Intergenerational mobility has attracted growing attention in sociology and economics as an important component of inclusive growth, sustainable poverty reduction, and greater equality. While previous research has carefully examined temporal and geographical dimensions of intergenerational mobility, the network dimension, in particular the structure of peer interactions, has untapped potential for theoretical and empirical advancement. The available research on the consequences of network dynamics for social stratification has concentrated on homophily as the key principle structuring friendship networks, showing that SES-based homophily reinforces social stratification (DiMaggio and Garip 2012; Manzo 2013; McPherson, Smith-Lovin, and Cook 2001; Tsvetkova, Wagner, and Mao 2018). This line of work demonstrates that homophilous selection and sorting, in combination with spillover effects in networks, concentrate resources, and create accumulated advantage on one side and deprivation on the other.
However, previous work has overlooked a paradoxical possibility: homophilous clusters stratified by SES create the possibility of bridge ties spanning socioeconomic boundaries, which carry the potential for resource flows across social strata and consequently attenuate the stratifying effects of homophily. While friendships that bridge between socio-economic boundaries also occur in networks without SES homophily, there is a decisive ecological difference between cross-SES ties in homophilous networks and networks without SES homophily. In a counterfactual world that is stratified but without homophily, cross-SES ties connect unequal nodes. In a world with stratified homophily, cross-SES ties connect unequal clusters (Burt 1992). The paradox of SES-homophily is that as clusters become more unequal, ties between clusters become more consequential. Specifically, we hypothesize that homophily-stratified network clusters create ecological conditions that alter the mobility consequences of cross-SES ties and thus change the way cross-SES friendships are experienced. A deeper understanding of those conditions is the primary goal of the proposed research. To test the hypothesized impact on stratification, we propose to examine the contexts in which cross-SES friendships emerge, measure the extent to which these friendships promote resource spillover across social strata, and whether this resource flow has long-term consequences on social mobility.
Despite its intuitive appeal, this hypothesis has been difficult to test at the micro level due to three data limitations: First, nearly all individual-level data on social inequality is obtained from surveys, which either do not include information on peers or rely on respondents’ estimates of peer features rather than direct measures. Second, cross-SES friendships are comparatively rare (about 25 percent of friendships in the Add Health surveys), which limits the ability to measure ecological effects, i.e. how the consequences for mobility depend on the structural context in which these friendships emerge. Contradictory results in the literature on peer influence are indicative of such contextual dependencies (Patacchini, Rainone, and Zenou 2017). Third, the data is rarely longitudinal, making it difficult to tease apart the causal direction. Do ties to higher SES peers promote social mobility or do mobility aspirations motivate tie formation to higher SES peers? Most importantly, available longitudinal data has not been collected long enough to examine peer effects on long-term career trajectory. Thus, we do not know whether cross-SES friendship in high-school has a lasting impact on social mobility.
With the release of the new Add Health wave (Harris and Udry 2018), a large dataset including full network information running long enough to examine the consequences of network dynamics for outcomes later in life became available. Waves 1-3 followed students in grades 7 to 12, collecting information on academic achievement, socioeconomic background, and friendship networks. Wave 5, released in 2020, contains socioeconomic outcomes of the students at ages 31-42.