Automobile Bikes Light Rail Urbanism

Transit Mode Share Trends Looking Steady; Rail Appears to Encourage Non-Automobile Commutes

» Results of the 2009 American Community Survey show major declines in carpooling, significant increases in biking.

Just how effective have new investments in transit been in promoting a shift of Americans towards public transportation? Has the recent livable communities movement resulted in increased commuting by bike or by foot?

The Census’ American Community Survey, released at the end of last month with the most recent 2009 data, provides a glimpse of what can change over nine years. These data are approximations in advance of the much bigger (and more accurate) sample set that is Census 2010, whose results will be released next year. The information detailed here applies to commutes only, not all trips.

By looking at America’s 30 largest cities — from New York to Portland — we can get some idea of how people are choosing to get to work, and how patterns are changing based on the availability of alternative transportation modes. I have chosen not to analyze metropolitan regions as a whole because I want to focus on the effects of improvements to transit systems and increasing walkability, two characteristics common to center cities but not necessarily to their suburbs. This biases the information, especially for places like Washington or Boston, where the central city represents a relatively small percentage of the overall regional population.

Nevertheless, the data demonstrate a number of interesting trends. Most notable are the huge declines in carpooling and large increases in biking noted over the largest cities. As the chart below shows, over the past nine years, carpooling’s mode share decreased on average by 25.9% and biking’s share increased by 58.5% (note that these are percent changes, not point changes, which are documented in a chart at the bottom of this article). The declines in carpooling were matched with a slight uptick in single-person driving, a 1.5% increase, and a decrease in transit share of 6.4%. These mode shares are not the same as total modal use; it is possible for transit ridership to increase even as modal share goes down (for instance, if city population increases), and vice-verse.

% Change in Mode Share, 2000-2009, Averaged Across America’s Biggest Cities
 Total AutoTotal Non-AutoDriving AloneCarpoolingTransitBikingWalking
All Cities-3.4-2.01.5-25.9-6.458.51.8
Cities with no rail-1.4-6.32.9-23.9-8.040.3-2.7
Cities with rail, but no major new rail investment-
Cities with major new rail investments-
Non-Texas Cities with major new rail investments-5.95.4-0.1-

Overall, the percentage of people commuting by automobile declined by 3.4%, and the mode share of those using non-automobile modes decreased by 2.0%. It was possible for both to decline because of an increase in people not traveling to work at all but telecommuting.

Though these numbers show little change in use for automobile and transit overall, they do provide some clue as to the effects of rail investments. When comparing cities that have no rail lines with those that have existing lines or have invested in new ones, a correlation between rail and transit use is apparent. Cities with no rail saw far smaller declines in automobile mode shares than their rail counterparts; they also saw declining non-automobile mode shares, compared to increases in the rail cities. These differences were especially considerable when considering rail cities outside of Texas; excluding them, transit saw no mode share change, whereas single-person commuting by car decreased (albeit by a minuscule amount).

This may indicate that rail lines can play an important role in encouraging the population to try modes other than the automobile. The non-automobile mode share, which includes transit, biking, and walking, is particularly interesting from this perspective because it may reflect the number of people choosing to live in areas where it is acceptable to use transportation other than the private car. Is this conclusive evidence that rail works better than bus service to encourage people out of their cars? Not necessarily, but it’s certainly a part of the overall equation.

Looking city-by-city, modal share changes reflect some overall trends. Automobile usage continues to decrease in the nation’s older, densely developed cities: The places recording the largest declines in overall car share were, in order, Washington, New York, Boston, San Francisco, Seattle, Portland, and Chicago. Those with the largest declines in non-automobile share were largely sprawling cities, including, in order, Columbus, Houston, Dallas, Fort Worth, Las Vegas, and Nashville.

% Change in Mode Share, 2000-2009 in America’s Biggest Cities
 Total AutoTotal Non-AutoDriving aloneCarpoolingTransitBikingWalking
El Paso-2.414.44.3-35.02.547.826.5
Fort Worth-1.5-16.44.7-29.91.5-18.2-31.5
Las Vegas-0.1-13.75.5-27.5-28.5-10.718.7
Los Angeles-
New York-12.63.3-5.6-
San Antonio-0.6-9.94.2-24.6-12.0-11.7-6.2
San Diego-1.5-13.23.5-31.4-12.414.6-19.6
San Francisco-9.66.2-3.9-
San Jose-2.1-2.10.0-13.3-21.343.132.8

The places recording the largest increases in transit modal share were Nashville, Washington, Austin, Seattle, Los Angeles, Charlotte, and Boston. All but Austin, Boston, and Nashville have spent hundreds of millions of dollars investing in expanded rail transit systems; Boston already has a large one. Portland, unsurprisingly because of its municipal investment decisions, had the largest modal increase in bike usage, but other cities less known for biking like Baltimore, Detroit, Philadelphia, and Chicago also saw significant increases as well.

How can we explain the significant public transportation mode share declines in Houston and Dallas, two cities that invested considerably in their respective rail transit systems? Both saw increases in ridership of their transit systems between 2000 and 2008: Houston saw a 1.05% increase, Dallas a 11.7% jump. Those increases, however, were entirely lost by 2010, which has been a terrible year for transit in the two cities. At the same time, their city populations increased by 15.7% and 9.3%, respectively; transit improvements couldn’t keep up. This may be because of poor choices in public transportation investments or de-densification in the urban cores of these cities (or annexation, spreading the population out), but either way these are not model cities for transit investments.

I’ll conclude with the below table, which documents mode share in 2009 in the biggest cities of the United States. As the chart shows, automobiles have a majority share in all cities except New York, Boston, Washington, and San Francisco. Unsurprisingly, these are dense cities and the places in the United States with the most complete transit systems.

2009 Mode Share in America’s Biggest Cities
 Total workersTotal AutoTotal Non-AutoDriving AloneCarpoolingTransitBikingWalking
El Paso25587590.15.079.810.
Fort Worth33189492.32.880.611.
Las Vegas24568588.86.377.910.
Los Angeles174841977.615.767.110.511.31.03.4
New York373191728.765.823.55.354.90.610.3
San Antonio60644690.25.578.811.
San Diego62612684.97.476.
San Francisco43707346.445.138.97.431.83.010.3
San Jose44298088.56.076.412.

Louisville, the nation’s 29th-largest city, is not included here because it merged with the surrounding county, significantly changing demographics, in 2003. I have calculated “averages” not in terms of total trips, but city-by-city; thus modal share in Portland is considered just as important as that in New York, despite the latter being much bigger. Note city classification in the first table based on changes during the 2000-2009 period:

  • No Rail: Austin, Columbus, Detroit, El Paso, Fort Worth, Indianapolis, Jacksonville, Las Vegas, Memphis*, Milwaukee, Nashville*, San Antonio.
  • No new significant rail investments: Baltimore, Boston, Philadelphia*.
  • New significant rail investments: Charlotte, Chicago, Dallas, Denver, Houston, Los Angeles, New York, Phoenix, Portland, San Diego, San Francisco, San Jose, Seattle, Washington.

* The minimal nature of Nashville’s Music City Star means I won’t include it as a “significant” rail investment here. Nor will I include streetcar projects in such cities as Memphis and Philadelphia.

Update: Mode Share Changes 2000-2009 by Point Change
(One Point = 1% of All Commuting Trips)
 Total AutoTotal Non AutoDriving aloneCarpoolingTransitBikingWalking
El Paso-2.240.633.29-5.530.060.060.52
Fort Worth-1.38-0.543.60-4.980.02-0.02-0.54
Las Vegas-0.07-1.004.07-4.14-1.36-0.040.40
Los Angeles-2.921.321.31-
New York-4.132.11-1.39-2.742.090.14-0.12
San Antonio-0.54-0.613.19-3.74-0.46-0.02-0.13
San Diego-1.27-1.132.56-3.83-0.520.11-0.72
San Francisco-4.912.62-1.57-3.340.641.000.98
San Jose-1.90-0.13-0.04-1.87-0.870.270.47
Mode Share in Nation’s Largest Metropolitan Areas – 2009
Metro Area (MSA)Total WorkersDriving AloneCarpoolingTransitBikingWalking
Kansas City100355382.478.881.230.211.48
Las Vegas87444979.4710.333.180.341.78
Los Angeles580665573.6210.806.200.862.63
New York876535650.397.0230.500.406.28
St. Louis132869182.258.952.550.301.64
San Antonio91934879.3511.402.320.182.02
San Diego140641175.809.873.080.622.80
San Francisco208377561.8610.1914.591.544.40

Image at top: San Francisco’s Market Street, by Yonah Freemark

48 replies on “Transit Mode Share Trends Looking Steady; Rail Appears to Encourage Non-Automobile Commutes”

The final table, sorted by Non-Auto mode, shows how in terms of work travel, the Northeast Corridor is more or less a different country compared to most of the rest of the USA.

Throw in San Fran and it hammers home the point that while there is a lot at the policy level one can do to influence travel behavior, there’s really nothing like being an old coastal city that flourished well before 1900 to maintain a strong non-auto travel structure in 2010.

Patrick M is right on point. It’s not the age of the city that’s important, it’s when it grew. Portland does so well because it grew early (as did many coastal cities); much of the current city was already in place at the end of the streetcar era. Compare newer cities where 80-90% of the growth took place after 1950, and you’ll see the results in (auto) mode share. Even today, those neighborhoods developed before 1940 continue to have high transit mode share, regardless of income (often high) and even in the most car-dominant cities. It’s all about well-connected street networks in neighborhoods with at least a modest density and some notion of mixed land uses in the general vicinity.

The overall premise is interesting, but I can’t help but wonder if it would be more meaningful to show the percentile change instead. A 8.5% increase in the transit modal share for Charlotte sounds impressive, for example, until you get to the second table and see that all this has done is increase the transit modal share to 3.5%. A similar concept in reverse for transit-heavy cities.

(I find a similar problem exists in transportation policies, for example, to increase bike commuting by 50%, which sounds impressive until you realize that it means increasing the modal share from 2% to 3%, making little dent in auto travel.)

One other benefit of this approach is that the total net change should be zero, so it’s easier to see where the trips went if a sizable decline in auto share occurs (e.g., if the auto share declined by 5 percentile points, while transit increased by 4 and walking/cycling increased by 1).

Could you add a table on the metro area level, and not just central cities? In many American regions, transit looks good until you realize that it’s rarely used outside the core cities, people who commute to the core CBD excepted.

I’ve added the table (the last I’m going to add!) but I should point out that the reason I focused on the cities themselves, rather than the whole metro region, is that they are the places where investments in transit and other sustainable transportation infrastructure can really pay off. Though metro areas represent a larger percentage of the overall population than just the central cities, they are slower to respond to changes in transportation options and do not have the walkable environment in place for people to be able to adapt easily to alternatives.

While that may be true in some places, it doesn’t necessarily hold true in others. Many transit agencies operate at the county, or multi-county level.

Further, some cities (Los Angeles, for example) are much better represented by their metro area, as it’s just as important if someone is taking the bus in Hollywood (city of LA) as it is if they’re in West Hollywood (separate city, same metro)

“No new significant rail investments: Baltimore, Boston, Philadelphia.”

Why exclude major BRT? At around 25,000 riders a day, Bostons Silver Line (including 3 underground stations) is certainly a major investment.

And Detroit, Jacksonville and Vegas all have rail lines via people movers or monorails. While ridership is tiny, compared to, say, NYC, it’s still a decent amount when looking at a bus line in Oklahoma city as a comparison point.

At about 20,000 riders a day, Id say the Vegas monorail counts as a major rail investment, and it opened in 2004.

But that ignores the monorail. Of course, it seems like all official ridership collections exclude the monorail….

IMO, theres no difference between the vegas monorail and elevated heavy rail like the miami system. Both have similar pros and cons.

Ask residents if they think the MBTA Silver Line is an improvement over the Washington St Elevated that was torn down decades ago.

Or, look up the Sierra Club’s 20 page report on the Silver line and look at the picture on the back cover of the report, which as always, is worth 1000 words.

What was there before was nothing at all. But the industrial area at the waterfront didn’t really need expensive mass transit improvements….

The airport branch, while being a bit more functional, is hardly worth it, given the Blue Line-Red Line service.

i think the author’s strict rail criteria works fine in this case.

alternatively, one could group exclusive and partially exclusive ROW systems and thereby include light & heavy rail, BRT, & ferry, while excluding shared ROW systems (bus, streetcar). along those lines, the modes share similar rider catchment areas (approx. 1/2 mi and 1/4 mi, respectively)

Yonah, do you know how the ACS asks respondents to specify their mode? It’s my understanding that it requires respondents to pick one and only one mode that represents their commute. This troubles me because it prevents us from understanding those commuters for whom mode choice varies on a daily/weekly/seasonally basis (as it does for me personally – transit/walk, and for many colleagues – often auto/transit/bike). Maybe this would all be a wash in terms of the overall stats, but maybe not — especially since “auto” is such a large share, if a relatively small part of those people actually sometimes use transit it could significantly boost the “actual” transit share.

Perhaps more importantly, the real value of transit and the other non-auto modes is that they are available to be used when needed. If these stats only indicate those who mostly use non-auto, they undercount the percentage of mostly auto commuters for whom non-auto modes are still “relevant” — which is an important point to make politically. For instance, if a person drives most days because their main office is in the suburbs, but takes the train 1 or 2 days a week when they need to work downtown (or bikes only when the weather is nice, etc.), they show up as auto-only and the fact that transit (or bicycling) is relevant for them is lost in the data.

Is this correct?

Yeah, I agree that this is a problem with the ACS data. Many more people occasionally use a non-auto mode, and there’s seasonality to commuting as well—people are much more likely to bike in the summer, for instance. There is also a lot of noise in ACS information for biking since the share is so small. With only 0.5 percent, that translates to just a few dozen respondents in most large cities.

A major thing that strikes me with ACS commuting information is that walking is somewhere around 6× as popular as biking in most areas, so even a 150% increase barely makes a dent in most places. Biking can be very effective for getting people around since it’s usually faster than transit, but it has its drawbacks too (lockable storage on a bike is rare, for instance).

Other things that make me think: For metro areas, the highest mode share for transit, biking, and walking are many multiples of the lowest mode shares, but the highest carpool mode share is about 15%, while the lowest is about 7%—only a 2× difference. However, in most places, carpooling is the most common option other than driving alone.

From the first table, we should probably be really disappointed that carpooling has declined so much. It does seem like a lot of trips have simply disappeared, though. A combination of telecommuting and a poor job market in 2009, I suspect.

A major transportation survey in the Toronto area (the Transportation Tomorrow Survey) asks respondents about the trips they made on a specific date. In theory, that is supposed to balance out respondents that occasionally take a different mode — if five different respondents take transit one day a week, it’s likely that (on average) one will take transit on that particular day.

The ACS asks people to name the mode they used primarily in the previous week. It doesn’t have the granularity of asking about a specific day, but it’s probably the most consistent thing there is in a mailed survey, as opposed to a phone survey. Asking about a day may produce bias because people may not remember so far back, and asking about the day before the survey may produce bias because people may answer the survey only on specific days (say, weekends).

Part of the problem with any sort of broad based analysis here is that many cities have increased service levels with some investments and then cut service elsewhere. Maybe the proper analysis would be to measure transit VMT rather than “rail investment”.

Wonder what Jarrett Walker of Human Transit would say about this? He seems to have a distinct anti-rail, pro-bus bias.

I think that he would say that the densities of almost all American cities support bus more effectively than rail. I was struck in light of the arc tunnel cancellation by the discovery that the one bus lane through the Lincoln tunnel takes more commuters into Manhattan than NJTransit rail does into Penn Station. Many of those buses moreover are run by profitable private companies. Apart from a hope to improve rail so as to enable major densification in New Jersey (where I live) and provide fossil fuel free transit, it would seem that expansion of bus infrastructure might more effectively get commuters into Manhattan.

David, the reason the Lincoln Tunnel is so productive is that the buses get to hog two full blocks of prime real estate. Building more capacity for that is going to be even more expensive than the ARC cavern.

Many NJTransit train riders use PATH or the ferries to get into Manhattan. More buses was one of the alternatives examined before the decision was made to build a rail tunnel. More buses means more highway all over the state. Means another road tunnel into Manhattan. Once the buses get to Manhattan there would have to be a new bus terminal. The cheapest option is more rail.

It would be more accurate to say that Jarrett’s bias is in favor of effective transit without regard to technology. I’ve been reading his blog for quite a while and attended one of his talks. He would be in favor of motorized skateboards if that were cost (dollars, time, and space) effective and got the job done.

I think you’re right Ted, that’s exactly what he would say. But, I wonder, does he rank his restaurants based on calories per hour and ignore flavor / experience?

I don’t find these statistics particularly encouraging. If you can forgive the characterization it seems like on the margin (work commutes as opposed to all travel) of the margin (city limits as opposed to metro area) of the margin (non-Texas cities) they show that transit market share was kept flat instead of falling. This reminds me of the grim statistics for Portland that show no significant modal shift away from driving alone despite decades of investment. It is encouraging to see the revival of some urban living and the increases in walking and bike commuting. These however seem to be serving a niche audience.

Did you look at the numbers for Portland? Here they are:

* The City of Portland had a non-auto share of 22.9% in 2009, which puts it in ninth place out of the thirty cities. Not just a “niche audience”.
* Between 2000 and 2009, Portland achieved a 3.59 percentage point (not percent) growth in non-auto mode share – the third highest growth rate of the thirty cities.
* Over the same period, Portland achieved a 2.09 percentage point decrease in drive alone mode share – once again, the third largest such decrease among the thirty.
* Car pooling fell in every city; Portland’s decline of 3.34 percentage points was in the middle.
* The greater Portland MSA had a drive alone rate of 71.63%, the seventh lowest rate of the thirty Metro areas.

Portland definitely outperforms other cities of its size and age.

I was referencing this:

It’s true that Portland outperforms some similar cities, though notably not Seattle, but the truth remains that despite major investments in light rail, an urban growth boundary and major support for densification Portland is a fundamentally car-dependent city with transit little used and only moderate densification in the core. These statistics moreover cover only work trips. When one looks at all travel the mode share for autos rises to 90% even in New York, 80% in Paris. I find my transit idealism suffering in the face of statistics.

I don’t know where you’re getting your statistics, but at least in Paris, the mode share in terms of all trips (not km of travel) is:

For the city itself:
– 16.3% cars/motorcycles
– 33.9% transit
– 49.9% walk/bike

For the whole region:
– 45.3% cars/motorcycles
– 20.5% transit
– 34.2% walk/bike

The majority of trips in the region, as these statistics show, are done in modes outside of the private automobile.

Thanks Yonah. I believe I got that number from an analysis of kms travelled which certainly masks the greater opportunities to take trips on foot and by bike in most of the Paris region.

There still seems to be a significant barrier in the US to development of new transit-oriented neighborhoods. Where I live in Princeton, NJ the developers are suing to try to build at greater densities (or any density as it’s now mostly open parking) next to the Junction train station, but it seems that it’s harder to encourage TOD in cities such as Portland that mostly developed around the automobile. At a minimum I hope that New Urbanist design catches on as the one such new neighborhood we have in central Jersey is wildly popular, but that’s still not really designed as TOD except insofar as it allows local walking and biking.

Some X-Y plots of different data columns and some statistical correlation probably needs to be put into effect here. Staring at these numbers, I’m becoming very concerned about how much the carpool share has declined. In most places, the most popular mode shares are ranked as:

1. driving alone
2. carpooling
3. transit
4. walking
5. biking

That’s basically the opposite of what’s desirable from a carbon footprint perspective:

1. walking
2. biking
3. transit
4. carpooling
5. driving alone

Well, biking is more efficient than walking, but ideally you want to be close enough so you don’t have to bike. They also take up a fair amount of space, so truly constrained locations may put transit up higher. But hell, I think we should plaster that second list to the door of every planning office in the country. However, to get there, every mode other than driving alone has to be encouraged. We want to see carpooling decline someday, but probably not right now.

The Ft Worth commuter rail line tends to carry people away from the city in the morning peak and back in the afternoon (most are going to Dallas). There are commuters to downtown Ft Worth, but it’s not the dominant pattern. For the most part, Ft Worth is like any suburban (bedroom) community on a commuter rail line.

First, I wouldn’t overinterpret any of this data. ACS data have had significant quality control problems. For example, in Oakland, they showed several thousand housing units disappearing between one year and the next! The data simply aren’t as good as in the “real” decennial Census, which isn’t perfect but is better.

With that caveat, there were some interesting distinctions between the cities and their metros. In metro Los Angeles, for example, the drive alone share is 74%, it drops to 67% in the city of Los Angeles. That’s hardly the target, but it does mean 600,000 people primarily commuting by means other than driving in the city where “everybody drives.” In Portland, the gap is even wider 72% driving alone in the metro but only 62% in the city. If you calculated a “Metro outside city” drive alone share the difference with the city would be even wider.

To me, it speaks to the ongoing importance of a smart growth as well as a transit agenda.

In the most sprawling cities, like San Antonio, there wasn’t much of a difference.

It’s hard to see how. You’d think if anything that the recession would lead to more carpooling. Presumably the recession would have affected the baseline number the most, and lead to overall falls in both drive alone and other commuters.

It is surprising no one has noted how it seems that most new bike users switched from transit.

Also surprising that the impact of fare increases is not considered. Base fares approaching $3.00 are not trivial to most people. Sure, the cost of driving also rose, but if you have a car with gas in it, hopping in and heading out feels as cost-free as ever.

Why no figures on Houston..?According to statistics published in Metro-Magazine (A transit industry publication), on the 100 largest transit systems, Houston has increased from 14 to 12.

I am a frequent user of METRO during off-peak and during peak travel periods, it would appear ridership as in many other cities of Houston’s size, does leave plenty of room for improvement along lines of equipment utilization, primarily in the need for smaller buses to run middle of day, NON-peak travel times. Metro revamped bus schedules February 14, 2010 on many routes, and has made changes on other routes since early summer 2010.. Light-Rail ridership during peak travel times often runs at crush loads 220+ per LRV unit, with multiple units of up to 450 capacity. There is need for new cars to supplement this fleet, as they near heave overhaul dates, the big problem, Metro ran afoul of FTW Buy America requirements and permitting the contractor to modify it’s contract after bids were opened..!

Metro will need a re-authorization election before 2014 (Sooner than later as delivery date of NEW LRV’s will take 2+ years from date contract is awarded..! ).

It’s difficult to use 2000 as the start date. For many metropolitan areas, 2000 was a highwater mark for employment, and therefore congestion was at peak levels. The 2000 Census came will before the demise of the Dot Com boom and unemployment in San Francisco, LA, New York, and Atlanta was less than 5%. In this inflated job market, commute congestion was significant. As a result, people looked to alternative means of transportation to get to their place of work. One only needs to remember the crowded BART trains from the East Bay to San Francisco to understand the impact of the economy on ridership. Conversely, 2009 represents a low employment level with peak hour congestion down significantly. The fact that cities have made any progress during these very different economic periods implies a lasting trend towards reduced personal vehicle commutes.

The heart of the midwest has been gouged out of this, specifically ones with rail like Minneapolis, Cleveland, St. Louis, and Pittsburgh. Not to mention cities making strides like Cincinnati and Kansas City. Makes it feel like no matter how much effort these cities make, nobody cares as much as when Charlotte (or whatever) sneezes. Just saying that the midwest always is overlooked — like not just a little bit.

What is the definition of non-auto and transit ? It would help to now excatly what is included in these sections. Thank you.

[…] I couldn’t find any modal share stats for walking in Groningen, but a recent Dutch government report gives a 25% utility (i.e. not recreation) walking share for Utrecht, and a 30% share for Rotterdam, Den Haag and Amsterdam (PDF here, see page 20). Why would Groningen’s walking share (and it’s a very walkable city) be so much lower than other Dutch cities? At “1 or 2 percent” the walking share in Groningen would be lower than Detroit. […]

Leave a Reply to Michael Farrell Cancel reply