The findings of the research discussed here were the focus of a recent FMCi event titled, "Networks of opportunity: Social capital and economic mobility," that took place August 3rd. Watch the recorded webinar here.
Improving economic mobility is a central goal for public policy. But how? Mobility is a complex, multidimensional issue, related to policies in education and training, housing markets, safety net efficacy, family stability, the labor market and much more. The latest Opportunity Insights report from Raj Chetty and colleagues suggests that one ingredient that may trump all the others: Friends.
Drawing on a massive dataset, comprising the social networks of 72.2 million users of Facebook aged between 25 and 44 years, Chetty and his team are able to assess how far social networks influence economic mobility. The richness of their data permits analysis at a very granular level, down not only to zip codes, but individual colleges and high schools. Their two papers Social Capital I: measurement and associations with economic mobility and Social Capital II: determinants of economic connectedness have just been published in Nature along with supplementary data here and here. As usual they have also created an interactive and public use version of the data.
The findings are striking and certain to have a profound impact on discussions of economic mobility. The headline finding is that at the community level, cross-class connections boost social mobility more than anything else, including racial segregation, economic inequality, educational outcomes, and family structure.
Creating more connections across class lines – either through greater economic integration of our institutions and neighborhoods or more opportunities for cross-class social engagement – looks to be the most promising route to improving rates of upward economic mobility in the U.S.
Here we summarize seven key findings from the research, as well as some implications for policy. Watch a discussion this Wednesday with Chetty and Stroebel, as well as Harvard’s Robert Putnam, AEI’s Scott Winship, and our colleague Camille Busette.
- Friendship networks are strongly class based, especially at the top
People are most likely to be friends with people of a similar socioeconomic status (SES), especially at the top of the ladder. As the figure below shows, one in three of the friends of those in the top SES decile are from the same decile, compared to just 24% from the whole bottom half of the income distribution:
As the team notes, there is an almost linear relationship between the SES of an individual and the average SES of their friends:
“A one percentile point increase in one’s own SES rank is associated with a 0.44 percentile point increase in the SES rank of one’s friends on average. The relationship is almost linear between the 10th and 90th percentiles of the SES distribution, with a slope of 0.41 in that range. The slope rises to 0.98 between the 90th and 100th percentiles, which shows that the highest-SES individuals tend to have particularly high-SES friends.”
2. Rich people make friends of college classmates, poorer people make friends of neighbors
There are also marked class differences in how and where people make friends. For high-SES people, college plays a much bigger role in the creation of friendship networks – mostly for the obvious reason that they are more likely to have gone to college in the first place:
For those from working class or middle-class backgrounds, neighborhood networks are much more important, followed by religious communities. As the team writes:
“Individuals with the lowest SES make about four times greater a share of their friends in their neighborhoods (residential ZIP codes) compared with individuals with the highest SES…Neighborhoods therefore play a larger role in defining the social communities of low-SES individuals, perhaps explaining why where one lives matters more for the economic and health outcomes of lower-income individuals than higher-income individuals.”
3. The Midwest really is friendlier than the northeast…
There are two factors that can influence the formation of friendships across class lines:
- lack of exposure to people of a different background (and particularly of lower-SES people to higher-SES people)
- “friending bias,” which means that even when there are people of different backgrounds around, friendships remain strongly class-based
The team assessed the degree of friending bias across the nation, and found some geographical patterns:
As they write:
“Friending bias is lower on average in areas with more high-SES exposure, with a correlation of about −0.2 across counties, but there are many exceptions to this pattern. For example, the northeast generally has high exposure but also high friending bias (that is, people with low and high SES in the northeast are relatively well integrated in schools and neighborhoods, but tend to befriend each other at lower rates).”
4. Economic connectedness is the only form of social capital that boosts mobility
Chetty and his colleagues examine three types of social capital:
- economic connectedness (EC), based on the extent of friendships across class lines
- social cohesion, based on measures of the thickness of social ties within communities
- civic engagement, for example as expressed by rates of volunteering
Strikingly, they find that only the first form of social capital – economic connectedness – is associated with higher rates of economic mobility:
The main measure of economic connectedness used by the researchers is the share of above-median SES friends that below-median SES people have, but they show similar results using quintiles too. As they write:
“On average, an increase in EC of 0.5 units (equivalent to raising the share of high-SES friends among low-SES people from 25% to 50%…) is associated with an 8.2 percentile increase in children’s incomes in adulthood. This is a large difference: for context, note that children with high-income (above-median) parents end up 17 percentiles higher in the household income distribution on average than children with low-income (below-median) parents the first form yields the greatest opportunities for mobility.”
It is indeed a large difference. The size of the dataset also allows the research team to use city-specific data to vividly show the connection between connectedness and mobility, for example comparing Minneapolis and Indianapolis:
“Low-SES individuals have a much larger share of high-SES friends in Minneapolis (49%, corresponding to an EC of 0.98) compared with Indianapolis (32%, EC of 0.65). Correspondingly, children who grow up in low-income families have much higher incomes in adulthood in Minneapolis than in Indianapolis. In Minneapolis, children reach the 43rd percentile of the household income distribution on average at age 35 years (US $34,300 in 2015), compared with the 34th percentile ($24,700) in Indianapolis.”
Place matters for mobility. But to a large extent this is because some places are more connected than others.
5. In fact, economic connectedness boosts mobility more than anything else
Drawing on earlier work, the Chetty team can compare the importance of social capital, in the form of economic connectedness, for economic mobility compared to other factors. The results here are perhaps the most remarkable of all. Drawing on a careful analysis teasing out the relevance of various factors, they find that economic connectedness is a stronger predictor of upward mobility than any other:
The team also finds that many variables that they have previously shown to be associated with upward mobility – such as income inequality and racial segregation – lose their predictive power once economic connectedness is taken into account. As they write:
“A lack of economic connectedness may be a key reason that upward mobility is lower in areas with larger Black populations and greater inequality.”
6. For upward mobility, it’s better to live in a more connected place than a richer place
What this means is that mobility rates are more influenced by the degree of economic connectedness in a neighborhood than by other factors, such as inequality, segregation or income. Zip codes with higher levels of economic connectedness have better rates of upward mobility, even if they are lower income:
Places with high levels of economic connectedness – i.e. the ones towards the top of the scatter plot – have higher rates of upward mobility (as indicated by the blue color), even when they have very different median household incomes. As the researchers write:
“These results imply that it is growing up in an area with high EC—rather than just around high-income people—that leads to better prospects for upward mobility”
7. Friending bias and economic segregation contribute equally to lack of connectedness
Which of the two barriers to economic connectedness – lack of exposure, or friending bias – matters the most? They seem to matter equally, according to some counterfactual scenarios:
“Both exposure and friending bias remain strongly predictive of counties’ causal effects on upward mobility, implying that moving to a place with greater exposure or lower friending bias at an earlier age increases the earnings in adulthood of children who grow up in low-income families.”
Scholars like Robert Putnam and our own colleague Camille Busette, Director of the Brookings Race, Prosperity, and Inclusion Initiative have argued for the importance of social capital – especially of the “bridging” kind expressed by economic connectedness – for economic opportunity. But this new research provides a wholly new empirical basis for this connection. This work suggests that the importance of relationships goes beyond obvious advantages such as job referrals: after all, the studies show that moving to a highly connected place even early in childhood makes a difference to long-term upward mobility.
Of course, there are limitations to the research. One issue is that like all measures of economic mobility, the data is somewhat backward-looking, especially in terms of evaluating institutions like high schools. Chetty and his colleagues provide some comfort on this point, however, showing very high correlations between Facebook connections among adults and Instagram connections among younger people. There is also a question of how far online connections reflect “real” friendships – but again there is good reason to believe that they reflect meaningful relationships, not least given the results of the studies.
What are the implications here for policy? The first is a broad one: social capital seems to matter a great deal, much more perhaps than was previously thought. Investing in measures to improve connectedness may have a much bigger mobility payoff than other policies, which may be more expensive and/or disruptive.
More specifically, the data allows policymakers at a local and even hyper-local level – right down to high school principals – to gain an insight into the main barriers to pro-mobility relationship building in their area or institution.
In some cases, the challenge will be one of integration: it is hard to build relationships with higher-SES people when there aren’t many around. Here the goal should be to desegregate schools, colleges and neighborhoods through housing reform, more progressive admissions policies and so on.
But in other cases, there may be a good mix of people but limited relationships across lines: here the challenge is one of interaction. The team provide examples of high schools redesigning commons spaces or classroom assignments to create more opportunities for mixing. Similar efforts could be important at a neighborhood level, for example through careful design of public facilities such as parks, libraries or transportation.
Chetty and his colleagues remind us that while interventions to support individuals, for example through educational investments or job programs are important, the web of relationships that people build along the way can be equally impactful. It is surely true that people can get by with a little help from their friends. But it also looks like their friends can help them to get ahead.
 The other principal investigators are Matt Jackson, Theresa Kuchler, and Johannes Stroebel.