Special Studies

Special Study #7:

The History and Development of the Social Security Retirement Earnings Test

by Larry DeWitt,
SSA Historian

August 1999

In his 1999 State-of-the-Union address, President Clinton announced his support for the idea of eliminating the Social Security Retirement Earnings Test (RET). The RET is unpopular with beneficiaries who want to have their retirement benefits and continue to work. Eliminating the RET has long been a favorite proposal in Congress. With the President's support, it is likely that the RET will soon undergo historic change. This paper examines the role and development of the RET; how it came to be part of the Social Security program, why it was created and how it evolved over the years.

The Social Security Act of 1935

The Social Security Act of 1935 contained two programs of economic security for the aged: Title I of the Act provided non-contributory, means-tested, old-age pensions, in the form of state welfare programs with federal funding. Title II of the Act, what is now thought of as Social Security, provided old-age insurance through a contributory social insurance scheme. The old-age insurance program required retirement from gainful employment as a condition of benefit receipt. For wage earners, this requirement was and is measured primarily by a test of earnings levels, hence the RET. The RET is simply the administrative form of the principle that one must be retired in order to collect retirement benefits from Social Security's old-age insurance program. The exact form of this requirement has changed considerably over the years, as we shall see, but the general principle has characterized the program from its inception.

The Retirement Test

In 1991 the House Ways & Means Subcommittee on Social Security held a hearing on the Social Security retirement test. Many of the witnesses testified in favor of repealing the RET, and several agreed with the observation of a senior Congressman from Arizona who stated: "Social Security, when it was created in 1935, sought to achieve two goals-moving older workers out of the work force to make way for younger workers, and to partially replace lost income due to retirement. Those goals were applicable in 1935, but are not in 1991."(1) Around the same time, a member of Congress from California authored a column in the Los Angeles Times in which he stated: "So how did this regressive and unproductive test become policy? It was born out of Depression-era reforms in 1935 to create jobs during the most grim economic period in U.S. history. Millions of older Americans were discouraged from staying employed. The earnings test reduced their overall income, making it unproductive for them to work beyond retirement age. This meant younger workers with families could have jobs."(2) A recent monograph for the Third Millennium group put it this way: ". . . the paramount short-term factor in Social Security's birth-the Great Depression . . . For the short term, it was considered imperative to get the elderly out of the job market. . . Social Security still carries out this Depression-fighting aim, despite the fact that there is no longer any goal to be served by pushing seniors out of the workforce." (3)

This idea about the origins of the RET is very common and is often advanced in discussions about the RET.(4) Essentially, it attributes the rationale for the RET to a set of unique economic circumstances prevailing during the Great Depression and to a supposed industrial policy of removing older workers from the workforce during that period. The gist of the argument is that since this set of economic circumstances no longer obtains, and the industrial policy objective is no longer desirable, there is no present rationale for the inclusion of the RET in the Social Security program.

The problem of the "superannuated worker" is a traditional one in industrial policy and pension theory. Since the onset of the Industrial Age, older workers began to find themselves at the end of their productive working lives, often without means of support. Such older workers often tried to remain employed and employers sometimes tried to find ways to move them out of the workforce. So the idea that this industrial policy objective underlies the RET, has an initial plausibility. But it is, as we will see, largely a historical myth which has grown over time and assumed the status of "common knowledge."

The retirement decision is a complex, multi-variate phenomenon, involving considerations of health, lifestyle, workplace circumstances and psychology, as well as economics. Whether any single economic incentive, such as the RET, could have a major impact on this decision is still an open question, with the available evidence inconclusive.(5) In any case, that question is beyond the scope of this paper. We are concerned only with whether the RET could plausibly have provided such incentive effects in the context of the mid-1930s, and whether or not such incentive effects were part of the intentions and goals of the program's designers.

The Purpose of the Retirement Test

Under social insurance theory, the rationale for the RET is simple and straightforward: one must retire in order to receive a retirement benefit because loss of earnings due to retirement is the insured condition. The risk-sharing inherent in insurance schemes rests on this principle. Under the alternative interpretation, the purpose of the RET is to force a choice on the aged worker: keep working and forgo the benefit or leave the workforce in order to collect a Social Security retirement benefit. According to the hypothesis, the RET was introduced into the Social Security Act of 1935 because the designers of the program were motivated, at least in significant part, by the industrial policy objective of removing older workers from the workforce in order to make room for unemployed younger workers.

This hypothesis is closely tied to the economic and social circumstances of the Depression. We can conceive of times when the demographics of the workplace are quite different. What becomes of this idea when there is low unemployment, high productivity, low numbers of older workers in the workforce and a small pool of willing older workers outside the workforce seeking to get back in? In other words, how would we view the RET in a world where we had no need to either encourage or discourage workforce participation by older workers? Presumably, there would be no need either to have or not have a RET in such a world, from the perspective of industrial policy. And yet, from the perspective of social insurance theory, we might still argue that the principle of the RET is applicable. We might still argue that both on grounds of risk-sharing, and of philosophical principle, that we should not pay insurance benefits to covered individuals who have not suffered the requisite loss of income.

This points up an important distinction. If superannuated workers remain in the workforce because they have no alternative means of income security, then providing a retirement pension fulfills an unmet social need. Certainly, much of the rhetoric of the social insurance movement suggests this as a motive. Also, the social insurance movement sought to prevent dependency in old age, meaning to prevent the social catastrophe of older workers leaving the workforce, for whatever reasons, without adequate means of post-employment income security. This too was a major part of social insurance theorizing and a major part of the rhetoric of the creators of the Social Security program. Neither goal has anything to do with industrial policy. On the other hand, if older workers remain in the workforce because they do not want to retire, but the offer of a retirement benefit induces them to do so, then this is an expression of industrial policy. So the question is, which of these visions of the world did the designers of the Social Security program have in mind in 1934-35?

The Depression and the Potential Impact of Social Security

The notion that the RET was put in the Social Security Act of 1935 in order to move older workers out of the workforce, rests on four premises:

  1. In the early 1930s there were large numbers of older workers in the workforce;
  2. There were large numbers of young unemployed workers who were being blocked from job opportunities by the presence of older workers;
  3. The provisions of the 1935 Social Security Act would serve as an effective incentive encouraging older workers to leave the workforce;
  4. This industrial policy objective was a major reason for adopting the RET in 1935.

We can begin by observing that the industrial policy interpretation of the RET is problematic in its first three premises.

In the first place, the aged had already been forced out of the workforce in disproportionate numbers by the economics of the Depression. Since older workers are generally the highest-paid workers in a business and often are thought to be less productive than younger workers, in many industries they were the first ones laid-off in the Depression. They were also the least likely to be re-hired for the same reasons. As a consequence, workforce participation for older workers was much lower than for younger workers.

The available data on unemployment during the early years of the Depression is severely limited. Before the creation of the unemployment insurance program, which became the institutional locus for the collection and dissemination of such data, the main statistics were produced by the decennial U.S. Census. The 1930 Census gathered data on unemployment, and produced two volumes of reports on the problem.

In 1930, according to the Census data, workers age 65 or older were barely 4.5 percent of the workforce (see Table 1).

Table 1: Workforce Percentage by Age Cohort (6)
Age Cohort 10-15 15-24 25-34 35-44 45-54 55-64 65-74 75+
Number 392,988 11,442,237 11,823,004 10,500,540 7,831,161 4,590,592 1,869,944 325,023
Percentage 0.8 23.4 24.2 21.5 16.1 9.4 3.8 0.7

The number of aged workers outside the workforce seeking to get back in was also very small. Of the 3,187,647 workers in 1930 out of work and seeking employment, barely 133,815 were age 65 or older.(7)

Older unemployed workers also stayed out-of-work much longer than younger workers (see Table 2).

Table 2: Time Unemployed Workers Out-of-Work, by Age Cohort (8)
  16-24 25-44 45-54 55-64 65 +
Under 6 mos. 25.4% 16.0% 12.0% 10.1% 7.7%
6-11 mos. 22.5 16.2 13.1 10.8 10.9
12-23 mos. 20.6 19.1 17.3 17.8 13.9
24-35 mos. 14.6 18.3 20.0 18.8 19.8
36-47 mos. 9.0 13.2 16.6 17.3 16.9
48 mos.& over 7.9 17.2 21.0 25.2 30.8


So older workers were not a major factor in blocking opportunities for younger workers or in competing with them for the job openings which did occur. So while we can perhaps conceive of circumstances in which large numbers of entrenched older workers are blocking workforce entry to younger workers, this was not in fact the situation in the 1930s, as far as the data available to program planners in 1935 indicated.

An even simpler point can be made about the ineffectiveness of the supposed incentives in the retirement program. In 1932 the average wage was $100 per month while the average Social Security benefit under the 1935 Act was expected to be $17.50 per month. Very few workers, who did not otherwise want or need to retire, would have traded a job for this retirement benefit. Moreover, benefits under the old-age insurance program were low and would remain at very low levels for many years to come. It was 1951 before the average Social Security retirement benefit exceeded the average old-age welfare pension benefit. So even if, in present circumstances, Social Security retirement benefits offer strong economic incentives encouraging retirement, this was not true in the early years of the program.

Another problem comes from the simple observation that the contributory old-age insurance system could not have an immediate impact on any workers since benefits were not scheduled to begin until 1942. The basic idea of the old-age insurance system was that it would build old-age security gradually, for future generations, so that they would not find themselves impoverished in the same way that the already aged were. As historian Edward Berkowitz said of the old-age insurance program: "It was not about the current depression so much as the next one."(9) It was the immediate relief provided by the Title I old-age pensions which was designed to help those already old in 1935. So if any provision of the 1935 Act could possibly have had an immediate impact on workforce participation rates of the elderly it would have been the non-contributory old-age pensions authorized in Title I of the Act, not the Title II Social Security benefits.

Finally, it is well to recall that the original Social Security program covered fewer than half the workers in America and the RET only applied to earnings in covered employment. Jobs in non-covered employment were therefore a ready alternative to older workers who might have wished to work and receive a Social Security retirement benefit. Non-covered earnings were first counted in the RET starting in 1954 and up to that point still about 40 percent of the available jobs were in non-covered employment. It thus seems implausible to suppose that the program's designers could have expected such a porous incentive to have much real labor force impact.

Despite these obvious shortcomings of the old-age insurance program as a mechanism for industrial policy, it is possible that, as a matter of historical fact, the founders of the program intended it to accomplish this objective. Perhaps they believed that one laudatory effect of the RET would be to provide an incentive for older workers to leave the labor force. Was this their intention? To answer this question, we must carefully examine the documentary record, including statements by the program's designers and the various official documents involved. We will see that there is some evidence which could be construed in support of this thesis, but the overwhelming evidence suggests this industrial policy objective had little or nothing to do with inclusion of the RET in the original Social Security Act of 1935.

Social Insurance in America

The modern approach to retirement security emerged out of the social insurance movement which began in Europe in the late 19th century. It was the European social insurance systems which were the starting point for the old-age insurance provisions of the Social Security Act. The first modern contributory social insurance program was designed by Germany's Iron Chancellor, Otto von Bismarck in 1881. By the time the U.S. was considering social insurance in 1934-35, as many as 34 European nations had existing social insurance programs. A key point to notice about these European systems is that they were designed and implemented long before the Depression. So the need to encourage older workers to retire to make room for unemployed younger workers, a presumed problem of the Depression, was not behind the design decisions made by the creators of these early systems.(10)

The European idea of social insurance began to be felt in America by the turn of the century. Probably the two most influential of the social insurance theorists were Abraham Epstein and I. M. Rubinow. Two other prominent academic experts in this area were J. Douglas Brown of Princeton University and Barbara Armstrong of the University of California. (11)

In 1913, Rubinow published his classic book on the subject, Social Insurance. This book influenced a whole generation of social insurance theorists. Rubinow's work was known to FDR who considered Rubinow to be the "greatest single authority upon social security in the United States." (12) According to Rubinow:

"All insurance is a substitution of social, co-operative provision for individual provision. Technically, this substitution of social effort for individual effort, is known as the theory of distribution of losses and the subsequent elimination of risk. . . the social advantages of distribution of loss are equally applicable to all forms of insurance, to commercial insurance as well as social insurance . . (13)

For the social insurance theorists, all social insurance benefits operate on the insurance principle, which requires that one receives the promised benefit when one suffers the insured condition. Social insurance insures against loss of income due to what President Roosevelt would come to call the economic "hazards and vicissitudes of life." So from the point of view of social insurance theory, the RET is simply an expression of the insurance principle in operation. There are very few explicit statements regarding the RET itself in the documentary record; but to the extent that Social Security's designers conceived of it as operating under insurance principles, it makes it likely that RET was put in the program to comport with these principles.

There were schemes in the Depression which did have a public policy goal of removing older workers from the workforce. The most prominent example was the Townsend Plan. The Townsend Plan wanted the government to provide a pension of $200 per month to every citizen age 60 and older, funded by a 2 percent national sales tax. Dr. Townsend's agenda definitely included the idea of encouraging older workers to step aside to make room for younger workers. The cover of the millions of copies of the Townsend Plan distributed throughout America contained the Plan's slogan: "Youth for Work-Age for Leisure."

The Townsend Plan was very influential in the 1930s. In fact, the Townsend Plan was one of the motivating factors inducing President Roosevelt to announce his own social insurance plan. FDR is reported to have observed: "We have to have it. . .The Congress can't stand the pressure of the Townsend Plan unless we are studying social security, a solid plan which will give some assurance to old people of systematic assistance upon retirement." (14) It is plausible to suspect that confusion between Social Security and the Townsend Plan is the original source of the persistent idea that Social Security had the objective of removing older workers from the workforce. (15)

There was also a movement for company pensions in the industrial sector in America starting at the turn of the century. (16) Industrial pensions are an interesting case because a familiar idea in early industrial policy was that the provision of retirement pensions had the laudatory effect of encouraging superannuated workers to retire, making room for the entry and promotion of younger, presumably more productive, workers. The industrial pension movement of the early 20th century was "sold" to employers, in part, on the basis of this idea. This "selling point" was needed because industry had to have special incentives-beyond the provision for unmet social needs-because to offer retirement pensions often put them at competitive disadvantages in relation to their competitors who did not bear this cost of doing business.

One clear contemporary example of a public retirement system which had an objective of removing older workers from the workforce was the Railroad Retirement System (RRS). Originally enacted in 1934, the RRS contained a provision for mandatory retirement at age 65, with possible one-year extensions to age 70 if both the worker and the employer agreed. The 1934 act was declared unconstitutional and two weeks after signing the Social Security Act into law President Roosevelt signed the new 1935 version of the RRS into law. Under the 1935 version, retirement at age 65 was not mandatory, but the worker's annuity was reduced 6 2/3 percent for every year he delayed retiring after age 65. This was intentionally done to encourage older workers to leave the workforce, making room for younger workers. This was a special problem in the railroad industry, where seniority rules put younger workers at insurmountable disadvantages.

Clearly, when public policy makers were intent on the objective of encouraging retirement of older workers, such as in the RRS, they knew how to do so, and the programs they designed contained explicit provision for this goal. That no similar features are found in the Social Security Act of 1935 again suggests that this was not one of the objectives of the program's designers.

The Committee on Economic Security

To design his social insurance proposal, President Franklin D. Roosevelt appointed a five-member cabinet-level Committee on Economic Security (CES), chaired by Secretary of Labor, Frances Perkins.(17) To support the work of the CES, an Executive Director was appointed (Edwin Witte of the University of Wisconsin) and a Technical Committee was formed, consisting of experts on loan from other Federal agencies and a handful of outside consultants. The old-age insurance plan was drafted by three consultants on the Technical Committee: Barbara Armstrong (a law and economics professor from the University of California); J. Douglas Brown (an economist from Princeton); Murray Latimer (chairman of the Railroad Retirement Board); with the help of three staffers: Otto Richter, a senior actuary; Marianne Sakmann, a research assistant; and Robert J. Myers, a junior actuary. These individuals are the principal authorities on the purposes of the provisions of the Administration's social insurance proposals.

The CES generated a ten-volume set of studies and reports, which was never published, along with a final summary report to the President which was 74 printed pages. The President transmitted to Congress the CES Report, a draft Economic Security Bill, and a covering message. In 1937, two years after passage of the Social Security Act, the Social Security Board published a 592-page book (Social Security in America), summarizing the 10 volumes of unpublished material. So, these documents--the unpublished 10-volumes of studies, the 74-page report, the draft administration bill, the President's cover message to Congress and the 1937 book--constitute the primary source documentation of the intentions of the program's designers on the CES.

A key figure in this discussion is of course President Roosevelt. The Social Security Act was his legislative initiative and it was, according to Frances Perkins, the proudest accomplishment of his administration. We must give primary weight to FDR's conception of what Social Security was designed to do. FDR was of course vitally concerned about the problems of relief. He wanted to get the nation back to work and he deployed many strategies to accomplish this, from the Civil Works Administration, to the Works Progress Administration, to the Civilian Conservation Corps. He was also concerned to provide immediate relief to the needy elderly. But in everything he said he made a clear conceptual distinction between the objectives of these various relief activities and the objectives of his old-age insurance proposals.

Even before becoming President, FDR was an advocate for social insurance. In 1931, in commenting on New York state's old-age pension program which he had recently signed into law, he made these observations:

". . . I am not satisfied with the provisions of this law. . . Our American aged do not want charity, but rather old age comforts to which they are rightfully entitled by their own thrift and foresight in the form of insurance. It is, therefore, my judgment that the next step to be taken should be based on the theory of insurance by a system of contributions commencing at an early age." (18)

In his message to Congress in June 1934, the President announced his intentions regarding Social Security:

"Next winter we may well undertake the great task of furthering the security of the citizen and his family through social insurance. . . I am looking for a sound means which I can recommend to provide at once security against several of the great disturbing factors in life--especially those which relate to unemployment and old age. . . . I believe that the funds necessary to provide this insurance should be raised by contribution rather than by an increase in general taxation. Above all, I am convinced that social insurance should be national in scope . . . leaving to the Federal Government the responsibility of investing, maintaining and safeguarding the funds constituting the necessary insurance reserves." (19)

In his transmittal letter to the Congress in January 1935 FDR summarized the essence of the proposal he was sending them:

"In the important field of security for our old people, it seems necessary to adopt three principles: First, non-contributory old-age pensions for those who are now too old to build up their own insurance. It is, of course, clear that for perhaps thirty years to come funds will have to be provided by the States and the Federal Government to meet these pensions. Second, compulsory contributory annuities which in time will establish a self-supporting system for those now young and for future generations." (20)

On the other hand, FDR did have some notion about the Social Security proposal's potential effects on the labor market. His strongest statement on the issues came in the first of his famous "fireside chats":

"The program for social security now pending before the Congress is a necessary part of the future unemployment policy of the Government. . . . It proposes, by means of old-age pensions, to help those who have reached the age of retirement to give up their jobs and thus give to the younger generation greater opportunities for work and to give to all a feeling of security as they look toward old age. . . . Provisions for social security, however, are protections for the future." (21)

Notice, however, that in this quotation the President refers to "old-age pensions" as having the hoped-for workforce effects, not the old-age insurance program. So he could not be referring, even obliquely, to the RET. It was the understanding of the program's designers that the old-age insurance program was intended to benefit not those already aged, but the younger worker who would be in a better position in the future to avoid the kind of impoverishment which the aged workers of the Depression-era were experiencing. The President and the CES believed the non-contributory Title I program (old-age assistance) was the only part of the Act which was designed to have an immediate impact on aged workers. As the Report of the CES summarized the intent of their proposal:

"An adequate old-age security program involves a combination of non-contributory pensions and contributory annuities. Only non-contributory pensions can serve to meet the problem of millions of persons who are already superannuated or shortly will be so and are without sufficient income for a decent subsistence. A contributory annuity system, while of little or no value to people now in these older age groups, will enable younger workers, with the aid of their employers, to build up gradually their rights to annuities in their old age." (22) In the 1937 book version of the work of the CES, we find the only explicit evidence for the industrial policy rationale for the RET:

"Since retirement from regular employment could be made a condition for receipt of benefits, the displacement of superannuated workers from the labor market would be accelerated. . . If old-age insurance benefits are paid to persons continuing in regular employment after attainment of age 65, they will become, in effect, a subsidy to older workers in competing in the labor market. Since the employment and wage rates of younger workers would be adversely affected by such subsidized competition, it was felt that old-age insurance benefits should be suspended in those cases where otherwise qualified persons were regularly employed. It was believed that the social advantages of encouraging retirement at age 65 far outweighed the objection that individual equities would be reduced when benefits were thus suspended. . . . In any event, the main purpose of the plan is to provide a partial compensation for the loss of earned income." (23) (Emphasis added.)

It will be useful to unpack this language a bit. The passage asserts three propositions:

  1. The RET will encourage older workers to leave the work force;
  2. The absence of the RET would encourage older workers to stay in the work force;
  3. The primary purpose of the RET is the social insurance principle of replacement of lost income.

We have already shown that proposition 1) is false, and we are discussing proposition 3) at some length. Let us consider, in passing, proposition 2).

As far as economic incentives are concerned, it is a more plausible argument to claim that paying retirement benefits to older workers, even if they continue working, would create a subsidy, enabling them to stay in the workforce. This is especially true when we consider that a common practice during the Depression-era was for employers to try and maintain jobs for their workers by reducing their pay or by splitting one job between two or more workers. In such an environment, a subsidy paid to older workers would make it that much easier for them to stay in the workforce. But to observe that the absence of a RET would have likely labor market effects in one direction, is not the same as concluding that its presence would have labor market effects in the other direction. A small subsidy to a wage earner could have incentive effects in encouraging continued employment whereas that same small amount might not induce someone to quit working and rely on it for their primary source of income. (24)

It is important to note that the 1937 book was not a strict summary of the unpublished CES studies, but was a partial summary, an updating, and a rationalization of the Social Security program in the light of the legislative changes introduced during Congressional consideration of the original CES proposals. None of the language quoted above is in the original CES Report or anywhere in the 10 volumes of unpublished studies. It appears for the first time two years after passage of the Act.

Following the introduction of the Administration's bill in Congress, the staff of the CES was tasked with developing additional papers in support of the proposals. One such paper, authored by Murray Latimer and included in the unpublished studies sometime in early 1935, explicitly made the argument that the old-age insurance program in general would have the suggested industrial policy effects:

"It has been said that the several titles of the Economic Security Act have no relation one to the other. . . . I wish specifically to call attention . . . to the connection between old-age insurance provisions and the unemployment compensation provisions. The population of this country is rapidly growing older. Unemployment among the aged group has been increasing for many years. . . . But many aged persons have remained in employment. By so doing they have pushed out of employment many younger employees. . . An old-age insurance program will solve this permanent problem of unemployment among our aged people. It will act as an incentive for many of them to leave their employment because they are assured of a livelihood in their old age without work, and because of this unemployment among younger workers is bound to decrease. . . . From that point of view, unemployment compensation and old-age insurance deal with two aspects of the same fundamental problem. Either one without the other would seriously impair the chances for achieving economic security." (25)

This was the first indication in any of the CES' work that, at least the staff, someone viewed this as one rationale for the retirement program. Notice however that Latimer is not attributing this rationale to the RET itself. He is suggesting that the provision for this unmet social need (income security in retirement) will have the laudatory additional benefit of improving the unemployment problem over the long-term.

In her 1932 book on social insurance, Barbara Armstrong devoted more than 60 pages to a discussion of the problem of old-age security, all in terms of provision for the unmet social need that old-age presented, without once mentioning any industrial policy objective of removing superannuated workers from the workforce. In her first paper on old-age insurance for the CES, written in August 1934 (and among the unpublished studies) under the title "Invalidity and Old Age Insurance" Armstrong wrote: "Risks insured against: . . . Loss of earning capacity through old age, old age being defined as the age of 65 years or over." (26) (Emphasis in original.). However, more than 30 years later, in an oral history interview, she made the most extensive claims for the industrial policy thesis of any of the principal players from 1934-35:

Q: "What about the question of a retirement test as one of the provisions of the bill?"

Armstrong: ". . .There were two objectives when they set up Social Security provision for old age. . . The first objective was to protect the person who had to retire and who therefore was going to be without current earnings income and to give him a substitute for the income he had lost because he had become superannuated. . . . Now, in addition . . . you had the problem of the unemployed, a tremendous unemployment, which, you know, was coincident with the Great Depression. And they did not want old people in the labor market. . . . Therefore, the objective was not only to protect the older worker from no earnings income, but it was also to get him out of the labor market. . . .but it had to be on retirement. And retirement means you've stopped work for pay. . . If you haven't retired you are not without income from work, ergo, you are not eligible for retirement benefits. . . This is social insurance. It's not commercial insurance. . . ." (Emphasis in original.) (27)

The Armstrong oral history provides some of the best evidence for the industrial policy thesis. Even so, there is much ambiguity in Armstrong's remarks. This discussion, which ranges over 10 typed pages in the transcript, begins with a question about the RET, but not all the remarks which follow are directed to the RET. Armstrong moves among half a dozen topics during these 10 pages and the link between the RET and the industrial policy is never made explicit. On the other hand, Armstrong's view of the insurance principle as a foundational principle in Social Security retirement benefits is very clear and unambiguous.

J. Douglas Brown became one of the leading theorists of Social Security in the years following the creation of the program. In 1972 he published his opus on the subject, entitled An American Philosophy of Social Security. Brown made a clear explanation of his views on the RET, without any mention whatever of the industrial policy thesis:

"While the established age for normal retirement is a fundamental issue in social security policy, it has not received the attention in recent years which another issue, the earnings-test, has gained. The contingency to be covered by old age insurance is not age alone but insufficient earnings from current employment because of old age. Old age insurance is indemnity insurance. It is the non-existence of any earnings from employment, or of sufficient earnings for reasonable support, which is indemnified. This aspect of the contingency to be compensated has been the focus of more misunderstanding since the Social Security Act was passed than any other feature of old age insurance." (28)

From his part, Robert Myers has unambiguously stated his view of the RET. In a 1996 oral history interview, Myers was specifically asked about the industrial policy rationale for the RET:

Q: I'm very interested in this issue of the retirement test because there's a very common, what I consider to be a myth, about the retirement test and I want to check if I'm right. It's now said by some people that the retirement test was put in the law in order to discourage older workers, to get them out of the labor force so that younger workers could take their place, and that was the intent of the retirement test. . . Is that an accurate description of what happened and how the retirement test came to be in the law or is that a myth that's evolved?

Myers: That's a myth that I've frequently pointed out. That was a by-product, perhaps, that people would retire, but my demonstration of why it's a myth is that the average benefit was going to be maybe $15 a month and the average wage was around $100 a month, so people weren't going to rush out, leave $100 jobs to get $15 benefits. And that wasn't going to remove very many people from the labor market. (29)

It would seem indisputable that for the CES the central purpose of the RET was that given in social insurance philosophy-that the insured individual was paid a benefit only when he/she suffered the insured condition. While there is some evidence that industrial policy objectives were viewed by some CES staffers as an important concern, it is not clear that this concern was linked to the RET, rather than to some other features of the overall program. Brown and Myers specifically deny such a link. Latimer was much concerned about the industrial policy objective, but he does not link it to the RET. Armstrong offers ambiguous statements on the matter. The only clear linkage appears in a brief statement from the 1937 book, which itself may be a bit of historical revisionism.

Congressional Consideration

The Economic Security Bill President Roosevelt transmitted to Congress in January 1935 contained the following language: "No person shall receive such old-age annuity unless . . . He is not employed by another in a gainful occupation." (30) This is the only language in the bill relating to the issue of retirement. There is no language of any kind related to any labor market effects of this or any other provision of the social insurance program.

The final version of the Social Security Act of 1935 contains this language on the subject:

"Whenever the Board finds that any qualified individual has received wages with respect to regular employment after he attained the age of sixty-five, the old-age benefit payable to such individual shall be reduced, for each calendar month in any part of which such regular employment occurred, by an amount equal to one month's benefit." (31)

The course of the RET provision through the Congress has been recounted by Edwin Witte:

"All actuarial calculations were made on the assumption that annuities would be paid only to employees retiring from active employment, and this was provided in the original economic security bill. The only person to object to this provision was Mr. Beaman. He raised the point that 'active employment' needed to be defined and rejected every attempted definition. . . .It apparently was not realized by anyone that this change would operate to again create a large deficit in the old age insurance fund; in fact, there was no discussion of the financial aspects whatsoever. When the subcommittee made its report to the full committee, its recommendation to eliminate the provision relating to retirement was adopted with discussion. As a result, no provision to this effect was included in the House bill.

. . . the Senate committee gave some attention to the requirement of retirement which was stricken from the original bill by the House Committee. . . . there was no difference of opinion among the members of the committee. In a public statement given out after passage of the House bill, the President included an item to the effect that retirement, of course, should be a condition of granting of any annuity. This seemed to be taken for granted by all members of the Senate committee . . . This amendment was adopted without any dissent. Later on it was quite readily agreed upon by the House conferees, who explained that the House committee had never understood that the amendment eliminating the requirement of retirement completely upset the actuarial calculations." (32)

This account by Witte is highly significant. It was a contemporaneous memorandum prepared by Witte, who was personally present in most of the executive sessions of the committees during the bill's consideration and was better positioned than anyone to assess Congressional intent. Witte makes clear that, in the relevant committees of jurisdiction, the only non-social insurance principles involved were cost considerations.

The House and Senate each held 3-4 weeks of hearings on the administration's bill during January and February 1935. The lead administration witnesses were Edwin Witte, Frances Perkins and, in the House, Harry Hopkins. Each of them discussed the provisions of the administration proposals, including passing mention of the RET, only in terms of standard social insurance theory. No mention was made of any supposed rationale involving labor market effects. Two staff members from the CES old-age security group also testified.

In his testimony before the House Committee, J. Douglas Brown makes the first mention of the objective of removing superannuated workers from the workforce. But not in the context of the RET, but rather by way of justifying the recommendation that the government pay part of the cost of the retirement benefits so that the pension amounts will be higher. It is the higher pension amounts, according to Brown, which will have the labor market effects. (Ultimately, this recommendation for a government contribution to the retirement benefit was not adopted by the Congress.)

In his testimony, Murray Latimer focused on the labor market issues in the legislation. Latimer, like Brown, only mentioned the objective of removing older workers from the workforce in the context of arguing for higher benefit levels.

A distinction may be helpful here. If one argues that the mere existence of a retirement benefit is the incentive encouraging older workers to retire, then this undermines the idea that it was the requirement of retirement in the form of the RET, that was the mechanism for accomplishing this goal. If mere social provision were an effective incentive, then the existence of the RET would be inexplicable under the industrial policy thesis. But under the social insurance thesis, there would still be a role for the RET since it would still be necessary to honor the insurance principle of indemnity against loss. Thus, program designers such as Latimer, Brown and Armstrong could all believe that retirement benefits would in time encourage older workers to leave the labor force and still believe there was a need for a RET to honor basic social insurance precepts. I suggest this was in fact their view of the matter. Failing to make this distinction, could lead someone to leap to the false conclusion that since these program designers had the industrial policy objective, and included a RET in the program, they must have do so in order to achieve the industrial policy objective. It is important to note this distinction because virtually all the rhetoric about the industrial policy objective was in fact commentary not about the RET but about the social provision of retirement benefits in general. This is a distinction not always made by commentators on this issue.(33)

To appreciate the significance of this distinction we have only to observe that many political figures would support the following argument: The RET creates an incentive for older workers to leave the workforce. This incentive is undesirable, therefore we should eliminate the RET. However, few people would support the following argument: Social Security retirement benefits create an incentive for older workers to leave the workforce. This incentive is undesirable, therefore we should eliminate the Social Security retirement program. (34) Clearly, the two factors are not interchangeable when considering the industrial policy thesis.

In his testimony before the Senate Finance Committee, Witte was asked to discuss the question of unemployment among workers age 65 and older. He had an extended colloquy on the subject, without ever mentioning the RET or any connection between this issue and the old-age insurance program. In her testimony, Secretary Perkins had a similar extended discussion with the Committee about the matter of the unemployed, but in this discussion nothing was said about the old-age insurance program. Both Witte and Perkins viewed the replacement of lost income under the unemployment insurance program as the vehicle in the bill for addressing the problem of unemployment.

In his testimony before the Senate, J. Douglas Brown also discussed the industrial policy objective of removing older workers from the workforce. However, he did so, not in the context of the RET, but in the context of social provision and the unmet need of an older worker with insufficient income to retire. (35)

In his testimony before the Senate Committee, Murray Latimer expounded at length about his far-reaching concerns regarding industrial policy and how they related to the old-age insurance program and the problems of unemployment. Latimer argued that social provision for retirement had the ancillary benefit of suggesting to young workers that older workers would be retiring rather than continuing in the labor force. He saw this as a "selling point" to help reconcile younger workers to the imposition of this new form of taxation. This "selling point" was the same one Latimer had been making for years when arguing for private sector industrial pensions.

At least one important member of Congress made passing reference to the industrial policy effects of the Social Security Act. Senator Robert Wagner (D-NY), principal sponsor of the bill in the Senate, offered a long explanation of the old-age insurance program, in straight social insurance terms, and ended with this observation: "And it must not be overlooked that industry will receive its full measure of benefit. The incentive to the retirement of superannuated workers will improve efficiency standards, will make new places for the strong and eager, and will increase the productivity of the young by removing from their shoulders the uneven burden of caring for the old." (36)

While it is significant that an important figure like Wagner gave some expression to the industrial policy thesis, it is important to notice that he also does not tie this concern to the RET, but rather attributes it in a general way to the provision of retirement benefits.

There was virtually no other discussion by other members of Congress, in either the extensive Committee hearings or the floor debates, of the industrial policy thesis. The Final Report on the bill in the House contained no provision for a RET and so there is no discussion of any rationale. The Senate Finance Committee Report did contain the provision and an explicit statement of rationale:

"A further important change in the parts of this bill dealing with old-age security which we recommend is the amendment to section 202 to the effect that old-age benefits shall be paid only to employees over 65 years of age who are no longer regularly employed. This was provided in the original bill but as the measure comes to the Senate it permits payment of old-age benefits to workers who have reached age 65 but who still continue in regular employment. This is an anomaly which we believe should not be permitted. There is no need for payment of old-age benefits to employees who continue in employment. This feature of the House bill materially increases the costs and would have necessitated additional taxes in future years. The amendment we suggest to section 202 will prevent anyone from drawing an old-age benefit while regularly employed. This will reduce the costs under title II by many millions of dollars in the course of the decades." (37)

In the Finance Committee's view, the RET is justified by appeal to costs and by appeal to social insurance precepts ("there is no need for payment of old-age benefits to employees who continue in employment"). But not because of any industrial policy related to removing older workers from the workforce.

So, as far as discerning legislative intent is concerned, the available evidence is that Congress enacted the RET due to considerations of social insurance principles, and perhaps costs, rather than to industrial policy.

The old-age social insurance program, with its RET, was enacted into law in August, 1935. Even before the first monthly benefit was paid, the RET provision was amended and it has been the subject of repeated amendments over the years.

The 1939 Amendments

The RET in the 1935 law, read literally, prohibited any Social Security payment when even a penny was earned in "regular employment." "Regular employment" was not explicitly defined. Even before many provisions of the 1935 Act went into effect, major changes were enacted into law in 1939. Wilbur Cohen, who was Edwin Witte's executive assistant on the CES and the first professional employee of the Social Security Board, has given us an eyewitness account of how and why this occurred:

"It was accepted both by the Committee on Economic Security in 1934 and by the Congress in 1935 that retirement from gainful work should be a prerequisite to the receipt of benefits. However, it is significant that neither the proposal drafted by the Committee on Economic Security nor the legislation enacted by Congress in 1935 gave any precise content to the conception of "retirement." In part this was owing to the lack of available statistical and economic information but more to the fact that a precise definition of the term was not immediately necessary, since such benefits did not become payable under the original act until January, 1942. The matter was thus left for regulations under the original law.

But the Social Security Board concluded that it would be unwise to allow such a vital matter to be left to the administrative discretion of the board through the issuance of regulations. The Board, therefore, recommended in 1939, and the Congress adopted, a proposal which had the effect of defining retirement as the receipt of less than $15 of earnings in a month from jobs covered under the insurance program. The Board felt that it was not administratively advisable to apply the test to earnings outside the coverage of the insurance system, since no adequate records would be available." (38)

It is significant that the $15 figure represented earnings at about one-third of the minimum wage mandated under the Fair Labor Standards Act of 1938, so it was a measure which necessitated substantial retirement. (The current retirement test exempt amounts, $800 per month for those under age 65 and $1,292 per month for those age 65-69, constitute 90 percent and 145 percent of the current minimum wage, respectively.) Placing a dollar figure on the RET was done primarily for practical administrative reasons, to clarify with some precision what was meant by "regular employment." This change did not signal any change in the fundamental social insurance principle underlying the RET.

The 1950 Amendments

With the coming of World War II, prices and wages increased dramatically over pre-war levels. However, throughout the War years, the RET had little effect since many older workers remained employed or returned to the workforce rather than retiring, in support of the war mobilization. Following the war, the nation's retirement and pension policies came under new scrutiny and the RET came under increasing criticism. There were obvious problems with the existing test:

  • wages levels had risen since 1939 and the $15 figure was outdated;
  • the "all-or-none" nature of the test was viewed as excessively harsh, it was thought that a more gradual reduction could facilitate the transition from full-time work, to part-time work, to full retirement;
  • some workers, especially among the self-employed, never retired and hence would never be in a position to collect a benefit for their contributions;
  • the RET was generally unpopular with those who wanted Social Security to be more like an annuity, payable simply by virtue of attaining retirement age.

In 1945 the House Ways and Means Committee appointed a panel of staff-level advisers to study, among other things, the RET. Their report, known informally as the Calhoun Report, concluded that elimination of the RET would "profoundly affect both the fiscal and conceptual aspects of the Old Age and Survivors Insurance." (39) Nevertheless, the study group recommended a number of liberalizations of the RET, including raising the exempt amount, applying an annual earnings test, modifying the "all-or-none" nature of the test and eliminating the test entirely for workers age 70 or older.

In 1947, the Senate Finance Committee appointed an Advisory Council which conducted a large and comprehensive study of the Social Security program. The 1947-1948 Advisory Council considered the RET in some detail. The Council's assessment was:

"In the opinion of the Advisory Council . . . the work clause should not be designed to encourage persons to cease all gainful work. The chief purpose should be to prevent the payment of benefits to persons who continue working for wages at or near the level of those earned during much of their working lives; such persons have not suffered the loss of earnings against which the system insures. . .

Many old-age insurance beneficiaries undoubtedly consider any work clause a hardship and restriction on their freedom of activity. In our opinion, the savings effected by a work clause for beneficiaries who are 70 years old or more would not be significant enough to outweigh the advantage of giving some recognition to the beneficiary's desire to receive benefits without qualification. . .

The social-insurance system of the future will probably have to take into account, more than does the present one, both the need for the economic contribution of the aged and their desire to make that contribution." (40)

This was clearly a philosophical break with the principles of social insurance as they had been understood up to that time. The Council was advocating making the Social Security retirement benefit an annuity for those age 70 and older. Even so, the rhetoric at the time did not describe this as an abandonment of the social insurance principle of compensation for actual loss. Other rationale about fairness and the popularity of the idea were the reasons offered for removing the RET at a certain age. The Council was suggesting what was, in effect, a political compromise to satisfy those who wanted to work and receive Social Security.

Another important policy innovation was introduced at the same time. The Council argued that so long as the worker did not have "wages at or near the level of those earned during much of their working lives," this was enough to satisfy the principle of the RET. This was the beginning of the practice of reducing the bite of the RET by increasing the exempt amounts. But if we accept the Council's interpretation of the insurance principle, then all the liberalizations in the exempt amounts over the years can be construed as fully consistent with social insurance principles. The elimination of the test at certain ages, however, cannot easily be so construed.

Notice also that the "industrial policy" argument has now been reshaped. It is now couched in terms of encouraging older workers to stay in the workforce because their economic contributions are desired. It is no longer a question about whether there are older workers in the labor force blocking the opportunities of younger workers. This is the real beginning of the contemporary view that our industrial policy should be to encourage workforce participation by older workers. This is the historical tipping point where the concern now becomes more with the absence of the RET than with its application. So even if we grant that the RET has labor market effects, this was an idea which really arose for the first time following World War II. It was not a widespread idea in 1935.

The Report of the 1947-1948 Advisory Council was received by the Social Security Administration (SSA) and was analyzed extensively and was the basis for SSA's legislative recommendations to the President. When President Truman sent his legislative proposals to Congress, however, he only included an increase in the RET exempt amount. He did not propose elimination of the RET at age 70 as the Council had recommended. Truman was motivated by cost considerations. The President had also recommended creation of a disability benefit and the provision of early retirement benefits for women. In view of these very expensive proposals, the cost of eliminating the RET could not be accommodated in the President's package.

The House Ways and Means Committee decided to drop the President's proposals for early retirement and disability benefits. With the money thus saved, it opted to eliminate the RET for workers age 75 and older. The Senate adopted the House position in conference.

In the 1950 amendments Congress agreed to eliminate the RET starting at age 75 and to increase the threshold from $15 to $50, but to retain the all-or-nothing nature of the test. Because the self-employed were covered for the first time under the 1950 amendments, an annual RET was introduced for non-wage workers. The elimination of the RET for those age 75 and older was rationalized as making the program more equitable for the newly-covered self-employed, who, it was argued, tend to not retire and hence they would be unable to benefit from their contributions.

The 1950 amendments were largely about changes intended to enhance the adequacy of the program. For example, the first ever cost-of-living increases were paid and the benefit formula was raised so that future benefits would start from this higher base. Coverage was also extended, setting the program on the road to virtual universal coverage.

In this context, the abandonment of a purist notion of the RET was driven by concerns over equity issues; by the common intuition that it seems somehow "unfair" for someone to contribute to a retirement system, and if they do not retire, to forever fail to realize any return on their contributions. This is of course the insurance principle in operation, but it has always been difficult to appreciate this principle when applied to retirement.

So by 1950, the nature of the Social Security retirement program had finally been altered on the issue of the RET. From that point on, it was established policy that the RET need not always apply. The "pure" version of social insurance had given way to the practical politics of public policy making.

The 1952 and 1954 Amendments

Following the outbreak of the Korean War, wages and prices again soared. The administration and the Congress realized that an adjustment in the RET earnings level was in order. In the 1952 Act the exempt amounts were raised (to $75 for monthly wages and $900 a year for the self-employed), but no other significant changes were made.

Shortly after taking office, President Eisenhower announced a comprehensive review of Social Security, including the RET. Despite wide-spread criticism of the RET, the President did not propose its elimination, but did recommend further liberalizations. In a January 1954 message to Congress, President Eisenhower suggested the monthly test should be replaced by an annual test under which the first $1,000 of earnings had no impact and earnings above $1,000 would result in a monthly benefit being suspended only for each additional $80 in earnings. In providing his rationale for these changes, the President made a sweeping criticism of the RET, in terms that have resonance with contemporary criticisms:

". . . present law imposes an undue restraint on enterprise and initiative. Retired persons should be encouraged to continue their contributions to the productive needs of the nation. I am convinced that the great majority of our able-bodied older citizens are happier and better off when they continue in some productive work after reaching retirement age. Moreover, the Nation's economy will derive large benefits from the wisdom and experience of older citizens who remain employed in jobs commensurate with their strength." (41)

As a result of these many criticisms, the 1954 amendments again made significant changes in the RET. Wages were put on an annual test, like self-employment income. This was a considerable liberalization because it meant one could earn over the monthly exempt amount for several months each year without any loss of benefits provided the total earnings stayed under the annual exempt amount. However, the test was still "all or none," with the full benefit withheld when the limits were exceeded. The exempt amount was raised again, to $1,200 annually. Non-covered earnings were counted for the first time. The age at which the RET no longer applied was reduced from 75 to 72. The argument again was one of equity; that lowering the RET cap would make the program fairer to self-employed farmers, who were being covered for the first time.

The 1960 Amendments

The 1960 Amendments brought another key change in the operation of the RET. For the first time, earnings over the exempt amount did not always produce a total loss of benefits. For earnings between $1,200 and $1,500 the reduction was $1 for every $2 of earnings. For earnings over $1,500 the old rule applied. This was done to alleviate the perverse result of earnings above the exempt amount sometimes yielding a net decrease in income, compared with simply earning up to the exempt amount. This was one of the inequities so criticized in the 1950s. From this point forward, the RET would utilize this principle.

Interestingly, neither the House nor the Senate version of the bill contained this provision as reported by the Committees of jurisdiction. The House bill had no provision regarding the RET and the Senate bill only contemplated an increase in the annual exempt amount from $1,200 to $1,800. During the House debate Ways and Means Chairman Wilbur Mills explicitly stated that no changes were suggested in the RET due to cost constraints. And since the bill was before the House on a "closed rule" no amendments were offered regarding the RET, even though several members spoke in favor of raising the exempt amount. The Senate passed the bill with the Committee provision unamended. During the Conference on the bill, the House receded to the Senate position on the exempt amount and added the provision for the $1 for $2 offset.

The 1972 Amendments

The 1972 amendments introduced another important innovation-the Delayed Retirement Credit (DRC). This credit increases the benefit amount for those workers who delay retirement past the Normal Retirement Age (NRA). The credit was initially set at 1 percent per year. This amount was raised to 3 percent in 1977. The 1983 amendments raised this credit to an eventual level of 8 percent (by 2009). For any month in which a benefit is paid, a DRC cannot be claimed. The cost of paying additional benefits to persons who continue working (if the RET were eliminated) would thus be offset by the loss of DRCs for such persons. From a cost perspective, the RET could be eliminated at the NRA in 2009 without any long-term programmatic cost.

According to Myers (42), DRCs were added to the law as a partial offset to the RET. The argument was one of fairness. It was argued that if program participants continue to work after 65, and forgo benefits due to the RET, it is only fair that they receive some additional compensation for their extra work.

This argument makes little sense in the social insurance view of the RET, since to withhold benefits in the absence of actual income loss is simply the insurance principle at work. Although DRCs can be seen as having an industrial policy objective (encouraging continued work) they do not fit neatly with the industrial policy thesis of the RET either since they imply that we have adopted two industrial policies with competing objectives (the RET encouraging retirement and the DRCs encouraging continued work). In the context of the RET, DRCs seem to make sense primarily in political or budgetary terms rather than in terms of underlying program rationale.

Also as part of the 1972 law, the $1 for $1 reduction was replaced entirely by a $1 for $2 reduction for all earnings above the exempt amount. This meant that no beneficiary would suffer a net loss of income due to additional earnings. (Because earnings were taxable and benefits were not, substituting a dollar of earnings for a dollar of benefits could result in a net decrease in income.)

The 1972 act introduced one other very important innovation. Previously, the RET exempt amounts were only raised when a specific act of Congress mandated an increase. The 1972 law put the increases "on automatic" by tying them to increases in average earnings, starting in 1975.

The 1977 Amendments

The 1977 amendments brought considerable legislative activity surrounding the RET. The House passed a bill eliminating the RET at age 65. The Senate passed a similar bill setting the end age at 70. The Conference accepted the Senate position and the final legislation ended the RET at age 70, effective in 1982. (An amendment in 1981 later pushed the effective date back to 1983 due to short-term cost concerns.)

The 1977 law also separated the cohort aged 62-64 from the 65-69 cohort, creating a more generous RET for the second group. This was accomplished by a series of ad-hoc increases for the older group, starting in 1978, after which the automatic increases would again apply-locking in place the advantage for the older cohort. This more advantageous treatment of those past Normal Retirement Age (NRA) continues to be a prominent feature of the RET.

A final change in 1977 eliminated the monthly earnings test for all years after the year of retirement. This "de-liberalization" was done to prevent the seemingly unfair result of workers with large annual incomes receiving full benefits for months in which they did not work, whereas workers with much smaller incomes could fail to receive any benefits at all if their income happened to be spread over the entire year. Both the Ford and Carter administrations had recommended this change and it was supported by the 1975 Advisory Council as well. This is one of the few changes made in the RET over the years that had the effect of making the test less generous.

Changes in the 1980s

The 1980 Amendments partly reversed course on the monthly test question. Eliminating the monthly test for years after the initial year could result in anomalies for certain self-employed individuals and for auxiliary beneficiaries when their benefits terminated. For example, if a student received benefits while in school but graduated and went to work half-way through the year, his annual earnings might make him overpaid for the months in which he legitimately received student benefits.

As a result, the monthly test was restored for certain auxiliary beneficiaries in the year of benefit termination as well as the year of initial entitlement, and earnings from self-employment derived from work performed prior to retirement, was not counted for purposes of the RET, except in the year of initial entitlement.

The 1983 Amendments increased the offset from $1 for every $2 of excess earnings to $1 for every $3-but only for the NRA cohort. Even though this change was legislated in 1983, its effective date was 1990. This was because the thrust of the 1983 Amendments was to address short-range solvency problems and Congress did not want provisions such as this one, which had a programmatic cost, from taking effect during the 1980s while solvency was being restored.

Legislation in 1988 made two minor changes regarding the RET in death cases. Previously, in a death case the exempt amount was prorated for the months in which the beneficiary was alive. The new rule granted the full annual exempt amount. Also, the higher exempt amount for the NRA cohort was granted if the beneficiary would have attained NRA during the year even if he in fact died before reaching NRA. These two minor changes were made to prevent overpayments after death in cases where the beneficiary had properly received payments based on the annual exempt amount and then had this annual amount reduced after the fact, as it were, because of his death.

1996 Legislation

Following the 1972 changes in the law, increases in the exempt amounts were no longer determined solely by legislative fiat but were indexed to increases in average wage levels in the economy. This had the dual effect of providing regular increases and of constraining the level of those increases. However, the Contract with America Advancement Act, passed in 1996, made another departure from the indexing by again making ad-hoc increases in the exempt amounts for the NRA cohort. A schedule of increases were programmed into the law which dramatically raised the exempt levels between 1996 and 2002. Indeed, the value of the exempt amount was more than doubled, by the estimates of the Congressional Budget Office. (43)

Also, this second set of ad-hoc increases for the NRA cohort had the effect of increasing the disparity between this cohort and the age 62-64 cohort. Under the 1977 law, the exempt amount for the 62-64 group was only 72% of that for the NRA cohort in 1995. Under the estimates in the 1996 law, the percentage would fall to 35% by 2002. This can be seen as a strong disincentive for early retirement. However, since the amount for the NRA group is fixed in law, and the amount for the 62-64 group fluctuates with average wage levels, it is possible that the differential will vary slightly from the estimates by 2002.

This doubling of the exempt level also will significantly reduce the relative impact of the RET for the NRA cohort. In 1995, the most recent year for which an analysis is available, 960,000 beneficiaries were impacted by the RET as a result of earnings for the NRA group. Of this total, 806,000 were beneficiaries in the age cohort itself (the others were younger auxiliaries), who had a total of $4.1 billion in benefits withheld as a result of the RET. These figures have risen very little since 1989 when 758,000 beneficiaries in the age cohort had an identical $4.1 billion in benefits withheld. (44) Because of the indexing of the exempt amounts, the relative impact of the RET has not changed much during this period. However, the doubling of the exempt level as a result of the 1996 changes, will eventually have a dramatic effect on the relative impact of the RET on the NRA cohort. In future years, there should be a significant drop (relatively) in the number of beneficiaries in the NRA cohort who are affected by the RET.

Also noteworthy in the context of the 1996 changes is the degree to which these changes were met with virtually universal support. Both Congressional leaders and the Administration joined in advancing the provisions, disagreeing only on the appropriate cost-offset in order to make the proposal deficit-neutral.

Significance of the Legislative Changes

The 1935 Social Security Act adopted the RET to honor the social insurance precept that the insured individual must suffer a loss of income to qualify for a benefit. Placing a dollar value on the RET in 1939 was an administrative clarification, not a policy change. Placing a dollar amount on the concept of "retired" was a technique for clarifying the test so that program administrators would not become hopelessly entangled in disputes with beneficiaries over whether or not they were in fact retired. Having accepted this principle, that retirement can be measured against a dollar standard, we can reasonably say that all subsequent increases in the exempt amount are likewise consistent with the original intent of the RET. They can be viewed, in the words of the 1947-1948 Advisory Council, as honoring the idea that the social insurance principle is "to prevent the payment of benefits to persons who continue working for wages at or near the level of those earned during much of their working lives; such persons have not suffered the loss of earnings against which the system insures."

Likewise, changing the offset from an "all-or-none" character to a graduated offset, as first done in the 1960 law, does not violate the original intent of the RET, broadly construed. The graduated offset is simply another adjustment mechanism to fine tune the test to the "wages at or near the level of those earned during much of their working lives" construction.

The differential treatment between the age 62-64 cohort and the NRA age cohort, introduced in the 1977 Amendments and expanded many times since, adds a new policy wrinkle to the operation of the RET. From that point on, policymakers have shown a clear and persistent tendency to treat the NRA cohort more favorably, either because the negative impacts of the RET are thought to be more onerous for this group, or to discourage early retirement for the younger group, or perhaps both. This preference is not always clearly articulated in these terms, especially the disincentive for early retirement, but these are twin effects from the differential treatment.

While the application of the exempt amounts and the offsets do not fundamentally alter the basic character of the RET, we certainly cannot say the same of the elimination of the RET at ages 75, 72 and 70, as provided in the 1950, 1954 and 1977 laws. These were fundamental shifts in the conception of the RET. For those at or above the exempt age, Social Security retirement benefits have become annuities rather than a replacement for lost earnings. Even so, it is pertinent to observe that these changes were not motivated by industrial policy concerns about workforce participation by older workers. Rather, these changes were introduced to provide greater equity to certain groups of workers and to achieve what would have to be described as a political accommodation with the natural desire on the part of beneficiaries to collect payments even though not retired.

The introduction of the Delayed Retirement Credits in 1972 is another important factor in the dynamics of the RET. In addition to creating incentives not to retire, the DRCs have the very significant effect of reducing and eventually eliminating the long-range program costs of eliminating the RET. This could well prove to be a decisive factor in future legislative activity regarding the RET.

Prospects for Future Changes

Proposals to lessen the impact of the RET, or even eliminate it entirely, have proven to be perennial favorites on Capitol Hill. When the House Ways and Means Committee was considering the 1960 changes to the law, no fewer than 175 bills were submitted by members seeking to modify the RET. This is a pattern that has persisted ever since. Although there is no direct tally of support, it is likely that a majority in the Congress would favor repeal of the RET if an acceptable way could be found to deal with the resulting costs.

Changes in the RET have cost impacts. Each time the exempt amounts are raised, or the offsets liberalized, program costs increase. The cost of the 1996 changes alone were estimated to be $7 billion over the 1996-2002 period. (44) But when the value of DRCs reach their full actuarial equivalent in 2009 the RET could in principle be eliminated without any long-range programmatic cost. Of course, there would still be the difficulty of absorbing the costs in the short-term. With the advent of large projected budget surpluses, it becomes easier to absorb the short-term cost by appeal to this future funding. This combination of political intent and favorable actuarial considerations, could make the prospects for elimination of the RET greater than at any time since the RET provision was created.


All peoples throughout human history have had to come to terms with the problem of economic security. That is, how to provide for individuals in their old-age, in event of disability or unemployment, or upon premature death of a family-breadwinner. With the coming of the Industrial Revolution, industrialized nations, like America, changed in fundamental ways. We became urban wage-earners, living in nuclear families, and life-spans increased dramatically so that increasing numbers of people lived to advanced ages. Economic security became irretrievably tied to some form of cash income.

Old-age, in particular, had long been a vexing problem in the developed world. There comes a time when most people grow too old for continued productive work, especially in a wage-based industrial economy. Generations of older workers found themselves unable to generate sufficient income to stop working. According to the CES, at the time of the creation of the Social Security Act, the majority of the elderly in America were living in some form of dependency. Retirement for the wide spectrum of average working Americans was virtually unheard of prior to the passage of the Social Security Act.

So there was a large unsolved problem of old-age dependency in America prior to Social Security. There was a natural human need to cease working at some point in later life. But in the industrial world, there was no practical way to meet this need. It was this dilemma, among others, that social insurance evolved to address. It was social insurance which first empowered significant numbers of typical working class Americans to retire at the end of a lifetime of work. Before the creation of the social insurance programs of the early decades of this century, retirement was a human imperative but an economic rarity. This awful dilemma was finally broken in America by the passage of the Social Security Act of 1935.

The creators of Social Security saw the new social insurance program as providing the means for older workers who wanted to retire but lacked the means to do so, to have the necessary economic security to retire in dignity. By and large, it was the philosophical premises of social insurance which informed their intentions and their program designs, not any industrial policy of making room for younger workers in a marketplace where they were crowded-out by older workers. The primary rationale for including a retirement test in the Social Security program was adherence to social insurance principles. The belief that the RET would have industrial policy impacts was not prevalent at the program's creation. Following enactment of the law, this idea became part of the folk-history of the Act.

It is historically incorrect to say that the Social Security RET was made part of the Social Security Act in 1935 in order to discourage continued workforce participation by older workers. The RET is part of the Social Security Act for the basic reason that Social Security was designed as an insurance scheme, which seeks to compensate covered individuals who suffer a loss of income due to retirement.

Of course the pure form of this principle was abandoned long ago. For many decades, public policy has decided that a compromise was appropriate between the principles of the social insurance program and the wishes of insured workers to have some level of continued earnings in their retirement.

It may well be that we have come to the point where elimination of the RET is the preferred public policy. We should however be clear that elimination of the RET is not a reversal of some out-dated industrial policy. It would be the abandonment of one of the principles of social insurance upon which the Social Security program was founded. The loss of this fundamental social insurance principle would no doubt be lamented by some and applauded by others.


The author wishes to thank Professor Edward Berkowitz of the George Washington University History Department and Barbara Lingg and the editorial board of the Social Security Bulletin for their constructive comments on earlier drafts of this paper. However, any remaining errors are the sole responsibility of the author.


1. "Testimony of the Honorable John J. Rhodes, III," in Social Security Retirement Test Hearing, op. cit., pg. 61.

2. Column by Congressman Packard, op. cit.

3. "Social Security: Smashing the Tinkering Approach," op. cit., pgs. 12-13.

4. The principal academic proponent of this thesis is William Graebner who makes much of some very scant evidence in support of this idea. Cf. Graebner 1980 and 1982.

5. The available analytical studies (cf. Leonesio 1990 and Michael Packard 1990) have often suggested a minimal effect from eliminating the RET. This may be due to the fact that increases in the exempt amounts over the years, as well as the liberalization of the offset amounts, and the lowering of the age caps, have all reduced the scope of the test. Even so, a substantial amount of benefits are withheld each year under the test, as well as some number of workers who forgo retirement because of the test (cf. Bondar 1993 and Kestenbaum, et. al. 1999).

6. 1930 U.S. Census, Volume 4, Occupations, Table 21, pg. 43.

7. 1930 U.S. Census, Volume 2, Unemployment, General Report, Table 6, pg. 13.

8. Data from Table 31, Social Security in America, pg. 148.

9. Berkowitz, op. cit., pg. 25.

10. Very few of the European systems contained an explicit requirement of retirement after the manner of the RET. However, these early systems generally conceived of the problem of old-age security both in traditional pension terms and in terms of what was called "invalidity" insurance, i.e., disability. Many of them were predicated on the assumption that the old-age problem was one of disability and compensated older workers using disability benefits, which had cessation or reduction of work as an eligibility requirement. Consequently, many of these systems could be accurately described as having a retirement requirement for old-age benefits, even though there might not have been an explicit provision along these lines.

11. Latimer, Brown and Armstrong were the experts chosen by the Committee on Economic Security in 1934 to advise it on the old-age insurance aspects of the President's economic security package. Rubinow was terminally ill at the time and Epstein was thought to be a difficult personality to deal with so both were bypassed by the Committee.

12. Quoted in Douglas, op. cit., page 14. President Roosevelt owned a copy of Rubinow's 1934 book "The Quest for Security." When he learned Rubinow was terminally ill, he autographed his copy of Rubinow's book and sent it to him with this inscription on the flyleaf: "For the Author--Dr. I. M. Rubinow. This reversal of the usual process is because of the interest I have had in reading your book." (Signed) Franklin D. Roosevelt.

13. Rubinow, op. cit., pgs. 3-9, 480-481, 302, 304.

14. Quoted in Perkins, op. cit., pg. 294.

15. Graebner takes the relationship between the Townsend Plan and Social Security and turns it on its head. (Cf. his 1982 essay, op. cit.) Graebner concludes that since the Townsend Plan intended labor market effects, and since Social Security intended to compete with the Townsend Plan, it follows that the Social Security program intended to have even stronger labor market effects than Townsend. This simply doesn't follow. FDR and the other creators of Social Security intended to compete with Townsend by offering a real plan in place of what they considered to be Townsend's fantastic scheme. For its part, the Townsend Plan was popular not so much because it offered to get older workers out of the labor force, but because it promised every American age 60 and older a monthly pension twice as high as the average worker's wage in 1935!

16. The industrial pension movement of the early 20th century had very limited impact on the problem of old-age security. By 1932, only 15 percent of the nation's potentially eligible workers were covered by any kind of company-provided retirement pension plan.

17. The other members of the CES were: Secretary of the Treasury Henry Morgenthau Jr.; Attorney General Homer S. Cummings; Secretary of Agriculture Henry A. Wallace; and Harry L. Hopkins, the Federal Emergency Relief Administrator and President Roosevelt's closest adviser.

18. Annual Message to the Legislature, 1931. In Public Papers of Franklin Roosevelt, Vol. 1, op. cit., pg. 103. Notice here that FDR has already begun drawing a sharp contrast between welfare-based pensions and a system operating "on the theory of insurance."

19. The Public Papers and Addresses of Franklin Roosevelt, Vol. 3, op. cit., pgs. 291-292. In this first formal statement on the issue, the President is clearly defining the Social Security program as a form of insurance.

20. The Public Papers and Addresses of Franklin Roosevelt, Vol. 4, op. cit., pgs. 45-46.

21. The Public Papers and Addresses of Franklin Roosevelt, Vol. 4, op. cit., pgs. 134-135.

22. Report To The President of the Committee on Economic Security, op. cit., pg. 25. Notice that in this quotation the CES uses the term "annuities" to describe the old-age insurance plan. This was the term used in the Administration's original bill. During Congressional consideration this expression was dropped.

23. Social Security in America, pgs. 198, 203.

24. Another key feature of the Social Security program probably had the effect of encouraging workforce participation by older workers. Under private pensions, employers had a strong disincentive to hire or retain older workers since private-sector pension plans contained age-differentials requiring higher premiums for older workers. This often put the employer in the dilemma of not hiring or retaining older workers or not having a company pension plan. The Social Security program, with its level-premium structure in which all workers were taxed at the same rate, thus gave employers a newly found incentive to retain older workers in their workforce.

25. Latimer, 1935, op. cit., pgs. 246-247.

26. Armstrong, 1934, op. cit., pg. 9.

27. Armstrong Oral History Interview, op. cit., pgs. 253-263.

28. J. Douglas Brown, op. cit., pg. 125.

29. Myers, 1996 oral history interview, op. cit.

30. Provisions of H.R. 4120 as introduced in the House on January 17, 1935.

31. Social Security Act of 1935, Sect. 202 (d).

32. Witte, op. cit., pgs. 159-160.

33. This is a distinction which Graebner does not make, with the consequence that he finds support for the industrial policy thesis in places where the present author does not.

34. Many people currently advocate changing the Social Security retirement program from a defined benefit to a defined contribution plan, or privatizing the program in whole or in part. However, virtually no one is advocating the elimination of retirement programs from American life to encourage older persons to continue working.

35. Cf. Hearings Before the Committee on Finance of the United States Senate op. cit., pg. 283-284. Brown makes use of the expression "uniform compulsory retirement method" when discussing the effects of the retirement program on unemployment levels. However, this phrase refers not to the RET but to the general character of the social insurance program as being of the contributory/compulsory type, as the rest of the passage indicates.

36. Remarks by Senator Wagner during Senate floor debate, June 14, 1935, Congressional Record, SENATE, pg. 9286. Graebner (cf. 1980, pg. 184-185) finds at least two other Congressmen who by his reckoning support the industrial policy thesis. Charles Truax (D-OH), wanted the Administration's bill to use age 60 as the retirement age, like the Townsend Plan. A careful reading of his remarks reveal that his criticism of age 65 may well have been in the context of the old-age pension provision, not Title II. William Sirovich (D-NY) spoke extensively, with great rhetorical flourish, on the House floor about the virtues of the bill, and recounted the entire history of civilization back to biblical days in framing his support for the Administration's bill. Graebner finds support for the industrial policy thesis embedded somewhere in Sirovich's soaring rhetoric. The present author can find no such indications.

37. Report of the Senate Finance Committee, op. cit., pg. 10.

38. Cohen, op. cit., pgs. 83-84.

39. Quoted in Cohen, op. cit., pg. 71.

40. 1948 Advisory Council Report, Senate Finance Committee print, pgs. 42-44.

41. Quoted in Myers, 1954, op. cit., pg. 14.

42. Cf. Myers, 1993, pgs. 251, 473-474. Myers also argues that the 8 percent DRC amount could be put in place earlier than 2009 and the RET eliminated at that point. Cf. Myers, 1993, pgs. 474-475.

43. The exempt amount for 2002 will be $30,000. The CBO estimate was that under prior law the amount would have risen to only $14,400.

44. Cf. CBO estimates, op. cit., pgs. 6-7.


Armstrong, Barbara Nachtrieb, Insuring the Essentials, MacMillan Co., New York. 1932.

Armstrong, Barbara Nachtrieb, "Social Insurance: Its Place in a Program of Economic Security and an Outline of the Main Coverage, Benefit, and Administrative Provisions of an Ideal Social Insurance System," unpublished CES study, August 1934. SSA History Archives.

Armstrong, Barbara Nachtrieb, Columbia University Oral History Interview, 1965.

Berkowitz, Edward D., America's Welfare State From Roosevelt to Reagan, Johns Hopkins University Press, 1991.

Bondar, Joseph, "Beneficiaries Affected by the Annual Earnings Test, 1989," Social Security Bulletin, Vol. 56, Number 1, Spring 1993. Pgs. 20-28.

Boyer, Arlynda Lee, "Social Security: Smashing the Tinkering Approach," monograph for the Third Millennium group. December 1997.

Brown, J. Douglas, An American Philosophy of Social Security: Evolution and Issues, Princeton University Press. 1972.

Cohen, Wilbur, Retirement Policies Under Social Security, University of California Press. 1957.

"Congressional Budget Office Cost Estimate for H.R. 2684." December 4, 1995.

Douglas, Paul H., "Dr. I. M. Rubinow's Contribution to Social Security," in The Forum Report, Fall 1970, Vol. 1., Series 1., pgs. 8-16.

Graebner, William, A History of Retirement, Yale University Press, 1980.

Graebner, William, "From Pensions to Social Security: Social Insurance and the Rise of Dependency," in Schacht, John, ed., The Quest for Security, The Center for the Study of the Recent History of the United States. 1982.

Hearings Before the Ways and Means Committee of the House of Representatives on H.R. 4120. Government Printing Office. 1935.

Hearings Before the Committee on Finance of the United States Senate on S. 1130. Government Printing Office. 1935.

Kestenbaum, Bertram; Shacleford, Michael; and Chaplain, Chris, "Effect of Benefits of Earnings at Ages 65 or Older, 1995," Social Security Bulletin, Vol. 62, Number 1, 1999. Pgs. 4-9.

Latimer, Murray Webb, Industrial Pension Systems in the United States and Canada, 2 volumes, Industrial Relations Counselors Inc., New York. 1932.

Latimer, Murray Webb, "Statement on Old Age Insurance Provisions of H.R. 7260," unpublished CES study, 1935. SSA History Archives.

Leonesio, Michael V., "The Effects of the Social Security Earnings Test on the Labor-Market Activity of Older Americans: A Review of the Evidence," Social Security Bulletin, Vol. 53, Number 5, May 1990. Pgs. 2-21.

Myers, Robert J., "Basis and Background of the Retirement Test," Social Security Bulletin, Vol. 17, Number 3, March 1954. Pgs. 14-17.

Myers, Robert J., Social Security, 4th Edition, University of Pennsylvania Press. 1993.

Myers, Robert J., "Social Security Administration Oral History Interview with Robert J. Myers," SSA History Archives. Available at http://www.ssa.gov/history/myersorl.html

Packard, Michael, "The Earnings Test and the Short-Run Work Response to Its Elimination," Social Security Bulletin, Vol. 53, Number 9, September 1990. Pgs. 2-16.

Packard, Ron, "Orange County Voices Social Security; Antiquated Earnings Test an Injustice to Senior Citizens," Los Angeles Times, Orange County Edition. Monday, April 9, 1990. Metro Section, Part B, pg. 11.

Perkins, Frances, The Roosevelt I Knew, Harper & Row, 1964 edition.

Report To The President of the Committee on Economic Security, Government Printing Office. Washington. 1935.

Roosevelt, Franklin D., The Public Papers and Addresses of Franklin Roosevelt, Vol. 1 (1932), Random House. New York. 1938.

Roosevelt, Franklin D., The Public Papers and Addresses of Franklin Roosevelt, Vol. 3 (1934), Random House. New York. 1938.

Roosevelt, Franklin D., The Public Papers and Addresses of Franklin Roosevelt, Vol. 4 (1935), Random House. New York. 1938

Rubinow, I. M., Social Insurance, Henry Holt & Co. New York. 1913.

Rubinow, I. M., The Quest for Security, Henry Holt and Company. New York. 1934.

Seager, Henry Rogers, Social Insurance: A Program of Social Reform, The MacMillan Company, 1910.

Social Security Board, Social Security In America, Government Printing Office. Washington, D.C. 1937.

Social Security Retirement Test, Hearing Before the Subcommittee on Social Security of the Committee on Ways and Means of the House of Representative. Serial 102-28. May 23, 1991. Government Printing Office. 1991.

Witte, Edwin E., The Development of the Social Security Act: A Memorandum on the History of the Committee on Economic Security and Drafting and Legislative History of the Social Security Act, University of Wisconsin Press. 1963.

SSA Historian's Office- August, 1999