Executive Director, Center on Budget and Policy Priorities
before the House Ways and Means Subcommittee on Social Security
February 15, 2000

I appreciate the opportunity to testify before the Subcommittee today. I am Robert Greenstein, executive director of the Center on Budget and Policy Priorities here in Washington, D.C. The Center is a nonprofit policy institute that works on an array of public policy issues, with a particular interest in matters of fiscal policy and policy impacts on low- and moderate-income families. The Center receives no federal grants or contracts.


The Earnings Test and Social Security Reform

I approach the topic of today's hearing with some ambivalence. On the one hand, as explained below, there is a case for eliminating the Social Security earnings test as it is applied to individuals who have reached the age at which full Social Security benefits are paid (sometimes known as the "normal retirement age"). On the other hand, the best course would be to deal with the earnings test as part of broader Social Security reform legislation, rather than to move legislation eliminating this test separate from — and ahead of — a broader reform package that restores long-term solvency.

Restoring long-term solvency to Social Security will necessarily entail some reductions in benefits or increases in payroll taxes unless policymakers pour in so much money from the rest of the budget for a number of decades as to make any hard choices unnecessary, a course of action that would likely cause serious fiscal problems a few decades from now. This point holds true for both privatization and non-privatization approaches to restoring long-term Social Security solvency.

Eventually, we will have to make some tough choices. Doing so will not be easy politically. Having elimination of the earnings test as part of such a long-term solvency package would make the package more politically palatable. By contrast, separate action now to eliminate the earnings test, outside of a long-term solvency package, would likely make such a package somewhat less attractive and still harder to pass. It also is difficult to argue that changes in the earnings test should be made on their own but that other changes for which the evidence may be more compelling — such as changes in the Social Security widows' benefit to reduce the high rates of poverty that old widows face — must wait for long-term solvency legislation.

I recognize, however, that Congress may wish to proceed nonetheless to address the earnings test this year. Accordingly, the main body of this testimony addresses various issues related to the earnings test and to potential legislation to eliminate it.


Should the Earnings Test Be Eliminated?

I believe the answer to this question depends on how the earnings test is eliminated. Done in a sound way, elimination of the earnings test is likely to represent a positive, if modest, improvement. Done in an unsound manner, eliminating the earnings test would likely turn out to be a net negative, causing a significant increase in poverty among elderly widows.

Fortunately, the Ways and Means Committee has before it a piece of legislation that eliminates the earnings test in an appropriate manner. This is H.R. 5, introduced by Rep. Sam Johnson. If the Subcommittee resolves to move forward this year with earnings test legislation, H.R. 5 would be the bill to pass.


Misunderstandings About the Earnings Test

The earnings test is one of the most misunderstood aspects of Social Security. There are three common sources of misunderstanding about the earnings test:

  1. Although many refer to the Social Security earnings test, there are, in fact, two earnings tests, not one. Separate tests apply for individuals who have reached the normal retirement age and for individuals who begin to draw benefits early (i.e., before the normal retirement age, most commonly at 62). As discussed below, the two earnings tests raise very different issues. Decisions need to be made separately about what to do regarding each of the earnings tests.
  1. The earnings tests do not impose a tax on earnings. To be sure, a beneficiary's earnings do cause his or her Social Security benefits to be reduced. But on average, the Social Security benefits that are reduced while a beneficiary is working are restored after the beneficiary stops working or reaches age 70, whichever occurs first. Once the earnings test ceases to apply to a beneficiary, the monthly Social Security benefit the beneficiary receives for the rest of his or her life is increased above the level the beneficiary would have received if he or she had not been subject to the earnings test. For a beneficiary with average life expectancy, the extra benefits received after the earnings test ceases to apply will about exactly equal the benefits lost due to the earnings test.(1) The Social Security benefit structure is purposely designed to produce this result. (Technically, this result will be achieved for the earnings test above the normal retirement age starting in 2005, when legislation Congress passed to achieve this result is phased in fully. Today, the subsequent benefit increases compensate for most, but not all, of the benefits that a beneficiary with average life expectancy loses as a result of the earnings test.)
  2. The fact that the earnings test is not a tax on lifetime benefits — and does not reduce benefits on average — is the reason why eliminating it has no effect on long-term Social Security solvency. If the earnings test truly were a tax on benefits, eliminating it would increase total Social Security benefits paid and hence worsen Social Security's long-term financial picture.

  3. Elimination of the separate test that applies to those who begin receiving Social Security benefits early (i.e., before the normal retirement age) would significantly increase poverty among the very old. People who begin to draw Social Security benefits at age 62 now receive about a 20 percent lower monthly benefit for the rest of their lives than the benefit they would receive if they began drawing benefits at the normal retirement age. This actuarial reduction affects the benefits their widows later receive, as well. If the earnings test for early benefit receipt is eliminated, more people will begin claiming benefits at age 62, with the result that more beneficiaries — and ultimately more widows — will be receiving reduced monthly benefits while they are very old. Those who are very old often no longer have other income sources and may have largely exhausted their assets; many such individuals must live largely or entirely on their monthly Social Security check.

There is little question that early claiming of Social Security benefits at age 62 increases poverty. Economists Jonathan Gruber of M.I.T. and Peter Orszag of the University of California at Berkeley recently looked at the average Social Security benefits of very old widows. They found the average benefits of widows whose spouses had begun claiming Social Security benefits before 65 were below the poverty line, while the average benefits of widows whose spouses waited until the normal retirement age to collect benefits were nearly $2,000 above the poverty line. They also found that a significant portion of this difference in average benefits was due to the decision regarding whether to begin receiving benefits early or to wait until the normal retirement age.(2) Some preliminary work indicates that eliminating the earnings test for people who draw benefits early would increase the number of elderly poor people by some hundreds of thousands.

Research on the Effect of the Earnings Test on Work Effort

Economists Gruber and Orszag also recently completed a review of the research literature on the effect of the earnings test on decisions about whether to work. They reported that most of the academic literature "suggests that, contrary to popular impression, the test has little effect" on the degree to which seniors work. They noted, however, that one recent study suggests the earnings test does have a more significant effect on the work decisions of individuals who have passed the normal retirement age, although they cautioned that this study has methodological limitations. Their overall conclusion is that, based on the available research, "The earnings test has some, but a relatively modest, effect on overall work activity" for those above the normal retirement age.(3) Their review found no evidence in the research literature of any significant effect on work activity of the earnings test that applies to early retirees.


Policy Implications

The research suggests that among those who have reached the normal retirement age, eliminating the earnings test may cause a modest increase in work effort. The Social Security actuaries report that eliminating this earnings test would have no effect on long-term Social Security solvency. Eliminating this earnings test therefore seems a sensible policy step. That is what H.R. 5 does.

The separate test that applies to early retirees is a different story. There is no evidence that eliminating it would cause an increase in work effort. But eliminating this test would cause a significant increase in poverty among the very old, especially among widows in their 80s and 90s. In addition, the Social Security actuaries have reported that eliminating the earnings test for early retirees would move the date of Social Security insolvency forward from 2034 to 2033 and modestly reduce the trust fund's projected assets.

Eliminating the test for early retirees also would have one other adverse effect — it would encourage increased claiming of Social Security benefits at age 62. As we move toward the difficult demographic decades that lie ahead, when the U.S. population will be aging and Americans will be living longer, encouraging more people to start claiming Social Security at age 62 is one of the last things we should be doing.



Elimination of the earnings test is not so pressing a national need that it must be accomplished immediately. The more pressing need is fashioning Social Security solvency legislation. Solvency legislation inevitably will entail tough choices. Fashioning and enacting such legislation will be easier if earnings test elimination can be packaged with it. For this reason, the most prudent course is to wait.

If, however, Congress is intent on moving forward this year, it should adopt H.R. 5, which eliminates the earnings test for those reaching the normal retirement age. It would be most unwise also to scrap the earnings test that applies to early retirees, a step that H.R. 5 wisely does not take.

End notes:

1. Those with above average life expectancies are somewhat over-compensated; the application of the earnings test actually increases their lifetime benefits. Those with below-average life expectancies are somewhat under-compensated.

2. Jonathan Gruber and Peter Orszag, "What to do about the Social Security Earnings Test?," Center for Retirement Research, July 1999.

3. Gruber and Orszag, op. cit. Gruber and Orszag observe that one study (by Robbins and Robbins) concludes that the earnings test for beneficiaries who have reached the normal retirement age has very large effects on work effort. Gruber and Orszag note, however, that this study has been sharply criticized by Social Security experts, most notably in a critique published by the Office of Research and Statistics of the Social Security Administration, for a series of errors that make its results unreliable.