This paper studies the cyclicality of human capital accumulation using a lifecycle RBC model with two types of heterogeneity: age and productivity in learning. Households live for 60 periods. Within the same cohort, there are two types of agent: high and low type. High types accumulate more human capital than low types given the same amount of time spent in education. In particular, the productivity in learning is calibrated for both types using US data on hours worked, earnings and time spent in education. Empirically, low types are the individuals with at most a high school diploma, while high types are those enrolled in or graduated from any post-secondary education program.

When a negative technology shock hits the economy, agents spend more time in education and accumulate more human capital. This is due to lower opportunity costs of education and a lower rate of return to physical capital investments. Therefore, the education sector acts as a buffer sector and allows agents to compensate for the reduction in labor income by increasing the human capital stock. However, high-productivity agents and old individuals use the education sector significantly less. This is due to the fact that they have already accumulated a higher amount of human capital before the crisis. They are very efficient at work and it is more expensive for them to reduce hours worked in order to study more. Further, the rate of return to human capital investments is lower for them. In reality this can be explained by the fact that an economic crisis affects certain categories of workers more than others. Young and low-productivity individuals, in fact, face even harder labor market conditions. Young people do not have experience in the labor market yet. Low types are less efficient at work than high types because of the lower human capital stock. Therefore, when the crisis hits the economy, these categories benefit more from the education sector and the substitution between physical and human capital.

The model has also some implications regarding the volatility puzzle. Labor supply volatility is higher in the model with human capital compared to the baseline RBC model. Further, labor supply volatility in the model matches the empirical estimate quite well for high types but not for low types. This result suggests that the volatility puzzle may be due to heterogeneity among agents.

Keywords: human capital accumulation, business cycles, labor supply.

JEL classification: J24, E32, J22