Volume 16, Issue 3 - October 2016
Annual RRC meeting highlights new findings
The Affordable Care Act (ACA) has yet to show an impact on labor supply through mass retirements, adjusting payroll taxes might encourage longer work lives, and reverse mortgages (HECM) appear to be helping seniors recover from financial shocks. These were some of the findings presented at the 18th Annual Retirement Research Consortium Meeting at the National Press Club in Washington, D.C. Funded by the Social Security Administration, the two-day, August event included presentations by 21 investigators and 21 discussants, as well as a welcome from SSA Office of Research, Evaluation, and Statistics Associate Commissioner John Phillips and lunchtime talks by SSA’s Acting Commissioner Carolyn W. Colvin and Axel Börsch-Supan from Germany’s Max Planck Institute for Social Law and Social Policy. More than 400 people attended in person.
Michigan Retirement Research Center had six projects on this year’s schedule: “The Affordable Care Act as Retiree Health Insurance: Implications for Retirement and Social Security Claiming” (Alan Gustman presenting), “The Dynamic Effects of Health on the Employment of Older Workers” (Eric French), “How Home Equity Extraction and Reverse Mortgages Affect the Financial Well-being of Senior Households” (Stephanie Moulton), “Adjusting the Payroll Tax to Promote Longer Careers” (John Laitner), “Cohort Changes in Social Security Bene?ts and Pension Wealth” (David Weir), and “Working Conditions and Sustainable Work at Older Ages: An International Perspective” (Jeffrey Wenger).
The Affordable Care Act as Retiree Health Insurance: Implications for Retirement and Social Security Claiming
Labor economists have been concerned that the ACA would reduce employment, especially among those nearing retirement, as workers would no longer need to rely on employer-provided health insurance. If the ACA does make retirement more attractive, it could undermine policies meant to encourage longer work. Gustman presented research, conducted with Thomas Steinmeier and Nahid Tabatabai, that bridges conflicting findings on the ACA’s effects found in recent analyses, such as Levy, Buchmueller, and Nikpay, 2015, and retiree health literature.
Gustman and team (GST) focused on three major groups of employed individuals: those with employer-provided health insurance, but without it once retired (about 40 percent of Ns); those with employer-provided health insurance while working and while retired (18 to 20 percent); and those with no employer-provided health insurance while working or in retirement (a little more than 40 percent). “We should find those who have health insurance on the job, but not in retirement are the ones whose retirement has changed the most as the result of ACA,” Gustman said.
Data came from the Health and Retirement Study (HRS) and was subject to a three-part analysis. Even after GST’s detailed work, they didn’t see observable, ACA effects on retirement claiming or expected retirement dates. Within their model, the group most likely to be swayed by ACA incentives would be those with health insurance while employed, but not in retirement. GST found their reduced work to be “quite modest, amounting to an increase of half a percentage point in the percent retired.”
Gustman pointed out that there may be larger reasons for such small results: The population is still learning about ACA; the time required to adjust savings may not have been long enough; ACA’s startup problems might have scared people off; or specifications might require consideration of all sources for retiree health benefits.
“It may be that employer offerings of retiree health coverage will decline, which would increase the size of the vulnerable groups. It could be that some states that didn’t put in Medicaid will put that into place,” he said. “So the bottom line is we don’t see any observable effects. We don’t see any effects on expected retirement dates. And in the very long term, our model tells us these effects aren’t going to be very large. Nevertheless, you, of course, want to redo the estimates for actual retirement outcomes and expected outcomes when the 2016 HRS data become available.”
In his response, Boston College researcher Matthew Rutledge mentioned that SSA has funded efforts to make GST’s model publicly accessible. “The structural model that Alan, Tom, and Nahid use in this paper is something that they very generously shared with the Center for Retirement Research (at Boston College) as an effort to try and make this a public good,” he said. “We are trying to get that up online and make sure that lots of researchers can share in this great resource.”
The Dynamic Effects of Health on the Employment of Older Workers
Eric French presented work done with fellow Institute for Fiscal Studies colleagues Richard Blundell, Jack Britton, and Monica Costa Dias (BBCF). The group looked at the effects of health on older workers’ employment by estimating “how transitory and permanent health shocks affect employment over time.” The group’s health and labor supply model works to differentiate between short-term (for example, a simple fracture) and long-term (a degenerative disease such as ALS) poor health, and the variances in how short- and long-term illness can effect return to employment.
“These dynamic links are important,” said French. “For example, if we think that after getting a health shock you don’t necessarily immediately drop out of the labor market, but you invest less in your human capital, and it maybe causes you to retire a little bit earlier, these dynamic effects would be important in a way that wouldn’t necessarily be captured in the simpler cross-sectional estimates.”
The researchers applied their model to the HRS and the English Longitudinal Survey of Aging (ELSA). “The key difference between the two data sets is the accent of the respondents. Outside of that, pretty much the same,” said French. For his presentation, French focused on subjective and objective health measures from about 2,500 low-education men.
Subjective health measures are frequently criticized due to justification bias: “Perhaps because they’ve been fired, or something like that, they might justify it by saying ‘I was unhealthy,’” said French. The group dealt with this by instrumenting subjective health factors with objective measures of health.
BBCF estimated a health process that allowed for transitory and permanent health shock. Then they estimated the employment responses to those shocks. The dynamic model found that health is highly persistent; transitory health shocks don’t have much impact on employment, but permanent health shocks do; and current employment is a strong predictor for future employment, even after accounting for the persistence in health.
The researchers found the relationship between present employment and present health surprising. “Now you might think that that would be the highest covariance, but actually it’s not,” French said. “A better predictor of whether you’re working today than your current health is your health in the past…Again this is consistent with this view that these permanent health shocks kind of knock you down, but only slowly…[health shocks] accumulate over time.”
The dynamic model, French concluded, suggests “that the effect of health upon employment is about three times as big as what you typically get out of a simpler type of estimation strategy.”
How Home Equity Extraction and Reverse Mortgages Affect the Financial Well-being of Senior Households
Stephanie Moulton from The Ohio State University presented research done with OSU’s Donald Haurin and Amazon’s Maximilian D. Schmeiser (HMS) on the effects of home equity extraction on senior households. “Eighty percent of seniors own their home,” Moulton said, “and home equity is one of the primary assets,” comprising half the net worth for the median senior. Aside from selling and moving, seniors can extract equity through cash-out refinancing of a first mortgage, revolving home equity lines of credit (HELOCs), closed-end second loans (HELOANs), or federally insured reversed mortgages (HECMs). Each method has different credit requirements, costs, and repayment terms that may differentially affect borrower credit outcomes.
The primary feature of HECMs, and part of their appeal, is that borrowers do not make payment on the loan until out of the home. If the amount of the mortgage exceeds the value of the home, federal insurance covers the difference and protects from negative equity. Borrowers may opt for a lump sum payment or, more commonly, a line-of-credit structure that allows home owners to access equity on an as-needed basis. Only about a million reverse mortgages have been written to date, with the peak years being 2008 and 2009.
Moulton and co-authors used data from the Federal Reserve Bank of New York/Equifax Consumer Credit Panel (CCP) to look at financial well-being, particularly credit outcomes for seniors. HMS identified people 62 and older in the CPP who extracted home equity between 2008 and 2011, then supplemented that data with their own credit panel dataset of borrowers who originated HECMs during those years. The researchers also included people from the CPP who did not extract equity during the relevant years. Each of these groups looks different at baseline (see Moulton’s chart, above).
“We compare credit profiles two years before somebody originates an extraction loan and three years after, so we’re following you for a period of six years,” said Moulton. “We want to see how your credit changes over time.”
The credit trends reveal the whole story. “The reverse mortgage borrower…has a credit shock prior to extraction, and their credit card debt actually increases prior to extraction, and then drops off dramatically after extraction and stays low,” Moulton said. “On the other extraction channels, we don’t see the spike in revolving debt prior to extraction.
“We do see credit card balances spiking right before folks get a reverse mortgage, then declining directly thereafter,” Moulton said.
The researchers also saw sharp drops in credit scores and increases in overdue payments right before HECM borrowers initiated a reverse mortgage. The probability of foreclosure decreases for HECM borrowers post-extraction. Those who chose cash-out refinancing showed an increase in foreclosure.
Adjusting the Payroll Tax to Promote
“The existing tax system may distort labor supply toward more leisure. If we wanted to mitigate that distortion, of course we could cut taxes across the board, but that’s really expensive in terms of revenue to pay for government services,” said MRRC director John Laitner, who presented work in progress with Arizona State University’s Dan Silverman. “This paper looks at the possibility of a more focused kind of tax reduction.”
Laitner and Silverman’s life-cycle model looks at extensive margins, or length of career, rather than amount of time worked during a week. They then apply it to data from the Consumer Expenditure Survey and HRS linked to Social Security earnings records to look at simulated OASI tax reductions of various amounts applied at various ages, and whether they would encourage people to work longer. The model draws heavily from Laitner and Silverman’s 2012 work — the difference is in their analysis of the model, which accounted for unobserved retirement ages and retirement due to disability.
Laitner shared the results of their simulations so far. If the payroll tax ends at age 64, too many people have already retired and the resulting career length change is very small. “But if you move the end back to 62, you can get about a third of a year,” Laitner said. “If you do it at 60, you can get about a half a year, and at 58, about three-quarters of a year.”
Laitner suggested that getting someone to work an extra year would bring an added year of income tax to the federal coffers. “There’s the big social gain in a policy like this,” he said. He also pointed out the “fly in the ointment. If I cut the tax off at age 60 and I had a guy who was counting on working to 64, he’ll get four years where he’ll get a pure transfer by cutting the payroll tax, and then maybe on the margin he’ll work one more [year],” said Laitner. “There will be some revenue loss and we’ll have to think about making that up. But the efficiency gain is in the background.”
Discussant Eugene Steuerle (The Urban Institute) particularly appreciated the model’s attention to the margins. “The model is very interesting because it really does address this additional efficiency issue of whether there are certain margins where you can get maximum bang for the buck,” he said. “I think that’s one of the unique contributions that John and Dan have made to this debate.”
Cohort Changes in Social Security Bene?ts and Pension Wealth
Health and Retirement Study principal investigator David Weir presented recent findings from the study. Co-authored with University of Michigan’s Chichun Fang and Charles Brown, the presentation focused on four recently added cohorts: people who were 51-56 in 1992, 1998, 2004, and 2010. “This is the way we keep HRS fresh,” Weir said. “If we want to continue to understand what’s happening to people before retirement, we need to add new people, and we do that every six years, adding six new birth cohorts.”
Among the newest cohorts, preretirement health has declined. The fraction of people who have reported fair or poor health has risen and the number of people on SSDI has almost doubled from 1992 to 2006. Weir noted that the increase in DI claimants may have economic, as well as health, roots.
Weir described the extensive lengths HRS historically has gone through to get private pension information. “For 2010, something new came along,” he said. “The Department of Labor created a searchable website that posted not just the information in the electronic version of Form 5500, but the entire filing, which includes attachments that often include a lot of plan detail…this made our ability to match much better.” They have gone from being able to match about a third of private pension plans to matching 90 percent.
A key aim for the HRS has been to assess retirement preparation. While IRAs and DC plans have gone up, the decline of DB plans has resulted in a loss of pension wealth across cohorts. Higher earnings and covered earnings have led to somewhat higher social security wealth, but household wealth has been flat from 1992 to 2010. In 1992 and 1998, wealth at 51 to 56 was about a quarter of total lifetime earnings. This, Weir noted, is close to the suggested replacement rate.
“As lifetime earnings have continued to grow (in 2004 and 2010), retirement wealth has not kept pace,” he said. “You may be surprised, but shouldn’t be, that retirement preparation looks better in the lower part of the income distribution because they’re relying on Social Security, which for [that group] is a pretty good replacement rate.”
Weir also broke down the lifetime earnings and retirement wealth findings by race/ethnicity and gender. The retirement wealth drop has been particularly bad for African-Americans, while Hispanic wealth has remained essentially flat. Because women in these cohorts have spent more time in the labor force, their lifetime earnings have gone up rapidly, although their retirement wealth has flattened. “It’s men where you’re seeing the real cratering of retirement wealth, usually due to that loss from DB plans,” said Weir.
He suggested that the findings point to policy changes that would encourage employers to offer and workers to participate in DC plans. He also pointed out that Social Security annuitization is playing an increasingly large role in retirement security. “This depends on solving solvency,” said Weir. “Can someone please save Social Security?”
Working Conditions and Sustainable Work at Older Ages: An International Perspective
RAND’s Jeffrey B. Wenger presented preliminary work, part of a larger project, done with Harvard University’s Nicole Maestas, RAND’s Kathleen J. Mullen and David Powell, and UCLA’s Till von Wachter.
“The population aging is stressing social insurance programs,” Wenger said. “This is true not just in the United States, but it’s true throughout the world, especially the industrialized world.” He presented descriptive information on what work sustainability looks like internationally and what kind of conditions encourage workers to stay employed longer.
Drawn from 35 countries, European Working Conditions Survey (EWCS) data from 2010 found that traits such as shift and night work, tiring/painful positions, working to tight deadlines, lack of latitude, poor fit between work and other commitments, lack of social support at work, poor career prospects, and job insecurity helped defined unsustainable jobs. In 2015 Wenger and team fielded a new U.S. survey from the American Life Panel (dubbed the American Working Conditions Survey — AWCS) using the same EWCS questionnaire so that data would be directly comparable between the two.
The researchers found that, in general, U.S. workers tend to do less unsustainable work overall. One question in particular showed a stark difference: In the EWCS, 62 percent of those 50 and older responded positively to “No training provided in the last 12 months” versus 38 percent in the AWCS. U.S. workers also had less fear of losing their job in the next six months (14 versus 10 percent) and were less likely to work in painful or tiring positions (43 versus 36 percent). European workers were less likely to do shift work (16 versus 27 percent for respondents 50 and older) and work at high speed (28 versus 33 percent).
“One of the things I found particularly interesting is this notion of adverse social behaviors…This is an index measure of seven different factors that combine [experiencing] verbal abuse on the job, unwanted sexual attention, threats, humiliating behaviors, physical violence, sexual harassment, or bullying. When we combine all those things up, people in European countries were less likely to report those things.”
Wenger then focused on flexibility, a job characteristic particularly prized by older workers. While U.S. workers are more likely to “blur the boundaries” by working during free time (45 percent EU versus 50 percent U.S.), they also are more likely to be able to take an hour or two off during working hours (25 percent EU versus 40 percent U.S.).
“American workers appear to have a lot more flexibility in the timing of work and the task autonomy as compared to Europe,” said Wenger, “but they’re also much less likely to have solid boundaries between work and private life. These two things seem to perhaps interact with each other.” The researchers plan future investigation into whether that is sustainable.
Presentation slides and summaries are available on the MMRC website (www.mrrc.isr.umich.edu). Working papers on the meeting’s presentations are expected this fall.