Why is Bicycle Commuting Down? More theories from the comments

I got several excellent comments from various people to my post about why bicycle commuting has fallen since 2014.  Some comments disputed some of my thoughts while others provided interesting theories of their own.  I’ll summarize these responses.  Here is the graph again:

I indicated the source as “US Census Data,” but I should have been more specific as one comment suggested that since the “census” is only done every 10 years that these numbers are completely made up.  These numbers come from the American Community Survey and they are updated annually.

The Decrease Isn’t Really Real

A small number of responses were of this type.  They said that the ACS is a survey and there is insufficient data to reflect a real decline.  It’s theoretically possible that the numbers for the last 3 years are erroneous and that the number of bicycle commuters is actually growing, but I doubt it.  The ACS is commonly used by cycling advocacy groups to point to successes – see this report by the League of American Bicyclists (LAB) which comes from the ACS data.

In general I think the LAB’s approach is a good one – attempt to measure municipalities’ cycling friendliness score and then use the ACS data to report successes.  This can be a powerful argument for cycling advocacy – sprinkle some cycling infrastructure improvements and watch it grow.

Given the success in building cycling infrastructure in lots of areas in the US, are we bumping up against a theoretical ceiling for cycling?  I don’t think so but I admit these ACS numbers make me less sure.  One comment suggested that looking at bicycle sales numbers as evidence that the decline is an anomaly. A quick search for those numbers doesn’t look optimistic to me.

Commute Times are Increasing

A few comments suggested looking at workers’ commute times which are, in fact, increasing over time.  Interestingly I couldn’t find any data on commuting distances.  I would imagine that there should be data that analyses congestion over time by looking at commute times and commute distances over time.

By the numbers, all of the new workers over the last several years are choosing to commute by car.  They apparently aren’t bothered too much by the longer commutes/congestion or maybe they don’t feel like they have a choice.  As commutes get longer, at some point along that continuum the choice to bike to work is no longer a realistic option.

The Places Where Cycling Is Easy Are More Expensive

A few folks responded with this argument including economist Arnold Kling:

Perhaps rents went up in some locations, forcing some would-be bike commuters to move out of range

This is not something I thought of prior to writing the post, but this makes a lot of sense.  The areas where cycling infrastructure is excellent or improving are also affected by the housing crisis.  Could it be that local zoning laws are indirectly working against cycling?  Seems like a reasonable theory worth exploring.

The Connected Generation Can’t be Disconnected Long Enough to Ride a Bike

Also, it is much harder to text on your phone when cycling, versus Uber (or even sneaking in texts at stoplights). For the ‘connected generation’ on-road cycling may be an unacceptable interruption. If they want exercise they can use a stationary cycle.

Steve here

This seems logical to me although I have to way to verify it or gauge how important its effect is.  I’ve always thought of this as an advantage to riding a bike, not a downside.  I’m sure not everyone feels the same way.

Jonathan Haidt Responds on “Helicopter-ness”

I asked Jonathan Haidt (author of The Righteous Mind and Coddling of the American Mind) what he thought about the timing of the decline and the helicopter parent theory.  He responded:

lenore skenazy and i found some data on the sharp decline in biking to school for kids, i think it happened in the 1990s. 2015 would not be a special year, for adults; it was special on campus.  i would not interpret the 3 year decline for adults as anything yet, could be random fluctuation. unless you can get data broken down by age. if you find no change for those over age 30, but yes for under 25, then it is Gen Z. 

I looked for data on bicycle commuting and age via the ACS.  Oddly enough, you can get raw bicycle commuting numbers normally, but when you group by age they are lumped together in a single group with “Taxicab, motorcycle, bicycle, or other means.”  This category is increasing over the timeframe of the bicycle decline which as far as I can tell makes this impossible to isolate bicycle commutes unfortunately.  If anyone knows how to use the ACS to isolate bicycle commuters please let me know.

Here are a couple charts that show the difficulty.  Even though there is a drop in bicycle commuting, this larger group is increasing over time.  Bicycle commuters are only a small part of this group so it’s hard to tease out any effect here.

Randal O’Toole Seems to Doubt the Work from Home Theory

I knew Randal O’Toole was an avid cyclist, so I emailed him to see if he had any ideas.  Here’s how he responded:

Those are good ideas. Here’s an additional datapoint: people who work at home earn the most money. Drive alones are second, transit third, and “other” (which includes cycling) well below that. Walking is lowest.

So if work-at-homes captured cyclists, it was the high-end cyclists. I’m not sure how many of those there were.

A more significant phenomenon: the census form asks people how they “usually” get to work. The National Household Travel Survey found that people who usually drive in fact drive 98 percent of the time but people who say they usually bicycle in fact only bicycle 70 percent of the time. Maybe some of those people who bicycled, say, 60 percent of the time in 2014 went to 40 percent in 2017.

Randal said a couple things that surprised me here.  One is that cyclists as a group earn less money than the average worker.  I tried to confirm this with the ACS survey and it’s true that the group that contains bicycle commuters earn less than average, but I suspect that, like age, this is similarly confounded by grouping “Taxicab, motorcycle, bicycle, or other means” together.  A quick search shows that in Australia cycling commuters are wealthier than average.  I also discovered from survey data from the NC DOT that cyclists that use greenways strongly skew rich and white relative to the local population.

Randal’s theory about percentage of time commuting seems to jive with the housing costs theory previously mentioned.

General Safety Concerns/Helicopter Generation

A large number of comments suggested that cycling is, in fact, getting more dangerous over time due to drivers texting, more cars, more cycling-intolerant drivers, etc.  I agree that it seems there is a general sense among cyclists that of course it’s getting worse on the roads.  This has always struck me as somewhat of a moral panic.  I’ve had my share of cars purposely buzzing me on my bike and yelling crazy things at me but I don’t get the sense from my own experience that these incidents are increasing over time.

Cycling advocates consistently talk about the large number of people who WOULD bike if they could be made to feel safer on a bicycle.  This approach seems to have been successful in getting municipalities to devote resources to large cycling infrastructure projects.  The current generation seems to be especially convinced by appeals to safety but there’s a bit of a paradox.  If people believe cycling is TOO dangerous they’re not going to do it no matter how good the infrastructure is.  (Safety first!)  But if people feel that cycling is safe then there isn’t much demand for new infrastructure to be built.

Are the roads getting more dangerous over time?  I decided to try to answer that question.  It doesn’t feel like it to me but let’s look at the data.  The National Highway Traffic Safety Administration has kept statistics on crashes for a long time.  They produce an annual report that has all sorts of data on traffic deaths.

First off, I would have intuitively expected to see deaths from distracted drivers increasing over time (texting, obviously), but that doesn’t appear to be the case.  (Not sure why we see that sudden decrease from 2009 to 2010 – anyone know?)

OK, well how about deaths from drunk driving?  Are those increasing?  No – it turns out they’re way down.

What about motorists that exceed the speed limit and cause a crash that kills someone?  Getting worse over time?  No – those are down too.

Note that the charts above are pessimistic since I didn’t normalize on vehicle miles traveled which has been steadily increasing over time.  That is, if we accounted for miles traveled the charts above would look even better than they already do.  Here’s a chart of vehicle miles traveled by year.  (Note I adjusted the y-axis minimum for better readability.)

If we reflexively respond to safety concerns in a thoughtless way we sometimes end up with poor solutions that are sometimes worse than the problem.

Most Unique Response: Fade Out of the Lance Armstrong Effect

From my local cycling friend Brendan:

The post Lance Armstrong cycling era taking effect.

I think the high end cycling world has definitely taken a hit – at least anecdotally from my perspective in a cycling heavy area.  I’m reasonably confident (65%) USA Cycling could produce some reports based on road race participation over the last decade and it would show a steady decline.  Would this have an effect on bicycle commuting?  Probably but most likely the effect would be small.

Where is the Cycling Ceiling in the US?

It ultimately boils down to this for me as I alluded to earlier: how much room is there to grow cycling in the US?  Of course I don’t have the answer to this but these numbers make me think that perhaps we’re closer to the ceiling than I would have guessed.  And given the headwinds of safety consciousness, housing costs, the price of gas, and the rise of telecommuting, how big could gains from infrastructure improvements be?

I’m starting to think about a way to come up with an answer to this question on my own from some other data sources.  Look for a post soon.

Offer to Bet

If you believe the drop in bicycle commuting is temporary, I’m offering a bet at 1:1 odds that the bicycle commuting number won’t exceed the number from 2014 in years 2018, 2019, or 2020.  Any takers?  Send me a DM on Twitter: @davemabe.

Supposed Economic Benefits of Greenways

(I support greenways where they make sense.  In fact, I’m on the Greenways Commission for the Town of Carrboro.)

You don’t have to look hard to find outlandish claims made in support of various causes these days.  For example, see Trump’s “Trade wars are good and easy to win” claim.  Occasionally though I come across something so egregious that I just can’t let it pass without commenting.

The NC DOT funded a three year study to quantify the economic benefits of shared use paths (SUPs a.k.a. greenways).  Their conclusion?  For every $1 spent in trail construction (once), $1.72 is generated ANNUALLY.  That’s a pretty amazing statistic.  Let that sink in for a minute.  Is that really possible?  If someone claims they can easily make a 72% return on your money every single year, do you respond with healthy skepticism?  You should!  This is being reported as a return on investment.

Which SUPs Should We Look At?

Let’s take a look at the study and see how they arrive at these numbers.  Of course you can’t do a detailed study of every SUP in North Carolina, so you have to choose a sample that’s representative of all SUPs in NC.  So how did they decide which SUPs to study?

The shared use paths studied were selected
because they:
• Have a state or regional significance.
• Have good opportunities to capture
economic revenue.
• Were not impacted by construction,
significant maintenance, or detouring
during the project period.
• Are relatively ‘established’ (i.e. at least
5 years old with minimal adjacent
land use changes anticipated).
• Have the ability to demonstrate a
transportation function.
• Are geographically dispersed across
North Carolina.
• Are a good mix between urban and
rural areas.

In other words, they picked ones that they could use to show maximum impact.  You literally could not pick 4 SUPs that would show a better economic result.  Not a choice I would have made but I get it.  Just one page into the study and you can see where this is heading.

Gathering Data

They got several people together and went to each SUP and set up a survey station.  This required 187 people across 3 years.  They visited some SUPs just once and others every year.  So it’s easy to imagine the situation – you’re running or biking along the SUP and you come across some friendly people asking you if you want to fill out a survey.  Of course you can’t force people to take a survey so some choose to do it, some do not.  Having spent a lot of time exercising on paths and greenways, I can imagine the type of person that on average would take a survey and, on average, would choose not to take a survey.  The people that choose not to take the survey are on average more likely to be working out and not willing to stop.  Of course these trips don’t generate any income at all.  Excluding these SUP users are good for the final numbers!  Pretty convenient!

Questions on the Survey

The survey actually looks pretty good.  They ask some good questions and it’s pretty thorough.  Race, gender, salary, how many people in the group for this trip, how long is your trip, transportation type – all reasonable questions.  They also ask how much money they’re going to be spending on this trip and how many trips they typically take.  They use this to produce tables like this one – which breaks down expenditures by type and transportation method:

This seems fine, but as I look closely at I notice that runners sure are buying a lot of groceries!  11% of the 663 users that were running are spending $41 per trip on groceries.  18% are running to and from a restaurant to eat.  4% are on their way to buy some consumer products at $97 per trip at retail.  All in all, 35% of all SUP users who are running are doing so to go spend some money.  That seems extraordinarily high!

A Useful Result: SUP Users are Rich and White

I did learn one interesting thing from the survey results.  SUP users skew overwhelmingly rich and white.  The second largest income bracket for users is >$200K.  About 75% of SUP users earn above the US median household income of $59K.  This is not something I’d thought much about before but it makes sense.  I would have like to have seen an income/race breakdown by transportation type.

Users are overwhelmingly white at 83%.  This number might not sound too skewed until you consider the fact that this SUP is in Durham where 37% of the population are people of color.  I never realized that SUPs were this much of an amenity for the rich.  But, hey, this is also convenient for the purposes of the desired outcome of the study!

Other Economic Impact Results Touted

The infographic at the top of the post touts not only the amazing “spend $1 now and get $1.72 annually forever” statistic.  There are other stats that are asserted.

“A one time $26.7M capital investment in the four greenways” – ok first of all this number has to be way lower than the actual cost of the greenways.  I expect the land alone probably costs that much.  There’s obviously something I’m not understanding about that number.  The implication is that $26.7M is the total cost of the SUPs and I just can’t believe it’s that low given how many bridges are on the ATT and how expensive bridges typically are.  For example, in Carrboro they spent $1.3M on a quarter mile of greenway with a bridge in it.

“$48.7M in estimated business revenue from greenway construction” – a one time $26.7M investment produces $48.7M in business revenue from greenway construction?  Does that make any sense?

“790 jobs are supported annually through greenway construction” – this is truly an amazing statistic that seems so absurd that it doesn’t deserve a response.

If These Stats Are Correct, What Does that Mean?

If you really can spend $1 and get a $1.72 annually, then it would make perfect sense to cancel all of North Carolina’s public projects and simply build SUPs everywhere.  Sure there would be decreasing returns the more we built but there’s a tremendous amount of room to play with.  Even a 4% annual return would be excellent!

The fallacy of the study is that it implies that the greenway creates the economic activity.  It does no such thing. One category of SUP users made trips to restaurants.  If the SUP didn’t exist, would those people starve?  Would the $1.72 instantly evaporate if the SUP went away?  No – people would quickly find alternatives.

A good question you should always ask yourself is “compared to what?”  Imagine we used this same methodology to evaluate the economic benefits of a road through a similar study.  We survey people that drive on the road.  Where are they going?  How often do they use it?  There are a lot more road users per square inch of road than a SUP – it’s pretty easy to see that the $1.72 number would be DWARFED by the same number for a road.  A lot of people use the road to drive to work.  Do we say that the road created that job?

No that would be absurd.

I wonder how much we paid for this study – it makes us all look silly and probably ultimately shows more evidence against a greenway than for it.