Returns to Education: New Challenges for Models of Human Capital Investment – RECHC

To analyze schooling and career choices, economists use a human capital framework, in which workers are viewed as economic agents who can accumulate skills (human capital) either by attending school or by working. Labor market outcomes (in particular wages) are determined by the amount of skills accumulated by the agent in the past. Therefore, by deciding to achieve a higher educational level, the agent takes into consideration that the expected additional earnings over the life cycle will compensate for the cost of acquiring education. Following the seminal paper of Keane and Wolpin (1997), empirical economists have estimated dynamic structural models of human capital accumulation. This approach consists in modeling explicitly individual decisions about attending school and working (considering, possibly, working in different occupations). At each age, agents take the decision that maximizes their life-time satisfaction, considering that their present decision affects subsequent decisions and outcomes. The parameters of the behavioral model are estimated using observational data on individual schooling attainments, labor market transitions and wages in a revealed preference approach.

This project consists in developing human capital models that relax problematic assumptions, to provide robust results for policy questions that remain important (such as the impact of a change in tuition fees) in a more diverse context, and to answer new questions that arise.

In a first project, we will relax two assumptions: (1) agents’ unobserved heterogeneity type is fixed over time, and (2) sample attrition is random. We illustrate the importance of allowing agents’ type to change in the French context.

In a second project we relax the assumption that agents have perfect information about their own skills. We study this in the South Korean context, which has the highest rate of college enrollment in the OECD, while also facing high unemployment rates among the educated.

In a third project we allow for a large diversity of both educational options and job opportunities. We study this in the German context, characterized by a large set of post-secondary educational programs, including vocational-oriented programs tailored to specific occupations.

Preliminary at this stage.

Preliminary at this stage.

In most advanced societies, improving skill supply to the labor market is on the policy agenda. This is justified both by the fact that skill accumulation fosters innovation, and by the fact that the adoption of new technologies modifies the set of skills needed in the labor market. One powerful way to affect skill supply is through education, as education is an essential input in the production of labor market productive skills. Understanding how individuals make their schooling decisions and how such decisions affect labor market outcomes (earnings, job mobility) is essential not only for the economist, but also for the policy maker who has to decide which policy gives the right incentives to invest in education.

The objective of this project is to improve the state-of-the art modeling strategies of educational and labor market decisions and adapt them to features that are particularly important in today’s context, in which different educational programs prepare for a diverse set of occupations and in which desired policies aim to improve the match between individual skills and interests. To this end, we will estimate dynamic structural human capital investment models in three different institutional contexts. This will allow us to learn about individual behavior and simulate policies that aim to improve the transitions within education, from education to the labor market, and between jobs.

A usual assumption made in standard human capital accumulation models is that individual unobserved characteristics (ability or preferences) are stable over time. This assumption does not capture well the possibility that preferences might evolve as the individual progresses into the educational system by acquiring more or less specialized skills. In a first project, we relax this assumption in the French context, where individuals specialize at different stages of their educational progression. Our model also introduces a way to deal with endogenous sample attrition, a feature widely encountered in longitudinal survey data and that can cause sample selection biases.

Another usual assumption is that individuals have perfect information about their own skills when they make their schooling decisions. In a second project, we relax this assumption and allow individuals to be overconfident about their (labor market) prospects in the South Korean context, characterized by a high college enrollment rate and a high unemployment rate among the educated.

In a third project, we extend the standard model by allowing for a large diversity of both educational options and job opportunities to study how the type of post-secondary education affects occupational choices and wages. This project is conducted in the German context, characterized by a large variety of post-secondary educational programs, including vocational-oriented programs tailored to specific occupations.