How far should you live from work?

Rush hour, New York City

Thirty minutes at most, according to the wisdom of the crowds.

That comes from reams of data and piles of research that suggests commute times tend to cluster around this point. People tend to be good at weighing their options, economists think. If you live farther from work, you can usually afford a bigger house or apartment. But there’s a point where that journey becomes too onerous, and you are willing to sacrifice some of those desires to live closer to your job. That point on average seems to be between 20 and 30 minutes.

I was inspired to look into this further after seeing an article by Charlie Gardner over at his blog, The Old Urbanist. Gardner had mined the American Community Survey for average commute times in major metropolitan areas. Though there was a tight correlation between population and commute time (metros with larger populations have longer average commutes), the differences weren’t pronounced. They ranged from a low of 22.5 minutes in Kansas City to a high of 34.6 in New York City. That’s not a lot.

But before we ask why commute times hover in a tight band, perhaps we need to ask why people commute so far in the first place. Why not work next door? The answers may seem obvious, but what’s readily apparent to one person may not be to another. That’s why we examine these things scientifically. Well, in this case your hunch is probably correct. An older study by Martin Wachs and his colleagues at UCLA found, unsurprisingly, that people choose where to live not just based on commute times but also neighborhood characteristics, schools, and safety.

Now we can move on to the more curious question, why commutes tend to average 20-30 minutes. It’s not just limited to the United States, either. In the Netherlands, the average commute time in the early 2000s was about 28 minutes. Many European nations average about 35 minutes. What makes a half-hour so universal in terms of commuting?

It didn’t used to be that way. Average commute times in 1980 were around 22 minutes. Today, they’re around 25 minutes. Three minutes may not seem like much, but remember it’s an average. To increase an average by that amount, some commutes had to grow significantly to counter those that shrunk or remained the same. Now, keep in mind there is a lot of variation about those averages. Some people travel 2 minutes to work, others well over an hour. But on average, they have increased.

What’s causing that lengthening is higher job densities in major metros. Job growth is requisite to economic growth, and vice versa. As metro areas add more jobs, those jobs tend to be concentrated in business districts (after all, not everyone can work out of their homes). And as business districts fill up, commute times lengthen because the roads leading there become more congested. So when the economy booms, traffic slows to a crawl. I heard anecdotal evidence of this when I lived in San Francisco. People told me, if you think traffic is bad now, it was much worse during the tech boom of the late 1990s. When all those tech workers lost their jobs, gridlock practically evaporated, they said.

Subtle changes in urban form may also cause longer commutes. One study in the Netherlands and another in Quebec, found that polycentric metro areas—those with two or more cities, like Minneapolis-St. Paul—tend to have longer auto commute times. As cities grow and begin bumping into one another, such agglomerations are likely to become more common. It’s possible commute times may increase as well. While there may not be consensus on this point, I haven’t found any studies that claim changes in urban form will shorten commute times. That makes sense if you look at somewhere like New York City, which is both monocentric and dense. People may work a short distance from their homes, but traffic is so congested and public transit makes so many stops that commute times are still relatively long. Simply increasing density in some cities may shorten commutes for a brief period, but the honeymoon won’t last forever.

Which is a bummer, because for the most part people think their commutes are too long. A survey of 2,000 commuters in the San Francisco Bay Area reported that 52 percent of respondents said they commuted at least 5 minutes longer than they would like. Among that group, median commute times were 40 minutes, which is certainly longer than the region’s average. On the other hand, 42 percent said their commutes were just right (their median time was 15 minutes). Surprisingly, 7 percent felt their commute was too short (median of 10 minutes). But despite the fact that a majority think their commute is too long, most people said they didn’t mind it, so long as their trips were less than 100 miles.

That people don’t mind their commute may be why commute times refuse to shrink. People in the Bay Area survey who didn’t mind their commute said they agreed with statements like, “I use my commute time productively” and “My commute trip is a useful transition between home and work”, which supports anecdotal evidence I’ve heard that people enjoy the separation between work and home. Twenty to thirty minutes may be just enough time to unwind.

It’s not entirely universal, though. Tolerable commute times seem to lengthen when people switch from cars to mass transit. People may find that time more productive, or maybe the time seems shorter because driving can be stressful, while just sitting usually isn’t. Personally, I know I’m willing to commute longer by train than car. Another reason is because mass transit commutes tend to be more reliable in terms of duration (at least for trains). Not having to worry about traffic jams doubling your commute is a big advantage.

Regardless of mode, people seem to settle on an ideal commute time. And once they have settled, they don’t seem to stray from it. A study of two metro areas in Washington State discovered that commute times don’t change much when people move or switch jobs. The thinking is that if a person gets a new job that’s farther away, they are more likely to move. Plus, as people have moved to suburbia, some jobs have followed. It’s a two-way street. But that doesn’t mean employers can move to the burbs without consequences. If an employer moves and an employee doesn’t move as well, the employee is more likely to find another job. Companies looking to relocate simply to cut costs may find the high turnover that results more costly in the long run.

Commuting is a big part of our lives, so it makes perfect sense that it would affect so much of the world around us, especially the cities we live in. Take a dense city like New York that has oodles of jobs, and lots of dense housing close in. That density helps keep commute times reasonable. But somewhere like Tulsa that doesn’t have as many jobs doesn’t have as much need for density. Form follows function, and currently the freeways in Tulsa are functioning pretty well compared with New York.

That presents a real dilemma for urban planners, who have been striving to increase densities in cities across the board. One approach has been mixed-use development that blends retail, housing, and office space. That may help reduce trip times for errands and such, but it doesn’t preclude people from living in one mixed-use neighborhood and working in another. The reality is, we’re probably not going to change commute times. If we offer faster and better transportation, people will use it until it becomes overburdened. At which point they’ll just move closer to work. Attempts to influence urban form through design may not have much of an impact if jobs don’t follow.

Sources:

Cervero, R. (1996). Jobs-Housing Balance Revisited: Trends and Impacts in the San Francisco Bay Area, Journal of the American Planning Association, 62 (4) 511. DOI: 10.1080/01944369608975714

Cervero, R. & Duncan, M. (2006). ‘Which Reduces Vehicle Travel More: Jobs-Housing Balance or Retail-Housing Mixing?, Journal of the American Planning Association, 72 (4) 490. DOI: 10.1080/01944360608976767

Clark, W.A.V. & Davies Withers, S. (1999). Changing Jobs and Changing Houses: Mobility Outcomes of Employment Transitions, Journal of Regional Science, 39 (4) 673. DOI: 10.1111/0022-4146.00154

Clark, W.A.V., Huang, Y. & Withers, S. (2003). Does commuting distance matter?, Regional Science and Urban Economics, 33 (2) 221. DOI: 10.1016/S0166-0462(02)00012-1

Giuliano, G. & Small, K. (1993). Is the Journey to Work Explained by Urban Structure?, Urban Studies, 30 (9) 1500. DOI: 10.1080/00420989320081461

Levinson, D.M. (1997). Job and housing tenure and the journey to work, The Annals of Regional Science, 31 (4) 471. DOI: 10.1007/s001680050058

Schwanen, T., Dieleman, F.M. & Dijst, M. (2004). The Impact of Metropolitan Structure on Commute Behavior in the Netherlands: A Multilevel Approach, Growth and Change, 35 (3) 333. DOI: 10.1111/j.1468-2257.2004.00251.x

Schwanen, T. & Dijst, M. (2002). Travel-time ratios for visits to the workplace: the relationship between commuting time and work duration, Transportation Research Part A: Policy and Practice, 36 (7) 592. DOI: 10.1016/S0965-8564(01)00023-4

Vandersmissen, M.H., Villeneuve, P. & Thériault, M. (2003). Analyzing Changes in Urban Form and Commuting Time∗, The Professional Geographer, 55 (4) 463. DOI: 10.1111/0033-0124.5504004

Wachs, M., Taylor, B., Levine, N. & Ong, P. (1993). The Changing Commute: A Case-study of the Jobs–Housing Relationship over Time, Urban Studies, 30 (10) 1729. DOI: 10.1080/00420989320081681

Photo by Jekkone.

Related posts:

Tell me how much you drive, and I’ll tell you where you live

Urbanites leave the car behind, but not as often as you might think

Drive a lot? Housing density may not be to blame

Thanks to the support of readers like you, Per Square Mile remains independent and ad-free.

If you enjoy what you just read, please consider supporting the site with a monthly recurring donation as a Sustaining Member.

You can also support Per Square Mile with a one-time donation.

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>

  1. I’m a little surprised you didn’t mention Cesare Marchetti and his attempts to compare commute times and distances across all of human history. “Marchetti’s Constant” is the hour per day that people have long been willing to spend traveling.

  2. I should live as far from work as I decide to live because no one can judge all of the intimate factors involved in my decision.

    The author’s ignore the consistent, sad history of the failing of central planning. Recent research has shown that cities resemble living organisms, and suggests that the millions of people living in them reach much better decisions on crucial issues than small groups of bureaucrats on planning boards.

    It seems not everyone is familiar with that research or the failure of planning boards.

  3. You gotta read the fine print on that study and think carefully about what an MSA is. San Jose is NOT in the same MSA as SF and Oakland. This quite easily explains the discrepancy between the 40 minutes you found and the 30 in the study – it’s obviously totally bogus to think about average metro area commute times if you decide a priori that folks who commute from Fremont to Sunnyvale don’t count in your statistical set. Several other areas that have a common commute suffer from this flaw in the analysis, including Baltimore/Washington and even Denver/Boulder.

    1. That study to which you’re referring was a sample of the Bay Area which included San Francisco, Concord, and Pleasanton, not the entire MSA (a term which you’ll note I didn’t use). Read the original research.

  4. I like the overall direction of your argument but and somewhat sceptical of your use of mean commute time and not median. Especially, at the point when you discuss change over time I would have liked to have known how the median has moved. Without that, I am afraid I am unable to agree with your assumption that the ‘crowd’ is doing anything like what you are giving it credit for.

    1. The difference between means and medians all depends on the underlying data. They may be the same, though I can’t be certain as I didn’t author the original studies. I just report ‘em.

      1. If you have not seen the median data then I am afraid much of first half of your article (wisdom of crowds bit) is speculative.
        Rest of it is interesting and thought provoking.

  5. The 30 minute commute time is sticky, but there are a lot of good reasons why we might want cities that have people commuting short distances slowly rather than long distances quickly. Short slow commutes use vastly less energy than long fast ones. The cities which support short slow commutes gobble up less surrounding land, and require much less infrastructure per capita, and can support convenient, cost-effective transit. In many cases, short slow commutes can even be done on foot or by bike, if we open up the streets to human use.

  6. Your observations are good. Both sets of authors who find that “polycentricity” increases commute times and distances, do not claim that their findings overturn the rule established by Peter Gordon et al (i.e. that urban decentralisation is accompanied by stability in travel times). The Quebec paper points out that their findings indicate that further research is necessary to establish whether the Gordon et al Law does not apply to smaller cities.

    Both papers may “control out” the reality that differing rates of dispersion of different types of households is a co-dependent variable with dispersion itself. Obviously a household with 2 income earners and children, is not going to be able to locate efficiently relative to both jobs and the childrens schools and
    other family-related amenities. These households will be over-represented in the “dispersed” areas and under-represented in the central areas, and “controlling”
    for household type may unfairly submerge this reality.

    In freer markets, jobs follow workers who have lower housing costs, rather thanv the other way around. If development is “fragmented” and land costs low, the
    cost of “efficient” location relative to any one job or amenity or new agglomeration is never an obstacle. Strict zoning, and strictly “contiguous” development, prevent these effects, as do higher land costs.

    In a growing city, travel times get longer anyway, which both sets of authors correctly state. The dispersion of jobs will be “lagging” the dispersion of the workforce, and the travel-time balancing mechanism will be taking time to “catch up”. What we perhaps do not know, is how much longer trip times might have been had a growing city remained strictly monocentric. We probably do not have any example of a growing city that is NOT “dispersing”. Planned “monocentricity” will tend to restrict growth per se, which will work to the advantage of “work trip time” statistical comparison outcomes but little else.

    It will also make quite a difference, whether the “dispersion” is taking place into low-density and uncongested “dispersed” locations, or whether the
    “dispersed” locations are themselves “high density” anyway through strict urban planning – which is certainly the case in the Netherlands. Growth has been
    strictly planned for decades, contiguity and high density mandated (at least in the case of each “node” even if the nodes are not immediately adjacent to each
    other) and land costs are high.

    It is an extraordinary achievement of “the free market” in the less-restricted US cities, that these cities achieve very affordable housing; high discretionary incomes after housing costs; high rates of marriage and child-bearing; high rates of growth; AND have stable travel times that are comparable or superior to cities whose policies have made housing unaffordable; reduced the rates of marriage and child-bearing; reduced the rate of growth; and increased congestion and local air pollution.

    In so far as the “anti-sprawl” policies can be said to have “succeeded”; eg in reduced energy consumption; the mechanism is not urban form, but reduced discretionary incomes once the much-more-expensive housing is paid for. Spending on everything non-housing will be lower, not just on transport and energy.

    In fact, in cities where the cost of urban land is much higher, there is a kind of effect where people and busineses are “priced out” of efficient locations, in contrast to the cities with low land costs, where the cost of buying a home closer to any particular job or desirable amenitiy is always far LESS of an obstacle. There is quite a lot of academic research regarding the UK economy, for example, that finds that agglomerations are prevented from developing because potential new participants are “priced out”. We can expect to see rivals to “Silicon Valley” emerge in the lower-land-cost parts of the USA, because Silicon Valley is now a location that potential new entrants are “priced out of”. This was not the case when Silicon Valley got started; if the kind of policies that make California’s cities so expensive had been in force longer, Silicon Valley would have happened somewhere else.

  7. See the article by Thomas J. Christian in the Journal of Urban Health: ‘Trade-Offs between commuting time and health-related activities’.

  8. Have you read Robert Lang’s “Edgeless Cities”? He would quarrel with your point that commercial space is limited and centric. In fact the distribution of jobs has dispersed with adaptation to traffic, and the suburb-suburb commute is now more important than the suburb-city commute in many cases.

  9. Is there any information about governmental efforts to encourage living close to one’s workplace? I’d like to hear of it — I don’t seem to be able to find anything.