Aggregate Evidence on the Link Between Age Structure and Productivity

This chapter examines the relationship between the age structure of the workforce and aggregate productivity. It is well known from the labor literature that there is a robust relationship between years of experience and income. If workers are paid their marginal product then this suggests a relationship between worker productivity and age. In the aggregate, therefore, we should expect changes in the age structure of a population to be correlated with changes in productivity. Feyrer (2007) finds a strong and robust correlation between workforce age structure and total factor productivity. The impact of age structure on productivity is much larger than is estimated by microeconomic evidence on the relationship between wages and experience. The magnitude of the result does not appear to be driven by reverse causality from productivity to workforce age structure through immigration or participation rates. This suggests that the social return to a workforce with a particular experience profile is higher than the private return to experience. This chapter examines the nature of this externality in several ways. To begin, I explore changes in the age distributions of the US workforce for several subcategories of worker. One mechanism through which the age distribution could produce large externalities is through innovative activity. I examine the evolution of the age distribution of patent holders in the United States over time. Another mechanism through which the age distribution of the workforce may be important is through changes in management. Lucas (1978) suggests that the quality distribution of managers may play a large role in determining output. Using census data, I explore how the entry of the baby boom cohorts into the workforce changed the composition of the managerial workforce over time. For this reason, it may be useful to identify the scope of the externality. Does it extend to smaller geographic units like US states and localities? This is an