Employer matching and 401(k) saving: Evidence from the health and retirement study Authors: Gary V. Engelhardt, Anil Kumar Persistent link: http://hdl.handle.net/2345/4215 This work is posted on eScholarship@BC, Boston College University Libraries. Chestnut Hill, Mass.: Center for Retirement Research at Boston College, 2004 EMPLOYER MATCHING AND 401(K) SAVING: EVIDENCE FROM THE HEALTH AND RETIREMENT STUDY Gary V. Engelhardt* Anil Kumar CRR WP 2004-18 Released: May 2004 Draft Submitted: April 2004 Center for Retirement Research at Boston College 550 Fulton Hall 140 Commonwealth Ave. Chestnut Hill, MA 02467 Tel: 617-552-1762 Fax: 617-552-1750 http://www.bc.edu/crr *Gary V. Engelhardt is an associate professor of economics at Syracuse University and a senior research associate at the Center for Policy Research. Anil Kumar is a research associate at the Center for Policy Research. The research reported herein was performed pursuant to a grant from the U.S. Social Security Administration (SSA) to the Center for Retirement Research at Boston College (CRR). The opinions and conclusions are solely those of the authors and should not be construed as representing the opinions or policy of the SSA or any agency of the Federal Government or of the CRR. The authors would like to thank Dan Black, Chris Cunningham, Bill Gale, Erik Hurst, Brigitte Madrian, John Moran, and seminar participants at Syracuse University, University of Chicago, and University of Virginia for helpful discussions and comments. © 2004, by Gary V. Engelhardt and Anil Kumar. All rights reserved. Short sections of text, not to exceed two paragraphs, may be quoted without explicit permission provided that full credit, including © notice, is given to the source. About the Center for Retirement Research The Center for Retirement Research at Boston College, part of a consortium that includes parallel centers at the University of Michigan and the National Bureau of Economic Research, was established in 1998 through a grant from the Social Security Administration. The goals of the Center are to promote research on retirement issues, to transmit new findings to the policy community and the public, to help train new scholars, and to broaden access to valuable data sources. Through these initiatives, the Center hopes to forge a strong link between the academic and policy communities around an issue of critical importance to the nation’s future. Center for Retirement Research at Boston College 550 Fulton Hall 140 Commonwealth Ave. Chestnut Hill, MA 02467 phone: 617-552-1762 fax: 617-552-1750 e-mail: [email protected] http://www.bc.edu/crr Affiliated Institutions: American Enterprise Institute The Brookings Institution Massachusetts Institute of Technology Syracuse University Urban Institute Abstract Employer matching of employee 401(k) contributions can provide a powerful incentive to save for retirement. We examine the effect of matching on 401(k) saving accounting for non-linearities in the intertemporal budget set. We use detailed administrative contribution, earnings, and pension plan data from the Health and Retirement Study and estimate that the elasticity of contributions with respect to matching is 0.15-0.27 overall, with sixty percent of this effect on the participation margin and the remaining forty percent on the intensive margin. The estimated after-tax cross-price elasticity of 401(k) contributions with respect to IRA saving is -0.60, which suggests 401(k)s and IRAs are substitutes in tax-deferred saving. We find no evidence of endogenous worker sorting based on the discount rate to plans that offer matching. I. Introduction As 401(k)s have come to dominate the pension landscape, researchers and policy makers have given increased attention to the impact of plan characteristics on retirement saving decisions.1 One important characteristic is whether and to what extent the employer matches employee contributions. A typical match might be 50 cents for each dollar of contribution, up to a maximum percentage of pay, say, 6 percent. Although much of the discussion by the popular press and policy makers presumes employer matching raises saving, there is actually strikingly little consensus among researchers. Some studies have found that increases in the match rate raise 401(k) saving (Papke and Poterba, 1995; Clark and Schieber, 1998). Others have found that it is not the match rate per se that matters, but whether the firm offers a match at all (Bassett, Fleming, and Rodrigues, 1998; Papke, 1995; Kusko, Poterba, and Wilcox, 1998). That is, providing a match raises 401(k) saving, but an increase in the level of the match rate (conditional on providing a match) does not. Finally, still other studies (Munnell, Sunden, and Taylor, 1998; and GAO, 1997) have suggested that, conditional on being eligible for a match, an increase in the match rate lowers 401(k) contributions, which, when interpreted in the context of a simple two-period model of saving, suggests that the income effect dominates the substitution effect from the higher rate of return matching provides.2 A central shortcoming in this literature has been the failure to exploit the fact that employer matching based either on a multiple match-rate schedule or caps on the 1 This includes work on automatic enrollment (Madrian and Shea, 2000, Choi, Laibson, Madrian, and Metrick, 2001, 2002), investment in company stock (Poterba, 2003), portfolio choice and trading in 401(k) plans (Benartzi and Thaler, 2001; Agnew, Balduzzi, and Sunden, 2003). 2 Throughout the paper, we refer to 401(k) saving and 401(k) contributions synonymously as per period flows. In a multi-period model, this would suggest the income effect dominates the substitution and human wealth effects (Summers, 1981). 2 generosity of the match induces kinks in the intertemporal budget constraint. As has been long recognized in the study of taxation on labor supply, reduced-form estimates of behavioral elasticities are biased and inconsistent unless these kinks are accounted for explicitly (Hausman, 1985; Moffitt, 1990). Indeed, the presence of kinks may reconcile some of the findings of previous studies. For example, the provision of a match may raise 401(k) saving if the substitution effect dominates, but variation in match rates may not matter if employees are bunched at kinks. We make five important contributions. First, we lay out a theoretical model and specify a life-cycle consistent structural econometric specification based directly on the first-order conditions for 401(k) saving. Second, we circumvent difficulties with measurement error in 401(k) contributions and matching incentives that have plagued previous studies by using administrative data from three sources: contributions from W-2 earnings records provided by the Social Security Administration (SSA) and Internal Revenue Service (IRS); detailed matching formulas from pension Summary Plan Descriptions (SPD) provided by the employers of HRS respondents; and, a combination of covered earnings histories for 1951-1991 and W-2 earnings for 1980-1991 from SSA, pension SPDs, and pension benefit calculators to construct public and private pension entitlements and accruals. Our sample consists of 1,042 individuals in 1991 eligible for 401(k) plans in the Health and Retirement Study (HRS). Third, we account for kinks in the estimation using a variant of the differentiable budget set methodology of MaCurdy, Green, and Paarsch (1990) based on kernel regression. Fourth, we estimate ad hoc reduced-form empirical models similar to past studies, and are able to replicate many of the puzzling findings from the previous literature in our data: the existence of a matching 3 program raises 401(k) saving, but conditional on offering a match, higher match rates have no effect or lower contributions. In contrast, our structural instrumental variable Tobit specifications suggest that the uncompensated elasticity of 401(k) saving with respect to the match rate is 0.15-0.27 overall, with sixty percent of this effect on the participation margin and forty percent on the intensive margin. We estimate little impact of income on contributions. The estimated elasticity of contributions with respect to the net hourly wage and the relative after-tax price of 401(k) versus IRA saving are 0.40 and -0.60, respectively, the latter of which confirms the intuition that 401(k) and IRA saving are substitutes. Fifth, we test whether individuals with low discount rates work at firms with higher employer match rates (Ippolito, 1997), often used as a criticism of the quasi- experimental “eligibility experiment” approach to identifying 401(k) saving effects (Poterba, Venti, and Wise, 1994, 1995). We find no evidence of worker sorting based on discount rates. The paper is organized as follows. Section II lays out the theoretical model that directly motivates the empirical work. Section III lays out the econometric framework and construction of the key variables. Section IV describes the data. Section V discusses the identification strategy. The empirical test for endogenous worker sorting is in section VI. Section VII discusses the estimation results. There is a brief conclusion. II. Theoretical Framework Previous studies, summarized in Table 1, have failed to exploit the fact that multiple-match rate schedules and caps on matching induce kinks in the intertemporal budget set. For example, Figure 1 shows the budget set in a simple two-period model of consumption typically used in undergraduate textbooks. Let c and c be consumption 1 2 4 in periods 1 and 2, respectively, and y be first-period earnings. For simplicity, assume a 1 single marginal tax rate, q, in the first period, no taxes and earnings in the second period, 401(k)s are the only form of saving and contributions are tax-deductible. Assume the firm matches contributions at a fixed rate m up to qmatch of contributions, and a limit on contributions, qmax. Then, as shown in the figure, the budget constraint is akdb. From zero to qmatch contributions, the slope of the budget constraint is - (1+m)(1+r)/(1- q), from qmatch to qmax, the slope is - (1+r)/(1- q), and beyond qmax, the slope is zero. At qmatch, there is a kink point, k. It is obvious from the figure that, in the absence of matching, the introduction of an uncapped match has standard substitution and income effects.3 However, this may not occur if the match is capped. That is, compensated changes in the match rate, m, may not induce changes in 401(k) saving if individuals are bunched at the kink point, k, and standard income and substitution effects are not well defined. Instead, standard income and substitution effects on each budget segment are defined by the slopes given above and the virtual incomes, yv1 and yv2. In addition, even though saving involves the substitution of resources across time, previous studies have not couched their analyses in formal models of intertemporal choice. This means that previous estimates cannot be interpreted as estimates of life- cycle-consistent uncompensated demands for 401(k) saving necessarily, because the empirical specifications may not be consistent with underlying utility maximization. So, while previous studies have been quite informative descriptive analyses, they say little 3 In fact, if the match rate and tax rate are zero, the figure collapses to the standard undergraduate textbook figure, in which the slope is just - (1+r). 5 about how 401(k) saving may respond to prospective changes in employer matching or what the optimal match rate should be to achieve a target saving objective. In contrast, we estimate econometric models of contributions that are consistent with life-cycle theory and incorporate non-linear budget sets explicitly. The budget sets individuals actually face are substantially more complicated than the one depicted in Figure 1 for a number of reasons. First, they may have multiple kinks due to variable- rate matches. Second, there may be multiple kinks because there are multiple marginal tax rates and contributions are tax-deductible, so that making a contribution may change the marginal tax rate. Surprisingly, none of the previous studies have accounted for the effect of taxation on 401(k) saving. Third, 401(k) plan participants also can save through other vehicles, such as IRAs and non-tax-deferred financial asset saving. Finally, there may be goods other than consumption, such as leisure, that enter utility, and a change in the match rate may induce intratemporal substitution across goods. To motivate the empirical work, we use a theoretical framework that incorporates these additional facets of behavior. Specifically, utility is derived from consumption,C, leisure, l, depends on a set of demographics, Z , and is weakly separable. The consumer lives from the beginning of working life, period 0, until death in periodT. The lifetime is composed of two parts. From period t +1 toT , the consumer is retired and chooses consumption, C, to maximize the present value of utility. In retirement, no hours of labor are supplied to the market, so leisure equals the time endowment, Ll. From period 0 to t , the consumer works and chooses consumption, leisure, voluntary 401(k) contributions, Q401k, and IRA contributions, QIRA, respectively. Wealth is accumulated in five assets: 401(k) wealth, W401k, IRA wealth, WIRA, non-401(k) pension wealth, 6 WP, non-401(k)-IRA-pension wealth, W A, and the present value of Social Security benefits, WSS, where total wealth is WT ” W401k +WIRA +WA +WP +WSS.4 (1) Let y(WT) be the expected present value at period t of utility for the second t part of life, as viewed during the first part of life, and r the rate of time preference. When working, the objective function is max E ŒØ (cid:229) t (1+r)- tU(C ,l ;Z )+(1+r)-ty(WT)œø , (2) Ct,Qt401k,QtIRA,lt tº t=0 t t t t ß and the components of wealth evolve as follows WA =(1+r )WA +w (Ll - l )+B - C - Q401k - QIRA - T , (3) t t t- 1 t t t t t t t W401k = (1+r )W401k +(Q401k +MV + R +M R), (4) t t t-1 t t t t WIRA =(1+ r)W IRA +QIRA, (5) t t t- 1 t WP = (1+a P)WP , (6) t t t- 1 and WSS = (1+aSS)WSS. (7) t t t-1 In (3),r is the stochastic gross interest rate earned on assets between periods t- 1 and t, t and E in (2) is the expectations operator conditional on the information set W .5 B is t t- 1 t 4 Non-401(k) pensions consist of all defined benefit plans and defined contribution plans not governed by section 401(k) of the Internal Revenue Code. We do not explicitly consider the role of housing, which is subsumed into other wealth. Engen, Gale, and Scholz (1996), Poterba, Venti, and Wise (1996), Bernheim (2002), and Engelhardt (2001), among others, have discussed the potential interactions between housing and 401(k)s. 5 We assume that all non-pension wealth earns the same gross rate of return. There is no other uncertainty in the model, and there are no bequests. The mean age in our sample below is 55. Carroll (1992), among others, has estimated that most lifetime income uncertainty has been resolved by this age, at which point households have transitioned from buffer stock to life-cycle savers. Hence, we did not feel income uncertainty was central to the households under study and did not model it. Uncertainty in the length of life 7