UPDATED: February 20, 2023

What are the determinants of inequality? The first step in answering this question is defining exactly what we mean by inequality. A working paper by Chetty, Hendren, Kline, and Saez takes an interesting approach: it measures inequality based on the likelihood that a child born into a poor family will rise in the overall income distribution.

They call this measure “absolute upward mobility.” If absolute upward mobility is high, it means a child born into a poor family has a good chance of rising in the overall income distribution. If it is low, that means the poor child will likely be poor when she grows up.

The authors construct upward mobility for different cities. A city with a high score is considered more equal; a child born to a relatively poor family in the city has a good chance of rising in the income distribution. A city with a low score is more unequal, as a poor child is likely to remain poor as an adult.

The part of the study that interests us most is the correlation between their measure of inequality and other variables at the city level. In other words, what characterizes the most “unequal” cities?

It turns out that one variable is far more statistically powerful than any other variable: the fraction of mothers that are single. Here is the correlation across cities between upward mobility and the fraction of mothers that are single.

Cities with more single mother households have less upward mobility

Each dot in the picture above is a city. The vertical axis is the measure of upward mobility, and the horizontal access is the fraction of mothers that are single. Cities with a higher fraction of mothers that are single have lower upward mobility. The relation is very strong, as the regression line shows.

As the authors show in their Table IX, inclusion of this single variable in a regression knocks out many other variables we might think affect the likelihood of rising in the income distribution. It knocks out the Gini coefficient and the high school dropout rate, for example. The statistical power of this one variable is stunning.

Now, we are wading into some hugely controversial issues here. So it is crucial to be scientific about what this figure means. This is a correlation; it is not necessarily evidence of causation. It doesn’t mean that if we somehow paired up single mothers with a partner, then suddenly inequality would go down. Further, the channel through which inequality and family structure are related is complex. For example, the authors show that cities that have a large fraction of mothers that are single have less upward mobility even for families where both parents are in the home. So again, the channel is unlikely to be a direct effect of single mothers on inequality.

The authors speculate that the fraction of mothers that are single is capturing other measures of income inequality. As they write, “…the fraction of single parents may capture some of the variation in other factors, most notably the level of income inequality. Indeed, [cities] with greater inequality have significantly higher rates of single parenthood. Hence, the results in Table IX are consistent with the view that inequality affects children’s outcomes partly by degrading family structure and social ties in the community.”

In our view, the last sentence is a bit strong given the evidence. It seems to imply that income inequality causesfamily degradation. We don’t see evidence of that in this study.

But more importantly, it brings up the main challenge for empirical economists and other social scientists studying inequality. Those on the right of the political spectrum believe the relation in the chart above is a causal one: some factors changed social norms and led to the breakdown in family structure. A strong family structure is crucial for instilling the education and habits necessary to succeed in the workplace, the argument goes. As a result of the breakdown in families, income inequality has risen. This view is expressed most forcefully in Coming Apart by Charles Murray.

Those left of center argue the causality is the opposite. Rising income inequality was the trigger that led to the disintegration of families. Less-educated Americans, especially men, cannot find a decent-paying job and would have to seek other ways to earn more money. As Paul Krugman put it: “Still, something is clearly happening to the traditional working-class family. The question is what. And it is, frankly, amazing how quickly and blithely conservatives dismiss the seemingly obvious answer: A drastic reduction in the work opportunities available to less-educated men.” He concludes: “the social changes taking place in America’s working class are overwhelmingly the consequence of sharply rising inequality, not its cause.”

So which is it? Did changing cultural values degrade the importance of the two-parent family, which then led to income inequality? Or did the rise in income inequality and the lack of opportunities for less-educated men cause the disintegration of families?

We are still looking for studies that answer this question definitively. If there are studies out there that help us interpret this correlation, let us know in the comments or on twitter. We will happily update the post.

But if progress is going to be made, we believe it will be led by researchers using advanced empirical techniques to come to objective conclusions about causality. There was a firestorm of coverage of this study when it was released in January. Some of it was balanced and sensible. But a lot of it was extreme and ideological. It’s going to be very hard for people to remain objective when looking into this issue. But objective study is exactly what we need.

The authors of this study should be commended for adding data to a controversial issue. Now let’s see if we can get more traction in figuring out the causal chain.

UPDATE: here are a few sources we’ve discovered while writing the post (we will add more as we come across them)