skip to Main Content

Volume 17, Issue 3 - September 2017

Researchers examine longer work lives

This year, two major themes emerged from MRRC researcher presentations: extending work lives and the family dynamics of wealth (see accompanying story, Page 4). Four of the MRRC presentations touched in some way on increasing older adults’ labor — a development that would improve the Old-Age and Survivors Insurance (OASI) Trust Fund balance. Health, wages, working conditions, and work-life balance featured in these discussions.


Pamela Giustinelli (Bocconi University, Italy) presented her joint work with Matthew D. Shapiro (University of Michigan), “Using Subjective Conditional Expectations to Estimate the Effect of Health on Retirement.”

“In this project we quantify the causal effect of health on work, and we provide a novel strategy to simulate the effect of hypothetical changes to the health distribution of older workers on the population’s labor supply forecast at specified horizons,” Giustinelli explained.

The researchers drew their data from the Vanguard Research Initiative (VRI), which surveys people older than 55 who hold between $10,000 and $5 million in assets with Vanguard. Giustinelli and Shapiro focused on respondents who were working and in good health at the time of the survey. The group was asked about the likelihood of their continuing to work in two and four years under high- and low-health scenarios. Respondents also were asked the likelihood of entering high and low health status in two and four years.

Giustinelli presented findings for the four-year horizon:

· 76 percent of respondents anticipated they would remain in high health, while 24 percent would enter low health;

· 53 percent predicted they would remain at work, 47 percent would not be working;

· 59 percent would stay at work if they were to remain in good health;

· the probability of working in four years would decrecrease to 33 percent if people moved from high to low health.

The authors then used these estimates to examine how hypothetical medical advances that improved older workers’ health might change their labor supply. They found that only 12 percent of workers would move to low health and 56 percent would be working.

While the sample is limited in scope, the authors are using this novel method to construct a structural model of labor supply behavior under various health scenarios. In addition, the soon-to-be-available 2017 VRI wave will allow the researchers to look at actual two-year results. Additionally, Giustinelli and Shapiro added their questions to an experimental module of the Health and Retirement Study (HRS) in 2016, and will apply the methods used here to that data when it is available.


The research literature sees a 25-percent decrease in wages between ages 55 and 65 — right before retirement. This decrease adds gloss to the decision to stop working at full-retirement age. In “Understanding Earnings, Labor Supply, and Retirement Decisions,” Ananth Seshadri (University of Wisconsin-Madison) and Xiaodong Fan (Monash University) built a complex, human capital life-cycle model and used Survey of Income and Program Participation (SIPP) data to examine why the wage decline occurs and how workers behavior might change if various Social Security reforms were adopted.

“In structural models that I’ve written before, there are times where I have no idea what’s going on — this is not one of them,” Seshadri said. Here, “human capital rises, it reaches a plateau, then after the age of 60, it drops… You end up retiring only when your human capital is low enough.”

The researchers tried several policy levers, including removing the Social Security earnings test, decreasing the Social Security benefit by 20 percent, and raising the full-retirement age by two years.

Sheshadri presented a sample of findings for high school graduates. Reducing Social Security benefits by 20 percent resulted in higher labor supply in later life and a 5-percent increase in wages between 65 and 80. This is significantly different from previous literature, which simply re-estimated retirement behavior under the new policy using the same age-wage profile as in the baseline model. Here, the model captures how workers adjust their human capital investment and labor supply over the life-cycle. Delaying FRA by two years and removing the Social Security earnings test had much smaller effects.

Full results are coming in the working paper due in September.

Working conditions

Do the qualitative job characteristics (flexibility, meaningful work, pay, etc.) that appeal to older workers have quantitative effects on labor force participation? In “The Value of Working Conditions in the United States,” RAND researchers Kathleen Mullen, David Powell, and Jeffrey Wenger, along with Till von Wachter (University of California-Los Angeles) and Nicole Maestas (Harvard University) attempt to answer that question using data gathered through a stated-preferences module in the RAND American Life Panel survey. Through their choices, respondents put a dollar value on 10 job characteristics: full-time status, schedule flexibility, telecommuting opportunities, physical demands, work pace, independence, paid time off, working with others, job training opportunities, and societal impact.

The researchers found that job characteristics play an important role in compensation differences. Mullen presented some of the results:

· The observed 20 percent wage gap between men and women shrinks to 9 percent when job characteristics are included.

· Without job characteristics included, whites make 9 percent more than blacks. With job characteristics, whites make 22 percent more.

· The college educated earn 49 percent more than those without college, this increases to 70 percent when job characteristics are included.

· Workers 50 and older make 5 percent more than those 35 to 49 years old. With job characteristics, this increases to 12 percent.

· Workers value the 10 working conditions collectively at 64 percent of their actual wages.

“Our results suggest that understanding preferences over working conditions is going to be key… to understanding how older workers make decisions about work and retirement,” said Mullen.

Work-life balance

Work that interferes with private-life priorities also can encourage retirement. In their project, “Work-Life Balance and Labor Force Attachment at Older Ages,” researchers Marco Angrisani (University of Southern California), Maria Casanova (California State University-Fullerton), and Erik Meijer (University of Southern California) examine how work-life balance (WLB), work-to-life interference (WLI), and life-to-work interference (LWI) influence retirement decisions.

“In order to design and implement policies that can effectively keep people working longer,” said Angrisani, “we need to study and understand the complicated dynamics of labor supply decisions at older ages.”

Angrisani detailed how they used a sample of workers ages 50 to 79 from HRS data to estimate the effect of WLB on older workers’ employment decisions and to quantify the extent to which WLB affects labor responses to spousal health shocks. Because of gender differences in work preferences, Angrisani and team modeled women and men separately.

In the data, the researchers observed that a worker’s response to a spouse’s health shock varied with the level of perceived WLB. They found:

· employment transitions are significantly associated with WLB—mainly driven by WLI;

· there are differences in the response to WLB by gender and employment status;

· relative to the sample average, WLI increases retirement probability by 27 percent for male part-time workers; 16 percent for female full-time workers; and 26 percent for female part-time workers.

Angrisani pointed out that they couldn’t rule out that other factors might be driving the findings, but that they were suggestive of a positive, causal relationship between WLB and a prolonged tenure in the labor force. In future research, the authors plan to use state-level policy changes affecting work flexibility to further explore this relationship.

Back To Top