Do Catch-Up Contributions Increase 401(k) Saving? Matthew S. Rutledge Research Economist Center for Retirement Research at Boston College 17th Annual Joint Meeting of the Retirement Research Consortium August 6-7, 2015 Washington, DC Saving is heavily subsidized in U.S. • Traditional 401(k) and IRA contributions: Pre-tax. o But subject to payroll tax (short run). o Taxed at withdrawal (long run). ο§ Hence, “401(k) deferrals.” o Tax expenditure on 401(k)s of $62.3 billion in 2015. • Unclear if 401(k)s increase retirement saving. o Poterba, Venti, Wise (1995): Higher net saving. o Engen, Gale, Scholz (1994): Just reallocation. o Chetty et al. (2014): Most people (Danes) are “passive.” ο§ Even active savers mostly just reallocate. 1 How can we evaluate 401(k) saving? • National roll-out in 1980, so not a great natural experiment. o Some variation at industry or geographic level, but participants may differ in other ways. o Some variation in tax incentive by tax bracket, but only for marginal dollar of saving, and big dataset required. • Source of variation: Catch-up contributions. o Starting in 2002, age 50+ have higher 401(k) limit. ο§ 2015: $18,000 for age < 50, $24,000 for age 50+. o Previously-constrained have new tax incentive for saving. ο§ Important: Not everyone can afford it. 2 This paper • Joint with April Yanyuan Wu, Francis Vitagliano. • How sensitive are high savers to 401(k) tax incentives? • Did 401(k) contributions increase due to the catch-up contribution provision? o Triple-differences framework around catch-up adoption. o Uses Survey of Income and Program Participation (SIPP) data linked to tax records. 3 401(k) limits rose above and below age 50. 401(k) Deferral Limits by Age Year 1999 2000 2001 2002 2003 2004 2005 Nominal limit Catch-up limit $10,000 10,500 10,500 11,000 12,000 13,000 14,000 $0 0 0 1,000 2,000 3,000 4,000 Real (2005$) limit Age < 50 Age ≥ 50 $11,723 $11,723 11,909 11,909 11,579 11,579 11,942 13,027 12,737 14,860 13,440 16,542 14,000 18,000 Real YoY increase Age < 50 Age ≥ 50 -2.2 % -2.2 % 1.6 1.6 -2.8 -2.8 3.1 12.5 6.7 14.1 5.5 11.3 4.2 8.8 Source: Authors’ calculations from the Internal Revenue Service. 4 Our contribution • First causal analysis of catch-up contributions on 401(k) saving. o Existing studies: Mostly descriptive. • Uses admin data on 401(k) saving linked to household survey. o More accurate than household survey alone. o More info on demographics, economic circumstances than in admin data alone. • Uses natural experiment to examine responsiveness of savings to tax incentives. 5 SIPP-SSA linked data • SIPP 1990-2004 panels. o Demographics, education, kids, net worth, occupation, spouse’s work status, homeownership, region, pensions. • Social Security Detailed Earnings Record, 1978-2006. o Total earnings, deferred earnings. o Accessed via the SIPP Synthetic Beta. • Sample: Workers ages 46-53 sometime during 1999-2006. o Exclude if earning less than 4 quarters of OASI coverage or unable to work due to health condition. 6 Triple-differences model π¦ππ‘ = πΌ0 + πΌ1 πΆππΆπ‘ + πΌ2 π΄ππ50ππ‘ πΌ13 πΆππΆπ‘ πππ₯ππ‘ + πΌ3 πππ₯ππ‘ + πΌ12 πΆππΆπ‘ π΄ππ50 + π‘ π΄ππ50ππ‘ πππ₯ππ‘ πΌ123 ππ‘ πΆππΆ +πΌ23 π΄ππ50ππ‘ πππ₯ππ‘ + +πΎ′πππ‘ + ππ‘ + πππ‘ • y = total tax-deferred earnings, or contribution rate (deferrals/earnings) • CUC = 1 if catch-up contribution in effect (2002 or later) • Max = 1 if ever contributed near 401(k) limit previously 7 Identification strategy • Short window around 2002. • Short window around age 50. • Max contributors very dissimilar from non-max contributors. • But max contributors pre-50 and post-50 are conditionally identical except for catch-up contributions. 8 About 9 percent of sample ever contributes near the tax-deferred limit. Frequency Contributing Near the Maximum 401(k) Contribution Share of years at ages 46-53 in 1990-2005 Never Near max 1-5 years Near max 6-7 years Near max all 8 years Number 205,920 10,614 4,789 4,621 Percent 91.1 % 4.7 2.1 2.0 Source: Authors’ calculations from the Survey of Income and Program Participation Completed Data Files, 1990-2005. 9 Max contributors: Greater earnings, net worth, education Summary Statistics for Full Sample and Participants Ever Near Maximum Characteristic Earnings Net worth Male Married Ever had children Children ages 0-17 Children ages 18-24 Black College degree or more Homeowner Blue collar Number of observations Full sample $57,181 $199,571 52.6 % 72.9 80.6 40.2 24.5 9.2 32.8 80.5 30.5 156,423 Max contributors $162,531 $438,849 70.2 % 82.0 74.1 32.3 20.6 3.2 72.1 91.5 34.8 12,406 Source: Authors’ calculations from the Survey of Income and Program Participation Completed Data Files, 1999-2005. 10 Post-2002, deferrals among max contributors grow faster after age 50. Average Deferred Earnings by Age Among Workers Who Are Ever Near the Maximum $18,000 $15,000 $12,000 $9,000 1999-2001 2002-2005 $6,000 $3,000 $0 46 47 48 49 50 51 52 53 Source: Authors’ calculations from the Survey of Income and Program Participation Completed Data Files, 1999-2005. 11 11 Catch-up-eligible max contributors deferred $543 more than younger max contributors. Triple-Differences Regression Results Dependent variable Mean of the dependent variable Year ≥ 2002 Age 50+ Ever previously at 401(k) limit (Age 50+) × (Year ≥ 2002) (Age 50+) × (At limit) (At limit) × (Year ≥ 2002) (Age 50+) × (Year ≥ 2002) × (At limit) Number of observations Deferred amount 2705.9 248.5 *** 95.9 * 5604.6 *** -11.4 -460.5 * 917.1 *** 543.3 * 86,830 Note: *** p<0.01, * p<0.10. Source: Authors’ estimates from the Survey of Income and Program Participation Completed Data Files, 1999-2005. 12 12 Catch-up-eligible max contributors deferred $543 more than younger max contributors. Predicted Increase in 401(k) Contributions from 1999-2001 to 2002-2005, 2005 $ Max contributors over 50 $248 $917 $543 Max contributors under 50 $248 $917 $1,166 $1,697a Non-max contributors over 50 $248 $237a Non-max contributors under 50 $248 $248 $0 Year Year × max Year × age × max $1,000 $2,000 a For simplicity, the figure does not depict the impact of the (Year)(Age) interaction. This interaction reduces the predicted total increase in contributions by $11 for each of the two groups that include individuals age 50 and over. Source: Authors’ estimates from the Survey of Income and Program Participation Completed Data Files (1999-2005). 13 Calculating the elasticity • $543 is 4.6 percent of average real 401(k) limit for age-50plus. • On average, real 401(k) limit is 33 percent higher post-2002 for 50-plus, and 11 percent higher post-2002 for age < 50. ο 22-percentage-point relative increase for 50-plus • Elasticity of 401(k) deferrals with respect to change in the maximum deferral is 4.6/22 = 0.21. • Alternative specifications: o Individual ages: $898 extra, elasticity = 0.30 o Fixed effects: $1,020 extra, elasticity = 0.40 14 Summary • Higher catch-up limit provides a source of variation to test sensitivity of 401(k) deferrals to tax incentives. • Max contributors age 50-plus defer an additional $540-$1,020 more than max contributors slightly younger than 50. • The implied elasticity of 401(k) saving with respect to taxdeferred limit is 0.21-0.40. 15 Conclusion • High-contributing 401(k) participants are sensitive to tax incentives to save. • The study does not address whether: o Overall retirement saving is higher, or if the extra deferrals are simply reallocated from post-tax saving vehicles. o Participants contributing substantially less are as sensitive to tax incentives to save. 16