  Their model works well in a densely-populated urban work environment.  It’s not clear it will work well out here in the sticks.  So they need every advantage they can get.


Profile of the typical Bikeshare member

In the suburbs, Bikeshare use consists overwhelming of trips by individuals with annual memberships.  Only small portion are for “casual” users who swipe a credit card for an individual short-term rental (e.g., tourists.)

Capital Bikeshare surveys its members on a regular basis.  Their  2016 member survey report (click for the .pdf here) summarizes the characteristics of the typical member, and forms the basis for this section of the analysis.

To summarize that report in a single paragraph, people use Capital Bikeshare when it is the fastest and/or most convenient way to get from A to B.  And as a consequence, the target market for this in the suburbs is going to be quite small.

First, about 90% of users said they used Bikeshare because it was faster or more convenient than their alternatives (page 15, validated separately on page 28).  Second, the vast majority both live and work near a Bikeshare rack (page 8).  If “near” is a quarter-mile or less, then 77% live near a rack, and 86% work near a rack.  Unsurprisingly, almost 90% access Capital Bikeshare by walking to the nearest bike rack (page 29).  Third, about two-thirds joined Capital Bikeshare so they could get back and forth to work (page 23).  Fourth, a large fraction of them have very short commutes, with nearly 40 percent of users living within three miles of their workplace (page 43).  Finally, on-order-of half of them don’t have access to a car or their own bike (page 9), and correspondingly, the most common way to get to work day-in, day-out was some combination of public transit and Bikeshare (page 44 ff).

Based on that, it really doesn’t matter where you put the bike racks, because this profile describes more-or-less nobody in the Town of Vienna.  Either our users are going to be quite different from the average, or we aren’t going to have many users.

The only thing about the profile that matches what I commonly see in Vienna is that about 15% of memberships are given to people as an employee benefit. And the typical users tend to be younger and lower-income than the Metro area average.  That kind-of matches what I see in terms of who bicycles down Maple Avenue now, or at least down my end of it (Maple and Nutley).  Mostly, it’s guys who look like they are getting back and forth to some kind of blue-collar job.  It’s always a guy, young-ish, no helmet, plain-vanilla sturdy (fat tired) bike, typically no lights.  For sure, the people who routinely ride down my end of Maple are not the spandex-clad $4000-bike crowd you see on weekends on the W&OD.

Separately, if the idea is to provide a link between Vienna and the Metro station, you have to put this in the context of extensive Fairfax Connector bus service on Maple Avenue. To be clear, the same Smartrip card used by Metro works on the Fairfax Connector buses.  Almost all trips within the Fairfax Connector system cost $2, reduced to $1.50 if you are transferring from Metro rail.  That includes unlimited bus-to-bus transfers within a two-hour time window.  Estimated travel time from Vienna Metro to (say) Maple and Center is lower on the bus than by bike, even at peak rush hour.

Even if you look away from Maple, many plausible Bikeshare targets are already linked to Metro with dedicated bus routes.  For example, there is a direct bus route from Dunn Loring Metro to the Navy Federal Credit Union (Fairfax Connector route 462).

So, whatever Bikeshare business you think will be generated between Vienna Metro and Vienna, you would be looking for a clientele that either a) refuses to ride the bus, or b) for whom the $1.50 one-way fare matters.  Given that Metro only serves areas farther in toward DC (i.e., areas with relatively high cost-of-living), it seems unlikely that people who live in the Metro-accessible areas, and pay the Metro fare to get here, would balk at the $1.50.

In that context, it is hard to imaging the high-volume destinations in Vienna for which Capital Bikeshare will qualify as “ the fastest and/or most convenient way to get from A to B”, when point A is the Vienna Metro.  Other than for people who, for whatever reason, do not want to ride the bus.  Or for the occasional users (e.g., tourists) who do not want to take the time to figure out the bus schedules.


Detailed data analysis

How has Bikeshare worked out in communities similar to Vienna?  That’s the question I tried to answer next.

Methods

Capital Rideshare produces a rudimentary data set showing every bike trip that takes place in their system.  Anyone can access it from their website.  The files are large (3.5 million bike trips per year), and have only sketchy detail (basically, starting point and time, and ending point and time).  Nothing on the file identifies individuals (e.g., there is no demographic information).

(As an aside, Github is the repository for some of the Capital Bikeshare data.  Github catalogs more than a hundred computer programming projects using that information.  None of those projects addressed my question:  What is a systematic way to find promising locations for Bikeshare racks.)

Let me start with a map of the Virginia portion of the Capital Bikeshare system, captured from the Capital Bikeshare website.  Each dot on the map is a Bikeshare rack.  The clustering of the dots in DC clearly shows the urban-centric nature of Bikeshare’s core market.

You have to begin with an understanding of how the Capital Bikeshare system is organized — what the regions are, at least here in Virginia.  In Virginia, everything from the East Falls Church metro and further toward DC is part of the Capital Bikeshare Arlington region.  Further south is an equally urbanized Alexandria region.

To be clear, while Capital Bikeshare has a rack at East Falls Church metro, that doesn’t serve the Falls Church downtown.  The destinations for trips to and from the East Falls Church bike rack are to and from stations further in toward the Arlington urban core.

Outside of those urban core areas of Arlington and Alexandria, there are clusters of bike racks around the Tyson’s Corner metro, and again in Reston.  I’m going to start by looking at those two clusters of racks, although I don’t think either of them is necessarily a great analog for Vienna.

A few purely technical notes:  You have to summarize the data by station number, not station name, as Capital Bikeshare sometimes renames its stations.   To count “trips” for a station (bike rack), I summed all trips that started there plus all trips that ended there, and divided by two. This method merely averages out the allocation of each rack-to-rack trip equally to the originating and destination bike racks.  (I.e., the total number of “trips” in this analysis will exactly equal the total number of rack-to-rack bike trips on the original Bikeshare data file.)  Finally, every point in Vienna is less than a half-hour bike ride from Vienna Metro.

Finally, Capital Bikeshare began service in Reston and Tysons in 2016.  So the 2018 data reflect their second full year of operation.

Results:

As expected, 99 percent of Bikeshare trips that start in the Reston cluster end there, and similarly for the Tysons cluster.  The point being, people are not using Bikeshare bikes to get to these remote clusters.  Almost without exception, Bikeshare trips are local within each cluster.

This is consistent with a system designed around 30-minute trips.  Most users are members (almost all holding annual memberships), for whom those trips are free.  Because few trips exceed 30 minutes, the Tyson’s and Reston clusters of Bikeshare racks are virtually isolated from each other, and from the rest of the system.

For the remainder of this, I toss out a handful of trips starting or ending outside of the Tysons and Reston clusters.

Second, three pictures show that most trips are by members (as opposed to casual, credit-card-swipe user), and most of those are for commuting to work.  Trips by members show a distinct Monday to Friday concentration and typical rush hour peaks.

So, as is true with the rest of the Bikeshare system as a whole, in Reston and Tysons, Bikeshare largely serves individuals who use Bikeshare as part of their commute to and from work.

My final point is that most of these Bikeshare racks are extremely lightly used, given how much they cost.  I tabulated the data for the individual stations, then flagged those that were used for less than one trip per day, on average.  I found that 69% and 75% of racks were used for under 1 trip per day, on average, in Reston and Tysons, respectively.  (See notes above to be clear about what I mean by a “trip”.)

The link below is for an Excel workbook with the summarized results and underlying computer programming.

Capital Bikeshare Reston Tysons analysis, corrected, 2019-03-09

The map at the top of this page shows total annual bike trips, by rack, and  grays-out the ones with under one trip per day,  on average, for 2018.

One way to characterize the level of use would be to count unique users on any given day.  Unfortunately, I have to guess at that, because the Bikeshare data contain no person-level data.  That said, it appears that almost all weekday use is commuters starting out and returning to the Metro station.  On the typical weekday, the entire Tyson’s Bikeshare network sees an average of under 12 trips (data not shown).  Under the assumption that those are all, in essence, round trips away from and back to the Metro, then, best guess, the entire Tyson’s Bikeshare network  serves about six people, on any given weekday. 

One final way to characterize the use rate is to show how many daily trips the average “slot” in a bike rack generates.   Based on that metric, the use rate for the Tyson’s bike racks is about 6% of the all-metro-area average.

A

Any business with high capital investment and low use is likely to have very high average costs.   Although I have only crude cost data, as reported (above) by Wikipedia, let me run with that to take a rough cost at average cost per trip in the Tyson’s Bikeshare cluster.  Taking the 2011 Wikipedia data above at face value, and rounding up for inflation, each rack costs $50K and each bike costs $1K to buy and $2K/year to maintain.  Let me depreciate the racks and bikes over 10 years.  Let me assume just one bike per rack, for the 12 Tysons racks.  So, 12 racks, 12 bikes, depreciated over 10 years, with $2K/year/bike maintenance cost.  I come up with an annual cost for the Tyson’s Bikeshare system of $85,000, and an average cost per bike trip of $25. 

For comparison, the same calculation yields an all-metro-area average cost of just over $3/trip.  Data for Arlington County (.pdf) works out to around $7/trip.  (Interestingly, after years of growth, the Arlington County report shows a decline in membership in 2018, which they attribute to competition from other “dockless” forms of transportation.)  A discussion of the public-policy economics of Capital Bikeshare can be found at this link.

That’s a very rough cut, sure.  But, roughly speaking, it would be cheaper just to buy them rides via Uber.  And that’s really not a surprise.  A use rate that’s about 6% of the system average should result in very high costs.  And it does.

Finally, it is worth nothing that the actual 2018 trip count in Reston is about one-fifth of what was projected in their original feasibility study.  (Page 52, Year 2 data, of this feasibility study for Reston Bikeshare (.pdf).

Nor should we expect to see substantial growth in volume over the near term.  We can use Bikeshare’s historical ridership data back to 2011 to find stations in their second full year of operation (as the Reston/Tyson stations largely were in 2018), and track them forward to see what sort of growth we should expect.  Based on historical averages, we would expect to see just 21 percent higher ridership at the Tysons and Reston stations by the year 2021 (three years after the 2018 baseline).

Conclusions

My main conclusion is that this appears to be a lot of investment for what is likely to be a very sparsely used system.  And, to be clear, it’s not just the money, it’s also the turf that they occupy.  If you take the best spots in Vienna, and place these racks there, and few use them, then you have to ask whether that space could have been better used.

My second conclusion is that maybe Capital Bikeshare needs a different model for the suburbs, one that does not rely exclusively on large bike racks.  When I see that the typical rack gets under one trip per day, the idea that racks must hold 10 to 20 bikes seems inefficient.  Plausibly, the model that they developed for the dense urban core market needs modification so that up-front capital costs (and ongoing maintenance costs?) are smaller, to match the overall lower trip volume at most stations in the outer suburban locations.

An alternative to that is that maybe we need something other than Bikeshare.  For the amount of money involved, it seems like there would be a lot of scope for action.  You can now read my initial look at dockless bikeshare here.  To a first approximation, that would cost the taxpayers nothing, but it is far from clear that dockess providers would have any interest in serving Vienna and similar suburban locations.

But by way of illustration, just consider what sort of home-made “dockless” system you could create for the cost of just one $45,000 Bikeshare rack.  That would be 300 hybrid bikes from Walmart, with lights, lock, and helmet.  (No helmet with Bikeshare unless you bring it yourself.)  You have to wonder if we couldn’t get 300 locks keyed alike, paint the bikes a funny color, let responsible people buy a key for $10, put up some bike racks around Vienna painted the same funny color, and be done with it.    If you are a key owner, and see one of the bikes, you can use it.  (This ignores the “casual” user, but Capital Bikeshare needs the casual (credit-card-swipe) users because they are lucrative, not because they account for a lot of trips.)

A crude “homegrown” system would some operating costs.  You might have to hire some high school kids to “rebalance” the system (move the bikes around) in their spare time.  The bikes would wear out, and some would be vandalized or stolen.  But even if you lost half the bikes every year, you’d have more bikes available, better point-to-point coverage, and vastly lower cost.

My third conclusion — a guess, really — is that nobody is much paying attention to the value aspect of this (bike trips per dollar invested) because the funding and operational arms of Capital Bikeshare are two completely separate entities.  As an economist, I find that sort of thing disturbing, and think that this investment needs to be assessed in terms of the value it brings (or does not bring) to the community.

To put it bluntly, Falls Church seems to do things a lot smarter than we do, when it comes to development.  Falls Church hired a consultant to study the issue, with a final report issued two years ago (.pdf).   Why doesn’t Falls Church have Capital Bikeshare?  Edit:  As it turns out, Falls Church is getting Bikeshare — paid for with Federal tax dollars.   Is it because they’d have to spend their own money to do it?  If so … I think this should give us pause, even if Fairfax County is paying for this.

Additional analysis that I may or may not do

My guess is, nobody will pay attention to a recommendation of “don’t do that, it costs too much for the apparent benefit”.  So, what I need to do next is to see if I can glean any information about what characteristics are likely to make a location well-used (other than being literally located at the Metro station.)

The only thing that appears, by eye, in the Workbook above is that the volume of trips at a location is largely driven by commuting.  Stations with high volume have a high proportion of work-related (commuting-related) trips.  And conversely, stations with low overall volume had a low proportion of commuting trips.

For Vienna, to me this suggests prioritizing work-related locations over leisure-related locations.  I note, for example, that the Reston library stop is one of the least-used stops in the entire Reston cluster of bike racks.  That would be consistent with a model where commuting is the principal driver of demand for Bikeshare rides.

Obviously, I could dig into this further.  The obvious next step would be to see how much total ridership would drop as I began dropping the least-used stations out of these two clusters (including all trips originating or ending at those stations).  This would begin to get at a cost-benefit tradeoff, with a goal toward sizing the Vienna network of stations with value in mind.

But at this point, I think I have done enough.  Although this was not the “systematic” model I had hoped for, I think that the analysis of use in our sister communities of Reston and Tysons has provided some guidance as to more- and less-promising locations for Capital Bikeshare racks around Vienna.

Further reading, Bikeshare feasibility studies.

City of FairfaxAnd the actual report (.pdf). Notable because it predicts very high usage at Tysons (which has not occurred), but suggests little demand within Vienna itself (page 16).  This is, I think, the only study that directly addresses Vienna.

Merrifield (Mosaic District), .pdf

Falls Church (.pdf)

Reston (.pdf), particularly interesting because they projected 45,000 trips in their second year of operation, which would be 2018.  They actually got a little over 8,000.  The vast gap between predicted and actual use in Reston should be a huge caution to anyone who is tempted to take these feasibility studies at face value.  (Or, possibly, that indicates that demand models based on more urbanized settings really do not work well for the less-dense suburban areas.)

Financial data for Arlington County bikeshare (.pdf) This is the only financial analysis I could find, but many feasibility studies around the country offered pro-forma (guesstimate) financial analysis.

A discussion of the public-policy aspects of Capital Bikeshare can be found at this link.

In general, Googling “bikeshare feasibility study” brings up such studies from all over the US.

See this page for my quick overview of dockless bike rental.