Do Households Increase Their Savings When the Kids Leave Home? Irena Dushi, Alicia H. Munnell, Geoffrey T. Sanzenbacher, and Anthony Webb U.S. Social Security Administration and Center for Retirement Research at Boston College 17th Annual Meeting of the Retirement Research Consortium Washington, DC August 7, 2015 Saving behavior when kids leave home has big effect on target replacement rates… Hundreds Illustrative Consumption by Age, as Percent of Income, Alternative Assumptions Regarding Kids Income 100% 80% 60% 40% Consumption cut when kids leave home Constant consumption 20% Social Security 0% 25 35 45 55 65 75 85 90 Age Source: Authors’ illustration. 1 …and explains why researchers disagree about retirement savings adequacy. • Research that assumes households do not cut consumption when kids leave generally finds households are not prepared (e.g. Munnell, Rutledge, and Webb (2014) using the NRRI). • Research that assumes households do cut consumption generally finds households are prepared (e.g. Scholz, Seshadri, and Khitatrakun (2006)). 2 Kids explain half the difference in the results between the two approaches. Percentage of Households Age 51-61 At Risk, 2004 40% 35.3% 30% 24.3% 20% 11.5% 10% 8.0% 0% Original NRRI NRRI adjusted NRRI adjusted Scholz and for optimal for optimal Seshadri (2008) drawdown drawdown + children Source: Alicia H. Munnell, Matthew S. Rutledge, and Anthony Webb. 2014. “Are Retirees Falling Short? Reconciling the Conflicting Evidence.” Working Paper 2014-16. Center for Retirement Research at Boston College. 3 What evidence do we have about how households actually behave? • Coe and Webb (2010) used Health and Retirement Study consumption data. o They find no evidence that total consumption declines when the kids leave home. o But sample size is small and data noisy. • Rottke and Klos (2013) used European data. o They find a modest decline in consumption and increase in saving when kids leave. 4 We examine savings instead of consumption. • For a given level of income and taxes, the amount not consumed is saved. • So testing for changes in savings is the flip side of testing for changes in consumption. • Study uses two separate datasets to explore the effects: o Health and Retirement Study (HRS). o Survey of Income and Program Participation (SIPP). 5 Advantages of our datasets. Advantages of HRS Data • Linked to W-2 data on 401(k) contributions. • Long panel, so observations pre- and post-kids leaving home. • Includes extensive control variables. Advantages of SIPP Data • Includes households of all ages. • Large sample size. 6 We estimate two equations: 1) pooled cross section (HRS and SIPP)… Si ,t 0 1 NoKidsi ,t 2 KidsGonei ,t 3Yi ,t X i ,t i ,t Where, • Si,t = 401(k) contribution rate • NoKidsi,t = indicator for households that never had kids • KidsGonei,t = indicator for households with kids who have left home • Yi,t = income • Xi,t = vector of controls including education, race and ethnicity, presence of a mortgage. 7 …and (2) fixed effects model (HRS only). Si,t - S i = a2 KidsGonei, t + a3 (Yi,t -Yi )+ g (Xi,t - Xi)+ ui.t Where, • Si,t - S i = deviation from average 401(k) • KidsGonei,t = indicator for households with resident kids during sample but who have left home. • Yi,t -Yi = deviation from average household income. • Xi,t - Xi = deviation in controls (only controls that change over time, e.g. age, presence of mortgage). 8 HRS households whose kids have left home are similar to those with resident kids. Selected Characteristics of Married HRS Households With Kids Number of kids Age when oldest kid born Age at first observation Education Less than high school Some college Black non-Hispanic Hispanic Homeowner Has mortgage Kid status during observation period Never resident Resident throughout Left home 2.8 3.2 3.1 23.2 28.2 25.2 54.6 52.3 53.5 10.6% 49.6% 7.2% 3.4% 90.7% 65.8% 10.8% 61.0% 14.1% 13.7% 91.4% 79.0% 11.9% 57.1% 9.8% 7.3% 91.7% 73.5% Source: Authors’ calculations from Waves 1-10 of the Health and Retirement Study. 9 HRS results for pooled cross section show small effect of kids on contribution rate. Impact of Kids on 401(k) Contributions as Percentage of Salary Not resident Kids not in home Never had kids 0.898*** (0.199) -0.148 (0.484) Definition of kids leaving home Not resident and Not continuously resident or not in school in school 0.446** 0.753*** (0.190) (0.194) 0.407 0.609 (0.488) (0.490) Note: Significance is indicated at the 1-percent level (***) and 5-percent level (**). Source: Authors’ calculations from Waves 1-10 of the Health and Retirement Study. 10 HRS results with fixed effects generally show no effect. Impact of Kids on 401(k) Contributions as Percentage of Salary Not resident Kids were in residence, but left 0.381* (0.225) Definition of kids leaving home Not resident and Not continuously resident or not in school in school -0.046 0.118 (0.197) (0.220) Note: Significance is indicated at the 10-percent level (*). Source: Authors’ calculations from Waves 1-10 of the Health and Retirement Study. 11 SIPP results are consistent with the HRS cross-section analysis – a small effect when kids leaving defined by age. Impact of Kids on 401(k) Contributions as Percentage of Salary Kids not in home (Base = Res. kids) Youngest 19-22 (Base = 0-18) Youngest 23+ (Base = 0-18) Never had kids Definition of kids leaving home Based on residence Based on age of in household youngest child -0.094 (0.103) -0.292** (0.148) 0.320** (0.150) 0.497*** 0.518*** (0.107) (0.105) Note: Significance is indicated at the 1-percent level (***) and 5-percent level (**). Source: Authors’ calculations from the 2001, 2004, and 2008 Survey of Income and Program Participation. 12 What effect do models predict? • The HRS and SIPP suggest households increase 401(k) saving by 0.3 to 0.9 percent of salary when kids leave. • The increase in saving is small compared to that predicted by models that assume consumption cut when kids leave. • For example, consider a household with 2 kids making $100,000 and contributing 6 percent to 401(k): o Those models imply an increase in contributions to the deferral limit of 18 percent when kids leave. o This increase of 12 percentage points is much larger than that indicated by our results. 13 Caveats • This finding is not the final word on the subject. • Households could increase savings in other ways, for example by paying down debt or increasing mortgage payments. • The response may be lagged even more than is visible in the HRS. • Finally, parents might still support kids who are out of house and school. 14 Conclusion and next steps • Evidence of only small increases in 401(k) contributions when the kids leave home. • Will investigate other forms of savings/debt reduction, such as increases in non-401(k) saving. • Subject to this caveat, our findings are more consistent with the model used in the National Retirement Risk Index – many households may be at risk. 15