Peer and Family Effects in Work and Program Participation Peer and Family Effects in Work and Program Participation Gordon B. Dahl Summary People don’t base decisions about their economic life solely on their own individually formed ideas and preferences. Rather, they’re influenced by the experiences of their peers and by social group norms. Gordon Dahl reviews the various ways family and neighborhood peer groups influence decisions to participate in the workforce and in government social assistance programs. These social spillover effects are hard to estimate because of the problems that economists refer to as reflection, correlated unobservables, and endogenous group membership. Dahl explains how researchers have overcome these challenges to produce credible estimates of the effects of family and peer groups on work and program participation. He reviews the most rigorous evidence to date and discusses possible mechanisms. Understanding neighborhood and family group influences is critical to thinking about policy, Dahl writes. The spillover effects on children, siblings, and neighbors can be just as important as the direct impact on parents and directly targeted peers, due to social multiplier effects. www.futureofchildren.org Gordon B. Dahl is a professor of economics at the University of California, San Diego. He is also an affiliated professor at the Norwegian School of Economics, the area director for labor economics for the CESifo Research Network, a research associate of the National Bureau of Economic Research, a research professor at the ifo Institue, a CESifo research fellow, a research fellow of the Institute for the Study of Labor (IZA), and a fellow of the Stanford Center for the Study of Poverty and Inequality. Elizabeth Ananat of Barnard College reviewed and critiqued a draft of this article. VOL. 30 / NO. 1 / SPRING 2020 107 Gordon B. Dahl t’s a longstanding question in social that could alter the child’s perceptions science: do families and other about the relative costs, benefits, and peers transmit cultures of work stigma associated with the two alternatives. and program participation? In this Information transmission or differential article, I review the evidence for investment could also occur as a result Itwo settings where these types of peer of having a parent receive government effects could be especially important: social transfers. On the other hand, characteristics assistance programs and a selected set of like poor health or reduced opportunities labor market outcomes. My focus is on could be correlated across generations, family and neighborhood peer effects. The creating mechanical intergenerational links effects of other cultural factors, such as that don’t reflect a behavioral response on ancestry and language, have mostly been the child’s part. studied using an epidemiological approach, and have been reviewed elsewhere.1 Of course, the United States has many Likewise, research on peer effects for other different social programs. Traditional groups, such as college roommates, and for welfare programs include Temporary other outcomes, such as crime, is beyond Assistance to Needy Families and the earlier the scope of this review.2 Aid to Families with Dependent Children. Other means-tested programs include the First, I’ll address the subject of Supplemental Nutrition Assistance Program intergenerational links in welfare use. (that is, food stamps) and Women, Infants, Academics and policymakers alike have and Children. Social assistance programs heatedly debated whether such links reflect also include the Earned Income Tax Credit, a culture of welfare. A Nobel Prize winner Unemployment Insurance, and Social in Economics, Gary Becker, expressed the Security Disability Insurance. In this article, belief that “mothers on welfare convey the I discuss peer effects for a varied but limited impression to their children that it is normal set of social assistance programs, based on to live off government handouts. In such an the availability of research. environment, it is difficult for children to place a high value on doing well at school Turning to family effects related to the and preparing for work by seeking out labor force, we find the rhetorical debate training on jobs and in schools.”3 However, less intense, at least when unemployment the fact that children with parents on is decoupled from welfare participation. welfare are more likely to be on welfare But people make similar arguments about themselves as adults doesn’t mean that the whether family members and other peer parents’ participation is what caused the groups influence how much individuals children to also participate. As the saying work and earn. For example, attitudes about goes, “correlation doesn’t imply causation.” traditional gender roles and the desire to fit into one’s group might affect a mother’s Still, the question has proven difficult to decision to work, especially after the birth of resolve. Parents’ participation in a welfare a child. But mothers in the same family or program isn’t randomly assigned. On the workplace are also likely to share common one hand, when a child has a parent who characteristics, such as similar levels of isn’t working and is on public assistance, income, which affect work decisions. 108 THE FUTURE OF CHILDREN Peer and Family Effects in Work and Program Participation It has proven difficult to estimate causality these caveats in mind, the best evidence to for these types of peer effects in work and date supports the idea that family members social assistance programs, given the well- and neighborhood peers play an important known problems of what economists refer role in decisions about work and program to as reflection, correlated unobservables, participation. and endogenous group membership (I define these terms below in the section “Challenges We know less about the mechanisms behind in Estimating Peer Effects”). It can also these peer effects. Several channels have be difficult to define the appropriate peer been postulated, most of which can be group and to access data that link members classified into four categories. The first of a peer group. But this is changing, both can be broadly defined as cultural factors, in the United States and even more in other including the transmission of preferences countries, where high-quality administrative regarding stigma related to program data collected by governments is increasingly participation, or the desire to conform available. to a group’s social norms. The second is information transmission, such as how to In this article I review recent advances in apply for a welfare program or how an the estimation of causal peer effects in the employer will react if a mother takes parental family and neighborhood contexts. A key leave. The third is direct interactions with takeaway is that the statistical methods used other similar individuals; for example, the to study peer effects aren’t equally credible. benefit of staying home could be greater if Recognizing this, I organize my discussion your friends aren’t working and also have by the statistical method used, rather than free time. The final category is changes to by type of question or peer group. Though the home environment, such as in family early studies documented clear correlations income or parental stress levels. Economists in both program participation and labor and sociologists have found some suggestive market outcomes, causality was tenuous. evidence consistent with channels in each of Recent research has identified causal these categories. effects using more convincing methods and better data. Taken together, these more empirically rigorous studies generally Preferences for work and indicate the presence of intergenerational program participation aren’t links and a strong influence of families and fixed at birth or formed in neighborhoods. isolation. The emerging evidence is compelling, but we should be cautious about how we interpret the findings. Just because spillovers—where Recent findings on peer effects, regardless of one peer influences another—may occur in the underlying mechanisms, have important certain settings and for certain populations policy implications. What children learn doesn’t mean they occur in other settings from their parents about employment versus and populations. Moreover, the existence governmental assistance could matter for of peer effects doesn’t mean that other the financial stability of a number of social contextual factors aren’t also important. With insurance and safety net programs. Similarly, VOL. 30 / NO. 1 / SPRING 2020 109 Gordon B. Dahl peers who serve as important information could matter is social custom or group transmission networks, or are influential in identity. People may be sanctioned for changing social norms, can amplify the effects behaving differently, or may simply increase of policy reforms that affect work and social their happiness by behaving like their peers. assistance programs. This is particularly true Information is another channel that scholars when information is scarce and perceptions discuss. These channels include learning are still being formed. Some of the evidence from family members and neighbors about indicates that these social interactions lead to how to sign up for a welfare program, what long-run effects that are substantially larger the requirements are, and what it’s like to than otherwise expected. be on the program. Similarly, peers could Possible Mechanisms provide insights about writing a résumé, job interviews, and proper work etiquette. When Most economists and sociologists would information is scarce, people can also learn agree that preferences for work and program from family members and peers about the participation aren’t fixed at birth or formed costs and benefits of work.5 Moreover, peers in isolation. The experiences of a person’s can serve as a network for job referrals.6 families and neighbors are key inputs into preference formation. Moreover, families and Peers could also matter if the benefits of neighbors could provide valuable information work or program participation directly related to both work and program depend on interactions with other similar participation.