Persistent poor neighbourhoods

Why can a neighbourhood stay poor for a century although almost everyone living there has been replaced?

Take a city tract that ranked among its poorest in 1930. Two world wars, a great migration, urban renewal, deindustrialisation and several waves of immigration later, scarcely a single family living there descends from those who lived there then. And yet in 2010 it still ranks among the poorest. Sampson's work on Chicago found exactly this: neighbourhood rankings on disadvantage stay remarkably stable across decades of near-total population turnover.

That is the puzzle worth sitting with. If poverty were simply a property of the people, replacing all the people should have moved the neighbourhood. It did not. So what is it that persists, if not the residents?

Begin with the obvious candidate. Perhaps poor people move to cheap places, so the place looks poor because of who arrives. That is true — but notice it is a mechanism, not a dismissal. Ask why the place is cheap, and ask what happens to a household that gets richer while living there.

Cheapness first. Housing is durable in a way almost nothing else in an economy is: a building put up badly in 1900, or beside a foundry, or on land a mortgage insurer once marked as hazardous, is still standing and still cheap in 1990. The original cause can be entirely dead and its price signature still alive in the stock.

Now the second question. A household whose income rises has options, and one of the things people buy with income is a different address. So the neighbourhood continuously exports its successes and imports whoever has least choice: the population is not a sample of who lived there, it is the residue of a filter. Do you see that this alone would keep average income low forever, even if the neighbourhood did each of its residents nothing but good?

But it does not stop there, because prices absorb everything. If a school improves, the rents around it rise — that capitalisation is well documented in studies comparing houses just either side of an attendance boundary. So an improvement in a place is partly converted into a price that screens out the people the improvement was meant to reach. The good neighbourhood becomes expensive; the cheap one keeps its function as the place available to those with nothing to bid with.

And there is a supply side. A thin tax base and a low expected return make both public and private investment scarce; insurers, lenders and shopkeepers price the area's history into today's decisions. Federal mortgage-insurance maps of the 1930s — the redlining maps — graded neighbourhoods, and boundary-comparison studies find differences in homeownership, house values and segregation on either side of those lines persisting for decades after the policy was abolished. A motorway driven through a neighbourhood in 1960 is likewise still there, still severing it, long after everyone who chose the route has gone.

So three engines run together — durable physical stock, a sorting filter that exports success, and capitalisation that converts any improvement into a price. None requires the residents to be the same people; each hands the disadvantage to the next set of occupants.

Two refinements matter, and the second is a genuine caution.

First, "the neighbourhood causes poverty" and "poor people sort into the neighbourhood" are not rival answers. They are two turns of the same wheel. Sorting supplies the population; capitalisation and disinvestment maintain the conditions; the conditions then shape the next generation's outcomes and feed the sorting again.

Second, how big the causal neighbourhood effect actually is remains contested, and honestly so. The core difficulty is that families are not randomly assigned to places, so any raw comparison confounds the place with whoever chose it. The best evidence comes from designs that break that link. The Moving to Opportunity experiment, which randomly offered vouchers to move out of high-poverty projects, found little effect on adults' earnings — but later work following the children found that those who moved young earned substantially more as adults, while teenagers who moved gained nothing or lost. Chetty and Hendren's study of millions of families who moved found the same shape: the benefit of a better area scales roughly with the number of childhood years spent in it.

That is a specific and rather demanding claim — neighbourhoods act on children, cumulatively, through exposure — and it is not the same as saying an address determines a life. The estimates are real but moderate, they vary by place, and plenty of individual variation lies elsewhere entirely.

• Why mixed-income redevelopment so often displaces the residents it was meant to help.
• Whether housing vouchers or place-based investment does more per pound, and for whom.

Every term the collection defines is gathered in the glossary.