I might not retire yet… if I am healthy

A retired couple sits on a dock next to a lake enjoying the sunset

Since the mid 1990s individuals in their 60s stay in the labor force in growing numbers in most advanced economies. Understanding this phenomenon is of great relevance for the policy discussion and potential reforms to the Social Security system in ageing economies. This is the topic studied in the Barcelona School of Economics Working Paper 1298, “Widening Health Gap in the U.S. Labor Force Participation at Older Ages,” by Tomaz Cajner, Javier Fernández-Blanco, and Virginia Sánchez-Marcos.

New cohorts are working more when old

While prime-age men have been participating in the labor force at lower rates, the participation rate of men aged 62-66 has increased by about 15 percentage points since the early 1990s. Further, men and women behave quite alike in this regard despite the large trend difference prior to 1990, as shown in Figure 1.  

Figure 1. Evolution of the Labor Force Participation Rate by Age Group and Gender.

Usual suspects

To shed light on what might be driving this phenomenon, the authors use data from the Current Population Survey between 1996 and 2019 and the Health and Retirement Study since 1992. Their analysis compares workers born in the early 1930s with those of later cohorts near the retirement age, namely those between 62 to 66 years old. 

They look into trend differences in participation rates by educational, occupational and health categories as well as the participation of the spouse. 

Where do the differences come from?

In more recent cohorts, they find that workers between 62 and 66 years old tend to have a higher degree of education, better health, and a spouse participating in the labor force. Specifically, compositional changes related to education, a spouse’s employment status, and health explain nearly a third of the increase in the labor participation rate of workers in this age group. These results, detailed in Table 1 below, hold for workers of both genders when comparing the 1934 and 1953 cohorts. 

Table 1. Compositional Changes between Cohorts

Nevertheless, eight percentage points of the difference in participation with respect to the base cohort remain unexplained. It might be that occupational changes have something to add to this, mainly if workers have shifted towards non-routine occupations in which the retirement age is expected to be higher. The authors explore this possibility finding that there is not much of a role for these mechanisms to explain the changes between cohorts. 

Healthy individuals driving the increase

The authors find that although there are no meaningful time-varying patterns in participation rates if we compare individuals’ across educational categories, marital status, and spouses’ employment status, a significant health gap in participation has been widening across time. More specifically, they find that the participation rate of those in good health from the 1948-1953 cohort has increased by 7.3 and 10.6 percentage points for men and women, respectively, when compared to the 1934-1936 base cohort. In striking contrast, they find that participation rates are lower for workers in bad health. Disability appears to play no role either.

What’s next?

How much do the recent Social Security reforms as well as the documented changes in life expectancy, medical expenses, lifetime earnings and labor productivity account for the heterogeneous patterns in labor force participation? This data work is part of a bigger quantitative project tackling such questions.