How Social Security's earning test, age and education affect the female labor supply

In the past, benefits of social security recipients may have been reduced if they earned too much income.  Since its inception in 1935, social security benefits have been subject to an “earnings test.”  Though the particulars of the law have changed over time, the earnings test taxes labor income and reduces the incentive to work. 

     Before the law was changed in 2000, the earnings test consisted of a lower and upper threshold.  If Social Security recipients’ wages were less than the lower threshold, their benefits were not reduced.   However, if the retirees’ income was between the lower and upper thresholds, they lost $1 of benefits for every $3 of wages they earned.  Finally, if recipients’ labor income exceeded the upper threshold, they lost all their benefits.  The earnings test was applied to all recipients, those who retired before their normal retirement age (NRA) and those who retired upon or after their NRA.  After 2000, the earnings test was no longer applied to retirees after the month they reached their NRA.

      This paper’s analyzes how this policy change affected the labor supply of married and single women.  Exogenous variables include marital status, age, ethnicity and the level of education.  If spouses’ wages are no longer taxed after they reach their NRA, household income increases and married women may work less.  If the income of single women is no longer taxed after they reach their NRA, their opportunity cost of leisure increases and they may work more.  Given their increased participation in the labor market, black, female retirees may be more likely to work than white women.  Women with more education would earn relatively more income and would be more likely to work given their higher opportunity cost of leisure.  Older women may be more likely to retire.

     This paper uses the difference-in-differences technique to quantify the effect of a policy change on economic behavior.  The data are monthly observations from the Current Population Survey.  Past papers have concentrated on how the earnings test affected the male labor supply and they have found mixed results.  The emphasis on the female labor supply is one of this paper’s contributions.  Also previous studies on the effects of the policy change in 2000 had relatively few observations after the regime change.  This study has more observations after the regime change and it includes the effects of the Great Recession, another contribution to the literature.