Designing Pension Plans for the Twenty-First Century
August 17 and 18, 2012
Jeffrey Brown and Scott Weisbenner, University of Illinois and NBER
Using a unique survey of participants in a large public pension plan that provides participants with a choice between a defined contribution (DC) and a defined benefit (DB) retirement plan, Brown and Weisbenner study what types of individuals choose DC plans. Holding all characteristics of the employer and the job fixed, they find sensible patterns with regard to economic and demographic factors: DC plan choice decreases with age, rises with the level of education, and is less frequent among groups (for example, police officers) for whom there are additional financial benefits to the DB plan. The authors further find that the ability to control for beliefs, preferences, and financial skills nearly triples the amount of variation in plan choice that they can explain, relative to using standard economic and demographic variables alone. Especially important are respondent attitudes about risk/return trade-offs, self-assessed investment skills, general and choice-specific financial literacy, and beliefs about plan parameters. The authors also note that beliefs about plan parameters are very important, even when these beliefs are factually incorrect. In general, people seem to making sensible choices based on what they believe to be true about the plans, but they do not always have accurate beliefs (and thus may not be making optimal decisions). Finally, they provide suggestive evidence that preferences on the attributes of the retirement system (for example., the degree of control provided) and perceptions of political risk are significant determinants of the DC/DB decision.
Robert Novy-Marx, University of Rochester and NBER, and Joshua Rauh
A large majority of public sector employees in the United States receive retirement packages in which, by contract, they bear no investment risk. Initial retirement benefits depend on salaries and years of service, and cost-of-living adjustments are either fixed, ad hoc, or linked to inflation. Novy-Marx and Rauh calculate the effect that introducing risk-sharing, either during retirement or the working life, would have on public sector pension liabilities. Introducing a variable annuity for the retirement phase, modeled on the Wisconsin Retirement System in which positive benefit adjustments are granted only if asset returns surpass 5 percent, would reduce unfunded accrued liabilities by around 25 percent while allowing all other plan features to be maintained. It would lower required annual contribution increases for full funding within 30 years by 44 percent. They then derive alternative policy measures that would have similar effects on costs. These include increasing employee contributions by 11.6 percent of pay in order to keep benefits unchanged; or giving employees a collective or DC plan with a 9 percent employer contribution for future service. They discuss these results in the context of models of life-cycle portfolio choice, which suggest that employees generally should prefer to take risk earlier in their lives rather than later.
John Chalmers, University of Oregon; Woodrow Johnson, U.S. Securities and Exchange Commission; and Jonathan Reuter, Boston College and NBER
The Oregon Public Employees Retirement System (PERS) uses benefit formulas drawn from both defined benefit (DB) and defined contribution (DC) pension plans and automatically pays retirees the maximum benefits for which they are eligible. Chalmers, Johnson, and Reuter use PERS administrative data covering January 1990 to December 2003 to study the impact of this hybrid plan design on employers' costs and employees' behavior. They find that the flexibility built into PERS is costly for employers to provide. The expected present value of the benefits owed to employees retiring under the hybrid plan during their sample period is 57 percent higher than it would have been under a traditional DB plan. Second, the hybrid plan distorts employees' retirement decisions. The simplest way to demonstrate these distortions is to note that as average retirement benefits increase above the level they would be in a traditional DB plan, average retirement ages fall. Two sources of exogenous variation in retirement incentives can be exploited: first, the use of stale returns to calculate retirement benefits between 1990 and 1999; and second, when PERS incorporates updated life expectancy tables into its benefit formulas effective July 2003, in an effort to reduce the level of underfunding. The authors find that employees respond to both types of retirement incentives. Third, there is evidence of peer effects, in that employees respond more strongly to their own retirement incentives when more of their coworkers face the similar incentives. The retirement waves that result from existing employees seeking to prevent declines in pension benefits are likely to increase the administrative costs associated with pension reform.
John Beshears, Stanford University and NBER; James Choi, Yale University and NBER; David Laibson and Brigitte Madrian, Harvard University and NBER; and Stephen Zeldes, Columbia University and NBER
Beshears, Choi, Laibson, Madrian, and Zeldes conduct and analyze two large surveys of individuals' hypothetical choices of retirement annuities. The researchers find that allowing individuals to annuitize only a fraction of their wealth increases that choice, relative to a situation in which choosing an annuity is an "all or nothing" decision. Very few respondents will choose declining real income paths over flat, or increasing, real payout streams of equivalent expected present value, and highlighting the effects of inflation increases the demand for cost-of-living adjustments. But framing this decision by focusing on flexibility, control, and investment risk significantly reduces annuitization. A majority of the survey respondents instead prefer to receive an extra "bonus" payment during one month of the year, funded by slightly lower payments in the remaining months. Ensuring later-life income, spending flexibility, and counterparty risk are the most important self-reported motives driving the annuitization decision.
Robert Clark, and Melinda Morrill, North Carolina State University
Studies examining workers' pension distribution choices have found that the tendency of workers is to select lump-sum distributions instead of life annuities. This choice, which seems to contradict economic theory, has been dubbed the Annuity Puzzle. Previous studies typically have used survey data based on respondent recall. Using instead administrative data from the North Carolina retirement system, Clark and Morrill illustrate a "reverse" annuity puzzle for public sector workers separating prior to retirement. Even when the present value of cashing out is higher, many separating workers maintain pension accounts. The distribution is larger in present value for three quarters of terminating workers, but only one third of them requested cash distributions within one year. The authors find that among vested separating employees, 33 percent chose to withdraw their funds within one year of separation. Non-vested workers, who only gain from maintaining their account if they return to public sector employment, still only withdrew within one year of separation 36 percent of the time. The evidence suggests that separating workers, particularly those with short tenure, may be forgoing important benefits because of a lack of knowledge, understanding, or accessibility of benefits.
Leora Friedberg, University of Virginia
The Great Recession led to massive job losses, even affecting jobs in state and local government (S&LG) and jobs held by older workers. Friedberg examines exit rates from different job sectors from 1993-4 to 2010-11 and finds that: 1) average exit rates for workers of all ages from the private sector and the S&LG sector are similar; 2) age patterns in exit rates differ -- before the Great Recession, S&LG workers generally had lower exit rates in their 40s and early 50s, but they had higher exit rates when young and old; 3) before the Great Recession, exit rates of middle and older workers in the private sector declined steadily, while exit rates at the same ages declined less in the S&LG sector, especially among those at middle ages; and 4) during the Great Recession, all but private sector older workers experienced increases in their exit rates, while only middle-aged workers in the S&LG sector experienced such increases. These patterns suggest that defined benefit (DB) pensions may play a role in explaining differences in job stability at middle and older ages across sectors. While most discussions have focused on funding issues, public pension reforms also can affect the S&LG workforce, leading to more exits of middle-aged workers and fewer exits of older workers, and in turn affecting payroll and health insurance costs.
Richard Disney, University College London, and Rowena Crawford, Institute for Fiscal Studies
Disney and Crawford examine the determinants of ill-health retirements among police officers in the police forces of England and Wales between 2002-3 and 2009-10. Differences in ill-health retirement rates across forces are statistically related to the area-specific stress of policing and to force-specific differences in human resources policies. The authors describe a series of reforms to police pension plans; in particular a shift in the incidence of financing ill-health retirement through pension plans from central government to local police authorities that occurred in the mid-2000s. They show that these measures had a significant impact on the level of ill-health retirement, especially among forces with above-average rates of ill-health retirement. They investigate whether residual differences in post-2006 ill-health retirement rates across forces are related to their differential capacities to raise revenue from local sources, and find that local police authorities were prepared to raise precepts to finance such retirements.
Julie Agnew, College of William and Mary, and Joshua Hurwitz, Boston College
As more and more public pension systems are shifting away from a defined benefit only framework, the complexity of the financial decisions facing public employees is increasing. This raises some concerns about the financial literacy of participants and their ability to make informed decisions. While surveys addressing financial education in private plans are available, little is known about what types of education and advice are offered in public plans. Agnew and Hurwitz present new results from the first National Public Pension Financial Education Survey. They focus specifically on primary defined contribution and hybrid plans. Their results indicate that some form of education or advice is offered by every surveyed plan and that the sponsoring entity is actively involved in the development of the programs. However, it appears that legal uncertainties related to advice and education may be a problem for a few plans. In addition, more rigorous evaluation methods to test programs are needed.
Jeffrey Smith, Virginia Military Institute, and James West, Baylor University
Using data on officers and enlisted members of the U.S. military from 1981 through 2011, Smith and West analyze the effects of future retirement benefit eligibility upon the decision of whether to remain in the military to quality for retirement benefits upon completing 20 years of active duty service. They find that the generosity of retirement benefits is strongly correlated with the decision to qualify for benefits, even given the large discount rates found in previous work.
Alicia Munnell, Jean-Pierre Aubry, and Josh Hurwitz, Boston College, and Laura Quinby, Harvard University
Public sector defined benefit pension plans are based on final earnings. As such, these plans are back-loaded; those with long careers receive substantial benefits and those who leave early receive little. In their paper, Munnell, Aubry, Hurwitz, and Quinby first discuss the design of state and local defined benefit plans, document the extent to which traditional public sector final earnings plans are back-loaded, and explore the extent to which the incentives may reflect the preferences of employers. Next they show how participation in final earnings plans affects the lifetime resources of state and local workers of various tenures as compared to private sector workers. Then they present plan-level data on the flows of participants out of the plan by age and tenure and explore the extent to which plan design -- specifically, vesting periods, mandatory participation in a defined contribution plan, and Social Security coverage -- affects the probability of vesting and the probability of remaining to the earliest full retirement age once vested. They find that complete reliance on delayed vesting and final earnings plans is detrimental to many public employees. Hence, the recent trend towards hybrid arrangements is a positive development, not only for risk sharing between taxpayers and participants but also for a more equitable distribution of benefits between short-term and career employees.
James Farrell, Florida Southern College, and Daniel Shoag, Harvard University
Public employee pension plans are increasingly transitioning towards defined contribution and hybrid models. This shift has provoked fears that individual employees will manage their funds poorly. Farrell and Shoag use new data to describe the universe of public plan investments and to compare investment behavior in public DB and non-DB plans. Using information on thousands of individual investors in Florida's DC plan along with new data on the investment behavior of thousands of public plans, they explore how plan design affects investment allocations, the distribution of investment outcomes, and the prevalence of common investment biases.
Edward Glaeser, Harvard University and NBER, and Giacomo Ponzetto, CREI, Universitat Pompeu Fabra
Glaeser and Ponzetto present a model suggesting that public sector pensions may be generous in part because pension promises are less easily observable than promises about more direct forms of compensation. The shrouded nature of public pensions in particular presents one explanation for why they are typically far more generous than private sector pensions. This model also predicts that pensions will be more generous when public sector workers live in the community, or when pre-funding requirements are lower. Because union workers are likely to live in big cities, the authors speculate that moving to centralized control over big city pensions may be particularly likely to reduce generosity. They then use the logic of their model to understand the public pension picture in Massachusetts, California, and Texas. They find that Massachusetts has moved steadily to centralize its control, initially because of a desire to create uniformity and to reduce the ad hoc nature of particular individual pensions.