Working Paper Series no. 679: Countercyclical School Attainment and Intergenerational Mobility

Educational choices are endogenous to aggregate economic conditions. Whenever it is more difficult to find a job, the earnings foregone while at school decrease. At the same time, without those earnings, or with reduced parental earnings, it might be more difficult to finance post-compulsory education. Existing evidence shows that, on average, education tend to be counter-cyclical, meaning that changes in opportunity costs dominate ability to pay considerations, with cohorts exposed to adverse economic conditions in critical ages obtaining significantly more schooling. However, it is not obvious how the fluctuations in schooling across cohorts induced by the business cycle will be related to parental occupation. Credit constraints are larger for low-skill families, but the change in the optimal level of schooling induced by changes in opportunity costs might be larger for children of these families. In addition, the number of students at the margin where economic conditions change the optimal level of schooling might also differ by parental occupation.

The empirical question we address in this paper is whether the link between parental and educational attainment significantly differs across cohorts exposed to different conditions when deciding to pursue post-compulsory schooling; and if it does, whether this translates into long-lasting differences in the relationship between parental occupation and labor market performance.

This is an important question for a better understanding of the determinants of intergenerational mobility (IGM) in income within a society, since skill acquisition is one of the main channels for the transmission of economic advantage across generations (Solon, 1999). Moreover, it provides new insights on the long-run effects of exposure to business cycle fluctuations at critical ages.

We address this question using a large sample of labor force survey data on 23 cohorts across 96 French provinces (départements), exploiting variation in the unemployment rate in the individuals’ province of birth at age 16, and information on labor market outcomes, educational attainment, and parental occupation. Crucial in our empirical exercise is the use of local-level variation in economic conditions, allowing us to net out time-specific unobserved heterogeneity, including national trends and policies; the use of the unemployment rate by province of birth, allowing us to rule out geographical sorting due to economic conditions at the moment of choosing education; and the use of the unemployment rate at the moment of finishing post-compulsory education, allowing us to rule out simultaneity problems between the unemployment rate and the cohort’s education decisions. Our measures of schooling attainment are indicators for holding post-compulsory and college degrees. Our measure of labor market outcomes, both for the children and the parents, is binary – white vs. blue collar occupation –, and is strongly correlated with contemporaneous earnings and educational attainment, which we do not observe for parents.

We find that cohorts deciding on post-compulsory education in bad economic times are more educationally intergenerationally mobile, meaning that their educational attainment is less correlated with having a white-collar father. We find an analogous effect in the labor market: these cohorts are also more occupationally intergenerationally mobile, meaning that their probability of obtaining a white-collar job is less correlated with having a white-collar father.

We frame our analysis and interpret the findings using a one-factor model of selection into education where education affects the probability of access to white versus blue-collar occupations and within-occupation earnings. The model allows us the decompose the business cycle implications of occupational IGM into a component that is driven by the endogenous responses of educational attainment and a component that reflects changes in occupational IGM due to changes in the occupational returns of education. We find that the former component appear much more potent.

Our results suggest that although recessions might tend to increase inequality (Bonhomme and Hospido, 2017), due for instance to the skill-biased nature of unemployment, the lack of labor market alternatives pushes especially the children of the low skilled to obtain more education, resulting in higher IGM in education and labor market outcomes within the treated cohorts.

Since we exploit variation based on the unemployment rate at the province of birth, one possible concern is about differential spatial mobility between birth and age 16 (and hence exposure to economic conditions) by parental occupation. We account for this possibility by using a Two-Sample 2SLS estimator, finding that it is unlikely to explain our findings, and if anything, it reinforces the results.

Finally, we provide evidence that gap between white and blue-collar children in terms of enrollment is negatively affected by the unemployment rate faced at age 16, but that this effect changes in sign at age 23 – i.e. the enrollment decisions of children of white collar workers are relatively more countercyclical with respect to unemployment faced at age 23. These results, displayed in Figure 1, are quite intuitive as they suggest that individuals with blue-collar parents adjust relatively more through enrollment in post-compulsory secondary education while children of white-collars use higher education as a buffer against local economic conditions relatively more.