Is Higher Income Inequality Associated with Lower Intergenerational Mobility?

A lot of our political debate boils down to questions about equality of outcomes versus equality of opportunity. But it turns out that they’re pretty closely related. Take a look at the chart below, which is from a terrific recent speech (with charts!) by Alan Krueger:

The horizontal axis shows the Gini coefficient, which is a summary of the degree of income inequality for each country. I think of this as a measure of inequality of outcomes. The United States sits out there on the right, which says that we have high inequality, which I bet that doesn’t surprise you.

The vertical axis shows a measure of intergenerational mobility, which summarizes the relationship between your income and your parents. A score of zero means that we have equality of opportunity — the kids of rich people earn as much as the kids of the poor. A high number means that the rich parents have rich kids and poor parents have poor kids. The U.S. has a score of 0.4 which means that, on average, you pass on 40% of your economic advantage to your kids: if I earn $100,000 more than you, then on average, my kids will earn $40,000 more than your kids. So I think of this as a measure of inequality of opportunity. You’ll notice that the U.S. also scores high on this measure. Americans are often surprised to learn that in the land of opportunity, your life outcomes are largely determined by your parents.

It’s striking just how closely related inequality and mobility are. And it’s political dynamite.  Why? If income inequality in one generation can be linked to unequal opportunity in the next, then income inequality can’t just be dismissed as the politics of envy. My bet is that this chart that will launch a thousand papers, as economists try to sort out just what these linkages are. Whatever the answer, it will transform our thinking about inequality.

But in the political arena, the first instinct is to deny. And so not surprisingly, this chart has led to a wonk-fight.  If you want the dirty details, here’s Scott Winship’s critiqueMiles Corak’s replyWinship’s counter, and Corak’s response. I score this fight for Corak. It’s not even close: he’s the leading figure in international comparisons of mobility, and he’s put together the best data around. Other authors and other datasets yield the same conclusions.

There’s a broader lesson here for what we learn from these sorts of data debates. Basically, Winship has a bunch of complaints about how the data are constructed — and many are valid. He says that it’s difficult to construct internationally comparable measures of income inequality – that the chart should use inequality from an earlier era, and that only some types of inequality would generate immobility. He also points out that mobility is difficult to measure: the data come from different countries with different researchers using different methods. It’s a standard play from the wonk-fight playbook: throw lots of mud at the data, and hope that this leads people to mistrust the conclusions that follow.

Here’s the thing: his criticisms actually strengthen the original finding.

Think about it. Imagine how strong the “true” relationship must be if it shows up even when using only rough proxies for the “true” levels of inequality and immobility. In light of Winship’s criticisms, the high correlation in this chart is all the more remarkable.  If his gripes are correct, then graph understates the correlation between inequality and mobility.

Now there’s one more possibility to consider.  If mis-measurement of inequality were related to mis-measurement of mobility, then the chart could reflect correlation in these measurement errors. But the two measures come from different datasets, different researchers, and different methods — so that’s pretty unlikely. And the measurement errors involved aren’t big enough to drive a correlation this strong.

Predictably enough, I spent yesterday reading lefty blogs trumpeting Corak’s analysis, and right-leaning blogs who didn’t want to believe the inequality-mobility link, endorsing Winship. But both missed the bigger picture implications. Either you’re convinced by Corak that the data can be trusted, and that they show there’s a strong link between actual inequality and actual mobility.  Or you believe Winship that the data are a pretty poor proxy for what’s really happening, and so there’s actually a very strong link that’s being disguised by imperfect data.


Something that seems to have been overlooked in all this is the difference between the availability of opportunity for economic mobility, and whether people choose to take advantage of those opportunities.

In this country we have a lot of opportunity. (I've seen this at first hand, because I've gone from being flat broke and homeless to probably 4th quintile of income and fifth quintile of wealth.) These opportunities are, however, mostly taken advantage of by immigrants - check for instance the percentage of foreign-born engineers & scientists earning $100K+ salaries. Mainstream American culture strongly motivates against education (especially in geek & nerd subjects, where good money is practically guaranteed). It pressures people into living above their means, buying on credit, not saving. For instance, check the savings rates and consumer debts of those countries with higher mobility: I believe you'll find a strong correlation there.


Milton Recht

College educated parents are more likely to have college educated children and the US ranks number 1 among OECD countries for having a college educated workforce. College educated workers earn more than non-college educated workers.

According to a 2011 OECD report, the rate of increase for college educated among new entrants into the workforce is growing faster in other OECD countries than in the US. Among new entrants to the workforce, the US drops to 15th versus number 1 for its entire workforce.

In the non-US OECD countries there is and has been greater growth of college educated workers, which means they will earn more than their parents, which makes the statistics appear that there is more upward mobility.

Once the growth rate of college educated workforce peaks in non-US OECD countries as it has in the US, the income mobility rates among all OECD countries including the US will equalize.

Source: Education at a Glance 2011, OECD Indicators



Seems like a pretty poor argument to me. If we accept Winship's claim that the data are a poor proxy, this certainly has a chance of masking the supposed 'very strong' link, but it also has a chance of making a trend appear when there isn't actually one there. Simply put, bad data can either weaken true relationships OR create fake ones. It's complete nonsense (not to mention bad statistics) to pretend otherwise.

That being said, it doesn't seem terribly surprising that there should be this kind of correlation when you think about access to education, inheritance, etc. so my guess is that the relationship is real.


Bad data can only create fake relationships when the badness follows some relationship. To argue a correlation isn't real, you need to show some plausible mechanism why errors are correlated. Truly random badness/noise can only weaken a correlation.
Looking for plausible mechanisms why errors might be correlated is an important part of science. If such a mechanism is found, a new experiment should be conducted to test it.

James B.

Wow, somebody should do something about that! Maybe we should pass the "Equalization of Opportunity" bill ... that would surely make things more fair. Who is John Galt?
... what was that short-lived "sarcasm" punctuation mark?

Seriously folks, why are we looking at this country by country ... we need productive people in the U.S. and if they find a better opportunity to be productive they will leave the U.S. Productive people should reap the benefits of their productivity. That may sound greedy or harsh, but the fact is ... it is the productive capacity and economic might of the U.S. that gives us the luxury of taking care of unproductive people in a charitable fashion ... NOT equalization of opportunity for everyone. We seem to have lost sight of that.

Michael Samsel

It seems that this graph shows that if through social policies all adults make the same in a given country, that their children when they become adults will all make the same, as long as the children remain in that country. Is that profound? In such a situation, behavior is irrelevant to income. Only where behavior influences income, and disparity is possible, is the question of inter-generational mobility relevant. The behaviors rewarded may not be fair or rational always, but they of course run in families--nature, nurture, or direct instruction-- and so of course disparity co-varies transgenerational income inertia.

The real need is for a way for anyone with a reasonable set of behaviors to make at least a good wage, and not tamper with basic liberty


I question the ideological motives of anyone who attempts to make inferences from a measure such as the GINI index. It has zero practical construct validity. Its effective implication is that everyone in the world values absolute income the same. Its missingness of data is obscene. It is a strictly quantitative indicator and therefore ignores how different cultures value income differently- if at all in the Western sense. It does not capture urban and rural income differences within a country, nor across countries. Etc. etc.

I think it would be absurd to suggest that the US’s GINI score and Singapore’s GINI score are in any way comparable . . . or, China and Denmark’s. There is no reasonable, nor meaningful, qualitative interpretation. Wake up! It’s Nonsense.

alex in chicago

What is the alternative? If you confiscate wealth, what is the incentive of people to not frivolously spend money prior to their death?