Know your long commutes

Apples and oranges

This page is compiled from a series of blog posts written in September and October of 2011.

The most frustrating thing about the latest Census report is that it lumps together car commutes and transit commutes. Many people (for example, the Consumerist, the the Asbury Park Press, and the Los Angeles Times) recognize that an hour on transit is easier to deal with than an hour behind the wheel, but nobody seems to have bothered to separate them out. As far as I know, I'm the first.

Here are the ten cities with the longest commutes, lumping together all modes like the L.A. Times and Marketwatch did, following the example of the Census Bureau themselves (PDF). I've put in the overall Mean travel time, but I've also included a column for "Drive time" that includes only car, truck and van trips, and a column for "Transit time" that includes only transit trips. The extra time that transit users spend is labeled as the "transit time cost":

You'll notice that with their large transit mode share, the average 50-minute transit time brings up the average New York City commute four minutes, and the 67-minute transit time from Bremerton-Silverdale brings their commute time up two minutes. But the small number of transit commuters in San Juan, Riverside and Atlanta actually bring the overall travel time down by a minute each. If we just look at drive time, we get a different picture:

This puts DC first, and New York City ninth, in drive time. Baltimore, Atlanta and Yuba City are all ahead of Chicago. Bremerton-Silverdale has slipped to 27th place, and tenth place is now taken by Vallejo-Fairfield, California, formerly eleventh.

In the past, this "longest commute" story has been used to drum up sympathy for drivers stuck in traffic and glorify small towns. Well, not in New York. Here in New York, we're in the top ten for drive time, but we're not number one. We're behind DC, San Juan, Atlanta and Chicago, as well as the crazy exurban drivers in Poughkeepsie, Riverside and Yuba City.

About those long transit commutes...

When people talk about long transit commutes, it's usually with an attitude of pity for the economically disadvantaged transit riders. But when I looked at the long commutes tallied in the 2010 American Community Survey, the second-longest commute, from Poughkeepsie, jumped out at me. You see, I've ridden that train many times. It's not bad. You can sleep on the train:

These are not the typical "long transit commuters" that are normally invoked in pity-oriented articles; they're well-dressed white professionals. So that got me wondering about the incomes of transit riders, and sure enough the American Community Survey has them. Here are nine metro areas where transit riders have higher incomes than drivers, and Ponce, Puerto Rico makes ten:

So here we see that the median transit commuter in Idaho Falls made $61,214 in 2009, while the median single-occupant driver made only $25,607. Torrington, CT was more dramatic because the incomes were higher: transit commuters made $82,431 while drivers only made $41,540.

First, a note on these figures. The American Community Survey (PDF) asks, "How did this person usually get to work LAST WEEK? If this person usually used more than one method of transportation during the trip, mark (X) the box of the one used for most of the distance." So if you drove 30 miles of a 61-mile commute, you would count it as a transit commute. Even if the 30 miles of driving took 45 minutes.

The answers to those questions are tabulated in Table B08301 on the Census Bureau's American FactFinder. I combined those with Table B08121 for the earnings data, and B08136 for the aggregate travel time.

To get the list above, I eliminated those cities where the ACS had no data for transit rider income or population, and those where the margin of error was more than 25% of the median transit rider's income. I have marked in pale red the metro areas where the difference in median income levels is below the margin of error for the transit riders. In those cases, the survey is not reliable enough to tell us whether there really is a difference in median income between the two groups, so we can treat them as basically equal incomes. Here are the population and commute time figures:

There are no commute time figures for Kingston, but the commutes are roughly the same as Poughkeepsie-Newburgh-Middletown. These two areas, plus Torrington, Trenton, and Bridgeport-Stamford-Norwalk, form the outer ring of suburbs around New York City. In all these places the rich people are the ones who can get jobs in the city, and the lower-income ones have to work nearby. Some of the rich commute to the city by car, but many take comfortable commuter trains or buses. Local transit is pretty crappy, though, so the people who work locally have to drive - and in many cases spend a large chunk of their money on driving expenses. In Trenton and Stamford, there is enough high-paying "job sprawl" to bring the incomes of the drivers up.

When I first looked at this list, I was sure that Idaho Falls was an error. People making $60K a year, spending 69 minutes a day on a bus in Idaho? But it turns out that a major employer in the area is Idaho National Laboratory, which runs a fleet of comfortable motor coaches to bring people to work. Apparently there's not much of anywhere to live around the labs, so all these nuclear physicists live in Idaho Falls. Roughly a third of the 4,000-person staff takes advantage of the opportunity to travel with their colleagues on free buses rather than falling asleep on the desert roads by themselves. This has been working well for sixty years, but recently the lab decided to replace the local stops with park-and-rides, essentially forcing every worker to drive partway. They seem to be spinning it as an overall win-win, but it's forcing the individual workers to spend more time and money on fuel and emissions (and perhaps even buy a car that will get driven a few miles and then sit in a lot all day) to save the lab a little time and money.

Bremerton-Silverdale, as mentioned previously, covers the area across the Puget Sound from Seattle, and the commute basically involves a comfortable hour on the ferry. Vallejo-Fairfield is an exurb of San Francisco, with bus/BART and ferry connections, as well as buses to Sacramento. I haven't been able to figure out what's going on in Ponce, but the average transit trip is only nine minutes longer than the average drive.

In sum, we shouldn't feel too bad about some people with long transit commutes. Many of them are pretty well-off, and they get a chance to take a nap or play solitaire. Of course there are plenty of poor people with long transit commutes, often involving multiple buses. Those are people we should work to help.

Travel time by Combined Statistical Area

One issue that came up when I posted the previous sections on "cities" (census-defined Metropolitan and Micropolitan Statistical Areas) where the income of the median transit rider is higher than that of the median driver is the fact that many of these MSAs area exurbs of larger cities. In fact, many of the MSAs in the top ten were part of the outer ring of New York City suburbs. I thought it would be interesting to look at Combined Statistical Areas instead: areas that the Census Bureau judges to be part of the same metropolis.

There's Idaho Falls at the top again, and the Chicago CSA (which combines the Chicago MSA with Kankakee and Michigan City) coming in second. I still haven't figured out what's going on with Ponce. Then come the other big cities: Boston, San Francisco and Washington. New York as a whole (combining Torrington, Kingston, Poughkeepsie, Trenton and Bridgeport with the New York City, Nassau-Suffolk, New Haven, Newark and New Brunswick MSAs) drops to number 10 on the list. Some big US metro areas are missing from the top ten: Philadelphia has dropped to #19 and Houston is at #26. Los Angeles, Dallas and Atlanta are replaced by Seattle, Portland and Youngstown. Yes, in Youngstown, that infamously shrinking city with its nice clockface transit service and a frequency-based map (PDF), transit riders only make a few thousand dollars less than drivers. To be honest, I don't know what's going on there.

In all these cities, the travel time ratio is pretty similar: except for Idaho Falls and Ponce it's between 1.55 and 1.81. Seattle, San Francisco and New York have ferries. Boston and DC, in addition to the above three, have commuter rail. The mode share doesn't seem to affect either the earnings ratio or the time ratio.

The main thing I take away from this list is how close together these CSAs are, with the exception of Idaho Falls, and how low the earnings ratios are. That suggests that in an entire metropolis there are going to be pockets with a high percentage of poor transit riders lorded over by a small elite of drivers, and those will typically balance out the Torringtons and Bremertons of the area. But that doesn't mean that there's no inequality. Here's the bottom ten CSAs:

Here we've got five Midwestern rust belt cities, four mountain Southern cities, and Syracuse. Drivers make three to four times what transit riders make. This is the more typical pattern that everyone expects.

None of these cities have rapid transit. Detroit has the People Mover and Little Rock has a heritage streetcar, but those don't actually go from homes to jobs. Syracuse used to have Ontrack, but that was shut down in 2007; ridership had been declining because it didn't go from homes to jobs.

I still can't figure out why there's so much inequality in Detroit, Syracuse, Lafayette and Lansing, but not in Youngstown. Maybe people driving to New Castle to get the bus to Pittsburgh? But in the 2005-2009 ACS, the mean drive time is 22 minutes and the mean transit time is 32 minutes. If anyone has an explanation, please tell me.

Wealthy transit riders by county

Since the distribution of incomes seems to flatten out across a region - or at least, to reach a maximum of about 1 - it seems that the diversity happens at the county level. Accordingly, I tabulated a list of the counties with the highest ratios of transit incomes to driver incomes:

There are some surprises here. What's going on in Saint Lawrence County, way up by the Canadian border? Is it something like we saw in Idaho Falls? I'm guessing that it's faculty commuting to SUNY Potsdam or Clarkson University. In any case, there are only 92 transit riders in the county, so maybe one of them can email me.

The case of Cherokee County, Georgia, is more straightforward. Just about all 457 of them pay $125 a month to ride an express bus into Atlanta several times a week. A massive HOV/BRT plan was shelved in 2009.

Litchfield County, Connecticut is the same as the Torrington Micropolitan Statistical Area discussed in my earlier post, and Ulster County, NY is the same as the Kingston μSA. The Illinois counties of Kane, DuPage and McHenry are all suburbs of Chicago reachable by Metra. Stafford and Spotsylvania counties in Virginia are southern suburbs of Washington, DC served by Virginia Railway Express, and Bristol County, Massachusetts is a suburb of Boston served by the MBTA commuter rail.

The 39 counties with an earnings ratio greater than 1 include suburbs of just about every metro area in the country: Fort Bend, TX (Houston, 1.71), Howard, MD (Baltimore, 1.69), Kitsap, WA (Seattle, 1.50), Chester, PA (Philadelphia, 1.39), Dakota, MN (Minneapolis-St. Paul, 1.28).

One other thing: of course you don't want your county to have too low an earnings ratio, because that means that all the poor people are being pushed onto transit. On the other hand, you don't want it too high either, because too many poor people driving means your local transit system sucks. A ratio of 1 is ideal; that's what you'd get if everyone took transit. Broadly speaking, it seems like there's an acceptable range from 0.698 (Multnomah County) to 1.25 (Bergen County), including those closest to 1, Norfolk County and Middlesex County, Massachusetts.

What "choice riders" will put up with

There's a school of transit planning that divides passengers into "captive riders" and "choice riders." The captive riders are your stereotypical transit riders who are either incapable of driving or unable to afford cars. They're physically disabled, mentally disabled or poor. Often they're some combination of the three, and maybe even immigrants or visible minorities as well. Many people support government-funded transit as a form of charity, but like any form of charity there are people who feel it's a waste.

The "choice riders" are people who can drive and can afford cars, but choose to take transit instead. Many transit planners treat them like kings who must be pleased at all costs. They get big, cozy seats on their buses and trains, and parking at the station. Lots of parking. The thinking is that if they are displeased they will just drive instead.

There are a number of problems with this line of thinking. First of all, it glosses over the fact that not all transportation choices are equal. I've identified at least four kinds, which I call the Single Trip, Habits, Investments and Subsidies.

Second, there are also "carfree by choice" households that chose not to make Investments in cars, and thus are "captive riders" from a practical point of view. We are able-bodied, of sound mind, with driver's licenses, and could theoretically afford to buy and maintain a car, but we choose instead to go on European vacations, or pay off our debts, or do any number of other things with the money. We are everywhere, but we tend to be concentrated in walkable cities with good transit. This was the core of my critique of the Eric Morris post I linked to at the top of this post. In fairness to Morris, he was referring to data from a 2001 John Pucher article (PDF). Also in fairness to Morris, it was stupid of him to take an article based on nationwide data and try to apply it to New York City without even considering the possibility that there might be differences in distribution of riders and choices.

The biggest shortcoming of that approach is a limited understanding of the choices people make. We are not just choosing transit or no transit. We have goals, the main one being to get to work. We are choosing between transit and driving. The key is whether transit helps us accomplish their goals better than driving.

It's true that bad service, high fares, low frequency and long travel times will factor into someone's decision to drive instead of taking the train. But only if it makes transit a bigger hassle than driving. The NHTS data that Pucher and Renne cite, and the ACS data that I discussed this past week, show that even middle-class white people will sit on a bus for a long time. Why do they do that? Because driving is worse.

There are many ways that driving to Manhattan, or Stamford or White Plains, in the morning rush hour is a pain in the ass. Traffic on the inbound highways is often backed up for long distances, even on I-684. You can't read or write, or play video games, without putting yourself and everyone around you in more danger than you already do by just being behind the wheel. You can't relax, sleep or drink alcohol, even on the way home. When you get to the city, unless your job provides free parking you have to circle for a space - or more likely, pay for one. Really wealthy suburbanites can afford to rent a parking space, but most commuters have something else they'd rather do with the money.

The conventional wisdom says that "choice riders" won't put up with all these transit hassles. Well guess what? In the greater New York area they put up with them, because the other choice is worse.

I want to make sure you don't just dismiss this as, "This is New York, and other places are different." It's not just New York, it's also Bainbridge Island and Idaho Falls and Vallejo. It's anywhere that driving is difficult and/or expensive.

All these places may have some pre-existing feature that makes driving a pain. In New York it's the rivers and the density. In San Francisco it's the bay, and in Seattle it's the Puget Sound. In Idaho it's the desert and the nukes.

But what about other places that have natural obstacles? DC has the Potomac. Miami has bays. Los Angeles has mountains. Phoenix has the desert and Los Alamos has nukes. Why don't they have transit commuters from their wealthy exurbs?

Idaho is a weird case. Mostly, I think, it has to do with "trip-chaining." In a place that makes it easier to get just about anywhere by car than by transit, commuting by car makes it easier for the commuter to chain trips, so they can drop the kid off at school on the way to work, or pick up the dry cleaning on the way home. I don't think they have a school or a dry cleaner's in the middle of the desert on the way to the labs, so it makes sense to take the bus to the lab and back, and take care of the other errands separately. In Los Alamos, by contrast, most people live in town with the labs, but if they drive from Santa Fe they can stop in Pojoaque or Tesuque on the way.

Parking availability could have something to do with it, as Morris suggests. I wonder if the government limited the amount of parking available at the Idaho National Labs. In the other cases, it would be interesting to see if there's any correlation between the availability of parking in the various central business districts associated with these areas and the transit mode share among upper-class commuters. Certainly New York and San Francisco have limited parking available.

The answer also has something to do with freeway revolts. New York stopped Lomex, Westway and the Mid-Manhattan Expressway, and tore down the West Side Highway. San Francisco tore down the Embarcadero Freeway and part of the Central Freeway. Vashon Islanders stopped a bridge across the Puget Sound. These have made it harder to get to jobs by car - or avoided making it easier. In DC, Miami, Phoenix and LA, the approach seems to have been to build as many highways and parking lots as possible. It's no wonder that people drive.

Returning to the original point: the "choice riders" made their choice based on the relative value of transit as opposed to driving. Transit providers only have to bend over backwards to please the "choice riders" if someone is simultaneously making it easier for them to drive. And even then, at best they will only postpone the inevitable.