What We Didn't Know About Multigenerational Mobility: In Conversation with Gary Solon

Better income data has uncovered surprising insights about income mobility and income inequality, says the noted researcher and economics professor.

Date Posted

14 Feb 2016


Issue 14, 14 Feb 2016

Income Mobility: Lower but more stable than previously believed

Thanks to better data and methods, researchers over the last few years have found that there is much less intergenerational mobility than previously thought. Inequality is transmitted from one generation to the next to a greater extent than was believed on the basis of the research of thirty years ago.

The belief among researchers thirty years ago that we live in very mobile societies is really a delusion that came from having poor data. For one thing, single year income measures tend to be imperfect measures of longer run income. On top of that, we are subject to a lot of reporting error because if you ask people in a survey how much money they made last year, you often get fairly inaccurate answers. Such measurement noise tended to obscure the patterns that are actually there. Thanks to having better data and analysis, we now understand that we were underestimating intergenerational persistence in the early research.

However, my own research has also found that intergenerational mobility over the last three decades seems to be fairly stable in the US. This was shocking, because I had expected that since inequality has risen a lot in the US, mobility would have declined. But the data shows otherwise. One speculation about why this is the case might have to do with a great expansion in public programmes intended to help fight poverty, including providing more services to children coming from poor families. So it’s conceivable that some of those programmes had a sufficient impact to offset some of the opposite tendency towards lower mobility.

Intergenerational Mobility: Differences and persistence not fully understood

Some work has been done on differences in income persistence across the income distribution in the United States. While people were expecting a poverty trap, there seems instead to be more stickiness at the other end: a ‘wealth trap’ if you will. There are probably more rags to riches cases than the other way around. Anecdotally, there seems to be better safety nets for the offspring of the wealthy. For instance, although my father was from a poor family, I had the chance to go to Harvard, an elite university. An extraordinary percentage of the students there were from a few elite private schools. Some of them spent all their time drinking and so forth, yet I have a hunch that those people are not doing so badly today, even if they haven’t changed their work habits all that much. They have enough going for them, including their family wealth and connections. They have more buffers against misfortune.

The question is: what can we do to level the playing field more, not in the sense of worsening the oppourtunities of well-off people, but finding things that are worth doing for children from disadvantaged families.

A brand new study by a team of Harvard and Berkeley researchers in the US, who have been able to get access to a large pool of income data from the tax authorities, has also been able to look extensively at intergenerational mobility quite accurately down to the city level. They seem to have found substantial differences in intergenerational persistence across cities. The study itself takes only an exploratory look at the characteristics that are most correlated between city differences and intergenerational persistence. However, they have observed that cities with more inequality tend to also have more intergenerational persistence. Other particulars are more difficult to understand, such as the correlation between declared religiosity and mobility, as well as the correlation between intergenerational persistence and the percentage of families in the cities headed by a single mother. One can speculate wildly about these correlations, but it is difficult to know exactly what is going on.

There are also substantial cross-country differences in international persistence that are somewhat correlated with cross-country differences in inequality. Norway, for instance, has both less inequality than most other rich countries and also more intergenerational mobility than most other rich countries. Canada, a neighbouring country culturally similar to the US, has only a bit less inequality than the US, but much more mobility. Again, one can speculate that there are some connections to how progressive public policy is in these countries, but these are loose relationships; you can see the patterns but the precise factors remain fairly mysterious.

While people were expecting a poverty trap, there seems instead to be more stickiness at the other end: a 'wealth trap'.

There is also a distinction between absolute and relative mobility. What I have been referring to so far is relative mobility, which is the connection between one’s position in the income distribution within one’s own generation and what the parents’ position was. So you could have a situation where your parents were at the 20th percentile and you’re also at the 20th percentile, but because the whole income distribution shifted upward, your absolute material wellbeing may be way ahead of your parents’.

Level The Playing Field: Some public policies are worth pursuing, but we need to be clear which

So far, the only clear-cut policy insight from this research is that we have less actual intergenerational mobility than we believed. One argument that is sometimes made is that we still have a fair society because everyone has the same opportunity to succeed — I think this is somewhat debunked by the better evidence on intergenerational mobility in the US. The playing field is clearly not altogether level. There is a clear difference in the average prospects for someone from a rich family versus someone from a poor family. The other argument that has been made is that even though the playing field is not level, it doesn’t mean public policy can do much good.

The question is: what can we do to level the playing field more, not in the sense of worsening the opportunities of well-off people, but finding cost-effective things to do that are worth doing for children from disadvantaged families. We should be looking at particular public policies — and there has been interest in a wide range of possible things ranging from improving nutrition for pregnant women and young children, to free school programmes and educational reforms, to more comprehensive healthcare and so on. Do these initiatives actually make a big difference in improving the opportunities of disadvantaged children? How large are the benefits relative to the cost of such programmes? While it would be difficult to make a hard-headed cost-benefit analysis of these programmes, I do think that is the next step. I feel like I’m in complete agreement with what I’ve been hearing from public officers and researchers in Singapore, which is that we need to go on to the practical questions of what policies might or might not be effective and worth pursuing.

It is remarkable that a society that’s becoming so affluent is actually expressing so much concern about growing inequality. The impression I am getting is that there is more concern here in Singapore than I hear about in the United States. A very large part of the American identity is viewing the country as the land of opportunity — and it is in many ways. Certainly the United States is a country where immigrants have come to seek opportunity. But it is striking how much stratification there is within the country’s own income distribution. More recently, results from researchers like me have been showing that actually there’s statistically more intergenerational persistence in United States than in some other rich countries. There is a sort of cognitive dissonance there.

The Power of Numbers: Good policymaking demands good data

In US research on intergenerational mobility, the huge breakthrough was the creation during the 1960s of two long running longitudinal surveys — the Panel Study of Income Dynamics and the National Longitudinal Surveys of labour market experience — which for the first time gave us fairly accurate intergenerational income data for representative samples from the population. The work that I and other researchers have done would not have been possible without these surveys.

The point of empirical research in economics and elsewhere is to update our 'priors' — i.e. our beliefs and assumptions.

However, it takes a very long time to start a survey like that and then get an intergenerational perspective. The survey has to have been running for at least a quarter of a century to get you there. This highlights the potential power of administrative data — such as data from the tax system, or the pension system — which is being collected on an ongoing basis for programme reasons. In the US, Harvard’s Raj Chetty and his team have been able to get research access to tax data: their work will result in landmark studies that will make a big difference for understanding income mobility in the United States.

One thing I’ve been hearing from researchers in Singapore is concerns about access to data. Of course, there are valid concerns about protecting confidentiality, but there are potential gains to be made from responsible research use of such data, as long as a mechanism can be found to make this available to researchers in a way that does not compromise the rights of individuals. Access to such data is easier in Sweden, which is part of why I’ve sometimes collaborated with friends in Sweden and worked on Swedish data to look at intergenerational mobility and related issues. To some extent, I’m surprised at the ability researchers have to get data in Sweden that no one in the United States would ever be able to access in our own country, and there have been important research payoffs to doing this. In Singapore, there are probably opportunities to exploit administrative data to make a lot of progress on understanding these issues better here.

I have tremendous respect for the potential power of statistical techniques and good data analysis for informing us about what policies work in what ways. In fact, progress on just more accurately describing intergenerational mobility, using better data and sensible statistical work with the data, can help improve our understanding of the societies we live in and let us learn more about how to improve them. My advisor in graduate school put it thus: the point of empirical research in economics and elsewhere is to update our ‘priors’ — i.e. our beliefs and assumptions. Sometimes research corroborates what you already believed, and it makes you more certain of it: that’s also valuable. But in my own experience I often expect to find something when I start a research project but am stunned by the findings and have had to change my thinking.

This keeps us honest and in the end we understand better what’s going on, thanks to the new evidence. That’s how we learn.


Gary Solon is the Eller Professor of Economics at the University of Arizona. Most of his research involves empirical analysis of labour market issues. concentrating on the behaviour of labour markets over recessions and expansions. Some of the other topics on which he has conducted influential research are the intergenerational transmission of socioeconomic status, earnings dynamics over the life cycle, and microeconometric methods.

Professor Solon spoke at the MTI-CSC Economics Speaker Series on “Social Mobility Across Generations: International and Singapore Perspectives” on 27 March 2014. He shared his views in conversation with ETHOS Editor-in-Chief Alvin Pang and CSC Senior Researcher Leon Lim.

Back to Ethos homepage