May/June 1999
Social Security Finances Are in Fine Shape

By David Langer

For many years the public, the Congress, and the President have come to believe the Old-Age, Survivors, and Disability Insurance program, better known as Social Security, is faced with a severe financial problem. However, my analysis of the actuarial data produced by the Social Security Adminstration (SSA) during the past 20 years, has led me to conclude that the imputed financial problems are the result of projections resulting from faulty actuarial assumptions underlying the projections.

The actuaries have relied almost exclusively on macroeconomic speculation based on a dismal view of the future economy. The actuarial assumptions produced are thus overconservative, generate higher costs than warranted and create a large imbalance of the payout of benefits and expenses over income over a 75 year period. Using more realistic financial projections would have shown Social Security to be in balance, with income expected to adequately cover all payments due for the period.

The current deficiency as it is now measured is 2.19 percent of the wages that are subject to the Social Security payroll tax, or 1.10 percent each for workers and employers. Based on appropriate assumptions, such deficiency vanishes.

A great deal is at stake in whether there is in fact a 2.19 percent deficiency. Consider the following:

• A number of powerful financial institutions wish to convert Social Security from a system that provides a federally guaranteed defined benefit for life at retirement that is based on a worker's wages and employment period and includes automatic cost of living increases, to one whose benefits are based in substantial part on the performance of individual investment accounts. The latter approach has come to be known as privatization. Under one key proposed version, I estimate the financial institutions would realize income of about $240 billion in the first 13 years, as well as the enormous power that will come from managing what will ultimately be trillions of dollars. The leading argument for privatization is that it will correct the financial trouble facing Social Security.

• In addition to diverting a portion of Social Security taxes to individual accounts, the current benefit levels under the privatization schemes would be substantially reduced by cutting the benefit formula directly, or by raising the retirement age, or both. The privatization proponents deem this essential to "save" Social Security. Such reduction however, however, also enlarges the surplus that has been emerging annually from the Social Security trust fund. The surplus, along with that from certain other trust funds, is improperly classified under what is called the "unified budget," as an offset to the federal government's general budget deficit and currently totals around $100 billion.

• For example, if, as recently announced, the general budget showed a deficit of $30 billion, the public would be informed that there is a surplus of $70 billion under the unified budget. This accounting practice enables Congress and the President to authorize additional spending or a tax cut when there is in fact no actual income to offset them. The $100 billion surplus therefore becomes part of the national debt to be repaid with interest. The accumulated surplusses to date that are now part of the national debt are nearing $800 billion. This is expected to rise a couple of trillion dollars on the basis of the conservative actuarial assumptions now used to project Social Security costs. No small potatoes. • The federal treasury borrows the surplus directly from the Social Security trust fund, thereby avoiding the need to borrow from the private market. This keeps interest rates down, primarily benefiting major borrowers and the stock market. While this has not been significant in the past when the amounts in the trust fund were small, it is obviously becoming a noticeable factor in the world of finance.

Background on Actuarial Assumptions
Actuaries need to make cost projections for a variety of programs, such as life, health, and casualty insurance, private and public pension plans, and social insurance programs. As part of the calculation process, it is necessary to make assumptions about economic and demographic variables, such as the future investment yield and mortality rates. Private programs, such as life insurance and pension plans, tend to use conservative (high-cost) assumptions to build up greater reserves as a precaution against a major adverse event, such as bankruptcy or a poor economy. That way there will be adequate funds available to cover the benefit values accrued to date should the program have to be terminated..

A national social insurance program such as Social Security, on the other hand, doesn't share these concerns. One reasonably expects that the federal government, as opposed to a private employer or insurer, will continue indefinitely. Workers and employers are required by law to participate and thus cannot opt out of paying the Social Security payroll taxes. Further, if the economy sours as it did during the depression, the taxing power of the government will be available to sustain benefit payments. In comparison, the purchase of a life insurance policy and the continuation of premium payments are entirely voluntary; the insurer can't make you do either and may even go under.

The greater certainty of the federal umbrella eliminates the need for Social Security actuaries to make conservative financial projections. That is, there's no need to accumulate large reserves, and more realistic actuarial assumptions should thus be employed. The use of assumptions that contain a conservative or liberal bias can, in fact, raise questions. The former may make the program appear to be in poor shape when it isn't, while the latter may give the impression the program is financially strong and benefits can therefore be raised when they should not be.

But then how does one go about developing "realistic" actuarial assumptions for the SSA's actuaries' traditional annual projection of benefit costs and income for periods of up to 75 years? Two thoughts come to mind: It is not possible to make any reasonably realistic financial projection over that period, except by pure chance, so why bother? On the other hand, if the 75 year projections are deemed necessary (as in fact they are) for policy reasons to guide the president and Congress, they'll feel they have a professionally generated and therefore neutral basis for decision making, even if a shaky one. How should one choose the assumptions?

To assist actuaries in making the choice, the Actuarial Standards Board of the American Academy of Actuaries publishes guidelines in the form of Actuarial Standards of Practice (ASOP's) that contain these relevant provisions.

• ASOP No. 27 points out the need to take into account the purpose of the measurement and appropriate recent and long-term historical economic data. It also notes that, Because no one knows what the future holds with respect to economic and other contingencies, the best an actuary can do is to use professional judgment to estimate possible future economic outcomes based on past experience and future expectations...

• ASOP No. 32 emphasizes that the actuary should consider the actual past experience of the social insurance program, over both short- and long-term periods, also taking into account relevant factors that may create material differences in future experience. Moreover, If assumptions differ from recent experience...the report should discuss [the factors] that led to the choice of the assumptions used.

SSA's actuaries, therefore, have a fair degree of leeway in selecting assumptions for their financial projections. However, they must reasonably observe historical experience or, if they choose to depart from it, explain why. Now let's examine the specific approach used by SSA's actuaries.

The Assumptions
For the purpose of the board of trustees' annual report on the Social Security trust fund, the actuaries project the net of the income and costs of the program over a variety of periods up to 75 years. This is a fairly complex task requiring the adoption of assumptions about the future based on a host of economic and demographic factors (gross domestic product, real wage increases, the cost of living, mortality rates, and fertility rates). The discussion that follows will dwell on the 75 year period, since it yields the largest deficit, 2.19 percent of payroll subject to Social Security taxes, and is the one the privatizers have seized on to declare that Social Security faces a serious financial problem, if not bankruptcy. (A surplus of 0.61 percent arises from the 25 year projection and a deficit of 1.37 percent from the 50 year one.)

Consider the calculation of the principal cost item-benefit payments for each of the 75 years. For those already retired for age, assumptions are needed about the mortality rates of the retirees, spouses and other dependents, and the cost of living adjustment. For active workers, assumptions have to be made regarding future wages, the numbers who will survive to retirement age or become disabled or die, the number and age of their dependents, the changes in the cost of living and average salary, the unemployment rate, the interest rate to be earned on the trust fund, marriage and remarriage rates and the number born each year.

Analysis of the GDP Factors
The Gross Domestic Product (GDP) is the key economic assumption in estimating costs, since it is an indicator of the economic health of the country. The outlook for the GDP will also influence the level of four of the other six economic assumptions: the average wage, real wage differential, unemployment rate, and increase in labor force. (The other two are the consumer price index and interest rate.) As a result, the GDP factors projected by the SSA actuaries are a major determinant of how solvent Social Security will be estimated to be.

Historically, the GDP has been increasing at an average annual rate of 3.2 percent since at least 1930. The trustees’ 1998 annual report lists the actual GDP values each year starting with 1960, and their 3.3 percent average to date is about the same. The range of this actual average from 1960 to each of the past 20 report years, 1979 to 1998, has been fairly narrow, 2.9 percent to 3.5 percent. However, the SSA actuaries have been projecting a much lower 75 year future average intermediate GDP. It starts at 3.0 percent in 1979 (when the actual average was 3.5 percent), and dropping almost continuously thereafter to 1.5 percent in 1994, and it has remained at that level through 1998 (when the actual average was 3.3 percent).

The graph (GDP averages) compares the historical average GDP with the projected average GDP for each of the past 20 years on the three bases regularly reported- optimistic, intermediate, and pessimistic. The Intermediate level is normally the average of the two, and is presented as a kind of best estimate of the future; the other two represent possible scenarios on the low and high sides.

Note that the Optimistic projected average GDP in 1998 is 2.2 percent, one-third below the actual 3.3 percent average, whereas one might expect it to at least equal if not exceed the actual average. The 1998 pessimistic projected average is 0.7 percent, something that has never been seen in this country over any extended period, and it falls as low as 0.4 percent starting in 2050 and 0.3 percent in 2060. These values, outside the realm of a reasonable low-end, drag down the intermediate factors, distorting the intention that these be a fair middle ground.

How did the SSA actuaries arrive at their GDP factors?
First, let's examine to what extent the actuaries considered the actual experience over both short and long-term periods. For this purpose, the chart (GDP: Actual 10 and 30) compares the projected average intermediate GDP for each report year with the historical average of the actual GDP values for both the ten-year and thirty-year periods preceding each year. Note that the intermediate projected averages are well below the actual 30 year average values. The mean shortfall is 37 percent less over the full twenty years of reports and diminishes to 49 percent less in the last five years; the mean reduction from the actual 10-year value is 22 percent less for the full 20 years and drops to 40 percent for the last five years.

The most recent historical 10-year GDP average is 2.5 percent, while the 30-year average is 2.8 percent. In contrast, the intermediate GDP factors projected by the SSA actuaries average but 2.0 percent for10 years and then drop to a 1.4 percent average for the next 65 years-a 30 percent decline.

There is thus little apparent reliance on recent experience or longer-term experience. We need to look in the trustees' annual report for an explanation of why factors were chosen that differ so greatly from historical experience.

What one finds is a kind of rationale. For example, "The intermediate set of assumptions reflects the Trustees' consensus expectation of moderate economic growth throughout the projection period...[T]he annual growth in real GDP is assumed to average 2.0 percent over the short-range (1998-2007) projection period, a slower rate than the 2.4 percent average observed over the most recent historical ten-year period (1988-1997). This 0.4 percent slowdown is mostly due to slower projected growth in labor force and employment."

After 2007 there is slower GDP growth because of slower growth in the working age population as the "baby boom" generation approaches retirement and lower birth rates. A discussion then follows of the slower GDP growth being affected by the slower growth in labor force participation.

This explanation of future expectations in terms of macroeconomic elements is a venture onto a slippery slope. What's the likelihood of the success of the actuaries' predictions as opposed to simply relying on past experience?

Consider, for instance, the alternative view of the future proposed by demographic economist Diane J. Macunovich, an Associate Professor at Williams College, given at the 1995 National Academy of Social Insurance meeting on Social Security. She places her confidence in the work of her colleague, Richard A. Easterlin, in making long-term forecasts. Basically, Easterlin's theory states that a variable equal to the number of persons born in a particular period (birth cohort), relative to the number of parents in the same period (parent cohort), has a quantifiable effect on the wages of young males, fertility, marriage, divorce and labor force participation rates. There's a causal relationship as well to GDP growth, productivity, average wages, inflation, and interest rates. This enables projections as far out as perhaps 30 years.

The current indication, Dr. Macunovich notes, is that the Easterlin cycle is presently near a low point, and that forecasting over a 75 year period will tend to be pessimistic, or high cost. She believes that this has in fact occurred and has contributed to the actuaries to projecting the large 2.19 percent deficit.

On another front, additional differences in opinion arose from discussions of the experts on the Technical Panel on Assumptions and Methods appointed to assist the Advisory Council on Social Security in reviewing assumptions. The experts found themselves faced with a "fundamental uncertainty" about the direction of the economy, viewing both the historical record and their projections differently even after months of discussion.

Reasonable alternatives
SSA's actuaries thus gave little weight to actual economic history in developing actuarial assumptions for use in making their financial projections over the next 75 years and gave no real explanation for this deviation. They relied instead on macroeconomic scenarios based on an economy operating at a much lower average level than has been the case for nearly 70 years.

We have also seen that there are alternative points of view that can differ considerably, as occured among the members of the technical panel and Professors Macunovich and Easterlin. Even the Easterlin theory is deemed effective for a limited period of perhaps 30 years, leaving 45 years out of the 75 year projection period without any kind of crystal ball to illuminate them.

This suggests that, not having any better stars to guide us, it is best to place a greater degree of reliance on actual prior experience. The extensive period of data available on the GDP has the merit of being a blend of diverse phenomena, including a depression, recessions, periods of prosperity, World War II, the Korean and Vietnamese Wars, and globalization.

One possible approach would be to use a combination of history and the factors derived from the method adopted by the SSA actuaries, but giving appropriately greater weight to the former.

For example, the actual average GDP since 1930 is 3.2 percent and the projected average GDP the SSA'a actuaries used in its 1998 report is 1.5 percent . If we weight the past 3.2 percent by 80 percent and the projected1.5 percent by 20 percent, we get an average GDP factor of 2.9 percent. On the basis of an approximation given me by Harry Ballantyne, SSA's chief actuary, the 1.4 percent increase in the GDP from 1.5 percent to 2.9 percent would reduce the current projected deficit of 2.19 percent at the end of the 75-year period by 1.48 percent to 0.71 percent.

Continuing onward, there's a further reduction to be made of 0.47 percent due to the recent 0.25 percent downward technical adjustment in the CPI by the Bureau of Labor Statistics, which lowers the 0.71 percent deficit to 0.24 percent. The SSA actuaries deemed this to be in long-range actuarial balance, because it is smaller than their "minimum allowable balance" standard of 0.78 percent, found by taking 5 percent of the "summarized cost rate" of 15.64 for the full 75 year period.

Further support for the result of the above calculation, that the 75 year projection is now in long-range actuarial balance, can be found by noting that the optimistic set of assumptions generates a surplus position of 0.25 percent in comparison with the 2.19 percent deficit for the intermediate set. The average GDP assumption of 2.2 percent for the optimistic set is even less than the 2.6 percent used in the preceding paragraph. Since most of the other economic assumptions are related to the GDP, it is reasonable to consider that the optimistic set of assumptions is a more appropriate basis for projection than the intermediate set.

Using assumptions based on actual economic experience leads to the conclusion that there is no looming imbalance in Social Security's projected financial position and that the system is in fine shape. There's no reason to be agonizing over whether the program needs to be privatized to save it, or needs to make substantial cuts in benefits, or needs to raise the retirement age. There ought to be a debate, instead, over the method now in use for deriving the assumptions and about the wisdom of continuing to rely on the 75-year projection as a standard of solvency, given the use politicians have made of it.

Reprinted, courtesy of Contingencies, the magazine of the American Academy of Actuaries. May/June 1999