Consumption Theory and Gentrification

I recently saw an article about location decisions by education in the Netherlands. The article discusses the impact of rail investment on different social classes, and claims that,

A recent study by Teulings et al. (2018) uses microdata to quantify the differences in the willingness to pay for particular locations between the high and low educated (omitting the medium education level) (Figure 2). It shows willingness to pay for the job availability (based on the locally available transport infrastructure to commute to these jobs) and urban amenities such as parks and historic scenery at the location. The highly educated (right panel) are very sensitive to the quality of a location.

The claim is that educated people prefer central cities, in this case Amsterdam, because of their consumption amenities. This is the consumption theory of gentrification, which holds that the process of gentrification is caused by a middle-class taste for urban amenities. However, this theory appears incorrect, on several levels. The references cited in the paper for location decisions do not really bear out consumption theory. Moreover, the history of gentrification strongly suggests that, if consumption amenities are at all involved, then they have been stable for at least a hundred years.

Instead of consumption theory, the best explanation is that location decisions are about jobs. Certain cities have higher production amenities, especially for the middle class, leading the middle class to preferentially move to them to obtain higher-income jobs. The choice of neighborhood is then driven by access to skilled jobs, usually in the CBD but sometimes also in new job clusters. If there’s gentrification, the cause is insufficient housing in closer-in areas, leading to spillover to adjacent neighborhoods.

The references

The internal reference cited for it in the paper, the work of Coen Teulings, Ioulia Ossokina, and Henri de Groot, breaks down willingness to pay higher rents in expensive cities (i.e. Amsterdam) based on job access and several consumption amenities. The paper’s headline numbers superficially bear out consumption theory: table 7 on page 23 says that job availability is only responsible for 38% of low-education people’s variance in willingness to live in expensive cities and 28% of high-education people’s variance; the rest comes from amenities. However, a closer reading suggests that this is not really about consumption amenities.

First, that 28% of middle-class location choice comes from job access does not mean 72% comes from amenities. Observed consumption amenities are only 18% (and only 14% for low-education workers); the rest is unobserved amenities (30%), which are a residual rather than any identified amenities, and covariances between jobs and amenities.

Moreover, the consumption amenities listed are proximity to restaurants, monuments, parks, and a university. Is a university really a consumption amenity for the middle class? This is unlikely. Graduates don’t really have the same consumption basket as students. Instead, what’s more likely is that universities provide skilled employment for a particular set of high-education workers (namely, academics and other researchers), who are willing to pay extra to be near work; academic job markets are so specialized that access to non-academic high-education work isn’t as important. Of course, universities also have extensive working-class employment, but a university janitor can get a similar job at a non-academic environment, and therefore has no reason to locate specifically near a university rather than another source of work, such as a hospital or office building.

Finally, there is a second reference in the article, reinforcing its claims about location decisions with American data. This is a paper by David Albouy, Gabriel Ehrlich, and Yingyi Liu. Albouy, Ehrlich, and Liu’s text does not endorse consumption theory – on the contrary, their discussion blames “policies and regulations that raise rents by creating artificial shortages in housing supply” (pp. 28-29). On the question of consumption theory, the results of the study are inconclusive. They do not look directly at amenities that critics of gentrification typically implicate in causing the middle class to displace the poor; the amenities they consider include mild climate, clean air, and a sea view.

The history of gentrification

The word “gentrification” was coined in 1964 to describe the process in Islington. However, Stephen Smith has argued from looking at historical rents that the process goes much further back. He finds evidence of gentrification in Greenwich Village in the 1910s and 20s. Already then, the middle class was beginning to move into the Village, previously a working-class district. Jane Jacobs moved in in 1935. She was income-poor, as were many other people in the Depression, but on any marker of class, she was solidly comfortable: her father was a doctor, she herself was a high school graduate and had some college education at a time when most Americans had never gone to high school, and her job was in journalism, at the time a middle-class career path.

In the 1950s and 60s, this process continued in full swing, in the Village and other inner neighborhoods of New York, such as the Upper West Side, which unlike the Upper East Side was originally rowdy (West Side Story is set there). Developers were building taller buildings, to Jacobs’ consternation, to house the growing middle-class demand.

I focus on early gentrification in New York and not London, because in this era, the American middle class was fleeing cities. In the 1950s New York was poorer than its suburbs. The wealthiest strata of the city had decamped to Westchester and Fairfield Counties starting in the 1910s and 20s (in 1930 Westchester had 520,000 people, more than half of today’s level, and more than Long Island). Then in the 1950s and 60s this process spread to the entire white middle class, causing a population surge on Long Island, in New Jersey, and in the parts of Westchester and Fairfield that the rich hadn’t already settled in. Moreover, companies were moving out of city centers, often to be closer to the CEO’s home, including General Electric (which moved to the town of Fairfield in 1974) and IBM (which moved to Armonk in 1964). Middle-class taste at the time was firmly suburban.

A better explanation for the early history of gentrification in New York concerns the subway. Before the subway opened, the working class had to live right next to the Lower Manhattan CBD and commute on foot. The els did provide some options for living farther uptown, but they were slow and noisy (they were only electrified around the time the subway opened) and until the early 20th century the working class could not afford the 5-cent fare. This led to extreme levels of overcrowding just outside the CBD, most infamously on the Lower East Side. In 1900, most of Manhattan was open to the middle class but only the Lower East and West Sides were open to the working class. By 1920, the fast subway and the 5-cent fare (held down despite post-WW1 inflation) made all of Manhattan open to everyone, making it easier for the middle class to outbid the poor for housing in the Village.

Production theory

The paper cited at the beginning of this post does not profess consumption theory; it claims that both production and consumption amenities explain gentrification. However, the actual work within the paper leans heavily toward production. It looks at the effect of opening a new rail line from the suburbs to Amsterdam, and finds that this leads to middle-class displacement of lower-education residents, who have less use for the train service. This is also consistent with what working-class residents of some Parisian banlieues think: a newspaper article I can no longer find cites people within Seine-Saint-Denis complaining that all Grand Paris Express will do is raise their rents.

In addition to being more consistent with Dutch and American evidence and American history, production theory benefits from not relying on special local explanations for a global trend. A process that began at similar development levels in the US and Western Europe is unlikely to be about American race relations. Even Tokyo is seeing gentrification in the sense that industrial waterfront areas are redeveloped, if not in the sense of mass displacement seen in New York, London, and other cities with stricter zoning.

There is no burning middle-class desire to live near poor people – quite the opposite, in fact. When the middle class does begin gentrifying a neighborhood, it’s because it offers convenient access to jobs. The same is true on the level of an entire city: San Francisco did not magically become a nicer place to live in when the current tech boom began – if anything, rising rents have led to a homelessness problem, which makes professional workers uncomfortable. A city that wishes to forestall gentrification will make it easy to build housing in the areas with the best job access, in order to encourage people to have short commutes rather than seeking increasingly marginal neighborhoods to move to.


  1. Michael James

    Fair enough.
    Just a point about San Francisco. The change was not actually fomented by tech locating in the city (versus Silicon Valley 60km away) but its workers preferring to live in SF. At the turn of the millenium a colleague got a professorship at Stanford in a prized department but chose to live in SF and commute daily. At the time this would have been almost unheard of, as the Palo Alto area would have been highly prized urban environment (I never thought so but that’s me…). After years passed he eventually chose (this guy had the choice) to move to UCSF. I am sure others, especially those working in the tech that was closer than the San Jose cluster (ie. in Foster City, Brisbane, Haywood etc) as the commute was not so bad (train and even back then they had secure facilities for bicycles). I recall that then-Google VP Marissa Mayer bought a showcase penthouse in SF to live in and commuted (maybe by helicopter, I don’t know …). An earlier generation of Mayers (ie. rich beyond dreams) would have moved into Hillsborough or its ilk (of course they are rich enough to have both if need be).
    So this increase in tech workers within SF was initially a worker-driven thing, not a local jobs thing. Including by second- or third-generation of companies, ie. in which the new-gen had previously worked down the peninsula but when they had the money and power to choose, moved themselves and their new companies into the city. Plus, the city began making room for them in the redeveloping areas south and east of Market, and the large Mission Bay redevo district.

    However I would agree that the gentrification of the Portrero area of SF was its access to those tech jobs nearby (mostly south of SF at that time), and the good transport options (both 280 & 101 and CalTran station). I had friends who worked in biotech (more in South SF & Foster City than Silicon Valley) and they could still (just) afford to buy in Portrero. There would have been resentment as it was a spillover area for lower-SES who wanted to live in the Mission district.

    • Joey

      Peninsula cities have seen huge increases in housing prices lately (and they were expensive before). And most of them build approximately zero new housing. Maybe there are people who choose to live in SF and commute, but there seems to the case that many people would live closer if they could.

    • Alon Levy

      This explanation is common in SF, but it’s not really correct. The tech companies founded in the 1990s and early 2000s located in Silicon Valley, and at the time SF was expensive but not nearly as bad as today. In 2006, when I got grad school admission offers, Columbia offered me $20,000 in annual stipend and Berkeley $18,500, which was proportional to living costs; today, the market rate in Berkeley is higher than within walking distance of Columbia and considerably higher than within two subway stops of Columbia.

      The big change this decade is that big companies now locate in SF proper and not just Silicon Valley. Twitter, Slack, Airbnb, and Uber are all headquartered in SF proper, not Silicon Valley. Airbnb is in Mission Bay, close enough to Caltrain that people can commute, but the other three are in SoMa, as an outgrowth of the SF Financial District (same process I characterize as more common in Europe than in the US). This, in turn, comes from the fact that commuting to Silicon Valley is a nightmare. Transit is for masochists only, and there’s so much traffic on the freeways that it’s only really convenient if you’re rich enough to live in Palo Alto, Menlo Park, Mountain View, and other very rich edge cities.

      The BEA has data on income flows of commuters by county. Go here and look for local area personal income -> gross flow of earnings. Outflow means total income earned by people working in the county living elsewhere, inflow means total income earned by people living in the county working elsewhere, net adjustment for residence means inflow minus outflow. Bedroom communities have positive net adjustment, job centers have negative net adjustment. As you can probably guess, the biggest negative number in the US is in Manhattan, by a huge margin. San Francisco and Santa Clara Counties both have negative net adjustments, but in the late 1990s they were about even (and in 2000 Santa Clara pulled ahead, at -17.2 billion dollars to SF’s -14.2), whereas in the last decade SF has had a bigger negative number, -28.4 billion to Santa Clara’s -22.6 billion in 2016.

      • Michael James

        Alon, I have just read your post a few times but still can’t quite get it. What you describe is exactly what I described:

        Twitter, Slack, Airbnb, and Uber are all headquartered in SF proper, not Silicon Valley.

        Those are exactly the next-gen techs I was referring to. They weren’t big when they were created and they were created in SF by people (I’d bet, without systematically checking) who had worked in existing tech in SV. There were still financial penalties of locating in SF but they wanted to be there, and their workers too wanted to be there rather than in dead sprawled suburbia down the peninsula, trapped into car-dependency. My point being that this does not conform to your notion of location of workers due to local jobs, but the reverse, even if today it may have become auto-catalytic and not so clear (and there may well be workers there reluctantly living in SF but hating it because of the cost and low-quality of housing, but that shouldn’t confuse the history). It is the same all over the world: every place wants these tech centres but there are only a few that are successful and they are in a handful of big(ish) cities because that’s where the workers (and certain other conditions) want to be.

        • Martin

          I think many people who live in Bay Area make frequent visits to SF whether it’s for dinner, events or just playing tourist. Much of it started before Lyft & Uber became widespread where it was easier to go out in SF, drink, and take Caltrain or BART home or crash in a hotel room. Neither of these were a good option in South Bay, East Bay and Peninsula. As this lifestyle continued, many people optimized their life to live where they have fun, and put up with a long commute. Living closer to work seemed pointless if you get home quickly, but going out is hard or limits you to nearby restaurants. Those without kids REALLY had little reason to be home early.

          I used to live 5 mins from work that I traded up for 1 hour commute via Caltrain + shuttle, and despite the commute, my life has been significantly happier since then. I now finally found a job in SF, so I walk to work, meet up with friends for HH afterwards anytime.

          Most of my friends who are in their 20’s, 30’s and 40’s, would much rather prefer a long work commute if it means living in SF. That’s largely what’s driving gentrification – this demand of living close to your playground. Some of my friends are tied to South Bay and are annoyed that we never visit them and they’re the ones that have to drive up to SF. But in reality, suburbs have little to offer once you have a taste of city life.

          • johndmuller

            Children seem to be a large factor in choosing a place to live. In the US at least, the central core is often not considered the best place for raising children. In addition to a somewhat higher level of personal risk involved in city life, the public education system is generally much better funded in at least selected suburbs than in the center city. This often leads to couples sticking it out in the center city enjoying those amenities until their children reach whatever age that they consider requires them to seek out a more desirable school system.

            Of course, having sufficient money to go all out private schooling can overcome this problem, but the higher degree of personal risk still remains and parents will worry about their children in that regard more in the city. Some problems, such as proscription drug abuse, are becoming endemic in the suburbs as well, but good old fashioned street crimes like muggings are still much more popular in the urban setting.

            These issues have little to do with employment, unless the commuting alternatives are absolutely impossible. I suppose they fall under the consumption side if they have to be factored into that dichotomy.

    • Alon Levy

      I just saw this paper tweeted into my feed. It appears that the studies showing that gentrification doesn’t cause displacement suffer from two drawbacks:

      1. In New York, these studies look at gentrification in the 1980s and 90s, when the city was still recovering from its 1970s population losses and there were vacant buildings all over the city, especially in then-gentrifying, now-fully gentrified neighborhoods. Now that the city’s population is higher than ever, the situation may have changed.

      2. Very low-income people in the US move at very high rates no matter what. Sometimes their moves are upward, e.g. they get better jobs, and sometimes they’re downward, e.g. they get displaced to a farther-out neighborhood. One study cited in the above link looks at Philadelphia and finds that low-income residents of gentrifying neighborhoods aren’t likelier to move than low-income residents of nongentrifying neighborhoods, but when they do move, it’s likelier to be to a lower-income neighborhood.

      • adirondacker12800

        “to a lower income neighborhood” is probably an artifact of gentrification. They move into the neighborhood before it starts to gentrify, when median household income is 60 percent of the county’s or state’s median, they stay for years and move out when the median income in the neighborhood is 125 percent of median, to a neighborhood where it’s 80 percent – it’s a lateral move or a step up.

  2. Martin

    I dont think this explanation works very well to understand the changes we have seen over time.

    We have to big changes. A large flow outside the cities in the 1940s-1960s (and continuing much later in many places). And a more modest inflow post 1990. Neither seems to be explained by the distribution of jobs. It was the jobs that followed the commuters outside the city core, not the other way around. Similarly the change back into the cities also do not seems to be driven by an increasing density of jobs in the period. In particular early sub-urbanization was entirely a middle class affair, where everyone still worked in the city center. If anything once again, it is the jobs that are following the residence patterns of the workers.

    I think it is indeed taste differences that can explain it, where urban consumption patterns were something to avoid in the post-war period, while now a smaller minority embrace it. Evidence of this in European cities is that there has been a clear shift in that the upper middle class thought it undesirable to raise children in the 1970s and 1980s in the city core, while this has been changed to a more positive valuation. I think such taste differences are of primary importance.

    The association with inner cities with poverty I think was a general trend with modernism, starting in the 1920s or so (see European cities, and London even earlier, as well as the sub-urbanization of western Tokyo), though the race perspective is indeed particular for the US.

    • Diego Beghin

      Focusing only on job density in the core isn’t enough. Job *access* matters as well. What else changed in the postwar period to make people move to the suburbs? Massive highway building. It became feasible to live in a far-out place with a reasonable commute. But this process can’t keep on going forever, because once a particular suburban area gets built out, it becomes exclusionary and extremely hostile to any extra density. So any new family moving to the suburbs has to move into ever more distant areas. At some point, congestion and extremely long driving distances make job access awful in any affordable suburb. This makes the city core more attractive to the middle class.

      I do think consumption preferences have some explanatory value: when you’ve ruled the parts of the city which don’t have good job access or which are unaffordable, you’ll pick the neighbourhood you like the most. But there are very few middle class people who would willingly suffer through a 2h commute just to live in a neighbourhood which has all the amenities they like.

      The association of the inner city with poverty isn’t universal, especially in the developing world. If you meant only the First World, Paris is a strong counter-example.

    • Alon Levy

      The distribution of jobs in the postwar era didn’t really change, but the ability to access them from the suburbs did. Mass motorization meant that the middle class could access city jobs without living in the city, which was not the case previously. In areas with well-developed commuter rail systems, we see this process begin earlier, with people commuting by train; I’ve read that Sydney’s urban form is low-density because it had good commuter rail early, so people could move to the suburbs even before cars.

      The return to the city trend has a couple of production-side explanations:

      1. As I explain in the post, working-class suburbanization (induced by the subway in New York) created space for the middle class to move in. In Paris, early removal of the working class from the city under Haussmann created the rich city, poor banlieues distinction well before rapid transit; in Stockholm I don’t know the history of when Central Stockholm became richer than the southern and western suburbs.

      2. Increasing traffic congestion has made the suburban lifestyle less tenable.

      3. The middle class has gotten more specialized. Middle-class jobs today are often specialized to one industry or cluster of employers: academia, tech, law, medicine. This makes city job clusters more important. I don’t think it’s a coincidence that within the middle class, the most specialized jobs, drawing from the entire world, i.e. academia and tech, tend to lead the return to the city, and have very low car ownership rates relative to income levels. Where such jobs are available in the suburbs (e.g. Silicon Valley before it became unaffordable to people making less than $200,000/year) or small towns (e.g. many American college towns), people in these fields do live a suburban lifestyle.

      • adirondacker12800

        The distribution of jobs in the postwar era didn’t really change

        Maybe for white collar symbol manipulators but it did greatly for people who worked with their hands. Past tense. The machines have been taking those jobs. On the other hand I remember I-287 being built. Some of them, their jobs and housing, have moved. On the third hand, places that used to be parking lots, in downtown Newark are now moderately tall office buildings. …. the ILGWU sponsored Penn South co-ops. I doubt there are very many UNITE HERE members on the waiting lists and they are probably in hospitality not the garment industry. The eastbound destination on one of the busier bus lines used to be Western Electric. It passed the RCA plants on the way. There are still a lot of people in New Jersey working in “electronics”, not as many of them, as a percentage, wielding soldering irons, as their used to be. .. I worked with someone who, when he was much younger, got a good job at Edgewater assembly. Went with the job to Mahwah Assembly. Mahwah assembly closed a long time ago. So did Edison Assembly, Linden Assembly and Tarrytown Assembly. IIRC correctly Tarrytown is redeveloping into condo, Linden and Edison are mixed use and they are still fighting over the toxic waste cleanup in Mahwah.

      • Michael James

        In Paris, early removal of the working class from the city under Haussmann created the rich city, poor banlieues distinction well before rapid transit

        Adirondacker is correct. Many may not realise it, Paris was/is an industrial powerhouse but all this industry was located in the banlieus so it certainly made sense for their workers to be located there. (A few of the big car plants, Renault in Boulogne-Billancourt and Citroën in the 15th (where today there is Parc André Citroën, hospitals etc), not to mention the food markets, hung out right into the 60s/70s but most such industry was also in the suburbs such as the Citroën plant at Aulnay-s-b (now closed). It was already underway and would have happened without Haussmann, as it did happen all over the world, driven by irresistible factors (scale, larger airports, trans-shipment points, railway marshalling yards, large industrial operations etc).

  3. rjmason

    I don’t find it unlikely at all for proximity to a university to be a consumption amenity!

    A typical university campus:
    (a) is a quasi-park with green space and water features, possibly better maintained and better policed than typical city parks
    (b) may have attractive architecture to look at
    (c) has performance spaces with plays, concerts, speakers, college sport matches, and other events
    (d) has a library which might be partly open to non-university users
    (e) has athletic facilities which might be partly open to non-university users
    (f) may offer night classes to the public

    I have personally consumed all these amenities at one time or another. There are lots of reasons to want to live next to a university even if you don’t work there.

    • Alon Levy

      I think you’re describing American college towns here; it’s not true of KTH or the universities here, and I don’t think it’s true in the Netherlands. To wit:

      a) Utrecht University doesn’t look like a quasi-park. The University of Amsterdam doesn’t, either; it has water features, but the city is replete with canals.

      b) The study I’m citing talks about historical monuments as a consumption amenity. The individual architecture might also matter, but are university buildings really prettier than other buildings in the cities in question? Google Earth tourism suggests they’re not, and here in Paris university buildings are on the ugly side (Jussieu, ugh).

      c) Intercollegiate sports really aren’t a thing here. Plays and concerts aren’t on campus, and have no reason to be near campus; I don’t know where musicians play in Amsterdam, but here the performance spaces are on the Right Bank, not in the Latin Quarter.

      d) I don’t know, are Dutch university libraries open to non-students? American, Singaporean, Swedish, and British ones often aren’t (and apparently at Cambridge even students only have partial access).

      e) Gyms aren’t such a rare facility that people will locate within a kilometer of a university for them.

      f) That’s not really an amenity for people who already have university degrees (who are what the study counts as high-education people).

      • rjmason

        Yes, my comments are based on experience primarily in the U.S. (and primarily in major U.S. cities, not little college towns).

        Here, and I expect in Europe also, there is variation among colleges and universities. Not all campuses are attractive, but some are beautiful. Some may be very private and sealed off from the public, but others are quite open and offer a lot to the neighboring community. One university may be famous and lend prestige to its neighborhood; another may be a community college with little prestige, but useful classes in auto repair.* So the amenities may vary from case to particular case. But if Dutch people TELL you that they value living next to universities, I have no trouble believing them.

        (* I don’t agree that continuing education is not an amenity for people with university degrees. If anything they may be bigger consumers of language classes, continuing professional education, swimming lessons and summer math camp for the kids, etc. etc. I don’t claim it’s a major factor in choosing where to live—that’s why I listed it last—but it counts for something.

        In the U.S., the presence of a university can also have first- and second-order effects on the quality of local children’s schools; children’s schools may be better because the university is there; and in the U.S. that IS a big factor, a huge factor, in choosing where to live.)

        • Alon Levy

          Which universities are you thinking about? Because I don’t think this is really true of Brown. Performers don’t go to Providence, they go to Boston. Yale has a twist in that it has a theater that gets film stars to act in its plays, but there, too, random music concerts don’t really go to New Haven.

          Both Providence and New Haven have bad public schools. Brown tries to take in students from local schools and so does New Haven, but I don’t think it’s well-known to the professional middle class, which from what I’ve seen moves to the suburbs or sends children to private schools; this is also true around Columbia.

          The study isn’t asserting that educated Dutch people say they value living next to universities. It’s asserting they value living next to universities (and it’s multiple regression, so this controls for jobs and the other listed consumption amenities). My claim is just that the jobs that mobile educated professionals take are sometimes disproportionately on or near campus, and moreover switching to a job far from campus is unlikely enough that they can make long-term plans based on having that academic or medical job.

  4. wanderer

    I have trouble seeing how access to jobs alone drove gentrification in San Francisco. There’s evidence of San Francisco rents rising faster than elsewhere as far back as the mid-1960’s (As a precursor the New York Times ran an article about gentrification–but obviously didn’t use the term–in the Telegraph Hill neighborhood in the 1930’s). But gentrification became a major issue in San Francisco starting in the mid-1970’s, as people started to rehab the famous Victorian houses. But over the decades, San Francisco’s share of regional employment has declined. The heaviest job growth has been in Silicon Valley and to a lesser extent along the 680 corridor in the East Bay suburbs. The San Francisco Financial District has remained a major employment center, but the changes in housing costs are somewhat different than the changes in jobs.

    It looks like both job access and consumption preferences are affecting the situation in San Francisco. That’s what led to the famous tech buses from San Francisco to Silicon Valley. Folks will say things like “I didn’t come all this way to live in Mountain View”.

    • Alon Levy

      What you’re missing is that in the last decades rents have been rising faster than elsewhere in Silicon Valley, too. San Francisco is cheaper to live in than Palo Alto and Mountain View. The trend of SF suddenly skyrocketing faster than San Jose and such is much more recent, as is the trend of SF tech employment growing faster than Silicon Valley tech employment.

  5. Pingback: Where are Transportation and Housing Politics Going? | Pedestrian Observations

Leave a Reply

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

You are commenting using your 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

This site uses Akismet to reduce spam. Learn how your comment data is processed.