Over the past 40 years, different regions of the US have pulled apart economically. Average incomes in a handful of thriving metro areas have risen quickly, while those in many parts of the country have stagnated. In 1980, only about 12% of the US population lived in metropolitan areas with incomes more than 20% higher or lower than the national average. By 2013, more than 30% of the population did.
In a new paper published in Social Forces, I explore the drivers of regional divergence. There are two main types of process that could be contributing to geographic divergence. The geographic distribution of people could be changing, such that more people at higher rungs of the income distribution are concentrated in one set of cities, while more people on the lower rungs are concentrated in a different set. This process of income sorting, which has been the focus of most research on regional divergence, could result from the movement of individual people or changes in which kinds of jobs are found where. Alternately, the shape of the income distribution itself could change, altering the earnings that people on different rungs take home without changing where they live.
To distinguish between these two processes, consider a hypothetical country with two cities. To start, both cities have similarly shaped income distributions, but City A is poorer, with a mean of $8, while City B has a mean of $12:
Now imagine the country goes through an episode of income sorting. This could be a result of high-income people moving from City A to City B, or it could result from City B doing better at creating high paying jobs. Whatever the reason, most of the jobs paying more than $10 end up in City B, while those paying less end up in City A:
The sorting process pulls the average incomes of the two cities apart: mean income in City A falls to $7.40, while that in City B rises to $12.60.
Now imagine that instead of sorting, the county experiences a rise in income inequality. The distribution of income ranks in each city stays the same, but everyone below the national average sees their income fall by $1, while everyone above it sees theirs rise by the same amount. This results in a very different change to the income distributions in each city:
Following the national rise in inequality, both cities see their income distributions stretch out, and there’s quite a bit more overlap. But the mean incomes still diverge by an amount similar to that in the sorting scenario: the average income in City A falls to $7.50, while that in City B rises to $12.50. Just looking at the averages, it would be hard to tell a given divergence was due to sorting or inequality.
To figure out whether the divergence in regional incomes in the United States is more due to sorting or inequality, I run a series of simulations based on Census micro data. I separate out the overall change in incomes within each city into changes in the income ranks of the people who live there and national changes in the dollars associated with each rank. Then I run a simulation where I hold income inequality at its 1980 level, while allowing sorting to proceed as it actually did. By recomputing standard divergence metrics under this hypothetical and comparing them with what actually happened, it’s possible to estimate the contribution of sorting alone to regional divergence. Doing the inverse gives an estimate of
When I run these simulations, it turns out that sorting by itself only accounts for about 23% of the divergence we’ve seen, while rising inequality at the national level accounts for about 54%. So it’s perhaps more accurate to think of the growing disparities between regions of the country as resulting from one national trend–rising income inequality–than from a collection of place-specific factors.