How much do demographic changes affect the overall LFP rate trend?
To quantify how much of the decline in the aggregate LFP rate is driven by changes in the composition of the population, we construct two hypothetical aggregate LFP rate scenarios. In this approach, we use Current Population Survey (CPS) microdata from the Bureau for Labor Statistics for 1976–2018 and split the population into 56 age-gender-education groups based on seven age, two gender, and four educational categories. Figure 2 shows the actual (blue line) and our two scenarios for aggregate LFP rates.
Our first scenario estimates how much the compositional shift towards an older population has changed the aggregate LFP rate. To construct this alternative rate, we fix the participation rate of each group at their levels from 2000, when the aggregate LFP reached its peak, and also fix the educational composition of the population at its 2000 level. We then use the actual age-gender population shares as weights. The red line in Figure 2 shows how much the aggregate LFP rate would have changed if the groups’ LFP rates and educational composition had remained at their 2000 levels but the age-gender composition of the population had evolved as in the data. Driven only by the changes in the age-gender composition, the resulting aggregate rate would have declined about three-fourths as much between 2000 and 2018 as it actually did. In other words, changes in the age-gender composition of the population contributed about three-fourths to the decline in the LFP rate.
In the second hypothetical scenario, we further examine how much the increased educational attainment of the population contributed to the increase in the aggregate LFP rate. For example, Daly, Jackson, and Valletta (2007) show how shifts towards a more educated population have affected wages and unemployment. For this purpose we only fix the groups’ LFP rates at their 2000 levels, but we allow the educational distribution of each age-gender group to evolve as in the data. The green line in Figure 2 shows that this alternative scenario for the LFP rate would have declined half as much as the actual rate and less than in the first hypothetical scenario. This indicates that increased educational attainment in the U.S. population between 2000 and 2018 undid a substantial part of the decline due to the aging of the U.S. population.
These hypothetical scenarios show that between 2000 and 2018 changes in the population composition substantially lowered the aggregate participation rate and its trend. Figure 2 also shows that the actual aggregate LFP rate in 2018 is lower than our composition-driven scenarios. This means that different demographic groups’ participation rates also changed and played a role in lowering the aggregate rate. We now turn to estimating the trends in the LFP rates of different demographic groups.
Our approach for estimating trends in demographic group participation
Estimating the trends for group-specific LFP rates requires us to first disentangle changes in the group rate into the slow-moving factors that are not affected by cyclical developments in the labor market and other changes that are driven by cyclical fluctuations. The former defines the trend in the group participation rate.
Researchers have developed age-cohort models of the demographic group trends (Aaronson et al. 2006, Aaronson et al. 2014, CBO 2018, and Montes 2018). In these models, the trend for the age-specific LFP rate of a demographic group defined by gender and education is the sum of an age effect (how old you are) and a cohort effect (when you were born). Typically, the time variation of the age-specific LFP rate is attributed to cohort effects and the age effect is considered fixed, but age-specific LFP rates vary quite a bit more than can be accounted for by cohort effects. For example, older workers participate at higher rates now than they did two decades ago because they live longer and healthier lives; young workers ages 16 to 24 participate at a much lower rate than in the 1990s in part due to higher school attendance rates. Augmenting a model with structural variables such as school enrollment and social security payouts helps capture the evolving age effects. However, such an approach requires a large number of these structural variables to estimate the historical trend, as well as numerous forecasts of these variables for forecasting the trend.
Our approach allows for time variation in the age effects of group-specific LFP rates. We simply assume that the age and cohort effects for each gender-education group vary over time. We then estimate the age and cohort effects using standard data filtering techniques. Our approach requires no structural variables in estimation or forecasting.
The trend in the aggregate LFP rate and its projections
We construct the aggregate LFP rate trend and its projection using our estimates of the trends of different demographic groups and the actual population shares. In the estimation, we use the CBO’s projections of the age composition of the population and fix the educational composition at the latest data point. Figure 3 shows the actual LFP rate and our estimate of its trend.
We find that the rate is at its trend of 62.8% in the first half of 2018. This is roughly consistent with the CBO’s estimate of the trend. We estimate that the aggregate trend will decline by approximately 2.5% in the next decade. Given that our method of estimating the group trends is different from the CBO’s, the close similarity of the projections of the trend for the aggregate participation rate implies that almost the entire projected future decline is driven by population aging.
In this Letter, we use a simple statistical model to construct the trend in the aggregate labor force participation rate and its projections over the next 10 years.
Our estimates indicate that the aggregate labor force participation rate is at its trend as of 2018. Combined with the low unemployment rate, this argues that the U.S. labor market is operating at or beyond its full potential. This finding is consistent with other recent work that finds that the decline in the LFP rate reflects longer-term developments (Valletta and Barlow, 2018, Daly et al. 2018).