Social Security Benefits
The Social Security Earnings Limit and Work and Retirement Incentives
Research Report
Alison Shelton, AARP Public Policy Institute
November 2000
Table of Contents:
- Description of the Earnings Limit
- What Labor Supply Incentives Are Created by the Earnings Limit?
- Analysis of Work Disincentives Created by the Earnings Limit
- Conclusions
In March, 2000, Congress voted to abolish the Social Security earnings limit for workers over the age of 65. As a result, persons over the age of 65 will receive full Social Security benefits regardless of how much they earn in wages or salaries. The change was made retroactive to January, 2000. An earnings limit still applies, however, to people between ages 62 and 64 who collect Social Security and earn wages or salaries over a specified exempt amount.
Policy debates about the Social Security earnings limit must consider several different sets of issues. One issue concerns the distributional consequences of the incidence of the earnings limit. A second issue concerns the incentives that the earnings limit may create for claiming Social Security before the normal retirement age, and the effect that this could have on widows.1
The subject of this paper is the work and retirement incentives that may be created by the earnings limit. When Social Security was established during the Great Depression, the earnings limit was designed to encourage older workers to leave the workforce. No benefits were paid to anyone who received labor income. Since then, the rules have been relaxed to allow some labor income with a partial reduction in benefits. The impact of the earnings limit on work incentives has come under growing scrutiny, because of increasing life expectancy and the current low unemployment rates.
Description of the Earnings Limit
The earnings limit reduces Social Security benefits when a beneficiary's labor income exceeds a certain threshold, called the "exempt amount." The earnings limit applies only to wage and salary income above the exempt amount; it does not apply to pension, dividend, or interest income that may bring total income above the exempt amount, because the FICA tax is paid only on wages and salaries. The exempt amount changes annually based on wage growth. In recent years, the following wage or salary amounts were exempt from the earnings limit:
Until the law was changed in 2000, the earnings limit reduced the Social Security benefit by one of two calculations, depending on the age of the beneficiary.
- For ages 62-64, the Social Security Administration withheld $1 in benefits for every $2 in wages above the exempt amount.
- For ages 65-69, $1 was withheld for every $3 in wages above the exempt amount.
The 50-cents-on-the-dollar benefit reduction for ages 62-64 is still in place, while the earnings limit has been lifted for age 65 and over starting in 2000.
A point that is not widely understood is that individuals who lose Social Security benefits under the earnings limit will eventually recapture all of their lost benefits, once they reach normal retirement age and over the course of a long lifetime. Benefit recapture is accomplished by recomputing the monthly benefit when the beneficiary reaches normal retirement age. The period required for recapture of benefits is computed actuarially. Thus, the earnings limit resembles a program of enforced saving more than it does a tax, although whether this forced saving is efficient can be debated. The recapture feature may be inequitable to certain racial or income groups that tend to have shorter life spans. (Gruber and Orszag, 1999)
Top of PageWhat Labor Supply Incentives Are Created by the Earnings Limit?
The earnings limit benefit reductions for workers aged 62-64 and 65-69 have been perceived as 50 percent and 33 percent marginal tax rates, respectively, on earnings over the exempt amount. This perception persists among beneficiaries, even though benefit reductions are returned to the individual during the course of retirement.
The earnings limit could have two opposite types of potential effects on the labor supply of people who are currently in the workforce, given the apparently common perception that the earnings limit is a tax on labor income. First, some workers may reduce their work hours in order to hold their earnings just under the earnings limit exempt amount. If the earnings limit were abolished, we would expect to see an unambiguous increase in the work hours of this particular group.
On the other hand, abolishing the earnings limit could actually induce some people to work fewer hours, not more. For Social Security recipients with wages above the exempt amount, lifting the earnings limit would increase their benefit. For workers who had not yet claimed Social Security, lifting the earnings limit could induce them to begin receiving Social Security. Thus, more income from Social Security would enable some people to hold their total income constant while reducing their working hours, if they desired, and if they had the flexibility to change their work hours. We could potentially see a reduction in work hours among two groups in response to lifting the earnings limit: (a) higher-income workers who are losing some or all of their benefits as a result of the earnings limit; and (b) workers who had not yet applied for Social Security benefits.
Which of these two effects - increased or reduced working hours - would predominate if the earnings limit were abolished is not immediately obvious for some categories of workers. Moreover, not all workers have flexibility to change their work hours.
Friedberg (1999; 2000) calculated that if the earnings limit were eliminated, the combined result of these two effects would be to increase by 5.3 percent the average hours worked by all 65- to 69-year-olds located at or above the exempt amount. Jonathan Gruber and Peter Orszag pointed out that the group of 65 to 69-year-olds located at or above the earnings test level in 1998 worked 64 percent of all the hours worked by 65 to 69-year-olds; therefore the aggregate increase in work hours among this age group would be 3.4 percent (Gruber and Orszag, 2000). These results encompass both positive and negative effects on work hours. For example, Friedberg found that workers at higher wages — those who lose the entire amount of their benefit to the earnings limit — would work four percent fewer hours, on average, if the earnings limit were lifted.
Using both graphical and econometric analysis, Gruber and Orszag (2000) concluded that the earnings test exerts no statistically robust influence on the work hours of men aged 66-69, if the marked trend over time toward earlier retirement is included in the model. In an earlier study, Honig and Reimers (1989) estimated that eliminating the earnings limit would increase labor supply by the small amount of 1.4%.
The decision to retire is very different from the decision to work greater or fewer hours. The earnings limit may induce some people to retire completely from the workforce, so that they can receive full Social Security benefits. If the earnings limit were eliminated, these people might remain in the workforce longer. Similarly, some retirees might decide to re-enter the workforce if the earnings limit were eliminated. The specific effect of the earnings test on retirement among people aged 66-69 was examined by Gruber and Orszag (2000), who found that if the earnings test did play a role in mens' decision to work or retire, that role was not large. For women, raising the exempt amount or lifting the earnings test entirely resulted in a large increase in womens' earnings. The result for women appeared to be largely explained by an increase in the number of women working, combined with a smaller increase in womens' work hours, although the results are not statistically robust.
Analysis of Work Disincentives Created by the Earnings Limit
This section of the paper presents some empirical evidence on whether the earnings limit may induce some workers to cut back on their work hours. The set of analyses presented in Charts 1 and 2 updates work that was originally conducted by Friedberg (1999; 2000).
The analysis presented here focuses exclusively on behavior around the earnings limit exempt amount. If some people respond to the earnings limit by reducing their work hours so that their wage or salary falls just under the exempt amount, then we would expect the data to show a "bunching" of individuals right below the earnings limit exempt amount. We would not expect to see this bunching for workers who do not yet collect Social Security.
These analyses use data from the Current Population Survey (CPS) conducted during the years 1996, 1997 and 1998. Three CPS survey years were used to amass sufficient observations to make the results statistically meaningful. For the purposes of the analyses that follow, state and local government workers were dropped from the CPS dataset, because some state and local governments do not participate in the Social Security system. Additionally, people who reported that they retired during the survey year were also dropped from the dataset. This was done because the presence of mid-year retirees could result in a large number of observations with wages below the earnings limit exempt amount, and muddle the behavioral interpretation of these lower wages. While some people may time their retirement so as to stay under the earnings limit exempt amount, it is likely that many more choose to retire at a particular point in the year because of poor health, the need to care for an ill spouse, or some other personally important event.
Following the methodology of Gruber and Orzag (2000), anyone who turned 62 or 65 during the survey year was also dropped from the dataset, because of ambiguity in respondents' ages. Finally, the analysis considers only men, because the data do not distinguish whether working spouses are getting benefits in their own right, or as dependents.
Charts 1 and 2 look for "bunching" below the earnings limit exempt amount among workers who also receive Social Security benefits. Workers are grouped into wage brackets of $1000 above and below the exempt amount. Chart 1 looks at workers aged 63-64. Chart 2 looks at workers aged 66-69. The shaded bars on the left of each pair represent workers who do not yet receive Social Security. The white bars on the right of each pair represent people who do receive Social Security and who may be subject to the earnings limit, if they have wage or salary income above the exempt amount.
Chart 1 shows some bunching of workers aged 63-64 who receive Social Security, right below the earnings limit exempt amount (the dotted vertical line). 2.8% of workers aged 63-64 are "bunched" at an income range that is between one dollar and $1,000 dollars below the dotted line representing the exempt amount. A further 2.1% of the group is bunched at an income range of $1,001 to $2,000 below the exempt amount. In contrast, only 1.6% of Social Security recipients had wage or salary income of up to $1000 above the exempt amount. Workers who do not yet receive Social Security, and hence would not be subject to the earnings limit (the solid bars), show no bunching below the exempt amount.
Statistical analysis shows that we can reject, with 99 percent confidence, the hypothesis that the equal numbers of workers age 63-64 are grouped in the wage increment $1000 above the earnings test as in the wage increment $1000 below the earnings test.
Chart 2 shows that bunching below the exempt amount also occurs for the 66-69 age group. 4.3% of workers who also received Social Security had wage or salary income within $1000 below the exempt amount, as compared to 2.6% of workers who had wage or salary income within $1000 above the exempt amount and also received Social Security. For these figures we can also reject, with 99 percent confidence, the hypothesis that equal numbers of workers age 66-69 are grouped in $1000 wage increments above and below the earnings test.
Conclusions
The earnings limit would appear to cause a small but measurable group of workers to reduce their labor supply so as to keep their total wage or salary earnings under the earnings limit exempt amount. Therefore, if the earnings limit were lifted, we would expect to see an increase in labor supply among some, although not all, of the workers who were "bunched" below the exempt amount. Some of the workers in this group would not increase their labor supply, because their full-time wage or salary income naturally fell into this wage bracket. It is also likely, however, that other workers would reduce their labor supply if the earnings limit were eliminated.
Several researchers have tried to estimate the net effect of removing the earnings limit. Friedberg (1999; 2000) found that if the earnings limit were lifted for workers age 65 and over, the work hours of the affected group would increase by 5.3 percent. Other researchers have found negligible, or very small, overall effects.
Footnote
1 These issues are discussed in separate Data Digests.
References
Friedberg, Leora, "The Social Security Earnings Test and Hours of Work," Testimony before the House Ways and Means Committee of the U.S. House of Representatives, February 15, 2000. See also Friedberg, Leora, "The Labor Supply Effects of the Social Security Earnings Test," NBER Working Paper No. 7200, June, 1999.
Gruber, Jonathan and Peter Orszag, "What to do About the Social Security Earnings Test," Center for Retirement Research, Boston College, July 1999.
Gruber, Jonathan and Peter Orszag, "Does the Social Security Earnings Test Affect Labor Supply and Benefits Receipt?" September, 2000.
Honig, Marjorie and Cordelia Reimers, "Is it Worth Eliminating the Earnings Test?" American Economics Association, Papers and Proceedings, May 1989.
Written by Alison Shelton, AARP Public Policy Institute
November 2000
©2000 AARP
May be copied only for noncommercial purposes and with
attribution; permission required for all other purposes.
Public Policy Institute, Public Affairs, AARP, 601 E Street, NW,
Washington, DC 20049
Pub ID: DD54