Nadia S. Karamcheva (Urban Institute), April Yanyuan Wu ( Mathematica Policy Research),

1. Does Social Security Continue to Favor Couples? Nadia S. Karamcheva (Urban Institute), April YanyuanWu (Mathematica Policy Research), and Alicia H. Munnell (Center for Retirement Research at Boston College) 16th Annual RRC Meeting Washington, DC August 8, 2014

2. Changing role of women: labor supply Labor Force Participation, by Marital Status Source: Authors calculations using U.S. Social Security Administration Modeling Income in the Near Term, Versions 5 and 6. Washington, DC.

3. Changing role of women: earnings Ratio of Median Wife’s to Husband’s Lifetime Earnings Source: Authors calculations using U.S. Social Security Administration. Modeling Income in the Near Term, Versions 5 and 6. Washington, DC.

4. Changing role of women: marital patterns Percent of Women Married, by Age - - - - - Projected Source: Authors calculations using U.S. Social Security Administration. Modeling Income in the Near Term, Versions 5 and 6. Washington, DC.

5. Changes in women’s benefit eligibility at first claiming Source: Authors calculations using U.S. Social Security Administration. Modeling Income in the Near Term, Versions 5 and 6. Washington, DC.

6. Despite gains in work years and earnings, the majority of current and future female retirees would still receive a higher benefit as a survivor Percent of Married Women Who Would Receive Higher Benefit as a Survivor Source: Authors calculations using U.S. Social Security Administration. Modeling Income in the Near Term, Versions 5 and 6. Washington, DC.

7. Research question • Does Social Security continue to favor couples if viewed from a lifetime as opposed to point-in-time measure? That is, do lifetime benefit/tax ratios yield a different answer than initial replacement rates?

8. Research approach The analysis will: 1) Examine trends in the ‘returns’ from OASI : • Three measures: benefit/tax ratio, net tax rate, share receiving positive transfers from OASI • Cohorts: Depression to Generation X • By gender, marital status and income distribution 2) Decompose differences into contributing factors

9. Preview of results • Decline in median individual benefit/tax ratios • 27 percent between the 1930s cohort and GenXers • Decline in median household benefit/tax ratios • 21percent between the 1930s cohort and GenXers • Changes vary by gender, martial status and income distribution • Factors explaining the decline in the returns from the system • Increased labor supply and earnings: (from 46% to 75 %) • Wife’s increased labor force activity explains >1/2 of the decline for couples

10. Data • Modeling Income in the Near Term (MINT) • Depression Era 1(DE1, 1931-1935) • Depression Era 2(DE2, 1936-1941) • War Baby (WB, 1942-1947) • Early Baby Boomers (EBB, 1948-1953) • Middle Baby Boomers (MBB, 1954-1959) • Late Baby Boomers (LBB, 1960-1965) • Generation X (GX, 1966-1975)

11. Three Measures of Returns • Lifetime benefit/tax ratio = , taxes include employee+employer share • Net tax rate = • Net transfer = 1 if Net tax rate<0, 0 otherwise

12. Changes in benefit/tax ratios – by Gender Median Benefit/tax ratios, Individual level Source: Authors calculations using U.S. Social Security Administration. Modeling Income in the Near Term, Versions 5 and 6. Washington, DC.

13. Changes in benefit/tax ratios –by Marrital Status Median Benefit/tax ratios, Household level Source: Authors calculations using U.S. Social Security Administration. Modeling Income in the Near Term, Versions 5 and 6. Washington, DC.

14. Changes in median net tax rate - Household Level Median net tax rate, Household level Source: Authors calculations using U.S. Social Security Administration. Modeling Income in the Near Term, Versions 5 and 6. Washington, DC.

15. Changes in share with positive net transfers - Household Level Share with positive net transfers, Household level Source: Authors calculations using U.S. Social Security Administration. Modeling Income in the Near Term, Versions 5 and 6. Washington, DC.

16. Decline is largest for couples with husbands’ earnings in top tercile DE 1 vs Gen X cohort, Household level Source: Authors calculations using U.S. Social Security Administration. Modeling Income in the Near Term, Versions 5 and 6. Washington, DC.

17. Explaining differences over time • Factors contributing to the changes over time: • Labor force participation • Marriage pattern • And…changes in FRA and claiming behaviors

18. Explaining differences between DE 1(1931-1935) and Gen X (1966-1975) cohorts: Percent of the overall change in median benefit/tax ratio explained by differences in the characteristics between earliest and latest cohort Source: Authors calculations using U.S. Social Security Administration. Modeling Income in the Near Term, Versions 5 and 6. Washington, DC. Note: We used DiNardo, Fortin and Lemieux (1996) semi-parametric re-weighting method to explain changes in Medians

19. Explaining differences between DE 1(1931-1935) and Gen X (1966-1975) cohorts (cont.) Among women, changes in labor force activity account for more than 2/3 of the difference in net tax rate and share with positive transfers. Decomposition: Individuals Source: Authors calculations using U.S. Social Security Administration. Modeling Income in the Near Term, Versions 5 and 6. Washington, DC. Note: We used Oaxaca-Blinder decomposition to explain changes in Means, and its application to binary outcomes developed by Yun (2004).

20. Explaining differences between DE 1(1931-1935) and Gen X (1966-1975) cohorts (cont.) Wife's labor force activity accounts for more than ½ of the change among married couples. Decomposition: Married Couples Source: Authors calculations using U.S. Social Security Administration. Modeling Income in the Near Term, Versions 5 and 6. Washington, DC. We used Oaxaca-Blinder decomposition to explain changes in Means, and its application to binary outcomes developed by Yun (2004).

21. Conclusion • While the OASI system continues to redistribute lifetime income from singles to couples and from men to women, the transfers appear to be shrinking over time • Women’s increased labor force participation and earnings have contributed significantly to the decline • Changing marital patterns have significant but small impact • Sensitivity analysis with alternative discount rates show similar results. Next Steps External validity check using HRS data Detailed decomposition of the changes in median ben/tax ratio and median net tax rate using DFL (1996) approach. 20