Who Chooses Defined Contribution Plans? Jeffrey R. Brown University of Illinois and NBER Scott J. Weisbenner University of Illinois and NBER NBER RRC Conference October 2006 1 Why care who chooses Defined Contribution (DC) plans? Numerous proposals to partially replace DB benefits provided by Social Security with personal retirement accounts In many proposals, participation in personal accounts is voluntary Benefit offset – participants must give up DB to have PRA To assess welfare implications and timing of expenditures of Social Security reform proposals it is important to understand who would opt to participate in personal accounts 2 Existing Research Large literature on 401(k) participation The counterfactual is different Consumption vs. saving as opposed to form of benefits Better to look at explicit DB vs. DC choice Employees at non-profit (Yang 2005) Faculty at NC universities (Clark, et al 2005) Also interesting to observe decision when it is not already on top of existing Social Security system (and for broader cross-section of employees) 3 Our Research Design Examine a sample of individuals that have to make a one-time, irrevocable choice between DB and DC plan This choice is important as: Combined employee & employer contributions are at least 14.6% of wages for these individuals Earnings from this employment are NOT covered by Social Security 4 Who is our sample? There are seven states whose public-sector employees do not participate in Social Security California, Colorado, ILLINOIS, Louisiana, Massachusetts, Ohio, and Texas We have administrative data provided by the State Universities Retirement System (SURS) of Illinois Established 1941, covers all faculty and support staff of Illinois higher public education including universities and community colleges 5 Decision faced by SURS-covered employees Within first six months of employment, are given opportunity to make a one-time, irrevocable choice among three retirement plans: Traditional DB plan: formula-based DB plan (or money purchase when more generous) Portable DB plan: somewhat less generous than traditional DB if retire from the system, but much more generous if take an early lump-sum distribution Self-Managed Plan: self-directed DC plan If make no decision within 6 months, participants are defaulted into the Traditional DB plan 6 Our Focus in This Paper What choices do individuals make? What worker characteristics are correlated with retirement plan choice? Next paper – why do individuals make their choices 7 Brief Background on SURS Prior to 1998, all SURS employees’ covered by the Traditional DB plan In 1997, Illinois legislature passed a law allowing participating employees a choice of three plans These new choices adopted by most employers sometime in 1998 Rules changing for employees starting after 7/1/05 Our sample is 45,000 new hires over period 19992004 8 Basics on the three options Some similarities across the three plans All have 5 year vesting for employer contributions All require employee contributions equal to 8% of salary All require annuities at retirement if individual wants retiree health care Important differences across three plans along: Rate of return Amount of employer matching contributions Handling of “refunds” for early termination 9 Traditional DB Plan The default plan for those who do not choose Retirement annuity is the highest amount calculated under two different formulas DB = 2.2% * Service * Final (4-year) Compensation Money Purchase = annuitized value of account balance (using generous Effective Rate of Interest) But benefit is subject to a maximum of 80% of final average pay In recent years, most individuals have had higher benefit under the money purchase formula Also receive spousal benefits at no additional cost 10 Traditional DB Plan While this plan is quite generous for those who retire from it, it is not as generous for those who leave the system early Two choices if leave early: Leave money in SURS to grow under MP formula Take refund: only receive employee contributions plus 4.5% interest – no employer contributions 11 Portable DB Plan Benefits calculated in same way as Traditional Except that one must pay for the survivor benefit Amounts to 7% to 13% reduction in benefits Big difference in treatment for those who leave the system before retirement and take refund If leave within five years, receive own contributions (+ cumulative returns from an Effective Rate of Interest) If leave with tenure of 5+ years, receive own contributions plus a dollar-for-dollar employer match 12 SMP (DC plan) Entirely participant-directed DC plan Participants choose from a variety of mutual funds provided by TIAA-CREF and Fidelity Balance in account depends on asset returns and vesting status If leave system in less than five years, lose employer contributions If leave system with tenure of 5+ years, leave with employee and employer contributions Employer contributions are 6.6% of salary or 0.825-for-one match At retirement, must annuitize if want health insurance – annuity rates inferior to SURS 13 Some Comparisons (for those hired before July 1 2005) Refund match Traditional 0% Refund int rate 4.5% Benefit if retire from system Portable 100% SMP 82.5% ERI (8-10%) Actual returns Max [DB Max [DB formula, Money formula, Money Purchase] + Purchase] Survivor benefits Annuitized account balance MP interest ERI (8-10%) ERI (8-10%) Actual returns Ben Max? 80% FAC 80% FAC None Annuity price Subsidized Subsidized Market Rate 14 Summary of Traditional Plan Traditional plan attractive for those with long careers under SURS DB formula tends to be attractive for those starting at older ages Money purchase formula tends to provide higher benefits for those that start at younger ages If leave service early under traditional, rational path is to leave money in SURS to benefit from money purchase option rather than taking refund 15 Portable vs. SMP? Benefits of Portable: Portable “match” rate is higher (8% vs. 6.6%) Portable “return” has averaged 8-10% nominal over past 20 years (and gains are implicitly locked in) Portable annuity prices extremely attractive DB formula provides a generous benefit “floor” Benefits of SMP Not subject to the benefit cap at retirement Less subject to political risk – but there isn’t any! 16 Portable vs. SMP For most employees in our sample, it would appear that the Portable is better choice Access to DB formula or MP if retire under system Higher match rate if refund from system High Effective Rate of Interest Would need SMP returns of 9-12% depending on time horizon to come out ahead with SMP Impairment clause of constitution minimizes political risk Two sets of simulations … 17 Portable vs. SMP: Refunds Compare refund amounts if leave system and take the money with you Assumes 5% real Portable Return SURS real rate over last 25 years is higher But no guarantees about this rate Assumes 60/40 portfolio returns from Shiller data over 1802 – 2004 with 30 b.p. expenses 18 Refunds: Portable 5% Real Return v. Historical Returns on 60/40 Portfolio Time Horizon 5 10 15 20 25 30 35 % times SMP beats Portable 28.5 38.9 43.8 44.8 43.2 37.7 36.6 19 Portable vs. SMP: At Retirement Then compare amounts if retire under the system after working from given age … $40,000 with 4% growth in salary 8.5% ERI and 8.5% SMP return (nominal) DB benefit discounted to retirement age using 1994 group mortality and 4.75% nominal interest rate 20 PV of Benefits at Ret. Age (* means exceeds 80% benefit cap) Start Age 25 30 35 40 45 50 55 Portable DB Formula 1,445 1,019 698 459 283 155 92 Portable M.P. Option 2,277 (1,500)* 1,414 (1,235)* 859 504 280 141 85 SMP 2,077 1,291 784 460 256 129 78 21 Portable v. SMP Summary Portable is the “Winner” under most (though not all) plausible scenarios SMP better only if returns large enough to: Beat assumed interest rate under SURS, plus Cover the contribution differential, plus Cover the annuity pricing differential Even when benefits hit cap under Portable, you can still take a refund to get higher amount Would lose ability to annuitize at favorable rates Would lose access to retiree health insurance 22 How Choices are Currently Framed SMP DC DC or DB? Portable DB Portable or not? Traditional Alternative Framing Traditional No Value Portability? Portable Yes SURS Or Self? SMP What Do People Choose? This project is focused on who chooses which option Next phase is a survey – why do they choose what they choose … (more on this later) 25 Our Data 45,000 observations from 1999 – 2004 Ranges from 6600 to 8600 per year 52% university, 40% community college, 8% other 55% academic, rest is staff, police, etc. 57% female 59% married Average age is 41 Of the cohort starting in 1999 35% left prior to vesting 7% left with vesting (as of April 2006) 26 Plan Choice, 1999-2004 SMP 15% Portable DB 19% Traditional DB 66% Enrollment in SMP (DC plan) 100 90 80 70 60 50 40 30 20 10 0 Full Sample Enrollment in SMP (DC plan) 100 90 80 70 60 50 40 30 20 10 0 Full Sample University Academic Enrollment in SMP (DC plan) 100 90 80 70 60 50 40 30 20 10 0 Full Sample University Academic University Academic $100K+ Enrollment in SMP (DC plan) 100 90 80 70 60 50 40 30 20 10 0 Full Sample University Academic University Academic $100K+ Jeff Enrollment in SMP (DC plan) 100 90 80 70 60 50 40 30 20 10 0 Full Sample University Academic University Academic $100K+ Jeff Scott Enrollment in SMP (DC plan) 100 90 80 70 60 50 40 30 20 10 0 Full Sample University Academic University Academic $100K+ Jeff Scott SURS Enrollment in Portable DB Plan 50 40 30 20 10 0 Full Sample University Academic University Academic $100K+ SURS The Default Option If no choice made in 6 months, defaulted into DB Vast majority of Traditional DB plan enrollment is via the default From 1999-2004, 56% of participants default into the Traditional DB plan vs. 10% that actively choose the Traditional DB plan Percent defaulting has grown over time Might it be rational to default? Plan to be here forever, want the Traditional, so no need to spend time filling out paperwork Plan to be here < 5 years, differences in returns are small 35 Percent of 1999 & 2000 Cohorts, by Tenure in Plan, that Defaulted into Traditional DB Plan 70 60 50 40 30 20 10 0 Left in < 5 Years Left in 5+ Years Still in System Empirical Methods 1. Series of linear probability models SMP Portable Traditional Any active choice Choose traditional (given traditional) Choose traditional (given active choice) 2. Multinomial logit of all four choices 37 Empirical Findings Academics more likely to choose SMP University employees more likely to choose SMP Community college employees more likely to actively choose traditional (unions?) Individuals who subsequently left with vested benefits more likely to choose Portable Higher earners more likely to choose SMP or Portable 38 Empirical Findings Women more likely to make active choices Married individuals more likely to make active choices Older employees more likely to choose traditional Younger workers more likely to default Evidence of strong “employer effects” Peer effects? Benefit office effects? 39 Next Steps Lots of information missing from administrative data that is relevant for retirement plan choice 2006/07 RRC project Focus groups Survey of members We would appreciate your input on survey questions and survey design 40 Hypotheses: Why Choose DC? Failure to understand benefit rules Optimism about financial market returns Excessive optimism about own investment prowess Value choice / control (Irrational?) concerns about political risk 41 Some Questions … Sources of information used when selecting plans How well can they describe differences across plans Expectations of financial market returns Confidence in investment ability Assessment of political risk Expectations about length of service under SURS Social Security and pension coverage of spouse Views about Social Security reform 42