Financing Social Security, 1939–1949: A Reexamination of the Financing Policies of this Period

Social Security Bulletin, Vol. 67, No. 4, 2007

Presented is an examination of the financing history of the U.S. Social Security system from the passage of the original law in 1935 up through the enactment of the 1950 Amendments to the Social Security Act. In particular, it focuses on the 1939 Social Security Amendments and the subsequent tax rate freezes enacted between 1939 and 1949. It examines the origins of these taxing policies and assesses the impact of the rate freezes on the long-range actuarial balance of the Social Security program during this period.

The author is the Social Security Administration (SSA) historian and is with the Office of Publications and Logistics Management, Office of Budget, Finance, and Management, SSA.

Acknowledgments: Lillian Liu of the Social Security Administration helped with the original research for this article. Professor Edward Berkowitz of the Department of History at George Washington University offered helpful comments on an earlier draft. An earlier version of this piece also benefited from the comments of several SSA reviewers, in particular, Steven Goss, Neta Warren, and Sheldon Baker of the Actuary's Office, and Paul Van de Water and Paul Hewitt of SSA's Office of Retirement and Disability Policy. An abridged version of this work was presented at the 2006 Policy History Conference of the Journal for Policy History and the Miller Center for Public Affairs, University of Virginia (June 1). The author is grateful to Professor Martha Derthick of the University of Virginia, who served as commentator on the work, for her very helpful comments and suggestions. Professor Mark Leff of the University of Illinois at Urbana-Champaign also generously gave of his time to read the abridged manuscript, and he offered several helpful suggestions for improving the piece. Additional SSA reviewers—particularly Linda Maxfield and David Timmons—commented extensively on a later draft and improved the resulting article.

The findings and conclusions presented in the Bulletin are those of the authors and do not necessarily represent the views of the Social Security Administration.


This article examines the financing history of the U.S. Social Security system during the period starting with the amendments of 1939 and concluding with the amendments of 1950. It reviews the program's financing policies during this period, and in particular, a series of tax-rate "freezes" enacted during this time. The tax-rate schedule codified in the Social Security Act of 1935 was prevented from taking full effect during these years and the rates were "frozen" at their 1935 level for 15 years. This article seeks to explain the policy context of these rate freezes and their impact on the program's long-range financial solvency.

Two major findings emerge from this research:

  1. One of the most basic tests of any policy proposal involving Social Security is the projected impact of that proposal on the program's short-range and long-range financing. It would be virtually impossible to propose any serious policy change without a certification from the Social Security actuaries regarding the potential impact of such change. Although Congress enacted the 1939–1949 rate freezes in eight separate legislative acts, the legislative history contains no useable long-range actuarial estimates to gauge the impact of the rate freezes on program financing. How and why such an anomalous circumstance could arise is explored here.
  2. Based on research in the archives of the Social Security actuaries, the author has been able to reconstruct the likely impact of these taxing policies and has discovered that throughout the period from 1939 to 1950, the Social Security program was almost certainly rendered out of long-range actuarial balance by the rate freezes. How such a circumstance could arise, without serious policy debate, is then examined by situating the rate-freeze decisions in the larger frame of Social Security policymaking during this period.


During the period starting with the passage of the Social Security amendments of 1939 and extending until the passage of the amendments of 1950, eight bills were enacted that had the effect of freezing existing tax rates at 1937 levels and preventing the tax schedule in the original Social Security Act of 1935 from taking full effect. No clear assessment was made available to lawmakers at the time of the impact of these rate changes on the long-range financing of the system. Moreover, the principles of long-range cost estimation incorporated into the 1935 law were not adhered to during this period.

Because the legislation after the landmark amendments of 1939 made few changes other than in tax rates1 before the important amendments of 1950, previous scholars have treated this period under study as if little of policy importance happened. Closer scrutiny suggests that the principles of Social Security financing were not followed as expected. In particular, it seems that significant financing policies were adopted without benefit of an assessment of their impact on the overall long-range solvency of the program. Using internal actuarial studies and other documents produced by the Social Security Board (the Board)2 during this period, the impact of the rate freezes can be reconstructed. This analysis provides documentation that the Social Security system was probably not in long-range actuarial balance3 during much, if not all, of this decade—a point that has previously gone essentially unnoticed.

Five factors combined to produce this anomalous period: (1) the ambivalent policy posture of the Roosevelt Administration toward the tax rate increases, (2) determined efforts by opponents of reserve-financing to limit the growth of the Social Security Trust Fund, (3) an apparent willingness on the part of Congress to enact legislation without benefit of long-range cost estimates, (4) a short-term cash-flow surplus in the system because of the booming economy of the war years, and (5) a previously overlooked institutional dynamic within the Board itself that acted to prevent the release of clear-cut long-range actuarial estimates during this period.

An Anomalous Period in Social Security Tax Policy

The amendments of 1939 were a major turning point in the Social Security program, in which the modest retirement program of the 1935 Act was transformed into a family centered social insurance scheme, and benefits were significantly expanded and liberalized. This legislative and political transformation has been the main focus of most previous work on this period. But the other, less studied, policy change during these years was a series of tax-rate freezes. Starting with the amendments of 1939, Congress enacted eight bills that cancelled scheduled payroll tax rates that had been codified in the 1935 Social Security Act. The net result was that the payroll tax rate, which was scheduled to triple between 1937 and 1949, was in fact frozen at the initial 1937 rate until 1950 (Table 1).

Table 1. Projected versus actual Social Security tax rates (employee and employer rates combined)
Calendar year 1935 law 1939 law Actual rates
1937 2.0 2.0 2.0
1938 2.0 2.0 2.0
1939 2.0 2.0 2.0
1940 3.0 2.0 2.0
1941 3.0 2.0 2.0
1942 3.0 2.0 2.0
1943 4.0 4.0 2.0
1944 4.0 4.0 2.0
1945 4.0 4.0 2.0
1946 5.0 5.0 2.0
1947 5.0 5.0 2.0
1948 5.0 5.0 2.0
1949 6.0 6.0 2.0
1950 6.0 6.0 3.0
SOURCE: Author's compilation.

These policy decisions were made in a climate in which the major policy players all declined to make an issue of the long-range impact of their policies. Since the amendments of 1950, proposed Social Security financing changes have been rationalized in terms of their impact on the long-range actuarial balance of the system. It is an almost unquestioned feature of contemporary Social Security policymaking that long-range actuarial cost projections must accompany any serious proposal for changes in program policies. Each year, the actuaries project program finances 75 years into the future and the program is said to be in long-range balance only if the trust fund assets plus projected income are within prescribed percentages of outgo during the 75-year estimating period. The use of a long-range estimating period was a principle established and insisted upon in the original Social Security Act of 1935. While this principle was adhered to in every significant legislative change from 1950 onward, it was absent during the period between these two legislative milestones—a period which encompassed the major amendments of 1939 and the tax-rate freezes of the 1940s.

Previous scholarship has been largely dismissive of the rate freezes, on the plausible assumption that the tax-rate freezes were not significant policy milestones but were more on the order of calm before the policy storms of the 1950s and beyond. Berkowitz (1983, 1986) touches on the rate freezes in passing; Tynes (1996) sees them as distant reflections of more salient political battles; Zelizer (1997, 1998) hits only the political highpoints of the story; Schieber and Shoven (1999) examine the debate between reserve and pay-as-you-go funding, but make no effort to assess the impact of the rate freezes. Even those scholars who focus on tax policy, such as Leff (1984), have tended to look at this episode primarily in terms of overall "new deal" tax policy. Even Leff's insightful and detailed look at the Social Security rate freezes (Leff 1988) missed important drivers of these decisions, such as the conflicted internal institutional dynamics within the Social Security Board that contributed to the absence of long-range actuarial estimates from the legislative process, and the crucial role the absence of long-range estimates played in facilitating passage of this legislation. And no scholar has yet attempted an assessment of the impact of these policy decisions on the underlying financing of the program during this period.

Analysis indicates that the amendments of 1939 cannot be properly understood without reference to the rate-freeze debates that followed them. Looking at those amendments in isolation—or only looking backward and comparing them with the 1935 law—is likely to lead to erroneous conclusions. For example, Neustadt and May (1988) asserted that the change in financing policy in the amendments is evidence that in 1939 the Roosevelt Administration abandoned the principles it had held in the 1935 law and decided to content itself with a merely "symbolic" commitment to the financing principles of the 1935 Act. This was not in fact how the Administration viewed the changes of 1939. In general, the Roosevelt Administration viewed the financing of the amendments of 1939 as a one-time deviation from the principles of the 1935 law—in the name of sealing a political deal with the conservatives—and intended to resume its commitment to the 1935 Act's financing principles after passage of the 1939 compromise.

The 1935 Law and the Issue of Reserve Financing

When President Roosevelt tasked the Committee on Economic Security in June 1934 with designing the Administration's social insurance proposals, he gave it one general proposition relative to financing as a guide in designing the Social Security program: he wanted a self-supporting program that would be funded by contributions from the workers who participated in the system and their employers. He did not want a traditional welfare program in which eligibility is based on need, nor did he want a system in which the general taxpayer was expected to pay a portion of the costs. This last stricture meant that the program was not to be funded, even in part, by general tax revenues.

The question of general revenues was embedded in a larger policy debate over reserve versus pay-as-you-go financing. The main advantage of a reserve is that it allows long-term payroll tax rates to be kept lower than they otherwise would have to be, because some portion of the income to the program can presumably come from the reserve rather than from current payroll taxation. But reserve financing requires near-term taxes to be higher than they would otherwise have to be, in order to build up the reserve. Pay-as-you-go offers the reverse pattern, with lower rates immediately, because benefit demands are lower at the start of a new pension system, along with the implicit obligation to higher rates in the future as demands on the system rise over time.

The ultimate financing plan put forward by the Administration was a partial-reserve scheme, with a significant portion of program financing coming from interest earned on the assets in the reserve. In fact, by 1980 tax receipts would only cover 60 percent of the benefit payments; the remainder of the benefit payout would have to come from the reserve.

Both President Roosevelt and Treasury Secretary Henry Morgenthau Jr. were firm supporters of the build-up of a large reserve, primarily because they saw a reciprocal relationship between a reserve and tax rates or the use of general revenues for financing. By using a large reserve, they believed that they could hold tax rates lower than would otherwise be necessary, and at the same time, avoid the use of general revenue subsidies as an alternative means of keeping payroll tax rates low.4

The use of reserve financing in the 1935 law was controversial from the start. Businesses were unhappy with the new taxes and wanted to find some justification for lowering them. Some economists were concerned with the contractionary macroeconomic effects of the higher tax rates at a time when the economy was still depressed. Many in Congress wanted lower tax rates here and now; and some were especially unhappy with the prospect of the federal government having access to this large source of funds, which they expected would be used to fund expansions of government that they opposed.

However this larger debate between reserve and pay-as-you-go financing is viewed, either approach is usually thought to require the use of long-range actuarial estimates. In pay-as- you-go, planners need to make long-range projections in order to specify the long-term tax rates to which they are committing future taxpayers by their current benefit policies. In reserve funding, planners need to make long-range projections to make sure that the reserve will in fact generate sufficient income to keep the system in long-range actuarial balance.

The actuarial estimates underlying the Social Security Act of 1935 provided cost estimates from 1937 up through 1980. The year 1980 was chosen as the end-point for the actuarial estimates so that there would be sufficient time for the youngest workers participating in the system to attain retirement age. This choice of estimating period was an effort to account for the impact of all program participants, young and old.5 At the end of this 43-year estimating period, the reserve would equal $47 billion. Even though the program had adequate financing on average during the estimating period, by 1967 the program would be spending more than it took in from payroll taxes and it would run a steady annual deficit from that point forward, were it not for the assets in the reserve.

The financing of the Social Security Act of 1935 was thus premised on three key policies: (1) the system used the partial-reserve approach to financing, (2) no general revenue subsidies were used, and (3) long-range actuarial estimates were provided to ensure that the system was actuarially sound. These were matters of high principle to the president and other key policymaking officials in 1935.

The Amendments of 1939

In 1937, in an effort to reopen the issue of the reserve, Senator Arthur Vandenberg (R-MI) persuaded the Social Security Board to jointly sponsor with the Senate Finance Committee an Advisory Council on Social Security. The scope of the council reflected the dual objectives of its sponsors: It was to consider the question of the reserves and also various types of program expansion. Vandenberg hoped to use the council to reduce the reserve, and Board Chairman Arthur Altmeyer hoped to use it to gain support for program expansion—and both of their objectives were achieved.

The Advisory Council report was issued in December 1938. It ended up recommending a huge expansion of the Social Security program by adding dependents and survivors benefits and moving up the start of benefit payments by 2 years. These recommendations were enacted into law in the amendments of 1939.

Although the amendments of 1939 significantly expanded the program, there were offsetting cut-backs in other features of the program, including the elimination of the lump-sum benefits from the 1935 Act and numerous changes in the benefit formula (Schmitter and Goldwasser 1939). The general effect of all the changes introduced in the amendments was to dramatically increase the generosity of the program in the early years (and hence, its costs), while reducing the level of benefits payable in later years. Although the benefit changes increased program costs in the near term (benefit payments were essentially doubled during the first 15 years following passage), they reduced costs in the long term such that the benefit changes were claimed to have no net cost, or even to result in some slight net savings.

During the council's deliberations, the continuing conflict between proponents of pay-as-you-go and reserve financing re-emerged. Edwin Witte, the former executive director of the Committee on Economic Security, and Altmeyer defended the reserve financing scheme, while a group of business representatives, led by Albert Linton, wanted to abandon the reserve and move immediately to a pay-as-you-go approach. Mixed in with this debate was again the issue of partial government financing of the system, which some in the business group favored as an alternative to higher payroll tax rates. Ultimately, the council finessed the dispute with a recommendation for an undefined "reasonable contingency reserve."

Although the Advisory Council recommended moving to a "contingency reserve," it would be reading too much into this recommendation to surmise that the council was advocating pay-as-you-go financing. Rather, this was a political compromise, developed to finesse the issue in the interests of unanimity in the council's final report.6

The Political Deal on Financing in the Amendments of 1939

As part of the political trade-offs around the amendments of 1939, the Roosevelt Administration agreed that the next scheduled tax-rate increase (in 1940) would be canceled, which would mean the tax rate would be frozen at the 1937 level for 1940, 1941, and 1942. The next rate increase would then be the 4 percent rate scheduled for 1943, and thereafter the 1935 tax schedule would pick back up. This rate cancellation would produce a loss of revenue to the trust funds of $825 million during the 3-year period, which would compound over time.

The freeze in the tax rate was put into the law at the insistence of the Republicans in Congress, as part of their efforts to both roll back taxes and to reduce the size of the reserve. In exchange for these concessions on financing, the Administration got the programmatic expansions it desired.

One key underlying the 1939 deal was the testimony of Secretary Morgenthau before the House Ways and Means Committee, during which he told the committee that he no longer believed it was important to build a large reserve (Committee Report 1939, 2111–2120). He said he now thought it sufficient that the trust fund reserve not be any greater than three times the highest annual payout expected during the next 5 years. Morgenthau characterized this much smaller reserve as a "contingency reserve," as recommended by the Advisory Council. This suggestion became the "rule of three" (or the Morgenthau Rule) and it would shape Social Security policy for the next decade.

Records indicate that congressional opponents of the existing tax rates and the large reserve were pleased with this outcome. One excited member of the Ways and Means Committee went so far as to call Morgenthau's brief testimony ". . . a very powerful presentation; in my opinion, one of the finest I have ever heard since I have been a Member of Congress" (Committee Report 1939, 2180). Senator Vandenberg proclaimed, "This is the healthiest thing that has happened in a long time" (Leff 1984, 281).

Morgenthau reversed his 1935 course in this fashion primarily because his agenda focused on federal taxing policy much more than on the financing of the Social Security system. Morgenthau was concerned about the possible contractionary impacts of increased payroll taxes at a time when the economy was in the doldrums following the recession of 1937–1938. He thought that by postponing the tax increases for 3 years, the economy would have time to improve and the deflationary impact of the rate increases would then presumably be lessened. There was in fact by this time a core of Keynesian economists at Treasury who pushed for Social Security taxing policy to be made subservient to larger efforts at counter-cyclical economic management (Leff 1988). Morgenthau also supported the idea of the Advisory Council that the program be allowed to make use of general revenues for a portion of its financing. Altmeyer (1966) would dryly observe, "This, of course, represented a complete reversal of his attitude in 1935."

Morgenthau's stated rationale for his reversal was tied to the argument he had made in 1935 against the use of general revenue funding. In 1935 he argued that because the program was not universal (only about 50 percent of the workers in the economy were covered by the 1935 law), it was unfair to tax general taxpayers to support a system from which all of them could not benefit. By 1939, it was known that coverage rates were actually somewhat higher than expected because of the "in-and-out" movement of some workers into and out of Social Security–covered jobs. In fact, the latest Board estimates showed that about 80 percent of all workers would earn some measure of coverage during their working lives (although not necessarily enough to be fully insured). So as part of the 1939 deal, Morgenthau supported the use of general revenue funding based on the argument that his earlier reservations about taxpayer fairness no longer applied.

Although this was Morgenthau's stated rationale, it is more likely that he was moved to embrace general revenue funding because of his perception that there was a reciprocal relationship between reserve funding, tax rates, and general revenues. By embracing general revenues, he could proffer a source of funding for future program costs that did not depend on the build-up of a large reserve, with its attendant up-front high rates of payroll taxation. Morgenthau's shifting back and forth on general revenues was thus being driven by his underlying concerns with fiscal policy, and not by any considerations about coverage.

The idea of the potential use of general revenues became a kind of implicit "blank check" so that if the financing of the program became problematic, one could assume that general revenues could be used to bridge any shortfalls. However, a key point here is that the use of general revenues was not put into the law in 1939, and it was not necessarily agreed to by all parties. Both the Board and the Advisory Council insisted that if the program were shifted to a pay-as-you-go basis, then there should be a statutory commitment to general revenues to guarantee adequate long-range financing. But their insistence was ignored, and the idea of general revenue financing was left suggested but not codified.

Thus a lawmaker could pose as being fiscally responsible (as having made provision for future costs) by alluding to this idea of the use of general revenues, without having to actually make an explicit commitment to such use and risk conflict with those who opposed general revenue financing of the program. The ambiguity in the commitment to the use of general revenues helped to seal the political deal, and the explicit inclusion of such a provision in the bill would most likely have killed the compromise because there was no consensus on the use of general revenue.7

One could also view the amendments of 1939 as a de facto decision to move to pay-as-you-go financing. A few observers supported this interpretation at the time, but the policy was far from being agreed upon. Administration officials, for their part, clearly viewed the amendments of 1939 as a one-time political deal, with the expectation of a return to the reserve build-up following the temporary rate freeze. Even Morgenthau gave no indication that he had shifted to supporting pay-as-you-go financing beyond what was needed for short-term fiscal stimulus.

Nevertheless, the financing of the amendments of 1939 clearly moved the program in the direction of pay-as-you-go financing by virtue of the fact that it dramatically reduced the size of the reserve and the resulting interest income to the program. To measure the size of this movement away from reserve financing, note that in 1955, under the original Social Security Act, 25 percent of the program financing would be from interest earned on the reserve; under the amendments of 1939 the corresponding figure was less than 9 percent (Senate Report 1939, Table 6).8

The Missing Actuarial Estimates

In light of the historical background here, perhaps it is not entirely surprising that in the legislative history of the amendments of 1939 there are no long-range actuarial estimates showing the net effects of the benefit and tax changes. The published actuarial estimates extended only for a 15-year period (1940–1955), showing a trust fund balance at that time about one-third the size of that under the 1935 law. In contrast to the 1935 estimates, these projections did not stretch far enough into the future to show the cost of the program for the younger workers then in the system. So the choice of a 15-year projection period was not rationalized in the manner of the 1935 estimating period.

Although no long-range data were presented, the actuaries reported that the trust fund would pass into a negative cash flow position after 1955. If the tables were extended past 1955, they would show that program financing would be inadequate for some period. One cannot escape the suspicion that this was the reason for the truncated estimating period. Had there been detailed actuarial estimates showing the program to be insolvent over the long run, it is likely this would have constituted a major obstacle to the political compromise of the 1939 law. The absence of such long-range estimates thus was a key enabler in the passage of the amendments. Perhaps this explains why no political actor in 1939 made an issue of the absence of these estimates.9

Even these shortened projections were the subject of some congressional discomfort, but not on the grounds of insufficiency—quite the contrary. Ways and Means Committee member Allen Treadway (R-MA) complained during the House hearings, "Aren't we borrowing trouble for future generations perhaps when we, by actuarial tables, look ahead for 16 years and use 1955? That is 16 years away . . . why not tend to the problem of today rather than 16 years hence? We have got a big enough job here looking after the present system, it seems to me." (Committee Report 1939, 2211).

Even though the Congress and the Administration presented no data to quantify the long-range impact of the 1939 changes, it is quite possible the system was no longer in actuarial balance. The Senate Finance Committee report on the bill soberly acknowledged such a possibility (Senate Report 1939, 18). The Ways and Means Committee report contained a more exultant acknowledgment. In a separate dissent to the formal committee report, several of the minority members of the Ways and Means Committee announced their dissatisfaction with the legislation, except in one key aspect, "We particularly commend the abandonment of the staggering and illusory $47,000,000,000 reserve fund . . . As a consequence of the abandonment of the $47,000,000,000 reserve fund, a 3-year delay in the scheduled increase in the old-age insurance pay-roll tax has been made possible . . . thus eliminating the immediate threat of higher pay-roll taxes . . ." (Committee Report 1939, 113–114).

Social Security Board officials represented the amendments as merely reducing somewhat the $47 billion reserve, but not sending it into negative territory. It is unclear what actuarial data they used to validate this claim. The only data presented to the Congress was an abbreviated set of actuarial tables showing trust fund transactions to 1955. The actuarial consultant for the Board, W. R. Williamson, testified that the new program was less costly than the existing program, but he did not answer the question of whether either the existing or the revised program was in actuarial balance in the long-run, under the most recent economic and demographic assumptions (House Committee Report 1939, 2473–2488).

Although the actuaries reported that the program would experience a negative cash-flow position after 1955, they made no published effort to estimate how long this negative cash-flow might last, and hence, whether or not the system was in long-range balance. And, most remarkable of all, no member of Congress was moved to ask.

The Impact of the War Years on Social Security Financing

World War II years were paradoxical ones for Social Security financing, producing an abundance of good news in the midst of one of the nation's gravest crises. During the war, unemployment was virtually zero (1.2 percent) as the nation mobilized. The mobilization meant that older workers returned to work or delayed their retirement and younger workers and women entered the labor force in unprecedented numbers. The net effect was that demand for benefits declined while tax revenues increased. Throughout the war years, benefit payments were running at only about 30 percent of what had been expected in 1939.

The Morgenthau reserve target (which was a cap on the reserve) was repeatedly exceeded during this period. For example, by the summer of 1942 the trust fund reserve stood at about 6 times the highest expected annual payout in the next 5 years, and by the summer of 1944 it was closer to 10 times the highest annual payout (Annual Trustees Reports 1942–1945). In simple absolute terms, the reserve was rising much faster than had been expected in 1939 (see Table 2).

Table 2. Projected versus actual trust fund reserves (dollars in millions)
Calendar year 1939 projections Actual reserves
1940 1,871 2,031
1941 2,127 2,762
1942 2,254 3,688
1943 2,651 4,820
1944 3,122 6,005
1945 3,506 7,121
1946 -- 8,150
1947 -- 9,360
1948 -- 10,722
1949 -- 11,816
1950 5,737 13,721
SOURCE: Data for 1939 projections from Senate Report 1939, Table 6:17. Actual reserves from Annual Statistical Supplement to the Social Security Bulletin, 1985, Table 14: 77.
NOTE: -- = not available.

From the exterior, the program appeared to be overfinanced. But Board officials argued that it was not. In the post-war period the economics would probably be the reverse of the wartime experience, with large numbers of postponed retirement claims being filed and with many workers leaving the workforce and many higher-paying defense jobs disappearing from the economy. By 1943, the Board estimated that there were between 500,000 and 600,000 fully insured retirement-age workers in the labor force who were deferring their retirement for the duration. So the "windfall" to the trust funds from the wartime economy would most likely be offset by an opposite effect following the end of the war. There was also a widely held concern among economists that the post-war economic adjustments would produce a depressed economy and perhaps even a return to a lesser version of the Depression of the 1930s. Thus, the Board argued, the long-range view required the nation to conserve the growing surplus because it would be needed to meet benefit obligations in the years after the end of the war. As it turned out, the Board was only half right. The predicted surge in retirement applications did appear. New retirement claims increased from 298,789 during the 1942–1944 peak-war years to 715,642 in the 1945–1947 post-war period. But the feared post-war depression never materialized.

The Subsequent Tax-Rate Freezes

On the eve of the next scheduled step-up in tax rates in January 1943 (under the 1939 law), Congress moved to repeal the increase. Legislation was introduced to cancel the 1943 increase and resume the schedule again in 1944. This time, the president and the Social Security Board strongly opposed the idea. In a letter to the chairman of the Senate Finance Committee, the president reminded Congress of their one-time deal in 1939, and the opportunity presented by the wartime economy, "In 1939, in a period of underemployment, we departed temporarily from the original schedule of contributions, with the understanding that the original schedule would be resumed on January 1, 1943. There is certainly no sound reason for departing again under present circumstances. . . . This is the time to strengthen, not weaken, the social security system. It is time now to prepare for the security of workers in the post-war years" (Congressional Record 1942, 7983–7984).

Board Chairman Altmeyer tried to frame the argument in more technical terms. He stated to the Congress, "The lower rates of contribution now in effect are only possible because the benefit load during the initial period of operation is a small fraction of what it will be in the later years. Moreover, the accruing liability which has been accruing for the payment of the future benefits is several times in excess of the amount in the existing trust fund. The actuaries have estimated that the present program may entail a level annual charge of as much as 7 percent of pay roll. On this basis the fund would already have a deficit of nearly $9,000,000,000 . . ." (Congressional Record 1942, 8005–8006).

This concept of the "level annual charge" (or level premium rate, as it was usually called) was the main long-range actuarial measure that Altmeyer was able to obtain from the Board's actuaries. However, a reading of the congressional hearing transcripts suggests that the members of the committees may not have fully understood this concept. At least they showed no concern at the report that the payroll tax rate was substantially below the level premium rate and that the trust funds were already in a form of deficit. As this was the main long-range actuarial measure available to Congress at this time, it is important to understand what such a measure means.

The level premium rate is the tax rate that would have to be charged throughout an estimating period in order for the system to be fully funded. So, for example, if the actuaries provided an estimate of trust fund transactions for the next 50 years, and they stated that the level premium rate for this period was 5 percent, this would mean that the tax rate would have to be set at 5 percent at the beginning of the 50-year period and maintained at that rate throughout the 50 years in order for the system to be in balance. That is to say, in order for the income/outgo of the system to balance over that 50-year period, a tax rate of at least 5 percent would need to be maintained throughout the period. There would of course be other ways to achieve this balance. One could set the tax rate lower than 5 percent in some years and higher than 5 percent in other years and still achieve balance. But the level premium rate is not a straightforward average, because the timing of the cost curve and the rate schedule is critical. One could, for example, set the rate at 3 percent for the first 25 years and 7 percent for the second 25 years (for a 5 percent average for the period), but this would not necessarily yield a balanced fund.10

The level premium rate is thus a crude measure of the adequacy of a tax-rate schedule in that a tax schedule in which the rate never rises to the level premium rate is one that can be presumed to be underfinanced. But the converse is not necessarily true. Even if the ultimate tax rate is greater than the level premium rate, this does not necessarily mean the program is fully funded. Again, it depends on the timing of the rate schedule and the cost curve. It appears that the deficit-to-date to which Altmeyer referred with his $9 million figure reflects this concept. At a given point in time after the tax rate has been running below the level premium rate since the start of the estimating period, an implicit deficit-to-date accrual can be assumed. So the level premium rate provides us a crude indicator of the program's long-run financial health.

Altmeyer's arguments were not persuasive to the members of Congress who embraced a simple syllogism: The Social Security program was taking in more than it needed to pay current benefits and more than was projected it would need for future obligations at that point in time, therefore, payroll taxes were too high. A coalition of highly motivated opponents of reserve funding, and rank-and-file members who found themselves hard-pressed to defend tax increases in the face of large surpluses, joined together for substantial majorities each time the tax-rate issue came up for a vote.

In all, the Congress enacted eight bills canceling scheduled Social Security payroll tax increases, covering the entire decade of the 1940s (see Table 3). President Roosevelt vetoed the 1944 repeal, but was overridden. Although he "reluctantly approved" the freeze for 1945, the president warned in his signing statement, "Two matters should be clearly understood. The Congress should realize that this bill deferring a statutory increase in contributions toward existing social security merely defers until next year the necessary fiscal receipts to pay the benefits. Also, it does not seem to me wholly sound to enact a tax law and then defer the taxes year after year" (Statement by President Roosevelt 1944).

Table 3. History of tax-rate freeze legislation
Legislation and data Impact on tax rates Senate vote
in favor a
(in percent)
House vote
in favor a
(in percent)
Amendments of 1939 (August 10, 1939) Cancelled rate increases for 1940, 1941, and 1942 87 99
Revenue Act of 1942 (October 21, 1942) Cancelled rate increase for 1943 100 98 b
Tariff Act of 1943 (December 22, 1943) Rate increase scheduled for 1944 postponed for 60 days c c
Revenue Act of 1943 (February 25, 1944) Cancelled rate increase for the remainder of 1944 75 69
Federal Insurance Contributions Act of 1945 (December 16, 1944) Cancelled rate increase for 1945 71 78
Revenue Act of 1945 (November 8, 1945) Cancelled rate increase for 1946 c c
Social Security Amendments of 1946 (August 10, 1946) Cancelled rate increase for 1947 c c
Social Security Amendments of 1947 (August 6, 1947) Cancelled rate increases for 1948 and 1949 c c
SOURCE: Author's compilation.
a. These vote totals are on passage of the overall bill in those cases where the freeze provision was embedded in a larger bill.
b. House vote on the Revenue Act of 1942 was on conference report. Provision not contained in original House bill. Vote taken by division.
c. Voice vote.

With the exception of the 1939 compromise, these rate cancellations essentially pitted the Roosevelt Administration and the Social Security Board against a bipartisan Congress. The last repeal bill was enacted in August 1947 and covered the 1948–1949 period. So as late as 1947, the Congress was persuaded by the vision of large trust fund surpluses that tax-rate increases still were not needed.

The Impact of the Morgenthau Rule

The presence of the Morgenthau Rule changed the political dynamic around the actuarial estimates. There was less expectation of long-range actuarial projections once the secretary of the Treasury suggested that the next 5 years were the yardstick for assessing the adequacy of the program's reserves. However, the Roosevelt Administration appeared chagrined by the use of the Morgenthau Rule beyond the life of the 1939 compromise. They intended the 1939 compromise to be a one-time deviation from the existing tax-rate schedule and were not contemplating that the Congress would use the Morgenthau Rule to justify a whole series of subsequent rate freezes.

The posture of the Treasury Department was especially awkward in the post-1939 freeze debates. Morgenthau tried to shift course with the president and the Social Security Board in opposition to the subsequent rate freezes, but with less than evident grace. In 1939, when Morgenthau was concerned about stimulating a sluggish economy, he wanted a rate freeze. In 1942, and subsequently, when he was more worried about inflation, he wanted the rate increases to go forward. But having broken the strong link between the reserve and the tax rates needed to build it, to argue for this shift in policy would be difficult. Abandoning the idea of the reserve in 1939 allowed Morgenthau to get his desired tax freeze. But to make a coherent argument in favor of the subsequent rate increases, he needed the commitment to the reserve as a rationalizing idea. Without a reserve build-up to justify rate increases, Treasury appeared to be less focused on issues of long-term financing of the Social Security system than on overall fiscal policy. But to admit using Social Security tax policy for these broader general aims of government fiscal management was to open a whole line of additional critique from those who thought that the Social Security system ought to be independent of these kinds of considerations.

The Role of the Trustees Reports

The amendments of 1939 created a formal Old-Age and Survivors Insurance Trust Fund and a Board of Trustees to oversee the operation of the fund. The law also required these Trustees to issue an annual report on the transactions of the fund, as a means of quantifying the program's financial commitments. The first such report was issued in 1941 and annual reports have been issued each year since. These reports, prepared by the actuaries at the Social Security Administration, are the touchstone of long-range financial planning for the Social Security system.

During the period from 1941 through 1950, the Trustees issued 10 reports on the actuarial status of the trust funds. It would be expected that these reports addressed the impact of the rate freezes on long-term financing. However, the data in the Trustees Reports during this period were ambiguous and made a clear assessment of the rate freezes difficult—which is a contrast to the greater long-range detail now produced in the Trustees' Annual Reports.

Some indicators in the reports suggested that the long-term financing of the system was in doubt following passage of the amendments of 1939. In several of the reports there is a general discussion of the problem of level premium rates and the observation that existing tax rates were lower than the level premium rates. In a subtle way, the actuaries were sounding warnings about the potential for the system to be out of long-range actuarial balance. In the 1944 report, the Trustees warned that the refusal of Congress to allow scheduled tax rates to rise as contemplated in the 1939 law was potentially placing the program in financial jeopardy (Annual Trustees Report 1944, 29).

These warnings, however, were muffled by three factors. First, they were couched in terms of using level premium rates, which is a less clear-cut way of expressing the issue of long-range actuarial balance than by using year-to-year tables showing dollars incoming, dollars outgoing, and the resultant trust fund balances. Second, there were multiple long-range projections in each of these reports with no guidance given as to which projection policymakers ought to utilize. And third, in most of the reports from this period, there are no tables of long-range data at all, only illustrative graphs showing a tax-rate line and a benefit-payout line crossing at some point in time. As previously discussed, whether the program is in long-range actuarial balance is a function of the timing of these two curves. One simply cannot tell from a visual inspection of the graphs whether or not the system is in long-range actuarial balance. This determination can only be made by looking at the detailed data used by the actuaries in constructing the graphs, and these detailed data are precisely what is missing from the Trustees Reports during this period.

It became necessary following the amendments of 1950 to make an unambiguous assessment of actuarial status because the Congress expressly stated it wanted the program to be "self-supporting" and wanted the annual reports to make an assessment of whether this goal was being met. Therefore three significant changes were introduced in the 1951 Trustees Report: (1) the Trustees included a table showing actual dollar figures for income, outgo, and trust fund balances for the long-range projection period; (2) the Trustees added, for the first time, a single intermediate set of estimates, and (3) the Trustees specifically identified the intermediate set of estimates as the one that Congress should use in setting tax-rate schedules and in assessing whether or not the program was "self-supporting." Although the other mechanics of the 1951 report are not dramatically different from those of earlier reports, these three changes gave policymakers a benchmark against which to make their policy decisions. It was therefore a major innovation from the perspective of Social Security policymakers, marking a watershed break from the actuarial practices prevailing during the 1939–1950 period and a return to a practice of estimating long-range actuarial projections more consistent with that used in the 1935 law.

Why Were There No Effective Long-Range Actuarial Estimates?

Four factors contributed to the absence of long-range estimates: (1) the temporizing of the Roosevelt Administration in 1939 and its unintended consequences in the subsequent freeze legislation; (2) the ad hoc nature of the rate freezes in the absence of a considered long-range financing policy; (3) an unusual staffing issue within the Social Security Board's Office of the Actuary; and (4) Congressional intent, reflected in the ease of enactment of the freeze legislation in the absence of long-range estimates.11

The Precedent in 1939

By promulgating the Morgenthau Rule in 1939 and providing only short-range estimates for the 1939 legislation, the Roosevelt Administration had inadvertently introduced a precedent that would be repeated in subsequent years. Thus the practice of providing only short-term detailed estimates had a methodological precedent, and this was certainly a factor in the subsequent freeze debates, as it was mentioned often by members of Congress (in the form of their support for the Morgenthau Rule).

Ad hoc Policymaking

The ad hoc nature of the policymaking during this period also provides some degree of rationale for the absence of effective long-range estimates. The 3-year freeze in the amendments of 1939 was part of a considered financing policy, even if it was done without benefit of long-range cost estimates. But the subsequent freezes were 1-year decisions—made on an ad hoc basis year after year. Making what appeared to be 1-year changes in financing might seem to not require the kind of long-range evaluations that one might expect when introducing a full-fledged financing policy. And one could argue that these ad hoc freezes did not represent a financing policy, but rather the absence of one. But after a series of ad hoc actions that had the effect of holding the tax rate unchanged at 2 percent from 1937 through 1949, it begins to look very much like a long-term policy. In any case, the actuaries were placed in a dilemma when trying to produce long-range estimates because they either had to decide to use the 2 percent rate as the basis of the estimates (which was not what the law dictated) or use the schedule in the law (which was repeatedly invalidated by the ad hoc actions).

An Unusual Staffing Issue

Staffing issues centered on the Social Security Board's selection of W.R. Williamson as the Board's actuarial consultant. Williamson had served briefly as a consultant to the Committee on Economic Security on the unemployment insurance subgroup, but he was not actively involved in the design of the retirement program. In fact, Williamson was opposed to the contributory, wage-related, model of social insurance adopted in the 1935 Act. Williamson was a proponent of what he called "social budgeting." This approach involved a universal, flat-rate benefit, unrelated to wage history or program contributions, and the only financing arrangement involved annual budget appropriations to cover annual program expenditures. Thus the irony here is that the chief actuary for the Social Security program advocated a form of social insurance in which actuarial estimates had little place.

Williamson's advocacy of social budgeting put him at odds with his peers and his superiors on the Board over the issue of long-range estimates. Williamson declined to provide unambiguous long-range cost estimates, even to his colleagues within the Board, arguing that uncertainty in estimating required that the actuaries only produce a range of numbers and then not speculate as to the most likely place in that range that future experience would land. Thus he refused to provide detailed long-range cost estimates for policymakers to use in crafting legislation.

Colleagues within the Board were frustrated by their inability to obtain useful long-range cost estimates and thought Williamson's point about actuarial uncertainty was being misused to justify a practice he preferred for other reasons. Some suspected that his personal antipathy to contributory social insurance was a motive behind his refusal to provide useable long-range estimates.

The actuary's office, under Williamson, would generally provide detailed short-range estimates and would only provide ranges of possible values for any long-range estimates. But policymakers need something more definite—a most likely set of numbers, or an average, or something, however imprecise, which can be used for such practical matters as setting tax rates. After all, tax rates and benefit amounts are fixed values, neither of which can be expressed as a range. And in setting tax rates and determining benefit amounts, Williamson's estimates were thus of limited practical use—which is one key reason that the actuarial estimates during this period failed to perform their role as a framing constraint to policymaking.

Williamson's peers in the Bureau of Old-Age and Survivors Insurance (BOASI) and the Bureau of Research and Statistics (BRS) constantly tried to get the actuaries to produce long-range estimates that were more usable for policymakers. In complaining about the draft of one particular actuarial study (No. 23), an exasperated Oscar Pogge (BOASI director) wrote, ". . . the use of the four illustrations is to present so wide a range in costs as to be of doubtful value to policy-makers. The range in some of the figures is so broad as to produce almost ridiculous results. . . ." Pogge's colleague, I.S. Falk (BRS director), complained about the same study and accused Williamson of intentionally fuzzing-up his estimates because of his personal antipathy to the program (Pogge 1947; Falk 1947). Even Board Chairman Altmeyer was frustrated by Williamson's refusal to provide him with information. Altmeyer felt that he needed to formulate program policy and despite being Williamson's boss, he too could not persuade Williamson to provide him with long-range actuarial estimates. In one exchange of memoranda, Altmeyer gave voice to the policymaker's lament when he complained to Williamson, " . . . as I have said to you before, I feel considerably handicapped in discussing actuarial cost estimates because of your unwillingness to indicate what you consider to be the most reasonable estimates" to which Williamson offered the actuary's rejoinder, "The existence of many interlocking factors, each of them with a considerable possible cost range in any dynamic functioning of our economy, makes any specific single cost estimate which implies definiteness of knowledge as to the future an undesirable thing to use, so our practice in actuarial studies has been to choose two reasonable sets of values for many of these factors and to determine two estimates, a low and a high, both lying within a wider range whose absolute limits we do not know" (Altmeyer 1943; Williamson 1943).

These internal debates (no matter who was right or wrong) revealed that the actuarial estimates in use during this period were seen by policymakers as seriously inadequate to their needs.12 The debates are also evidence of the general point here: The actuarial estimates used from 1939–1950 were unlike those in use before or after this period.

Congressional Intent

Some of the responsibility for the missing long-range estimates has to be put on the posture of the Congress. Despite Williamson's reticence, the committees of jurisdiction in the Congress could have insisted the Board's actuaries produce long-range estimates during legislative consideration of the amendments of 1939 and the freeze bills, if they had any interest in obtaining them. The Board's actuaries did perform a handful of internal studies during this period that, with a little effort, could be read as indicating the program was not in long-range balance. But none of these actuarial studies were made part of the legislative history of any of these eight laws.

During the period of the rate freezes most members of Congress appeared to be perfectly content with the absence of long-range cost estimates. In the 1944 freeze debate, a member of the Ways and Means Committee, Congressman Thomas Jenkins (R-Ohio), even introduced a proposal to freeze the tax rate at 2 percent permanently. He justified it by complaining that Social Security financing was too long-range. "Everything will be all right for 20 years," he told his colleagues. "Why should we worry about it? Sufficient to the day is the evil thereof." (Newspaper article 1944) Jenkins' viewpoint was probably shared by many members, although few would admit to it as openly. In any case, members repeatedly voted without long-range estimates and without complaint.

The Impact of the Rate Freezes on Social Security's Long-Range Actuarial Balance

Given the absence of long-range estimates during this period, what, then, can be said about the consequences of the tax-rate freezes on Social Security's financial prospects? This question has not been addressed in previous scholarship because of the absence of long-range actuarial estimates in the published legislative history. Even so, an approximate idea of the impact of the rate freezes can be found by looking at the published Trustees Reports; a small number of actuarial studies produced by the Board's actuaries; some internal memoranda prepared by the Board's actuaries for Altmeyer's use and other Board officials; and at least one published study.

The 1939 Freeze

In his testimony before the House Ways and Means Committee on the amendments of 1939, Williamson's presentation to the members of the Committee did not clarify the long-term financing issues (Committee Report 1939, 2473–2488). Williamson presented no dollar income/outgo data beyond 1955, but his actuarial tables included a statement of the probable level premium cost of both the existing and the contemplated program, under both the 1935 actuarial assumptions and a more current set of assumptions.

The first startling fact from Williamson's testimony was that the existing program, without any changes, was already out of long-range balance under the updated actuarial assumptions. The current estimates were that the level premium cost for the existing law might be as high as 7.88 percent of payroll. Because tax rates under the 1935 law came nowhere near this level premium rate, the existing program was no longer self-sustaining. This apparent long-range imbalance should have been a major issue for Altmeyer and President Roosevelt, for whom the self-sustaining principle was of prime importance in 1935. But this implicit insolvency in the existing program passed unremarked.

In support of the Board's repeated assurances about the costs of the 1939 proposals, Williamson's data showed the 1939 law to be less costly than existing law, under either the original or the updated assumptions. But here too a problem lurked. Under the updated assumptions, the level premium rate for the 1939 law was 6.60 percent of payroll. The combination of the existing tax schedule and this level premium rate meant that the 1939 program was out-of-balance under the updated assumptions as well, and these figures were computed by Williamson without considering any changes in the 1935 rate schedule.

It was evident from the discussion at the hearing that the members of the committee had not grasped the potential implications of Williamson's limited data for the question of the program's long-range balance. The committee members spent most of their time looking at the short-range dollar figures—which included year-by-year income, outgo, and net-balance figures. The level premium rate formulation appeared too abstract to engage the members' interest. Had Williamson presented the same type of date and dollar figures for the long-range as he did for the short-range estimates, it would have been much more difficult for the Congress to overlook the consequences of their decisions. Such data were available in the internal worksheets from which the actuaries extracted the level premium figures that Williamson reported to the committee. But the actuaries refused to release the long-range dollar data. Even other high-level officials of the Social Security Board could not get the figures. Two months after enactment of the amendments of 1939, G.R. Parker, the Board's regional director, sent a memo to the head of BOASI (his boss and Williamson's peer) asking for the long-range figures used in computing the level premium rate numbers Williamson used in his Congressional testimony. Williamson's deputy, D.C. Bronson, replied to Parker, refusing to provide the long-range figures, telling him, "As to the projection of costs beyond 1955 we do not have figures which are susceptible to sending you. We have of course worked on these far distant figures under various projections and assumptions. We do not feel, however, that it would be advisable to send them out although possibly at a later date some distillation of our results may be available" (Bronson 1939).

Williamson was also asked during his testimony what would happen if the tax rate were frozen for 3 years, as under the 1939 law. He replied that roughly "a billion dollars" would be lost causing the program to have a negative cash flow in 1942, but that following the rate step-up in 1943 the program would resume its glide path and would not again be in a negative cash-flow position until 1955. Asked by an alert committee member whether he was telling them that it did not matter if they froze the rate for 3 years, Williamson replied that over the long run a billion dollars more or less would get lost in the static of actuarial estimation. No Committee member thought to ask what would happen after 1955 when the program passed into a negative cash-flow.

The Subsequent Freezes

Expecting the 1939 deal to be a one-time affair, the Roosevelt Administration was shocked into action when in 1942 rumblings began to be seriously heard about freezing the rate for another year. In a letter to Congress in October 1942, Altmeyer warned that the system already had a $9 billion deficit, even without additional freezes; in other words, by this time the system should have accumulated that much additional tax revenue if it were operating at its level premium rate. This figure meant either that the program was insolvent, or at some future time tax rates would have to be raised sufficiently above the level premium rate to recoup this amount of "lost" revenue. As mentioned, this type of deficit figure is not a direct measure of long-range balance; it is a current measure of where the system ought to be presently if it is to be in balance in the long run. So whether the system was in balance in the long run and what the dollar value of its balance sheet might be in the future could not be determined from this measure.

A second similar set of estimates, obtained from Bronson (1943), showed the level premium range to be roughly 4 percent to 7 percent, and the dollar deficit to be between $5.4 billion and $13.5 billion. In November 1944 Bronson updated the dollar figures for the 4 percent and 7 percent level premium rates, reporting that the deficit to date had risen to between $6.6 billion and $16.5 billion.

Altmeyer reported all of these figures to the Congress, but to no avail. The actuaries continued to decline to produce detailed year-by-year projections showing income, outgo, and reserve figures in dollar terms. Expressed in the abstract shorthand of level premium rates or in the indirect measure of a present deficit, the long-range imbalance of the system provoked no Congressional debate.

The first complete long-range actuarial estimate of the costs of the amendments of 1939 was produced by the Social Security Board actuaries in a December 1942 study (Actuarial Study No. 17). At that point, the tax rate had been statutorily frozen up through 1943. The actuaries developed two sets of estimates, a low-cost and a high-cost estimate, and projected trust fund reserves for 50 years, from 1950–2000. The projections assumed the freezes would stop at that point and the 1935 tax schedule would resume in 1944. Under both sets of estimates, the program passed into a negative cash-flow position no later than 1980 but was out of long-range balance under only the high-cost estimates. So by December 1942, there was some after-the-fact indication that the program might have already been insolvent because of the 1939 legislation (see Table 4).

Table 4. Long-range trust fund estimates from Actuarial Study No. 17 (dollars in millions)
Net income Balance in fund
at end of year
Low-cost estimate
1950 1,446 14,380
1960 911 31,532
1970 418 47,644
1980 (96) 62,678
1990 (487) 76,578
2000 (610) 91,504
High-cost estimate
1950 1,147 12,570
1960 406 24,691
1970 -347 31,781
1980 (1,388) 30,844
1990 (2,206) 18,376
2000 (2,572) a
SOURCE: "New Cost Estimates for the Old-Age and Survivors Insurance System, with the Assumption of a Static Future Wage level." Actuarial Study No. 17, Social Security Board (December 1942).
NOTE: Estimates in parentheses indicate deficits.
a. Fund exhausted in 1999.

In September 1943 the Board's actuaries produced their second detailed evaluation of the long-range picture of the trust funds, using slightly updated demographic and beneficiary data. In this study (Actuarial Study No. 19) the actuaries concluded that if the tax rates were permitted to return to the 1935 schedule in 1944, that under both the high-cost and low-cost estimates the program would remain in balance through the year 2000. Three months later Congress acted to freeze the 1944 rate at the 1937 level, rendering this estimate outdated.

In one published study in October 1945, which appeared in a professional journal, a junior Board actuary presented data showing that if the tax rate were not raised beyond the 2 percent prevailing at that time, the long-range deficit in the program would be between $17 billion and $24 billion by 1980 and between $83 billion and $136 billion by the year 2000 (Immerwahr 1945). This article was the first published study that attempted to answer the question of what the long-range impact would be if the 2 percent payroll tax were pursued as a long-term policy (as some in Congress clearly intended). It showed unambiguous numbers in that the entire range, low and high, was in the red. At this point, if not earlier, it was at last clear that the tax-rate schedule in place at that time had rendered the Social Security program out of long-range actuarial balance.

In January 1946, the Social Security Technical Staff of the House Ways and Means Committee produced a study of the Social Security program, including the issues of financing. This document, the Calhoun Report, recommended that the Congress stop the practice of ad hoc freezes and adopt a newly rationalized contribution schedule under which rates would go up to 3 percent in 1947, 4 percent in 1957, 5 percent in 1967, 6 percent in 1977, and that a general revenue subsidy of one-third of costs be instituted thereafter. As part of justifying its recommendations, the Technical Staff produced a table, based on Actuarial Study No. 19, in which they showed that if the Congress held to the Morgenthau Rule's limit on the size of the reserve—or abandoned a reserve altogether—payroll taxes would have to be continually increased after 1964 until, in 1995, they would range somewhere between 8.4 percent and 10.1 percent of payroll (Technical Staff Report 1946). Their point was that a rate frozen at 2 percent was not viable in the long run. However, since their table also showed that the rate could be held at 2 percent for several more years before encountering a negative cash flow, the Congress opted to freeze the rate again later that same year.

The third set of long-range projections by the Board's actuaries was finalized in April 1947. This study (Actuarial Study No. 23) had the benefit of being able to examine the first seven of the eight rate-freeze actions, enabling it to develop four possible cost scenarios (see Table 5). Assuming that the frozen rate of 2 percent continued, the projections showed that under any of the four possible scenarios, the program would be insolvent—as early as 1965, or at the latest, by 1990.

Table 5. Long-range trust fund estimates from Actuarial Study No. 23 (dollars in millions)
Net income Balance in fund
at end of year
Low employment, low-cost estimate
1950 88 9,042
1955 (242) 9,417
1960 (558) 8,132
1970 (1,108) 586
1980 (1,759) a
Low employment, high-cost estimate
1950 (102) 8,163
1955 (499) 7,247
1960 (867) 4,247
1970 (1,738) b
High employment, low-cost estimate
1950 868 13,142
1955 512 17,979
1960 162 21,473
1970 (594) 23,553
1980 (1,475) 17,200
1990 (2,368) c
High employment, high-cost estimate
1950 506 11,446
1955 41 13,875
1960 (458) 14,009
1970 (1,665) 5,293
1980 (3,325) d
SOURCE: "Long-Range Cost Estimates for Old-Age, and Survivors Insurance, 1946." Actuarial Study No. 23, Social Security Board (April 1947).
NOTE: Estimates in parentheses indicate deficits.
a. Fund exhausted in 1971.
b. Fund exhausted in 1965.
c. Fund exhausted in 1990.
d. Fund exhausted in 1973.

In August 1947, the Congress acted to freeze the tax rate for the remaining years in the decade (1948 and 1949). However, this time, rather than making a simple ad hoc freeze, a new contribution schedule was enacted replacing the one in the 1935 law. Under this new schedule, the tax would remain at 2 percent through 1949, rise to 3 percent in 1950 and 1951, and rise to a maximum rate of 4 percent in 1952 and thereafter. Although the 1947 law promised an increase in tax rates in 1950, the ultimate tax rate was lowered from 6 percent under prior law to only 4 percent. In a supplement to Actuarial Study No. 23, produced after the law was enacted, the actuaries projected that the new tax schedule—which clearly represented a new long-range taxing policy—would produce negative cash flows in all four scenarios and a depleted fund in two of the four alternatives. The salient point is that this actuarial study was developed only after the passage of the law—it too was absent from the legislative history of the 1947 legislation. Thus, here again, Congress enacted a major change in tax policy without benefit of long-range actuarial estimates.

The end to this period in Social Security's financing came with the passage of the amendments of 1950. Just 3 years after the enactment of the 1947 tax schedule it was discarded and the financing of the program was re-rationalized from scratch, based on the newly modified program. The amendments of 1950 made major changes in the value of the benefits (increasing them by 77 percent on average). The changes were so profound that benefit computations from that point forward are referred to in Social Security Administration regulations as "New Start" computations.

This re-rationalization had several impacts on financing policy: A new tax schedule was put in place, long-range actuarial balance was certified for 50 years into the future, language was put in the legislative history insisting that the tax schedule be sufficient to make the program "self-supporting," the authorization to use general revenues to make up funding shortfalls was repealed, and the actuaries were required to redesign the Annual Trustees Reports to make a clear finding of whether or not the system was in long-range actuarial balance. From that point forward, long-range actuarial projections have always been available to Social Security policymakers. Thus ended this anomalous period in which major taxing policies were adopted without benefit of an assessment of their long-range consequences.


In the ordinary course of making Social Security policy, today's policymakers would expect to see annual long-range cost estimates showing the projected income and outgo to the system and its long-range actuarial balance. Such long-range cost estimates are a framing constraint on policymaking, limiting the ability of policymakers to adopt financing schemes that threaten the system's long-range solvency.

One of the most startling facts about the Social Security Amendments of 1939 is that this type of long-range data was not presented in support of the legislation. Congress enacted the amendments in the absence of any long-range actuarial estimates showing their potential long-term impact. The foreshortened estimates used in the 1939 legislation actually masked a very uncertain long-range financing picture and were an important policy departure from the actuarial practices surrounding the creation of the program in 1935. Moreover, following the amendments of 1939, seven additional legislative enactments blocked scheduled tax-rate increases from taking effect. For the entire decade of the 1940s, a scheduled tripling of tax rates was repeatedly deferred. These subsequent pieces of legislation were likewise enacted without benefit of any long-range actuarial estimates.

Although the question of long-range actuarial assessments was caught up in larger debates during this period over reserve versus pay-as-you-go financing and the use of general revenue funding, long-range actuarial estimates are required in order to quantify long-term financing commitments whether under a reserve or a pay-as-you-go theory of financing. The absence of effective long-range estimates during the period under study is thus anomalous under either approach, and it is a marked departure from the way financing policy has been framed in Social Security both before 1939 and after 1949.

The problem is not so much that Congress enacted freezes in the Social Security tax rates during this period. In some respects, it might have been a rational policy, given the economic bonanza of the war years and given the inclination of some to want to move away from reserve funding to a more obviously pay-as-you-go financing basis. The problem is that the rate freezes were enacted without benefit of a clear assessment of the long-range impacts of these decisions. Freezing previously scheduled tax rates is not necessarily bad policymaking, but making changes in payroll tax rates without an assessment of the long-range impact on the system can hardly be seen in any other way.

The absence of effective long-range estimates during these years is, in the author's view, an indication of policy failures in several respects. The advocates of pay-as-you-go financing did not meet the implicit obligations of this approach by setting a schedule of future payroll tax rates sufficient to fund the program. The Roosevelt Administration, while initially insisting on a large reserve and associated long-range actuarial projections, temporarily abandoned both principles as an expedient to easing passage of the political compromise underlying the amendments of 1939. Rank-and-file members of Congress found it easy to defer scheduled tax increases as long as there were no long-range data suggesting their actions were problematic, and they were apparently content to accept the absence of such data. The Treasury Department apparently was willing to abandon long-range estimates when the resulting tax policy was more consistent with their efforts to manage the larger economy. And the fortuitous financial windfall to the program from the wartime economy offered an irresistible temptation to policymakers to focus on the short-term surpluses in the system and turn their gaze from the demands of a longer view.

Prior scholarship has generally overlooked the significance of the absence of long-range actuarial estimates from the legislative history of the bills enacted during this period. Far from being a mere "technical" matter, the author suggests this absence of long-range estimates was a key factor easing passage of the amendments of 1939 and the subsequent freeze legislation. Moreover, the absence of long-range actuarial estimates was a key enabler that allowed policymakers to drive the program into probable insolvency, without explicitly acknowledging this possibility. Indeed, it may very well be the case that the Social Security amendments of 1939 rendered the program insolvent, a fact that was kept hidden from view by the absence of any long-range actuarial estimates in the legislation.

This review suggests that the amendments of 1939 are linked with the rate freezes of the 1940s by a common approach to financing policy, and that the changes in 1939 cannot be fully understood without positioning them within the context of this later historical period. While the benefit provisions of the 1939 law are linked backwards in policy to the provisions of the 1935 law, the financing provisions of the 1939 law are the starting point of a sequence of policymaking by which this law is linked to the subsequent rate freezes of the 1940s. One of the results of the present research is to place a new emphasis on this period from 1939 to 1949 as a significant one for Social Security policymaking and to correct the existing historiography, which tends to overlook this period.

From the perspective of the program's financing, this period features a unique combination of circumstances: the promulgation of the Morgenthau Rule and the precedent of the 1939 legislation using only short-range estimates, the reticence of the Social Security Board actuaries to provide long-range estimates, the skillful politics of the opponents of reserve financing in moving the program toward a pay-as-you-go basis by reducing the size of the reserve, the acquiescence of rank-and-file members of Congress in a series of tax deferral decisions, and the unprecedented financial windfall of the war years—all combined to produce this unusual period in Social Security policymaking. This combination of circumstances allowed policymakers to enact laws that probably placed the Social Security system in a long-term deficit, without ever adopting this underfunding as an explicit policy goal and without ever being forced to acknowledge that this underfunding was the probable consequence of their actions. But from the passage of the amendments of 1939 until the enactment of the amendments of 1950, the Social Security system was more than likely insolvent in the long run—a fact that has gone largely unnoticed.


1. There were some minor revisions to coverage rules during this period, but the net effect was to leave the scope of the program essentially unchanged.

2. Originally the U.S. Social Security Administration was known as the Social Security Board. It did not acquire its present name until 1946.

3. "Long-range actuarial balance" is a summary measure assessing whether the Social Security program's financing is sufficient to meet its projected benefit obligations over a long-range estimation period. Actuarial balance is computed by determining whether total tax revenues and trust fund assets are sufficient to meet total projected expenses over an extended period. Long-range actuarial balance is one of several alternative ways of characterizing the program's financial health. Thus it should be understood that even if the system is in long-range actuarial balance, it might still suffer substantial financial shortfalls during specific times within the estimating period. On the other hand, to determine that the program is not in long-range actuarial balance does not mean that it is unsustainable at a given point in time, but only that on average over a long-range estimation period it lacks sufficient sources of revenue to fully cover its commitments. In more commonplace usage, one might ask whether the program is solvent or insolvent in the long run. In this article, the terms "solvent" and "insolvent" are to be understood as informal synonyms for the summary measure of long-range actuarial balance.

4. The question of whether the trust funds reduce overall taxpayer burdens or whether they constitute a form of real savings is outside the scope of this study. These larger issues were indeed part of the debates over financing during this period, although the analysis presented here is independent of any particular view on these larger issues. Suffice it to observe that Administration policymakers viewed their financing options in the terms described here—that is to say, that the reserve was a device to reduce future payroll tax rates.

5. The current estimating period of 75 years was adopted in 1965 in an explicit effort to provide a projection period that encompassed not just the retirement of the youngest workers in the system, but their full period of benefit receipt as well.

6. Berkowitz (1983, 146–147) suggests that the council report contained no definition of what a "reasonable contingency reserve" might be precisely so that this ambiguous phrase could mean different things to different people—improving the prospects for a political compromise on the issue.

7. Because of their discontent with this ambiguity, officials from the Social Security Board lobbied for a specific provision authorizing the use of general revenues to be enacted into law as part of the 1943 freeze legislation (enacted in early 1944). This provision was law until it was repealed in the amendments of 1950, never having been used.

8. The 1935 income figure was a projected $2.5 billion in 1955, of which $640 million would be income from the reserve. The 1939 figure was a projected $2 billion income in 1955, of which only $169 million would be income from the reserve. Clearly, the program had become more pay-go in the short run; whether this was true also in the long run (beyond 1955) is impossible to say because of the missing long-range estimates.

9. Edwin Witte, almost single-handedly, strongly argued for the necessity of a large reserve fund (and implicitly certifying it through proposed long-range actuarial projections). But no one in the Congress or the Administration made an issue of the absence of long-range actuarial estimates from the legislative history of the 1939 law.

10. To see why this is so, assume that the total taxable payroll was $1,000 billion for the first 25 years of the period and $500 billion for the second 25 years. A split tax rate of 3 percent to 7 percent would yield $65 billion in tax revenue, whereas a single 5 percent rate for the full 50 years would yield $75 billion. So if the level premium rate was 5 percent, the split rate schedule would leave the program with a $10 billion shortfall, even though the tax rate averaged 5 percent over the estimating period.

(For simplicity's sake, the example does not include discussion of trust fund assets or interest income on those assets. However, when the actuaries set a level premium rate they generally considered trust fund balances and interest income as part of the computation of the level premium rate.) Before 1972, the level premium rate, as described here, was used in actuarial estimates. Starting in 1972, the actuaries shifted their methodology to an "average cost" technique, which is an arithmetic average of costs/income without regard to trust fund balances or interest income. This methodology is not the same as the pre-1972 actuarial technique, even though it is sometimes described using similar names. If trust fund assets are not included in the computation, the indicated "level premium" tax rate would be higher than otherwise required during periods of positive fund balances (Myers 1993, 416–417.)

11. Although they do not specifically address the issue of long-range estimates, both Berkowitz (1983) and Leff (1988) argue that Social Security's financing was part of an unsettled policymaking climate during this period, and perhaps we might think that the missing estimates could be explained by appeal to this unsettled policy climate. However, the principles of actuarial estimation were well-established long before this time (Myers 1954), and they were used in the 1935 Act and were used again starting in 1950. So the absence of expected principles of actuarial practice during this period cannot easily be explained by the generally unsettled nature of Social Security policymaking during this era.

12. Williamson, for his part, grew increasingly estranged from the Social Security program and from his peers. In 1947, he resigned to go into the private sector where he became an open advocate for his "social budgeting" model.


Altmeyer, Arthur J. 1943. Testimony before the Senate Committee on Finance on October 14, memorandum to W.R. Williamson, November 4. Manuscript in National Archives II, RG 47, Correspondence of the executive director, Social Security Board and the commissioner of Social Security, 1941–1948, Box 322, Folder 710.

———. 1966. The formative years of Social Security: A chronicle of Social Security legislation and administration, 1934–1954. Madison: University of Wisconsin Press, 103.

Annual Reports of the Social Security Trustees.1942–1945. For 1942, confer (cf). Table 3: 6; for 1943, cf. Table 4: 10; for 1944, cf. Table 6: 12—Note, the figure for 1944 is based on the average of the two estimates the Trustees provided; and for 1945, cf. Table 6: 13.

Berkowitz, Edward D. 1983. The first Social Security crisis. Prologue (Fall): 133–149.

———. 1996. Social Security and the financing of the American state. In Funding the modern American state, 1941–1995: The rise and fall of the era of easy finance, ed. W. Elliot Brownlee, 148–193. Cambridge: Cambridge University Press.

Bronson,D.C. 1939. Memorandum to G.R. Parker, dated October 26, subject: Federal Old-Age and Survivors Insurance Trust Fund—Your memorandum of October 18 to Mr. Corson. In NARA II, Correspondence of the executive director, Social Security Board and the commissioner of Social Security, Box 321, Folder 705.

Committee Report 1939. Social Security: Hearings relative to the Social Security Act Amendments of 1939, Committee on Ways and Means, House of Representatives, Volumes 1–3. Washington, DC: U.S. Government Printing Office.

U.S. Congress. 77th Congress, Second Session, Senate. 1942. Congressional Record (October 9).

Falk, I.S. 1947. Memorandum to Robert J. Myers, dated February 24, subject: Actuarial Study No. 23. In National Archives II, R.G. 47, Files of the Office of the Actuary 1946–1950, general correspondence.

Immerwahr, George E. 1945. Problems in federal Old-Age and Survivors Insurance. Transactions of the Actuarial Society of America XLVI(114): Part 2 (October).

Leff, Mark H. 1984. The limits of symbolic reform: The new deal and taxation 1933–1939. Cambridge: Cambridge University Press.

———. 1988. Speculating in Social Security futures: The perils of payroll tax financing, 1939–1950. In Social Security: The first half-century, eds. Gerald D. Nash, Noel H. Pugach, and Richard, F. Tomasson, 243–278. Albuquerque: University of New Mexico Press.

Myers, Robert J. 1954. The first United States government actuary and his successors. Transactions of the Society of Actuaries VI(16): 488–497 (October).

———. 1993. Social Security, fourth edition. Philadelphia: Pension Research Council and University of Pennsylvania Press.

Neustadt, Richard E., and Ernst May. 1988. Thinking in time: The uses of history for decisionmakers. New York: Free Press, 101–109.

Newspaper article. 1944. House Committee favors Dole over Social Security (paper unknown). Clipping in National Archives II, R.G. 47, Correspondence of the executive director and the chairman of the Social Security Board, Box 322, Folder 710 (November 28).

Pogge, O.C. 1947. Memorandum to Robert J. Myers, dated February 14, subject: Comments on Actuarial Study No. 23. National Archives II, R.G. 47, Files of the Office of the Actuary 1946–1950, general correspondence.

Schieber, Sylvester J., and John B. Shoven. 1999. The real deal: The history and future of Social Security. New Haven: Yale University Press.

Schmitter, Lyle L., and Betti C. Goldwasser. 1939. The revised benefit schedule under federal Old-Age Insurance. Social Security Bulletin 2(9): 3–12 (September).

Senate Report. 1939. The Social Security Amendments of 1939, Senate report no. 734. Washington, DC: U.S. Government Printing Office, 17 (July 7).

Social Security Administration. 1985. Annual Statistical Supplement to the Social Security Bulletin, 1985. Washington, DC: U.S. Government Printing Office.

Statement by the President. 1944. Manuscript in National Archives II, RG 47, Correspondence of the executive director, Social Security Board and the commissioner of Social Security, 1941–1948, Box 322, Folder 710 (December 16).

Technical Staff Report. 1946. Issues in Social Security: A report to the Committee on Ways and Means of the House of Representatives by the committee's Social Security technical staff. Washington, DC: U.S. Government Printing Office (January).

Tynes, Sheryl R. 1996. Turning points in Social Security. Stanford: Stanford University Press.

Williamson, W.R. 1939. Statement of W.R. Williamson, actuarial consultant to the Social Security Board. Social Security hearings relative to the Social Security Act Amendments of 1939. House Ways & Means Committee hearing report, 76th Congress, First Session, 3: 2473–2488.

———. 1943. Memorandum to Arthur J. Altmeyer, dated November 13, subject: Your memorandum of November 4 on the Senate Committee testimony. Manuscript in National Archives II, RG 47, Correspondence of the executive director, Social Security Board and the commissioner of Social Security, 1941–1948, Box 322, Folder 710.

———. 1947. Memorandum to Mr. Pogge, Mr. Perlman, Mr. Alpern, from: W.R. Williamson, dated February 18, subject: Costs in Social Security. Manuscript in National Archives II, R.G. 47, Files of the Office of the Actuary, Box 21.

Zelizer, Julian E. 1997. Where is the money coming from? The reconstruction of Social Security finance, 1939–1950. Journal of Policy History 9(4)399–424.

———. 1998. Taxing America: Wilbur D. Mills, Congress, and the State, 1945–1975. Cambridge: Cambridge University Press, 1998.