Why do we care about mode share?

The New York Times ran an op-ed the other day that helpfully demonstrated the pitfalls of lifestyle arguments in favor of urbanism, namely that they are annoying to everyone but the people making the argument.

The boys, like their father, are lean, strong and healthy. Their parents chose to live in New York, where their legs and public transit enable them to go from place to place efficiently, at low cost and with little stress (usually). They own a car but use it almost exclusively for vacations.


“Green” commuting is a priority in my family. I use a bicycle for most shopping and errands in the neighborhood, and I just bought my grandsons new bicycles for their trips to and from soccer games, accompanied by their cycling father.

These arguments – whether they’re about physical health, or “diverse” or “vibrant” or “creative” communities, or whatever else – are, at bottom, about telling people that they are lacking, and that in order to improve themselves they should become more like the author. In the 1970s, when city dwellers felt superior mainly because of their supposed cultural capital and were telling middle-class suburbanites to loosen up a little, that might have been obnoxious but harmless. In our current situation – when the city dwellers making these arguments are the economic elite (the author of this particular piece, Jane Brody, lives in gentrified brownstone Brooklyn, I believe) – it’s a lot more sinister. Brody talks about commutes as if their length and form were something that most people could freely choose, rather than something imposed upon them by their wages and the price of housing and form of development of their metropolitan area. She makes this a story about personal morality, rather than the constraints we choose to put on people through public policy.

This is related, I think, to the study about mode share in U.S. cities that got passed around the urbanist blogosphere recently. In virtually every instance, the study was presented like a sports power ranking, with the winning cities being those with the least travel by car (“city of Chicago ranks sixth among large U.S. cities for percentage of people either biking, walking, or riding transit,” is a typical formulation of the lede).

But why, exactly, do we care about mode share? The pettiest possible answer is that we do conceive of cars v. transit/biking as a sort of culture war, just like many committed drivers have alleged, and what percentage of people choose to drive or do something else is how we measure whether or not we are winning. This, clearly, is not a particularly edifying possibility. A better answer might be that we really do want everyone else to be more like us – to reap the benefits of non-car commuting, from being healthier (although, contra Brody, I spent my subway commute today scarfing down a pound of spaghetti) to polluting less – and this tells us how many people are enjoying those perks.

That’s much more reasonable, but still problematic in that, like the Times piece, it strongly implies that the issue is individual choice, rather than the circumstances that constrain that choice. The people who write for places like Streetsblog know that circumstances matter, but for the casual reader, articles about mode share makes those issues a sort of specialists’ background.

That’s too bad, because mode share does convey some important information about constraints. If we assume that, allowing for some cultural margin of error, most people will choose to get to work via whatever method they find most efficient and comfortable, then we can determine roughly what percentage of people in any given city have decent access to transit – access that’s at least in the same ballpark of convenience as driving – just by looking at what percentage of people actually use it. Obviously there are complications to this: since one major inconvenience of driving is cost, cities with high poverty rates may have mode shares that exaggerate their transit’s effectiveness, for example. And since transportation choice is basically zero-sum on an individual basis – that is, all that matters is the relative efficiency of each mode – you could get a lot of people on transit by making driving truly hellish, without providing decent service. (Although in the American context, I think there are vanishingly few places where that would be an issue.)

Moreover, if we care about mode share as a proxy for service effectiveness, then beyond a certain point – say, a quarter, a third, whatever, of commuters – you’re kind of done. It doesn’t really matter. If New York City, with one of the most comprehensive transit systems in the world, can only get 50% of its commuters on buses and trains, then surely most of the distinction between it and, say, Asian cities with much higher transit mode shares isn’t the quality of their systems (although they may be of higher quality), but the increased misery of driving in ever-denser places. The issue stops being whether we can get from 40% to 45%, but whether subregions of the metropolitan area have strongly varying mode shares, suggesting that you can only get decent access to transit if you live in the right place. And, of course, that is in fact the case.

But if what really matters is service levels and access – if what we’re trying to accomplish is giving everyone a level of service where transit is a viable option, for reasons outlined here – then why not just measure that directly? Why not have widely-disseminated statistics about the percentage of people in every metropolitan region who can walk to a transit stop? Or make a bigger deal about the number of people who can reach some given percentage of metro area jobs via transit in a reasonable time frame? I almost never see those numbers in urbanist conversations, and to the extent that I do, they’re sort of ghettoized into the “social justice” urbanist subculture.

But these seem like relevant numbers for “mainstream” urbanists, too. In fact, they seem a lot better than mode share. Generalized public arguments in favor of transit projects are more likely to benefit from language that suggests they’ll provide options, rather than language that suggests the ultimate goal of the policy is to force people out of their cars. Because, in fact, that’s what public policy should be about: making transportation easier for more people, rather than moralizing about the perfectly legitimate choices that people make, given their circumstances.

16 thoughts on “Why do we care about mode share?

  1. You’re making two criticisms that seem in direct opposition to each other. If the point is that it’s not about individual choice but about circumstances, then it makes perfect sense to talk about metro area mode share, because it tells us which metro areas have the circumstances for transit and which don’t. The same is true of metro area transit mode share growth: the growth in mode share in Vancouver and Washington tells us that their policies are producing more favorable circumstances and other North American cities should consider emulating the policies identified in connection with this transit growth.

    Also, when you say segregate the comment that “although they may be of higher quality” into a parenthetical comment, you’re brushing aside a real concern about the quality of transit in New York, coming from the exact sort of elitism you deplore. New York has good transit, if your life is centered on the Manhattan core and if you live in the inner half of the city. And indeed among people who work in Manhattan (regardless of where they live) the transit mode share is not 50% but 67%, with a substantial chunk of the rest walking. But if you live in the outer parts of the city, you’re pretty much screwed unless you work in Manhattan and commute at rush hour, because of a combination of two very much policy-related problems: the suckiness of commuter rail, and the suckiness of transfers between the IND and the IRT and BMT (to get from Jamaica to Coney Island without detouring through Manhattan you need to make 3 transfers: J-C-S-Q). Note also that the dropoff in transit mode share in New York as one works farther out is much steeper than elsewhere – Downtown Newark’s share as a job center is less than 30% – and this is again related to the suckiness of commuter rail.

    All of those things matter, and are captured perfectly in the mode share. In contrast, other measures, especially the ones you propose as alternatives, do not. The nearest bus stop to me is useless for me, because the bus line doesn’t go along the axis relevant to me or nearly all transit users in my neighborhood; fortunately, the second nearest bus stop is highly useful and is only one block farther away. People in American suburbs and Sunbelt cities usually have nominal access to transit in that the nearest bus is relatively close and gets them to work in an hour and a half, but this is not useful if the bus comes every hour, to say nothing of the fact that nowhere in the world are 90-minute commutes reasonable or normal. (Transit users in New York average 45.) It’s especially bad to say that those supposedly high-transit access, low-mode share metro areas just have very easy car use when we’re talking about places like San Jose and the Inland Empire, with long auto commutes and bad traffic. The Streetsblog community ignores these measures because those measures are bad and ought to be ignored.

    1. Those are all good points, and I think it’s particularly true that “closest transit stop” isn’t very useful without setting some sort of standard about the level of service that comes to that stop, and how that line interacts with the rest of the network. Which makes it too convoluted to have a lot of meaning to the average person.

      But I don’t think that last part at all contradicts what I said–I explicitly wrote that mode choice depends on a zero-sum evaluation, so you could have people using transit because driving is terrible, or you could have people driving in congestion-prone places like the Inland Empire because transit is practically non-existent. In that case, actually, using a different metric–how many jobs the average person can reach via public transit, say, in 45 minutes–seems to convey more information than mode share, which can only tell you which mode is less terrible.

      I also don’t think it makes sense to champion mode share as the best measure of equality to transit access, since, as we’ve both pointed out, it varies wildly from one region of a city to another. You can solve this problem by citing different subregions’ shares, or maybe you could make a map of it–but the bottom line is that you can’t get that information from a single number for mode share, and the percentage of articles on Streetsblog or elsewhere that talk about mode share by sector of the city is very, very small. You can, however, get some sense of that issue by just directly measuring the percentage of people with access to x% of jobs, or x% of the metro region, in a given amount of time, on transit.

      Finally, while you or I might understand that when we talk about mode share, we’re talking about a city’s infrastructure, and not the choices that individuals make, many decently-informed members of the public–including a good number of people with a moderate interest in urban issues and transportation, and certainly the voters and even decision-makers who need to be persuaded that investing in transit is important–do not. When we make mode share virtually the only macro number we talk about, I think we’re sending an unfortunate message about our goals, whether we realize it or not. That’s not to say we shouldn’t talk about it at all–it’s valuable, obviously. But at the very least, I think we ought to be very clear about why we care about it, and I think it would be better to bring in other numbers that are more illuminating about the options people have, and what we want those options to be.

      1. You’re still confusing those Brookings metrics with transit accessibility. They’re very crude metrics, and trying to refine them by changing your assumptions a bit won’t help. Maybe I can get to work on the bus in 45 minutes, but the bus is unreliable and comes every hour. Maybe my shopping errand trips are not walking distance as in transit cities but are spread throughout the metro area and I need to take a bus for 30 minutes to get to them, with a different bus for each shopping destination. Maybe because of parking minimums I have free parking at both ends of each trip for which I already pay $20,000 more for an apartment, reduced wages since the employer has to spend more on parking, and higher retail prices since the store has to pay rent on the parking spots. Maybe the walk from my apartment to the bus stop is on a fast multi-lane arterial road with 2′ sidewalks without snow shoveling in winter.

        The danger of looking at a correlate too much (assuming these access measures are even correlates – Brookings’ list of best transit access cities bears little relationship with high-transit usage cities) is that you start thinking the correlate is everything. It’s not. There are all these intangibles that make a difference.

        In contrast, you’re losing very little additional information when you measure raw transit ridership (or mode share, which to me is mainly a way of getting around the linked vs. unlinked trip problem). Yes, you can get higher transit ridership by making driving worse. So what? First, within each country, or for that matter within the first world, the difficulty of driving varies much less than the above intangible factors for transit. Second, some of those driving difficulties, such as congestion around dense CBDs, are actually an indication that the travel market in question is better-served by transit; conversely, in geographies with consistently low transit mode share and ease of driving, such as rural areas, this is an indication that the travel market is better-served by cars. And third, making driving worse improves transit: fewer cars means that development is more walkable, streets are pedestrian-friendlier, and retail clumps into clusters that can be served on one transit trip.

      2. I guess I find all this a compelling argument that the Brookings metrics are flawed and not to be used by themselves, but not that mode share isn’t. Of course there are intangibles–but that’s the case for mode share, too. Maybe City A and City B have equally bad transit, but City A has more poverty. And again, mode share inherently can’t tell you anything about geographic inequality.

        I suspect there’s also some of your technical/political thing going on here. I’m really not making an argument that mode share needs to be replaced for specialist analysis; I don’t know, and am certainly inclined to take your word that it’s the best measurement tool we have. I am making an argument that in public conversation, mode share has some really unfortunate tendencies–namely, to obscure geographic inequality and to strongly frame the debate around individual choices rather than the circumstances that cause people to make those choices.

        And yes, you can point out ways that other metrics miss something about what they’re trying to measure. But knowing that the average low-income person in the Chicago metro area can only reach 20% of jobs, or whatever, in 90 minutes on transit does tell me something about accessibility. And, in fact, I think it tells the layperson much more than the fact that only 10% of Chicago area residents use transit to get to work.

  2. Isn’t a key distinction between mode share and the metrics you mention (percentage of jobs that a low-income person could access via transit, or average proximity to a bus stop, etc.) that mode share is a number reflecting the actual choices made in real life, whereas those other metrics are hypothetical?

    To take the low-income example further, knowing that a low-income person could only reach 20% of jobs in 90 minutes on transit is definitely useful. But isn’t it even more useful to actually know what percentage of low-income workers chose to actually take it?

  3. That does seem like a key distinction, but I don’t think it means that mode share is always more useful. Mode share is probably the best way to get a quick heuristic about what proportion of people have access to transit service that is at least as convenient as driving to get to work. But because that’s basically a zero-sum comparison, it can’t tell me anything about exactly how onerous it is to take transit if you need to do that. If that’s the information I’m interested in, it seems much more useful to have a number that represents access directly, in the form of how many jobs a person of low income, or median income, or whatever, can access in a given amount of time, or how many other kinds of amenities, or whatever.

  4. “Transit-dependent” and “low-income” might overlap somewhat but they’re actually two very different things. Ridership is a metric that focuses on usefulness to people, and most relevant if the goal of transit-subsidy is to increase the systemic value of the network.

    Your metrics focus on usefulness to a place. People can move if it suits them. Yes, the low-income are highly constrained, but within those constraints they’re intelligent enough to make their own decisions about spending-on-transport vs spending-on-housing. If the goal of transit-subsidy is transfer to the low-income so as to ease those constraints, cash is probably a more effective alternative.

    1. I think the issue is this: ridership is a blunt metric in that it only tells you what percentage of people find it more useful to take public transit than to have a car, taking into account the extra costs of car ownership. It doesn’t tell you anything about what sorts of tradeoffs are actually involved in that calculation. Again, I’m coming at this more from a political perspective than a technical one; I think it’s much more intuitive for non-expert people to think about the tradeoffs in a place-based sense, as you put it, than by ridership. What, exactly, are people giving up if they don’t have a car? Put another way, how much access to amenities do people have to buy with their car payments? As far as the idea that cash transfers can make up for transit subsidies, I don’t think that’s totally off, but I do think that a) cash transfers won’t help people who can’t drive for non-monetary reasons, like the young, the elderly, or the disabled; b) cash transfers would fail to have the sort of land-use effects that improve access to amenities that transit can–in fact, pushing reliance on cars in large metros would eventually create enough congestion that it would constrict access to people who lived far away from amenity-rich areas, ie, the poor; and c) cash transfers targeted at the poor that are large enough to offset the cost of owning a car – which I’ve seen estimated at $10,000 annually – are just never going to happen. Public transit happens to be a subsidy that disproportionately benefits the poor but which is also politically possible because it benefits the professional and business classes too (at least in some cities).

      1. I understand where you’re coming from. The existence of transit has huge consumer surplus. What’s relevant to determining the best metric, though, is whether the *marginal extension* of the transit to additional service area has huge consumer surplus, particularly if the service-quality would suffer as a result. People (even low SES) won’t cut their car expenses unless that service-quality is very high, and people for whom transit is particularly important (whether cause of finances or the other reasons you listed) choose to live where high service-quality makes sense. Put another way, the existence of a transit system may save a working-class family $10K or have $10K in value to an elderly person, but expanding the transit system to neighborhood XYZ only has a fraction of that value, because the family/elder doesn’t *have* to live in XYZ. So the equivalent in cash-transfer isn’t $10K but a number much smaller than that (it’s the value difference between living in XYZ or the nearest-equivalent-except-transit-friendly hood). For all these reasons, service-quality ends up really important for urban design outcomes as well. Thus I focus on the actual quality of the service, and I feel that shows up strongest in ridership statistics.

        I understand that many cities have a problem whereby living expenses near transit are crazy, but that has more to do with development restrictions (which you’ve covered extraordinarily well elsewhere in the blog), and Chicago doesn’t really have that problem. We’ve got areas with excellent transit but scary pathologies, and areas with existant but crappy transit. I’ve no idea how to address the first issue, and I feel the second issue has more to do with ROW-priority and Metra’s dysfunction than financial subsidies.

        Focusing on the service area can in fact have a result opposite what you intend, because you dilute the service quality and prevent the great transit & development positive-feedback loop, thus greatly limiting the actual quantity of housing with access to high-quality transit.

      2. Yeah, I think we’re in total agreement that all things being equal, it’s a terrible idea to expand transit’s catchment area by radically reducing its service levels in any given place. I don’t know that I agree that Chicago doesn’t have a problem with development restrictions near transit – in fact, I definitely don’t agree. There is obviously enough demand to live near much of the Red, Brown, and Blue lines on the North Side that we could be building dozens and dozens of mid-rise buildings, but we don’t, because it’s illegal and unbelievably onerous to go through the PD process in an established residential neighborhood. And further out – I’m thinking about the Blue Line towards O’Hare – there’s lots of areas that are actually zoned SFH-only. Obviously there’s plenty of buildable land on the South and West sides, but many of those areas also have problems with amenities, safety and schools, as you pointed out.

        Agree about Metra. Holy shit. I think there are basically two things that need to happen for Chicago to have a more functional transit system; one is a BRT network in the city and suburban Cook County, and the other is a sane level of service on Metra. There’s no reason that we shouldn’t have at least half-hour frequencies during the week on most lines, and more than that in the city.

  5. Mode share is useful for indicating how useful public transit is vs cars for most workers. A small city where many live in a 45 minute bus ride to work (with low frequencies) while almost everyone has a cars because it’s low density and car commuting takes 20 minutes will have a low mode share but a decent access rate. It doesn’t sound like a particularly successful situation for transit, but it depends on what you think the point of transit is. Perhaps rather than just looking at jobs accessible by transit, you could compare jobs accessible by transit vs car for some given time

    NYC’s driving to work mode share is 29%, not 50%, with the car commutes disproportionately in trips with inconvenient transit. The rate isn’t out of line with any large European city, though of the course the suburbs are much lower. Alon Levy made all the comments I could make on the usefulness of mode share better than I could.

    1. Yeah, fair enough to all of this. I think Alon et al got the better of that argument.

      1. Though, I do agree there are some drawbacks of mode share.Take (using a NYC example, there’s probably something similar in the Chicago metro, but maybe a little less extreme) the railroad suburb Scarsdale, in southern Westchester. Transit mode share: 38%, fairly high though lower than most of NYC. Median Household Income: $250k/year. Almost all the transit usage is people using MetroNorth, mainly to well paying Manhattan jobs. It might be easy for many locals not to use a car to work, but for local job commutes not to Manhattan (possibly also not toWhite Plains, a suburban job center on the same train line) the transit isn’t all that good, though not horrible for American suburban standards (one bus line that goes east-west at hourly frequencies, another slow north-south bus every 15 minutes, and two circulators that loop around town to connect to the train, maybe to help lower parking demand). Ditto for many local errands, a spot on the edge of the outer boroughs there’d be more in walking distance and more comprehensive local transit. Of course, for those working in Manhattan they need thier car as much (doubt there’s many carless adults at that income). I did have a friend from college who lived in Scarsdale for a few years without a car, she lived with relatives (mainly to save in housing costs) in walking distance from the train station.

        In Long Island, if you look at a map of transit mode share, it reflects more the places that have the most Manhattan commuters rather than places that are easiest to live without a car (which all have a LIRR station). Of course, that fact that more people do work in the center city means cars are less needed, but it doesn’t say much about the neighborhood. Anyway, my point is transit use is often skewed heavily towards center city jobs, it’s not completely informative on the “how much inconvenience not having a car” scale. Car ownership might be a better measure, though is skewed by income and difficulty of owning a car. [The central districts of London have higher car ownership than similar districts of Manhattan, probably because Londoners are more likely to have a place to put their car, not really because a car is more necessary in London] If you’re interested in transit usage from an environmental or just reducing car traffic perspective, mode share is somewhat useful, from a social jusitce / inequality standpoint less so.

      2. Yeah, I think that last bit is pretty central to what I was trying to get at. I think I’m convinced that mode share is the single most useful statistic for either interest, but it’s far from sufficient, at least for the social justice aspect.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s