The Impact of the Age Distribution on Unemployment: Evidence from US States 
Economists have long understood that shifts in the age distribution can affect the aggregate unemployment rate. In particular, changes in the share of young people in the population should raise the unemployment rate, because young people tend to have higher unemployment rates than older people. For at least the last five decades, analysts have typically used a “shift-share” method to infer the age distribution’s influence on the unemployment rate. This method involves holding age-specific unemployment rates constant at a base level while labor force shares or population shares change over time (or vice versa). Some authors have pointed out that a potential problem with the shift-share method is its assumption that changes in the age distribution do not affect age-specific unemployment rates—that one age group's population share has no indirect effects on another group’s unemployment rate. This paper uses state-level data to revisit the age distribution’s influence on unemployment in the United States. It examines demographic effects across the entire age distribution instead of focusing on just the youth share of the population, as most previous studies do. It also uses nearly two decades of additional data on state-level age-group shares and unemployment that were not available for earlier research.
Key Findings
- The age distribution affects the unemployment rate in the same direction that a shift-share analysis would predict. For example, a large cohort of young people raises unemployment.
- However, there are also indirect effects of the age distribution that amplify the direct effects. Taken together, the direct and indirect effects generate a total demographic effect on unemployment that is larger than a shift-share model would predict.
Implications
As the US youth share has declined, estimates of the natural rate of unemployment have been revised downward. The paper’s results suggest that the downward revisions may reflect the impact of indirect effects that were not included in previous estimates. Similarly, when the youth share was rising during the 1970s, indirect effects may have raised the natural rate by more than the increase predicted by shift-share models. All told, demographics may have played a larger role in the historical evolution of US unemployment than has been previously understood.
Abstract
Economists have studied the potential effects of shifts in the age distribution on the unemployment rate for more than 50 years. Most of this analysis uses a “shift-share” method, which assumes that the demographic structure has no indirect effects on age-specific unemployment rates. This paper uses state-level data to revisit the influence of the age distribution on unemployment in the United States. We examine demographic effects across the entire age distribution rather than just the youth share of the population—the focus of most previous work—and extend the date range of analysis beyond that which was available for previous research. We find that shifts in the age distribution move the unemployment rate in the direction that a mechanical shift-share model would predict. But these effects are larger than the mechanical model would generate, indicating the presence of amplifying indirect effects of the age distribution on unemployment.