## Short-Range Actuarial Projections |
|||||

ACTUARIAL STUDY NO. 119
by Chris Motsiopoulosand Richard B. Tucker |
|||||

II. ASSUMPTIONS

Income to and outgo from the OASI and DI Trust Funds depend on many economic, demographic, and program-specific assumptions and methods. These include productivity, inflation, average earnings, fertility, mortality, net immigration, marriage, divorce, retirement patterns, and disability incidence and termination. Income depends on how these factors affect the size and composition of the working population and the general level of earnings. Similarly, outgo depends on how these factors affect the size and composition of the beneficiary population and the general level of benefits.

Estimates for the annual Trustees Report are prepared under a range of assumptions:

- Alternative I is characterized as
*low cost*—it assumes relatively rapid economic growth, low inflation, and favorable (from the standpoint of program financing) demographic conditions. - Alternative II is characterized as
*intermediate*—it is based on the principal economic and demographic assumptions set by the Social Security Board of Trustees, and it represents the Trustees' consensus estimates of the most likely future economic and demographic conditions. - Alternative III is characterized as
*high cost*—it assumes slow economic growth, more rapid inflation, and financially disadvantageous demographic conditions.

Sections II, III and IV of this study present estimates from the intermediate set of assumptions for the 2005 Trustees Report. Section V presents results based on the low cost and high cost alternatives.

Estimates for the Administration's Budget and the Budget's Mid-Session Review are prepared under one set of assumptions, based on principal economic assumptions set by OMB. Such estimates are not included in this study.

A. ECONOMIC ASSUMPTIONS

An econometric model designed by the Office of the Chief Actuary, Revenue Estimates and Economic Analysis Group projects values for key OASDI program-related variables based, in part, on the economic assumptions. These variables include total and average wages in OASDI covered employment, the number of workers in covered employment, automatic cost-of-living adjustments (COLAs), and average wages used for Social Security indexing purposes.

Tables II.1 and II.2 summarize respectively the principal economic assumptions and additional economic factors under the intermediate estimates, that are critical to the projections of the future financial status of the combined OASI and DI Trust Funds. In the discussion that follows, assumptions under the low cost and high cost alternatives are presented for comparison. For details regarding the alternative assumptions, refer to tables V.A1-V.A3 and V.B1-V.B3.

1. Productivity Assumptions

Total U.S. economy productivity is defined as the ratio of real gross domestic product (GDP) to hours worked by all workers.^{1} The rate of change in total productivity is a major determinant in the growth of average earnings. The annual increase in total productivity averaged 2.2, 1.2, 1.3, and 1.6 percent over the business cycles 1966-73, 1973-78, 1978-89, 1989-2000, respectively. The ultimate annual increases in productivity are assumed to be 1.9, 1.6, and 1.3 percent for the low cost, intermediate, and high cost assumptions, respectively.

For the intermediate assumptions, the annual change in productivity is assumed to decrease from 3.3 percent for 2004 to 2.0 percent for 2005 and 2006, then to an average of 1.8 percent for the years 2007 to 2009. Though declining, these changes are relatively strong by historical standards. After 2009, the annual change in productivity gradually decreases to the ultimate assumed level of 1.6 percent by 2013. For the low cost assumptions, the annual change in productivity decreases gradually from 3.2 percent for 2004 to the ultimate assumed level of 1.9 percent by 2012. For the high cost assumptions, the annual change in productivity decreases from 3.1 percent for 2004 to 0.5 percent for 2005. Thereafter, the annual change in productivity varies with the business cycle until reaching its ultimate growth rate of 1.3 percent for 2014.

2. Price Inflation Assumptions

Future changes in the Consumer Price Index for Urban Wage Earners and Clerical Workers (hereafter denoted as CPI) will directly affect the OASDI program through the automatic cost-of-living benefit increases. Future changes in the GDP chain-type price index (hereafter, the GDP deflator) affect the nominal levels of the GDP, wages, self-employment income, average earnings, and the taxable payroll.

Historically, the CPI has increased by an average of 4.5 percent for the 40 years from 1963 to 2003, while GDP deflator has increased by 4.0 percent during the same period. It should be noted that several methodological changes made by the Bureau of Labor Statistics in methods for computing the CPI since 1995 will tend to reduce the difference between the growth rates of these indices in the future.

The ultimate annual increases in the CPI are assumed to be 1.8, 2.8, and 3.8 percent for the low cost, intermediate, and high cost assumptions, respectively. These rates of increase reflect a growing belief that future inflationary shocks will more likely be offsetting and that future monetary policy will be more like the recent past, with its strong emphasis on holding the growth rate in prices to relatively low levels.

For each alternative, the ultimate annual increase in the GDP deflator is assumed to be equal to the sum of the annual increases in the CPI and a -0.3 percentage point price differential. This differential is based primarily on methodological differences in the construction of the two indices. Hence, the ultimate annual increase in the GDP deflator is 1.5, 2.5 and 3.5 percent for the low cost, intermediate and high cost assumptions, respectively.

For the intermediate assumptions, the annual change in the CPI is assumed to decrease from 2.6 percent for 2004 to 2.2 percent for 2005. Thereafter, the annual change in the CPI increases gradually to the assumed ultimate rate of 2.8 percent as of 2008. For the low cost assumptions, the annual change in the CPI decreases from 2.0 percent for 2005 to 1.7 percent for 2006, then rises to the assumed ultimate rate of 1.8 percent for 2007. For the high cost assumptions, the annual change in the CPI mostly increases from 2.5 percent for 2004 to 5.7 percent for 2009, then decreases to its assumed ultimate rate of 3.8 percent as of 2013.

3. Average Earnings Assumptions

The level of average (nominal) earnings in OASDI covered employment for each year has a direct effect on the size of the taxable payroll and on the future level of average benefits. In addition, increases in the level of average wages in the U.S. economy directly affect the indexation, under the automatic-adjustment provisions in the law, of the OASDI benefit formulas, the contribution and benefit base, the exempt amounts under the retirement earnings test, the amount of earnings required for a quarter of coverage, and under certain circumstances, the automatic cost-of-living benefit increases.

This covered earnings concept is closely linked to average U.S. earnings, defined as the ratio of the sum of total U.S. wage and salary disbursements and proprietor income to the sum of total U.S. military and total civilian (household) employment. The growth rates in average U.S. earnings can be broken down into the growth rates for total U.S. economy productivity and the GDP price index (see previous two sections), and the growth rates for other components, including average hours worked, the ratio of earnings to compensation (which includes fringe benefits), and the ratio of compensation to GDP.

The average annual change in average hours worked was -0.2 percent over the last 40 years. The ultimate annual rates of change for average hours worked are assumed to be 0.1, 0.0, and -0.1 percent for the low cost, intermediate, and high cost assumptions, respectively.

The average annual change in the ratio of earnings to compensation was -0.3 percent from 1963 to 2003. For wage workers, the assumed ultimate annual rates of change in the ratio of earnings to compensation are -0.1, -0.2, and -0.3 percent, for the low cost, intermediate, and high cost assumptions, respectively.

Thus, the ultimate projected annual growth rate in average U.S. earnings is about 3.9 percent for the intermediate assumptions. This reflects assumed ultimate annual growth rates of about 1.6, -0.2, 0.0, and 2.5 percent for productivity, the ratio of earnings to compensation, average hours worked, and the GDP deflator, respectively. Similarly, the ultimate projected annual growth rate in average nominal U.S. earnings is 3.4 percent for the low cost assumptions and 4.5 percent for the high cost assumptions.

Over long periods of time the average annual growth rates in average U.S. earnings and average earnings in OASDI covered employment are expected to be very close to the average annual growth rates in the average wage in OASDI covered employment (henceforth the average covered wage). Thus, the assumed ultimate annual growth rates in the average covered wage are 3.4, 3.9, and 4.4 percent for the low cost, intermediate, and high cost assumptions, respectively. For the intermediate assumptions, the annual rate of change in the average covered wage is assumed to rise from the estimated 3.8 percent increase for 2004 to an average of 4.3 percent over the 2005 to 2008 period, then to fall to an average of 4.1 percent from 2009 to 2013. Thereafter, the annual rate of change in the average covered wage moves to its assumed ultimate annual growth rate of 3.9 percent for 2014.

4. Assumed Real-Wage Differentials

For simplicity, real increases in the average OASDI covered wage have traditionally been expressed in the form of real-wage differentials—i.e., the percentage change in the average covered wage minus the percentage change in the CPI. This differential is closely related to assumed growth rates in average earnings and productivity, which are discussed in the previous section. Over the 40-year period, 1964-2003, the real-wage differential averaged 1.0 percentage point. The assumed ultimate annual average covered real-wage differentials are 1.6, 1.1, and 0.6 percentage point(s) for the low cost, intermediate, and high cost assumptions, respectively.

Based on preliminary data, the real-wage differential was 1.2 percentage points for 2004. For the intermediate assumptions, the real-wage differential is projected to rise to about 2.1 and 2.2 percentage points in 2005 and 2006, respectively. The real-wage differential is then projected to fall to 1.8 percentage points for 2007, 1.5 percentage points for 2008, and to the ultimate assumed differential of 1.1 percentage points (3.9 percent nominal wage growth less 2.8 percent CPI inflation) by 2014.

For the low cost assumptions, the real-wage differential is assumed to be in the range of 1.2 percentage points to 2.4 percentage points between 2004 and 2012, moving to the ultimate assumed real-wage differential of 1.6 percentage points thereafter. For the high cost assumptions, the real-wage differential for the short-range period is projected to fluctuate between -0.7 and 2.6 percentage points, eventually stabilizing at about 0.6 percentage point for 2013 and later.

5. Labor Force and Unemployment Projections

The civilian labor force is projected by age, sex, marital status, and presence of children. Projections of the labor force participation rates for each subgroup take into account the percentages of the population that are disabled or in the military, the levels of Social Security retirement benefits, the state of the economy, and changes in life expectancy. The projections also include a "lagged-cohort effect" that applies changes in participation rates for a cohort at a specific age (relative to earlier cohorts at the same age) to participation rates for that cohort at older ages.

The annual rate of growth in the size of the labor force decreased from an average of about 2.1 percent during the 1970s and 1980s to about 1.2 percent from 1990 to 2003. Further slowing of labor force growth is projected due to a substantial slowing of growth in the working age population in the future—a natural consequence of the baby-boom generation approaching retirement and the succeeding lower-birth-rate cohorts reaching working age. Under the intermediate assumptions, the labor force is projected to increase by about 0.9 percent per year, on average, through 2014.

The unemployment rate is in the most commonly cited form, the civilian rate. The total unemployment rate presented reflects the projected levels of unemployment for various age-sex subgroups of the population. The unemployment rate for each subgroup is projected based on a specification (consistent with Okun's Law) relating changes in the unemployment rate to the changes in the business cycle, as measured by the ratio of the actual to potential GDP. For each alternative, the total unemployment rate is projected to move toward the ultimate assumed rate as the economy moves toward the long-range sustainable growth path.

The ultimate assumed unemployment rates are 4.5, 5.5, and 6.5 percent for the low cost, intermediate, and high cost assumptions, respectively, and they are reached in 2014.

6. Domestic Product Projections

The real growth rate in gross domestic product (GDP) equals the combined growth rates for total employment, productivity, and average hours worked. Total employment is the sum of the U.S. Armed Forces and total civilian employment, which is based on the projected total civilian labor force and unemployment rates. For the 40-year period from 1963 to 2003, the average growth rate in real GDP was 3.3 percent, combining the approximate growth rates of 1.7, 1.8, and -0.2 percent for its components—total employment, productivity, and average hours worked, respectively.

For the intermediate assumptions, the average annual growth in real GDP is projected to be 2.7 percent over the short-range projection period (2005-14), a slower rate than the 3.3 percent average observed over the historical 40-year period (1963-2003). This slowdown is primarily due to slower projected growth in total employment. For the low cost assumptions, annual growth in real GDP is projected to average 3.3 percent over the decade ending in 2014. The relatively faster growth is due mostly to a higher assumed rate of growth in worker productivity. For the high cost assumptions, real GDP is assumed to fall in the last quarter of 2004 and the first and second quarters of 2005, resulting in a total decline in real GDP of 1.2 percent. After 11 quarters of recovery, a second recession, with a total decline in real GDP of 1.7 percent, is assumed to begin in the second quarter of 2008 and last 3 quarters. After the second recession, a moderate economic recovery is assumed through 2011, with continued modest economic growth thereafter. For the high cost assumptions, annual growth in real GDP is projected to average 2.0 percent for the decade ending in 2014.

7. Interest Rate Projections

Investment policy for the trust funds is set by law (see Interest in section IV. Trust Fund Income and Outgo for more detail). Non-marketable securities called *special issues* are issuable only to the trust funds and include *certificates of indebtedness*—short-term securities maturing within 12 months of issue—and bonds maturing 1 to 15 years in the future. Special-issue investments bear interest rates determined by a formula. The current formula sets the rate applicable in a given month to the average market yield on marketable interest-bearing securities of the Federal government which are not due or callable until after 4 years from the date the rate is determined. This formula became effective with the October 1960 rate. The rate is determined on the last business day of the month preceding the month of issue.

For the 10-year short-range projection period, nominal interest rates are projected based on changes in the business cycle and in the CPI. Under the intermediate assumptions, the nominal interest rate is projected to rise from 4.3 percent for 2004 to 5.6 for 2008 and 2009, and to 5.7 percent for 2010, reflecting a recovering economy along with a higher rate of inflation. Thereafter, the nominal interest rate rises to the ultimate assumed level of 5.8 percent for 2011. For the low cost assumptions, the average annual nominal interest rate is assumed to reach an ultimate level of about 5.5 percent for 2012. For the high cost assumptions, it is assumed to peak at 8.7 percent for 2010, and then decline to an ultimate rate of about 6.0 percent for 2013.

Monthly and average annual nominal interest rates are shown in table II.3, along with the annual effective rate earned by the overall assets of the trust funds in each year. This rate is calculated as the amount of interest earned during the year divided by the average level of assets for the year. Since almost all of the interest to the trust funds is paid at the end of June and December, an average balance is calculated for each calendar half-year based on detailed cash flow data of the trust funds. Dividing total interest earned by the sum of the average balances for each half-year produces the desired effective rate.

Effective rates can also be estimated by calculating the dollar-weighted rate of return for the year—or for each half- year period compounded to yield an annual rate. The dollar-weighted return is the interest rate that satisfies the following equation:

*assets _{bop}*+

In words, the desired effective rate equates (i) the assets at the beginning of the period plus the present value of trust fund income, to (ii) the present value of trust fund outgo plus the present value of assets at the end of the period. Income and outgo are discounted from various times throughout the month based on detailed cash flow data.

B. AUTOMATICALLY ADJUSTED PROGRAM AMOUNTS

The Social Security Act specifies that certain program amounts affecting the determination of OASDI benefits are to be adjusted annually, in general, to reflect changes in the economy. The law prescribes specific formulas that, when applied to reported statistics, produce automatic revisions in these program amounts and hence in the benefit-computation procedures. These automatic adjustments are based upon measured changes in the national daverage wage index (AWI)^{2} and the CPI.^{3} In this section, values are shown for program amounts that are subject to automatic adjustment, from the time that such adjustments became effective through 2014.

Tables II.4 and II.5 present the historical and projected values, under the intermediate assumptions, of the CPI-based benefit increases, as well as the AWI series and the values of many of the wage-indexed program amounts. Projections under the alternative assumptions are presented in tables V.A4-V.A5 and V.B4-V.B5.

Table II.4 includes:

- The annual percentage increases which have been applied to OASDI benefits under automatic cost-of-living adjustment provisions in the Social Security Act. The benefit increase for December 2004 is 2.7 percent, based on the increase in the not seasonally adjusted CPI-W from the third quarter of 2003 through the third quarter of 2004. The increase first applies to the December benefit, which is payable in January 2005.
- The annual levels of and percentage increases in the AWI. Under section 215(b)(3) of the Social Security Act, the AWI for each year after 1950 is used to index the taxable earnings of most workers first becoming eligible for benefits in 1979 or later. This procedure converts a worker's past earnings to approximately their equivalent values near the time of the worker's retirement or other eligibility, and these indexed values are used to calculate the worker's benefit. The AWI for 2003 is $34,064.95, and it is used to adjust most of the other program amounts that are subject to the automatic-adjustment provisions.
- The OASDI contribution and benefit base—the maximum amount of earnings subject to the OASDI payroll tax in the specified year. The OASDI contribution and benefit base for 2005 is $90,000. Employees and employers will each pay a tax rate of 6.2 percent on wages up to the wage base—that is, a person with $90,000 or more in wages will pay the maximum $5,580.00 in OASDI taxes in 2005. Refer to appendix A for details on determining the wage base.
- The retirement earnings test exempt amounts—the annual amount of earnings below which beneficiaries are not subject to benefit withholding. The retirement test exempt amounts for 2005 are $12,000 for beneficiaries under normal retirement age (NRA), and $31,800 for those at NRA. The latter amount is provided by Public Law 104-121. Under the retirement test, beneficiaries under NRA will have a $1 reduction in benefits for every $2 in earnings over $12,000. Public Law 106-182 eliminated the earnings test for workers attaining NRA through age 69 in 2000 or later. However, the limits still apply to all months prior to the beneficiary's birthday in the year he or she attains NRA, specifically a $1 reduction in benefits for every $3 in earnings over $31,800.

Table II.5 includes:

- The bend points used in determining a beneficiary's PIA are $627 and $3,779. For a retired worker retiring at age 62 in 2005, the PIA is determined as follows: (90 percent of the first $627 of AIME) + (32 percent of AIME in excess of $627 but not in excess of $3,779) + (15 percent of AIME in excess of $3,779).
- The bend points used in determining a beneficiary's maximum family benefit (MFB) are $801, $1,156, and $1,508. For a retired worker retiring at age 62 in 2005, the MFB is determined as follows: (150 percent of the first $801 of PIA) + (272 percent of PIA in excess of $801 but not in excess of $1,156) + (134 percent of PIA in excess of $1,156 but not in excess of $1,508) + (175 percent of PIA in excess of $1,508).
- The amount of earnings required to receive one quarter of coverage in 2005 is $920. In order to be insured for benefits, the beneficiary must acquire a requisite number of quarters of coverage.
- The "old-law" contribution and benefit base for 2005 is $66,900. This is the base that would have been in effect before the enactment of the 1977 amendments. The old-law base is used in determining special-minimum benefits for certain workers who have many years of low earnings in covered employment, and also in calculating OASDI benefits for workers who are eligible to receive pensions based on noncovered employment. In addition, it is used for certain purposes under the Railroad Retirement program, and in determining the maximum pension guaranteed under the Employee Retirement Income Security Act (ERISA).

Details on determining each year's average wage index and related OASDI program amounts are described in a separate actuarial note.^{4}

In addition to the program amounts affecting the determination of OASDI benefits that reflect changes in the economy, there are certain legislated changes that have affected, and will affect, benefits. Two such changes are the scheduled increases in the normal retirement age and in the delayed retirement credits. The scheduled changes in these two important items and their effect on benefits expressed as a percentage of PIA are shown in Appendix B.

C. DEMOGRAPHIC ASSUMPTIONS

The Demographic Analysis Group, Office of the Chief Actuary, prepares estimates of the Social Security area population based on the principal demographic assumptions set by the OASDI Trustees. Table II.6 summarizes these principal demographic assumptions under the intermediate estimates. In the discussion that follows, principal demographic assumptions under the low cost and high cost alternatives are presented for comparison. For details regarding the alternative assumptions, refer to tables V.A6 and V.B6.

1. Fertility Rate

During the 1976-90 period, the total fertility rate rose from 1.74 children per woman to 2.07. Since then, the total fertility rate has remained fairly stable, around 2.0 children per woman.

For the intermediate projection, the assumed ultimate total fertility rate of 1.95 children per woman is attained in 2029 after a gradual decline from the 2000 level of 2.06. In contrast, ultimate levels of fertility for low cost and high cost alternatives by 2029 are 2.2 and 1.7, respectively. Estimates are based on recent fertility rates and evolving trends in age-specific birth rates.

2. Death Rate

Since 1979, the age-sex-adjusted death rate declined at an average rate of 0.7 percent per year.

The age-sex-adjusted death rate is assumed to continue decreasing steadily during the entire projection period. The intermediate assumption predicts a total reduction of 42 percent from the 2004 level by 2079. In contrast, over the same period the low cost alternative predicts a total reduction of 20 percent; the high cost alternative predicts a total reduction of 62 percent.

Of the estimated 2.57 million deaths that occurred in 2004, 1.93 million (75 percent) were individuals age 65 or older. Since changes in mortality at these ages directly affect benefit payments, death rates for ages 65 or older are an integral part of the future financial status of the OASDI program. Since 1900, mortality at ages 65 or older has declined roughly 0.72 percent per year. Since 1979, the rate of decline has been less dramatic, about 0.47 percent annually. However, based on potential future changes in health care and life style, the assumed ultimate average annual rate of decline in age-sex-adjusted death rates for ages 65 or older is 0.67 percent under the intermediate assumptions, 0.29 percent under the low cost assumptions, and 1.17 percent under the high cost assumptions.

3. Life Expectancy

Life expectancy, or average remaining number of years expected prior to death, is calculated in two different forms, for two separate purposes.

Period life expectancy is calculated for a given year using the actual or expected death rates at each age for that year. It is a useful summary statistic for illustrating the overall level of the death rates experienced in a single year. It is thus closely related to the age-sex-adjusted death rate discussed above. Period life expectancy for a particular year may be viewed as the expected remaining life at a selected age only if it is assumed that there is no change in death rates after that year.

Cohort life expectancy truly answers the question "What is the expected average remaining lifetime for an individual at a selected age in a given year?" Cohort life expectancies are calculated using death rates not from a single year, but from the series of years in which the individual will actually reach each succeeding age if he or she survives.

Under intermediate assumptions, period life expectancy for a newborn increases from 74.6 years in 2004 to 81.6 years by 2079 for a male; and from 79.6 years in 2004 to 85.2 years in 2079 for a female. Period life expectancy at age 65 increases from 16.2 years in 2004 to 20.4 years by 2079 for a male; and from 19.0 years in 2004 to 22.8 years in 2079 for a female.

Under the low cost alternative, period life expectancy for a newborn increases to 78.3 years for a male and 82.1 years for a female, by 2079. Period life expectancy at age 65 increases to 17.8 years for a male and 20.4 years for a female, by 2079.

Under the high cost alternative, period life expectancy for a newborn increases to 86.1 years for a male and 89.2 years for a female, by 2079. Period life expectancy at age 65 increases to 23.7 years for a male and 26.0 years for a female, by 2079.

Cohort life expectancies are somewhat greater than period life expectancies for the same year. This is because death rates for any given age tend to decline as time passes and the cohort grows older.^{5}

4. Immigration

Total net immigration is assumed to be 1,000,000 persons per year for the intermediate projection. The assumed level of net annual immigration is the combination of 600,000 net legal immigrants per year and 400,000 net other immigrants per year. In contrast, annual net immigration for the low cost and high cost alternatives is estimated to be 1,400,000 and 722,500, respectively.

Actual future values for the demographic factors will likely exhibit cyclical fluctuations; the values assumed here are intended to represent the average experience for such cycles. Details of the various factors used to derive the demographic assumptions are described in a separate actuarial study.^{6}

D. PROGRAMMATIC ASSUMPTIONS

Table II.9 summarizes the principal programmatic assumptions under the intermediate estimates. In the discussion that follows, principal programmatic assumptions under the low cost and high cost alternatives are presented for comparison. For details regarding the alternative assumptions, refer to tables V.A7 and V.B7.

1. Covered Workers and Coverage Rate

A covered worker in a given year is any worker with some earnings in that year creditable for Social Security purposes. The projection of the number of covered workers is developed within the econometric model referred to earlier. The coverage rate summarizes the number of covered workers during the year as a percentage of the working-age population in the Social Security area. Table II.7 shows historical and projected counts of covered workers, and table II.8 shows analogous coverage rates.

Fully insured status is required of a worker for most types of OASDI benefits. This status is obtained by earning one quarter of coverage (QC)^{7} for each year after attainment of age 21 and before the earliest of (i) attainment of age 62, (ii) onset of disability, or (iii) death. The fully insured population is developed according to the following steps:

1. Using data from the CWHS (Continuous Work History Sample) on quarters of coverage, historical arrays are established, by sex and single year of age, which represent counts of newly insured workers, i.e. workers who have just attained sufficient quarters of coverage to be insured (for ages 62 or older, this means 40 or more quarters of coverage). Because of unavoidable lags in earnings data for recent historical years, the newly insured counts are revised each year for the three most recent years.

2. Newly insured counts are then divided by covered worker counts, to produce arrays of newly insured rates. For each sex and cohort, a weighted average of these rates is formed for the five most recent historical years. These weighted averages are then multiplied against projected covered worker numbers, resulting in an extension into the projection years of the newly insured worker arrays.

3. The newly insured population arrays are then accumulated, by summing over each sex and cohort from the year that fully insured status is attained, and applying survival rates. This produces a preliminary set of fully insured population arrays.

4. The preliminary fully insured arrays are divided, cell by cell, by arrays of the Social Security area population, to produce preliminary arrays of fully insured rates. The latter are smoothed in various ways, and also "corrected" with the use of actual in force data for ages 65-69 over recent historical years. For ages 70 or older, we rely completely on actual in force data from recent years, operating under the assumption that by age 70, virtually all fully insured individuals have chosen to receive a benefit. The result is the final set of arrays of fully insured rates, and companion arrays of fully insured population.

Overall fully insured rates are shown in table II.9, whereas a detailed treatment of the fully insured population and fully insured rates, by sex and age group, appears in tables III.B2 and III.B3. The overall fully insured rate has grown from 71 percent at the end of 1975 to 82 percent at the end of 2004. It is expected to reach 87 percent by the end of 2014. This growth is attributable largely to the increase in the female labor force participation experienced during the last two decades, and projected to continue throughout the next decade.

3. Disability Insured

Disability insured status is required of a worker for eligibility for a primary disability benefit and auxiliary benefits to family members. To be insured for disability benefits, a worker must accrue a sufficient number of QCs to be deemed fully insured, and in addition must have worked recently in covered employment. The number of required QCs varies by age, and ranges from 6 out of the last 12 quarters, to 20 out of the last 40. The disability insured population expressed as a percentage of the fully insured population is projected by age and gender based on recent experience and labor force participation rates, as described in section III.A. The disability insured rates shown are ratios of the disability insured population to the total population aged 15 to normal retirement age, as of December 31. Overall, the percentage of the population that is disability insured is projected to steadily rise throughout the short-range period.

4. Disability Incidence Rate

The general decline in disability incidence between 1975-82 is attributable in part to a "stricter" program. Following a period of very low growth in incidence from 1983-89, the DI program experienced a surge in disability claims beginning in 1990 and incidence rates rose significantly through 1992. The period 1992-2000 is characterized by robust economic expansion, low unemployment, and legislative restrictions on certain impairments. These and other factors have contributed to the decline in applications and awards over this period. Recent sharp annual increases in incidence rates over the period 2001-04 represented a notable departure from the experience of the preceding decade, which generally showed modest annual declines in the age-sex-adjusted disability incidence rate. These increases were likely due in large part to the slowdown in economic growth experienced during that period.

However, a special administrative activity undertaken by SSA beginning in 2001 has also contributed slightly to the upsurge in disabled worker awards. This special workload was the result of discovering a substantial number of current or former recipients of Supplemental Security Income (SSI) benefits whose disability-insured status under the DI program was not previously recognized. As this special disability workload is processed over the next several years, the resulting disability awards will contribute to temporarily higher incidence rates than would have been expected as part of longer term underlying trends. After the last of the special workload cases is processed (2010), the incidence rates are projected to drop back somewhat from recent levels, consistent with an assumed return to faster economic growth.

The intermediate estimates show an incidence rate of roughly 5.0 disabled workers per thousand insured by the end of the short-range period (2014); the corresponding estimates for the low cost and high cost alternatives are 4.2 and 5.9 disabled workers per thousand insured, respectively.

5. Disability Termination Rate

Most disabled worker benefits are terminated as a result of death or conversion to retired worker benefits. Recovery is a smaller yet more volatile termination category, subject to significant swings as it is influenced by new legislation and budgetary constraints. The downward trend in the overall disability termination rate is the result of two significant trends in the DI rolls that have developed over the years: (i) falling death rates, and (ii) the prominence of mental impairments, which has led to an increase in younger and physically healthier beneficiaries. The result has been fewer conversions to retired worker benefits, as well as fewer deaths.

In 2003, the termination rate fell sharply reflecting a large decline in conversions as the normal retirement age began to increase 2 months each year until reaching age 66. As a result, a portion of the annual conversions are deferred from one year to the next until the transition is complete by 2009. The overall termination rate (reflecting all causes) is projected to remain near the 2003 level before increasing back to higher levels in 2009 when the gradual increase in the normal retirement age ceases. For more details, see Terminations: Conversion in section III. Benefit Payments.

The intermediate estimates show a termination rate (reflecting all causes) of roughly 91 terminations per thousand disabled worker beneficiaries by the end of the short-range period (2014); the corresponding estimates for the low cost and high cost alternatives are 95 and 87 terminations per thousand disabled worker beneficiaries, respectively.

Further details on trends in disability incidence and termination are described in the next section. A complete discussion can be found in Actuarial Study No. 118.^{8}

^{1} Historical levels of real GDP are from the Bureau of Economic Analysis' (BEA) National Income and Product Accounts (NIPA). Historical total hours worked is an unpublished series provided by the Bureau of Labor Statistics (BLS), and is for all civilian and military wage and salary workers and the self-employed.

^{2}For a more precise definition and history of the indexing series, see Title 20, Chapter III, section 404.211(c) of the Code of Federal Regulations, and Actuarial Note No. 103: Average Wages for Indexing Under the Social Security Act and the Automatic Determinations for 1979-81 (Eli Donkar, May 1981).

^{3}Details of these indexation procedures are published annually in the Federal Register, and are also available on the Internet at www.socialsecurity.gov/OACT/COLA/index.html.

^{4}Actuarial Note No. 133: *Average Wages for 1985-90 for Indexing Under the Social Security Act* (Michael Clingman and Jeffrey Kunkel, September 1992).

^{5}For details on period and cohort life expectancies, refer to Actuarial Study No. 116: Life Tables for the United States Social Security Area 1900-2100 (Felicitie Bell, and Michael Miller, August 2002).

^{6}Actuarial Study No. 112: *Social Security Area Population Projections: 1997* (Felicitie Bell, August 1997).

^{7}In 2005, a worker receives one QC (up to a maximum of four) for each $920 of annual covered earnings. This dollar amount is indexed each year by increases in average wages.

^{8}Actuarial Study No. 118: *Social Security Disability Insurance Program Worker Experience* (Tim Zayatz, June 2005).