Mobility by Postcode

The standard defence of inequality is mobility. Large gaps are acceptable if anybody can cross them, because then the distribution describes positions rather than people, and a poor year is a stage rather than a fate.

It is a serious argument, and it turns on an empirical question: how much of a parent's position is inherited by their child?

The measure is intergenerational elasticity. An elasticity of 0.5 means half of a parent's income advantage relative to the average carries into their child's adult income; an elasticity of zero would mean where you start tells you nothing about where you end.

The comparative numbers are not what most people expect. Denmark, Norway and Canada sit around 0.15 to 0.2. The United States and the United Kingdom sit around 0.4 to 0.5: among the least mobile rich countries measured.

And surveys consistently find that Americans substantially overestimate mobility in their own country while Europeans underestimate it. Those beliefs, rather than the measured rates, predict support for redistribution, which is one of the more interesting findings in political economy: what people think the ladder looks like matters more politically than what it is.

Plot inequality against immobility across countries and you get the Great Gatsby curve, named by Alan Krueger. The slope is clear and upward: more unequal countries are less mobile.

The causal direction is not settled. Higher inequality raises the stakes attached to each rung and gives advantaged parents more resources to secure their children's position; or both may follow from the same institutions, such as how schooling is funded and how concentrated housing is by income. What the curve does undercut is the specific defence it addresses, because the combination of high inequality and high mobility is not on offer in the data. No country has it.

Then the American data got dramatically better, and the finding that came out was about geography.

Raj Chetty's team at Opportunity Insights linked tens of millions of anonymised tax records to the places children grew up, which allowed mobility to be measured not by country but by county and eventually by neighbourhood.

The variation within the United States turned out to be larger than the variation between countries. A child raised in the bottom fifth in Salt Lake City had roughly double the chance of reaching the top fifth as a child raised in the bottom fifth in Charlotte. Some American commuting zones are as mobile as Denmark; others are far below anything measured in a rich country.

That could be a fact about the people who live in those places rather than about the places. The team tested it the way you would want: by looking at families who moved.

Children whose families moved to a higher-mobility area did better as adults, and (this is the part that makes it a causal claim) the benefit was proportional to the number of childhood years spent there. A child who moved at four gained roughly twice what a child who moved at twelve gained. A dose-response relationship of that kind is very hard to explain by selection, because the families are the same families.

So it is about the place. Which raises the question of what about the place.

Five correlates emerged: less residential segregation, lower local inequality, better schools, more two-parent households, and higher measures of social capital. These are correlations rather than established mechanisms, and several are plausibly downstream of the same underlying factors, so the policy implications are less clear than the finding.

A later paper using Facebook friendship data sharpened one of them considerably. The strongest single predictor of upward mobility turned out to be economic connectedness: the extent to which low-income children have friendships with higher-income ones. Not the amount of community activity, not civic participation, but cross-class friendship specifically.

Which suggests that what segregation does is not merely concentrate disadvantage but sever the connections through which information about opportunities, norms and expectations travels. That is a mechanism you could act on, and it is a considerably more specific target than "improve communities".