You have probably had this moment: in a city you have lived in for years you turn down a street you have never walked, and suddenly feel you should not be here. Nobody says anything, nothing happens; it is just that the density of shops, the speed people walk at, who is looking at whom, have all switched to another setting.
Cities sort people into groups by address, and those groups then live startlingly different lives. This page takes apart three things: how the earliest researchers saw this, and what they missed; how much force "where you live" actually has; and where the original residents go when a district gets expensive. The answer to the last one runs against intuition, and the disagreement turns out to be about measurement aperture, not about which side you are on.
In 1920s Chicago the population had multiplied several times over within decades, and new migrants were packed into the ring of old housing around the centre. The University of Chicago had a Robert Park who had spent more than a decade as a newspaper reporter before turning academic, and his demand on students was famous: don't sit in the library, get out on the street, get the seat of your trousers dirty.
They did go — dance halls, cheap lodging houses, gang turf. This body of work became known as the Chicago School, and its working intuition was a good one: a city is not a pile of buildings, it is a machine that assigns people to positions. In the diagram of concentric zones Ernest Burgess drew in 1925, the ring pressed against the business district is the zone of transition — cheapest, most crowded, highest turnover.
A bit earlier, the German sociologist Georg Simmel had written in 1903 that the city dweller grows a blasé attitude: too much stimulus, the nerves cannot react to every item, so differences get flattened — not coldness, exhausted bandwidth. In 1938 Louis Wirth boiled the tradition down to three variables: size, density, heterogeneity. The more people, the more crowded, the more unlike one another, the more relations among acquaintances give way to relations conducted by function.
The trouble was their next sentence: they judged those crowded poor districts in the zone of transition to be "socially disorganised" — no norms, no organisation, loose sand.
The man who dragged that sentence across the floor was in his twenties. From 1937 to 1940, William Foote Whyte moved into an Italian-American neighbourhood in Boston's North End for three and a half years (he called it Cornerville in the book), spending his days with a man he called Doc and his corner-boy crew. He found the order there terrifyingly fine-grained: who should pay, who speaks first, who owes whom a favour, all of it ranked and stable.
The finest stroke is bowling. Whyte wrote the scores down game by game and found that how well a man bowled tracked his rank in the group — a low-ranking man who bowled well elsewhere simply could not deliver against his seniors; the others jeered, and he deflated. Same pair of hands, different setting, different score. Status is not an abstraction; it bent the ball on the spot.
The transferable point: "there is no order here" very often means "wrong instrument for seeing order" — the observer walks in carrying a picture of what order looks like, fails to find it, and reports its absence. (Herbert Gans reached the same conclusion in another district of the same city in 1962 and called his book The Urban Villagers.) The disorganisation vocabulary took heavy criticism later, and deserved it: it read a group's way of living as pathology. Evidence: the concentric-zone model is descriptive induction from a single city; Whyte and Gans are single-site long-term fieldwork, giving mechanism rather than distribution.
Within one city, life expectancy can differ by more than a decade between postcodes. Anyone can measure the gap; the hard step is the next one. Did the neighbourhood produce these outcomes, or were people whose trajectories already differed sorted into different neighbourhoods? This is not pedantry: where you live is never random — income, work, where your relatives are, whether you can get a loan, all take part in that sorting. So "children from poor neighbourhoods end up poor" proves nothing at all.
One route is to measure the property of the neighbourhood itself. In 1995 Robert Sampson, Stephen Raudenbush and Felton Earls surveyed 8,782 residents across 343 Chicago neighbourhood clusters. What they measured is called collective efficacy: whether neighbours trust one another, plus whether they are willing to act for the common good — if a few adolescents are hanging around the street during school hours, will anyone say something.
The result appeared in Science in 1997: violence was markedly lower in neighbourhoods with high collective efficacy, and the association held after controlling for poverty, residential stability, individual characteristics and prior levels of violence. But this is correlational — low efficacy could equally be the consequence of violence rather than its cause.
The second route is far harder: randomise directly. From 1994 to 1998 the US Department of Housing and Urban Development ran an experiment called Moving to Opportunity, enrolling 4,604 families living in high-poverty public housing across five cities and assigning them at random, one group receiving a voucher usable only in low-poverty neighbourhoods. That is rare in social science — "where you live" was actually randomised.
The early results disappointed a lot of people: in 2007 Kling, Liebman and Katz reported that adult employment and earnings did not improve. The turn came in 2016, when Chetty, Hendren and Katz linked the experimental roster to tax records and looked again once the children had reached their twenties. Those who moved before turning 13 were earning about $3,477 more per year in their mid-twenties than the control group (a control mean of $11,270, so roughly 31% higher), and attended college at higher rates; those who moved after 13 gained nothing, and slightly less than nothing.
That changed the shape of the question. A neighbourhood is not an on/off switch but more like a dose accumulating by the year: the longer you live there the more accrues, and a move in your teens leaves too few years to accrue. The early studies saw no effect because they looked too soon — and at the wrong people.
Two things stated honestly. The sample is families who volunteered to move out of high-poverty public housing, so extrapolate with care. More importantly: we know it works without knowing what it works through. School? Peers? Safety? Still unsettled — and different answers imply completely different policies.
In 1964 Ruth Glass, a sociologist working in London, coined a word for what she was watching: rows of working-class housing bought up one by one by the middle class, refurbished, and the social composition of whole districts replaced along with them. She called it gentrification. What drives it has had two explanations coexisting ever since.
One looks at capital: Neil Smith's rent gap, proposed in 1979 — when the ground rent a decayed district currently earns falls far enough below what it could earn, and the gap grows large enough, capital flows in to refurbish and capture the difference; who moves in is consequence, not cause. The other looks at people: David Ley's account is demand-side, a post-war middle class working in culture and knowledge whose tastes ran to old housing and walkable streets, and whose preferences pulled the capital after them. Both sides have presentable evidence and neither has beaten the other — the dispute is over weights, not over existence, and this page passes no judgement between them.
The counterintuitive part is one layer down. Intuitively, gentrification equals original residents being pushed out. But since the 2000s a body of work that tracks specific individuals has produced a different result: Freeman using national housing survey data, Ding and colleagues using a Philadelphia credit-bureau panel, Brummet and Reed using longitudinal census data — low-income residents of gentrifying neighbourhoods generally do not move out at higher rates than those in comparable non-gentrifying ones.
Don't conclude yet: the same body of work also found that the change in a district's composition comes mainly from the inflow side — college-educated people move in at a much higher rate.
Put those two findings together and the out-migration rate is aimed at only half the picture. Homes empty all the time — a job change, a marriage, a return home. What decides whether the district changes is who takes them next: if everyone who takes them can afford the new price, the next generation of the original residents can never get back in. Peter Marcuse gave this situation a name back in the 1980s: exclusionary displacement — the people kept out never appear on any out-migration list, because they never arrived.
So the disagreement here is first of all about aperture, not about sides: one measure counts the outflow, the other would have to count an inflow that would otherwise have happened. The same body of work also found that the low-income residents who do leave land in worse places, and that less-educated renters are hit harder. "The overall out-migration rate is not high" and "some people were hit very hard" can both be true at once.
Evidence strength: these are administrative data or credit panels able to follow the same individuals, considerably better than cross-sections, but none is a randomised experiment, and results vary a great deal between cities. Whether anything should be done about it is a value judgement, outside the range of this discipline — what can be stated clearly here is only this: which arrow is it that you mean to change.