Social Security Reform: Implications of Different Indexing	 
Choices (14-SEP-06, GAO-06-804).				 
                                                                 
The financing shortfall currently facing the Social Security	 
program is significant. Without remedial action, program trust	 
funds will be exhausted in 2040. Many recent reform proposals	 
have included modifications of the indexing currently used in the
Social Security program. Indexing is a way to link the growth of 
benefits and/or revenues to changes in an economic or demographic
variable. Given the recent attention focused on indexing, this	 
report examines (1) the current use of indexing in the Social	 
Security program and how reform proposals might modify that use, 
(2) the experiences of other developed nations that have modified
indexing, (3) the effects of modifying the indexing on the	 
distribution of benefits, and (4) the key considerations	 
associated with modifying the indexing. To illustrate the effects
of different forms of indexing on the distribution of benefits,  
we calculated benefit levels for a sample of workers born in	 
1985, using a microsimulation model. We have prepared this report
under the Comptroller General's statutory authority to conduct	 
evaluations on his own initiative as part of a continued effort  
to assist Congress in addressing the challenges facing Social	 
Security. We provided a draft of this report to SSA and the	 
Department of the Treasury. SSA provided technical comments,	 
which we have incorporated as appropriate.			 
-------------------------Indexing Terms------------------------- 
REPORTNUM:   GAO-06-804 					        
    ACCNO:   A60904						        
  TITLE:     Social Security Reform: Implications of Different	      
Indexing Choices						 
     DATE:   09/14/2006 
  SUBJECT:   Comparative analysis				 
	     Federal social security programs			 
	     Financial analysis 				 
	     Pension plan cost control				 
	     Pensions						 
	     Population statistics				 
	     Program evaluation 				 
	     Retirees						 
	     Social security beneficiaries			 
	     Social security benefits				 
	     Social Security Program				 

******************************************************************
** This file contains an ASCII representation of the text of a  **
** GAO Product.                                                 **
**                                                              **
** No attempt has been made to display graphic images, although **
** figure captions are reproduced.  Tables are included, but    **
** may not resemble those in the printed version.               **
**                                                              **
** Please see the PDF (Portable Document Format) file, when     **
** available, for a complete electronic file of the printed     **
** document's contents.                                         **
**                                                              **
******************************************************************
GAO-06-804

     

     * Results in Brief
     * Background
     * Social Security Currently Indexes Both Benefits and Revenues
          * Program Did Not Use Indexing until 1970s
          * 1977 Amendments Created Indexing Approach of Current Social
               * Indexing Lifetime Earnings to Wages
               * Indexing Initial Benefit Formula to Wages
               * Indexing Benefits to Prices for Existing Beneficiaries
               * Indexing Maximum Taxable Earnings to Wages
          * Various Reform Proposals Include Indexing Provisions
     * A Variety of Indexing Approaches Highlight International Ref
          * Retirement Indexing Approaches in Other Countries Generally
          * Indexing Approaches Aim at Containing Costs
          * Indexing Approaches Affect Both Current and Future Beneficia
     * Indexing Can Be Used to Achieve Desired Distributional Effec
          * Proposals to Index Initial Benefits Have a Proportional Effe
          * Indexing Approaches Could Also Be Modified to Achieve Nonpro
          * Indexing Applied to Taxes Would Have Adequacy and Equity Con
          * Revising COLA for Existing Beneficiaries Would Have Importan
     * Key Considerations in Choosing an Index
          * Choice of a Particular Index Implies Assumptions about the A
          * Stability of Economic or Demographic Relationships Underlyin
          * The Treatment of Disabled and Survivor Beneficiaries Poses C
     * Concluding Observations
     * Agency Comments
     * Microsimulation Model
          * Description
          * Assumptions and Limitations
               * 2005 Social Security Trustees' Assumptions
               * Distributional Effects Over Time
               * Pre-retirement Mortality
          * Description of Alternative Policy Scenarios
               * CPI Indexing
               * Mortality Indexing
               * Dependency Indexing
               * Scaling to Achieve Comparable Levels of Solvency over 75 Yea
          * Data Reliability
     * Benchmark Policy Scenarios
          * Criteria
          * Tax-Increase-Only, or "Promised Benefits," Benchmark Policie
          * Benefit-Reduction-Only, or "Funded Benefits," Benchmark Poli
               * Scope
               * Phase-in Period
               * Defining the PIA Formula Factor Reductions
     * Program Did Not Use Indexing until 1970s
          * Ad Hoc Benefit and Tax Changes Had Sporadic Effects
     * Indexing in 1972 Amendments Built on Previous Ad Hoc Benefit
          * Approach Used for Ad Hoc Benefit Increases
          * 1972 Amendments Introduced Indexing
     * Indexing Approach Introduced Potential Instability in Benefi
     * GAO's Mission
     * Obtaining Copies of GAO Reports and Testimony
          * Order by Mail or Phone
     * To Report Fraud, Waste, and Abuse in Federal Programs
     * Congressional Relations
     * Public Affairs

Report to Congressional Addressees

United States Government Accountability Office

GAO

September 2006

SOCIAL SECURITY REFORM

Implications of Different Indexing Choices

Social Security Indexing Social Security Indexing Social Security Indexing
Social Security Indexing ing Social Security Indexing Social Security
Indexing Social Security Indexing Social Security Indexing Social Security
Indexing Social Security Indexing Social Security Indexing Social Security
Indexing Social Security Indexing Social Security Indexing Social Security
Indexing Social Security Indexing Social Security Indexing Social Security
Indexing Social Security Indexing Social Security Indexing Social Security
Indexing Social Security Indexing Social Security Indexing Social Security
Indexing Social Security Indexing Social Security Indexing Social Security
Indexing Social Security Indexing Social Security Indexing Social Security
Indexing Social Security Indexing Social Security Indexing Social Security
Indexing Social Security Indexing Social Security Indexing Social Security
Indexing Social Security Indexing Social Security Indexing Social Security
Indexing Social Security Indexing Social Security Indexing Social Security
Indexing Social Security Indexing Social Security Indexing Social Security
Indexing Social Security Indexing Social Security Indexing Social Security
Indexing Social Security Indexing Social Security Indexing Social Security
Indexing Social Security Indexing Social Security Indexing Social Security
Indexing Social Security Indexing Social Security Indexing Social Security
Indexing Social Security Indexing Social Security Indexing Social Security
Indexing Social Security Indexing Social Security Indexing Social Security
Indexing Social Security Indexing Social Security Indexing Social Security
Indexing Social Security Indexing Social Security Indexing Social Security
Indexing Social Security Indexing Social Security Indexing

GAO-06-804

Contents

Letter 1

Results in Brief 3
Background 5
Social Security Currently Indexes Both Benefits and Revenues 7
A Variety of Indexing Approaches Highlight International Reform Efforts 18
Indexing Can Be Used to Achieve Desired Distributional Effect 24
Key Considerations in Choosing an Index 38
Concluding Observations 43
Agency Comments 43
Appendix I Methodology 45
Microsimulation Model 45
Benchmark Policy Scenarios 52
Appendix II Background on Development of Social Security's Indexing
Approach 59
Program Did Not Use Indexing until 1970s 59
Indexing in 1972 Amendments Built on Previous Ad Hoc Benefit Increases 62
Indexing Approach Introduced Potential Instability in Benefit Costs 65
Related GAO Products 68

Tables

Table 1: Key Indexing Approaches under Current U.S. Social Security System
13
Table 2: Summary of Indexes and Automatic Adjustments Proposed in United
States 17
Table 3: Characteristics of Earnings-Related Public Pension
Programs-Selected Countries 23
Table 4: Application of So-called Scaling of PIA Factors for Aged
Dependency Ratio Benefit Reduction Index: An Example Using 2050 PIA
Formula Factors 51
Table 5: Summary of Benchmark Policy Scenarios 53
Table 6: Summary of Benchmark Policy Scenario Parameters 58
Table 7: Percentage Increases in OASI Benefits, Prices, and Wages, by
Effective Date of OASI Change, 1950-1971 60

Figures

Figure 1: Social Security Benefit Formula Replaces Earnings at Different
Rates 10
Figure 2: Social Security's Earnings Replacement Rates for Illustrative
Workers 12
Figure 3: Indexing Changes with a Larger Proportional Reduction Have a
Greater Impact on the Distribution of Benefits, but Scaling to Achieve
75-Year Solvency Illustrates That the Proportional Effects Have Similar
Results 26
Figure 4: Proportional Indexing Changes Would Maintain the Progressivity
of the Current Benefit Formula, but Could Reduce Adequacy (Initial
Benefits in 2050) 28
Figure 5: Lower-Income Individuals Would Fare Comparatively Better under
the Progressive Application of the CPI Index than under the CPI Index
Alone (Initial Benefits in 2050) 31
Figure 6: Scaling the Progressive Application of the CPI Index to Achieve
Equivalent Solvency Demonstrates That Most Individuals above a Certain
Point Would Receive About the Same Level of Benefits (Initial Benefits in
2050) 32
Figure 7: A Onetime Payroll Tax Increase Would Spread the Tax Burden More
Evenly across Cohorts than Gradual Increases through an Index 35
Figure 8: The Growth of $1,000 Benefit under the CPI and Two Alternatives
Illustrate That Those Beneficiaries Who Receive Benefits Longer Will Be
Affected the Most 37
Figure 9: Inflation-Adjusted Values of the Maximum Taxable Earnings Level
before Automatic Adjustments, 1937-1975 61
Figure 10: Percentage of Total Covered Earnings below Social Security's
Maximum Taxable Earnings Level, 1937-2005 62
Figure 11: Changes in Average Wage Index and Consumer Price Index,
1951-1985 66
Figure 12: Social Security Workers per Beneficiary 67

Abbreviations

ABM automatic balancing mechanism

AIME average indexed monthly earnings

AWI average wage index

COLA cost-of-living adjustment

CPI consumer price index

CPI-E consumer price index for older Americans

CPI-U consumer price index for all urban consumers

CPI-W consumer price index for urban wage earners and clerical workers

DI Disability Insurance

GDP gross domestic product

GEMINI Genuine Microsimulation of social Security and Accounts

OASDI Old-Age, Survivors, and Disability Insurance

OASI Old-Age and Survivors Insurance

OCACT Office of the Chief Actuary

OECD Organisation for Economic Co-operation and Development

PENSIM Pension Simulator

PIA primary insurance amount

PSG Policy Simulation Group

SSA Social Security Administration

SSASIM Social Security and Accounts Simulator

This is a work of the U.S. government and is not subject to copyright
protection in the United States. It may be reproduced and distributed in
its entirety without further permission from GAO. However, because this
work may contain copyrighted images or other material, permission from the
copyright holder may be necessary if you wish to reproduce this material
separately.

United States Government Accountability Office

Washington, DC 20548

September 14, 2006 September 14, 2006

The Honorable Charles E. Grassley Chairman The Honorable Max Baucus
Ranking Minority Member Committee on Finance United States Senate The
Honorable Charles E. Grassley Chairman The Honorable Max Baucus Ranking
Minority Member Committee on Finance United States Senate

The Honorable Jim McCrery The Honorable Jim McCrery

Chairman Chairman

The Honorable Sander M. Levin The Honorable Sander M. Levin

Ranking Minority Member Ranking Minority Member

Subcommittee on Social Security Subcommittee on Social Security

Committee on Ways and Means Committee on Ways and Means

House of Representatives House of Representatives

The Honorable John Warner United States Senate The Honorable John Warner
United States Senate

The total long-term financing shortfall currently facing the Social
Security program is significant and growing over time, thereby making
system reform an important priority. Once the Social Security trust fund
balances are exhausted in 2040, annual revenue will be sufficient only to
pay about 74 percent of promised benefits, according to the Social
Security trustees' 2006 intermediate assumptions. Benefit costs are
projected to exceed income in 2017, and thus trust fund securities will
need to be redeemed. This will require increased government revenue,
increased borrowing from the public, reduced spending in the rest of the
government, or some combination of these. Redeeming these securities will
have an adverse impact on the federal budget much sooner than the 2040
trust fund exhaustion date. The total long-term financing shortfall
currently facing the Social Security program is significant and growing
over time, thereby making system reform an important priority. Once the
Social Security trust fund balances are exhausted in 2040, annual revenue
will be sufficient only to pay about 74 percent of promised benefits,
according to the Social Security trustees' 2006 intermediate assumptions.
Benefit costs are projected to exceed income in 2017, and thus trust fund
securities will need to be redeemed. This will require increased
government revenue, increased borrowing from the public, reduced spending
in the rest of the government, or some combination of these. Redeeming
these securities will have an adverse impact on the federal budget much
sooner than the 2040 trust fund exhaustion date.

Many recent reform proposals have proposed modifications to the indexing
currently used in the Social Security program. Indexing is a way to link
the growth of benefits and/or revenues to changes in economic or
demographic variables. For example, initial benefits can be set to grow
with changes in average wages or changes in prices. Modifications to
indexing seek to slow the growth of benefits or increase the growth of
revenues, either of which would improve solvency. However, indexing does
not guarantee that the program will achieve and remain in long-term
financial balance. Proposals that would modify Social Security's indexing
Many recent reform proposals have proposed modifications to the indexing
currently used in the Social Security program. Indexing is a way to link
the growth of benefits and/or revenues to changes in economic or
demographic variables. For example, initial benefits can be set to grow
with changes in average wages or changes in prices. Modifications to
indexing seek to slow the growth of benefits or increase the growth of
revenues, either of which would improve solvency. However, indexing does
not guarantee that the program will achieve and remain in long-term
financial balance. Proposals that would modify Social Security's indexing
implicitly pose the question of whether and how such adjustments could
provide a mechanism to keep the program sustainably solvent and minimize
the need for periodic rebalancing of the program's finances. At the same
time, how it is done can affect the distribution of benefits between low
and high earners and across generations of workers.

Given the recent attention focused on indexing as a critical component of
reform, this report examines (1) the current use of indexing in the Social
Security program and how reform proposals might modify that use, (2) the
experiences of other developed nations that have modified indexing when
reforming their public pension systems, (3) the effects of indexing
modifications on the distribution of Social Security benefits, and (4) the
key considerations associated with modifying Social Security's indexing.

To examine the use of indexing in the Social Security program and how
reform proposals might modify the indexing, we conducted a literature
review and reviewed recent Social Security reform proposals. To examine
the experience of other developed nations that changed indexing when
reforming their own national pension systems, we reviewed the academic
literature and documentation on other countries' public pension systems.
To analyze the effects of different forms of indexing on the distribution
of benefits, we calculated benefit levels for a sample of workers using a
microsimulation model (see app. I for a more detailed discussion of our
scope and methodology).1 For this analysis, we selected four well-known
indexing approaches to illustrate the effects on the distribution of
benefits.2 To describe the distributional effects of the different
indexing approaches, we used our model to simulate benefits for workers
born in 1985.3 Consistent with our past work on Social Security reform,
and to illustrate a full range of possible outcomes, we used hypothetical
benchmark policy scenarios that would achieve 75-year solvency either by
only increasing payroll taxes (which simulated "promised benefits") or
only reducing benefits (which simulated "funded benefits").4 To determine
the key considerations associated with various forms of indexing, we
reviewed the literature and talked with relevant experts. We have prepared
this report under the Comptroller General's statutory authority to conduct
evaluations on his own initiative as part of a continued effort to assist
Congress in addressing the challenges facing Social Security. We conducted
our work between July 2005 and August 2006 in accordance with generally
accepted government auditing standards.

1 We used the GEMINI model under a license from the Policy Simulation
Group, a private contractor. GEMINI estimates individual effects of policy
scenarios for a representative sample of future beneficiaries. GEMINI can
simulate different reform features, including individual accounts with an
offset, for their effects on the level and distribution of benefits. See
appendix I for more detail on the modeling analysis, including a
discussion of our assessment of the data reliability of the model.

2 These are consumer price index (CPI) indexing, dependency ratio
indexing, mortality indexing, and a so-called progressive indexing
approach that uses different indexes at various earnings levels. See
appendix I for a discussion of these indexes and our scope and
methodology, as well as GAO, Social Security: Program's Role in Helping
Ensure Income Adequacy, GAO-02-62 (Washington, D.C.: Nov. 30, 2001), GAO,
Social Security Reform: Analysis of Reform Models Developed by the
President's Commission to Strengthen Social Security, GAO-03-310
(Washington, D.C.: Jan. 15, 2003), and GAO, Social Security: Distribution
of Benefits and Taxes Relative to Earnings Level, GAO-04-747 (Washington,
D.C.: June 15, 2004).

                                Results in Brief

While the initial Social Security program did not use automatic indexing,
it is now a key feature of the program's design, as well as a central
element of many proposals to reform the program. Under the current system,
the indexing provisions that affect most workers and beneficiaries relate
to (1) the formula used to calculate initial benefits for new
beneficiaries, (2) the cost-of-living adjustment (COLA) for existing
beneficiaries, and (3) the cap on taxable earnings. The benefit indexing
provisions help maintain relative standards of living across age groups
and protect the purchasing power of benefits over time. Various reform
proposals have suggested changes to all of these provisions. For example,
for future beneficiaries, some proposals would index initial benefit
levels to keep pace with price inflation rather than wages. This would
result in gradually declining earnings replacement rates but maintain the
purchasing power of current benefit levels across age groups, assuming
wages grow faster than prices on average over time.5 Other proposals
accept slowing the growth of initial benefits, in general, but seek to
protect benefit levels for the lowest earners, consistent with the
program's goal of helping ensure income adequacy. Proposals to change the
annual COLA for existing beneficiaries generally focus on making it
reflect inflation levels more accurately, with the presumption that this
would result in lower benefits. On the revenue side, proposals to increase
the cap on taxable earnings generally seek to raise revenue from higher
earners and avoid increasing tax rates for all workers.

3 We focused on workers born in 1985 because all prospective program
changes under all alternative policy scenarios would be almost fully
phased in for these workers.

4 See appendix I for a complete description of our benchmark policy
scenarios.

5Earnings replacement rates measure the extent to which retirement income
replaces pre-retirement income for particular individuals and thereby
helps them maintain a pre-retirement standard of living.

Other countries' efforts to reform their national pension systems reveal a
diversity of indexing approaches. Countries with relatively high
contribution rates tend to focus on methods that reduce benefits to
address the financial solvency of their pension systems. Although most
Organisation for Economic Co-operation and Development (OECD) countries
index past earnings to reflect wage growth in computing initial retirement
benefits, some now use a price growth index (France, Belgium, and South
Korea), or an index that blends price and wage growth (Portugal, Poland,
and Finland). Some benefit formulas contain a measure of life expectancy
that reduces payments to new retirees in accordance with increases in
longevity (Sweden, Italy, and Poland). Changes to indexing approaches
abroad sometimes affect both current and future retirees. Germany includes
a "sustainability" factor that lowers pension amounts for both new and old
retirees when the number of workers paying into the system declines
relative to those drawing benefits. Similarly, other national systems rely
on automatic balancing mechanisms that modify both the future benefits of
workers and the benefits of current pensioners (Sweden, Japan).

Indexing can have different effects on the distribution of benefits and on
the relationship between contributions and benefits, depending on how it
is applied. Regardless of the index, adjusting the initial benefit level
through the benefit formula typically would have a proportional effect,
with constant percentage changes at all earnings levels, on the
distribution of benefits. However, indexing can also be used to achieve
specific distributional goals. For example, so-called progressive indexing
applies different indexes at different earnings levels to adjust benefits
of higher-income earners more than the benefits of lower-income earners.
Indexing payroll tax rates would have distributional effects across
generations, maintaining the existing distribution of benefits but instead
affecting equity measures like the ratio of benefits to contributions
across age cohorts. In this case, younger cohorts would have lower ratios,
because they would receive lower benefits relative to their contributions.
Finally, proposals that modify the indexing of annual COLAs for existing
beneficiaries would have adverse distributional effects for groups with
longer life expectancies, such as women, but these individuals would still
receive higher lifetime benefits since they live longer. In addition,
disabled worker beneficiaries, especially those who receive benefits for
many years, would also experience lower benefits because such proposals
would typically reduce future benefits, and this effect compounds over
time.

Indexing raises other important considerations about the program's role,
the stability of economic or demographic relationships underlying the
index, and the treatment of disabled worker beneficiaries. The choice of
the index implies certain assumptions about the appropriate level of
benefits and taxes for the program. Thus, if the current indexing of
initial benefits to wage growth was changed to track price growth, there
is an implication that the appropriate level of benefits is one that
maintains purchasing power over time rather than the current approach that
maintains replacement rates. The solvency effects of an index are
predicated upon the relative stability and historical trends of the
underlying economic or demographic relationships implied by the index. For
example, the 1970s were a period of economic instability in which actual
inflation rates and earnings growth diverged markedly from past
experience, with the result that benefits unexpectedly grew much faster
than revenues. Finally, since the benefit formula for Social Security
retirement and disability benefits are linked, an important consideration
of any indexing proposal, as with any other change to benefits, is its
effect on the benefits provided to disabled workers. Disabled worker
beneficiaries typically become entitled to benefits much sooner than
retired workers and under different eligibility criteria. An index that is
designed to improve solvency, for example, by adjusting retirement
benefits, could also result in large reductions to disabled workers, who
often have fewer options to obtain additional income from other sources.

                                   Background

Title II of the Social Security Act, as amended, establishes the Old-Age,
Survivors, and Disability Insurance (OASDI) program, which is generally
known as Social Security. The program provides cash benefits to retired
and disabled workers and their eligible dependents and survivors. Congress
designed Social Security benefits with an implicit focus on replacing lost
wages. However, Social Security is not meant to be the sole source of
retirement income; rather it forms a foundation for individuals to build
upon. The program is financed on a modified pay-as-you-go basis in which
payroll tax contributions of those currently working are largely
transferred to current beneficiaries. Current beneficiaries include
insured workers who are entitled to retirement or disability benefits, and
their eligible dependents, as well as eligible survivors of deceased
insured workers. The program's benefit structure is progressive, that is,
it provides greater insurance protection relative to contributions for
earners with lower wages than for high-wage earners. Workers qualify for
benefits by earning Social Security credits when they work and pay Social
Security taxes6; they and their employers pay payroll taxes on those
earnings. In 2005, approximately 159 million people had earnings covered
by Social Security, and 48 million people received approximately $521
billion in OASDI benefits.

Currently, the Social Security program collects more in taxes than it pays
out in benefits. However, because of changing demographics, this situation
will reverse itself, with the annual cash surplus beginning to decline in
2009 and turning negative in 2017. In addition, all of the accumulated
Treasury obligations held by the trust funds are expected to be exhausted
by 2040.7 Social Security's long-term financing shortfall stems primarily
from the fact that people are living longer and labor force growth has
slowed.8 As a result, the number of workers paying into the system for
each beneficiary has been falling and is projected to decline from 3.3
today to about 2 by 2040. The projected long-term insolvency of the OASDI
program necessitates system reform to restore its long-term solvency and
assure its sustainability. Restoring solvency and assuring sustainability
for the long term requires that either Social Security gets additional
income (revenue increases), reduces costs (benefit reductions), or
undertakes some combination of the two.

To evaluate reform proposals, we have suggested that policy makers should
consider three basic criteria:9

1. the extent to which the proposal achieves sustainable solvency and how
the proposal would affect the economy and the federal budget;

6 In 2006, workers receive 1 credit for each $970 of earnings, up to the
maximum of 4 credits per year. To be eligible for retirement benefits a
worker needs 40 credits.

7 These estimates are based on the Social Security trustees' 2006
intermediate, or best-estimate, assumptions.

8 Life expectancy has increased fairly steadily since the 1930s, and
further increases are expected. Increases in life expectancy vary by
gender, education, and earnings. Women, highly educated individuals, and
higher-income individuals generally experience greater life expectancy.

9 See GAO, Social Security: Criteria for Evaluating Reform Proposals,
GAO/T-HEHS-99-94 (Washington, D.C.: Mar. 25, 1999), and GAO, Social
Security: Evaluating Reform Proposals, GAO/AIMD/HEHS-00-29 (Washington,
D.C.: Nov. 4, 1999).

2. the balance struck between the goals of individual equity10 (rates of
return on individual contributions) and income adequacy11 (level and
certainty of monthly benefits); and

3. how readily such changes could be implemented, administered, and
explained to the public.

Moreover, reform proposals should be evaluated as packages that strike a
balance among the individual elements of the proposal and the interactions
among these elements. The overall evaluation of any particular reform
proposal depends on the weight individual policy makers place on each of
the above criteria.

Changing the indexing used by the OASDI program could be used to increase
income or reduce costs. Indexing provides a form of regular adjustment of
revenues or benefits that is pegged to a particular economic, demographic,
or actuarial variable. An advantage of such indexing approaches is that
they take some of the "politics" out of the system, allowing the system to
move toward some agreed-upon objective; they may also be administratively
simple. However, this "automatic pilot" aspect of indexing poses a
challenge, as it may make policy makers hesitant to enact changes, even
when problems arise.

          Social Security Currently Indexes Both Benefits and Revenues

While Social Security did not use automatic indexing initially, it is now
a key feature of the program's design, as well as a central element of
many reform proposals. Under the current program, benefits for new
beneficiaries are computed using wage indexing, benefits for existing
beneficiaries are adjusted using price indexing, and on the revenue side,
the cap on the amount of earnings subject to the payroll tax is also
adjusted using wage indexing. Reform proposals have included provisions
for modifying each of these indexing features.

10 For a discussion of individual equity issues, see GAO, Social Security:
Issues in Comparing Rates of Return with Market Investments,
GAO/HEHS-99-110 (Washington, D.C.: Aug. 5, 1999).

11 GAO-02-62 .

Program Did Not Use Indexing until 1970s

Before the 1970s, the Social Security program did not use indexing to
adjust benefits or taxes automatically. For both new and existing
beneficiaries, benefit rates increased only when Congress voted to raise
them. Benefit levels, when adjusted for inflation, fell and then jumped up
with ad hoc increases, and these fluctuations were dramatic at times.
Similarly, Congress made only ad hoc changes to the tax rate and the cap
on the amount of workers' earnings that were subject to the payroll tax,
which is also known as the maximum taxable earnings level. Adjusted for
inflation, the maximum taxable earnings level also fluctuated
dramatically, and as a result, the proportion of all wages subject to the
payroll tax also fluctuated. (See app. II for more detail.)

For the first time, the 1972 amendments provided for automatic indexing.
They provided for automatically increasing the maximum taxable earnings
level based on increases in average earnings, and this approach is still
in use today. However, the 1972 amendments provided an indexing approach
for benefits that became widely viewed as flawed. In particular, the
indexing approach in the 1972 amendments resulted in (1) a
"double-indexing" of benefits to inflation for new beneficiaries though
not for existing ones12; (2) a form of "bracket creep" based on the
structure of the benefit formula that slowed benefit growth as earnings
increased over time, which offset the double indexing to some degree; and
(3) instability of program costs that was driven by the interaction of
price and wage growth in benefit calculations. (See app. II for more
detail.) Within a few years, problems with the 1972 amendments became
apparent. Benefits were growing far faster than anticipated, especially
since wage and price growth varied dramatically from previous historical
experience. Addressing the instability of this indexing approach became a
focus of policy makers' efforts to come up with a new approach. As a 1977
paper on the problem noted, "Clearly, it is a system that needs to be
brought under greater control, so that the behavior of retirement benefits
over time will stop reflecting the chance interaction of certain economic
variables."13

12 One type of indexing took the form of automatic inflation adjustments
to the earnings replacement factors in the benefit formula. At the same
time, the earnings used in the formula were higher on average for each new
group of beneficiaries, partially because of inflation. See appendix II.

13 Lawrence H. Thompson. "Toward the Rational Adjustment of Social
Security Benefit Levels," Policy Analysis, Vol. 3, No. 4, Fall 1977.

1977 Amendments Created Indexing Approach of Current Social Security System

The 1977 amendments instituted a new approach to indexing benefits that
remains in use today. The experience with the 1972 amendments and double
indexing made clear the need to index benefits differently for new and
existing beneficiaries, which was referred to as "decoupling" benefits.
Indexing now applies to several distinct steps of the benefit computation
process, including (1) indexing lifetime earnings for each worker to wage
growth, (2) indexing the benefit formula for new beneficiaries to wage
growth, and (3) indexing benefits for existing beneficiaries to price
inflation.14 Under this approach, benefit calculations for new
beneficiaries are indexed differently than for existing beneficiaries, and
earnings replacement rates have been fairly stable. The cap on taxable
earnings is still indexed to wage growth as specified by the 1972
amendments.

  Indexing Lifetime Earnings to Wages

Social Security benefits are designed to partially replace earnings that
workers lose when they retire, become disabled, or die. As a result, the
first step of the benefit formula calculates a worker's average indexed
monthly earnings (AIME), which is based on the worker's lifetime history
of earnings covered by Social Security taxes. The formula adjusts these
lifetime earnings by indexing them to changes in average wages.15 Indexing
the earnings to changes in wage levels ensures that the same relative
value is accorded to each year's earnings, no matter when they were
earned.

For example, consider a worker who earned $5,000 in 1965 and $40,000 in
2000. The worker's earnings increased by eight times, but much of that
increase reflected changes in the average wage level in the economy, which
increased by about seven times (690 percent) over the same period. The
growth in average wages in turn partially reflects price inflation;
however, wages may grow faster or slower than prices in any given year.
Indexed to reflect wage growth, the $5,000 would become roughly $35,000,
giving it greater weight in computing average earnings over time and
making it more comparable to 2000 wage levels.

14 Wage indexing also applies to other provisions of the program that are
not part of the primary benefit computations. Such provisions include
earnings test thresholds, maximum family benefits, coverage thresholds,
and thresholds relating to disability insurance.

15A worker's earnings for a given year are indexed by multiplying them by
the ratio of the national average wage for the indexing year to the
national average wage in the year the income was earned. The indexing year
is the second calendar year before the year in which the worker is first
eligible-the year the worker reaches age 62, becomes disabled, or dies.
Earnings after the indexing year are counted at their actual value.

  Indexing Initial Benefit Formula to Wages

Once the AIME is determined, it is applied to the formula used to
calculate the worker's primary insurance amount (PIA). This formula
applies different earnings replacement factors to different portions of
the worker's average earnings. The different replacement factors make the
formula progressive, meaning that the formula replaces a larger portion of
earnings for lower earners than for higher earners. For workers who become
eligible for benefits in 2006, the PIA equals

           o  90 percent of the first $656 dollars of AIME plus
           o  32 percent of the next $3,299 dollars of AIME plus
           o  15 percent of AIME above $3,955.

For workers who do not collect benefits until after the year they first
become eligible, the PIA is adjusted to reflect any COLAs since they
became eligible. The PIA is used in turn to determine benefits for new
beneficiaries and all types of benefits payable on the basis of an
individual's earnings record. To determine the actual monthly benefit,
adjustments are made reflecting various other provisions, such as those
relating to early or delayed retirement, type of beneficiary, and maximum
family benefit amounts. Figure 1 illustrates how the PIA formula works.

Figure 1: Social Security Benefit Formula Replaces Earnings at Different
Rates

The dollar values in the formula that indicate where the different
replacement factors apply are called bendpoints. These bendpoints ($656
and $3,955) are indexed to the change in average wages, while the
replacement factors of 90, 32, and 15 percent are held constant. In
contrast, under the 1972 amendments, the bendpoints were held constant and
the replacement factors were indexed. (See app. II.) Indexing the
bendpoints and holding replacement factors constant prevents bracket creep
and keeps the resulting earnings replacement rates relatively level across
birth years. Indexing the benefit formula in this way helps benefits for
new retirees keep pace with wage growth, which reflects increases in the
standard of living.

Figure 2, which shows earnings replacement rates for successive groups of
illustrative workers, illustrates the program's history with indexing
initial benefits.16 Replacement rates declined before the first benefit
increases were enacted in 1950 and then rose sharply as a result of those
increases. From 1950 until the early 1970s, replacement rates fluctuated
noticeably more from year to year than over other periods; this pattern
reflects the ad hoc nature of benefit increases over that period. Between
1974 and 1979, replacement rates grew rapidly for new beneficiaries,
reflecting the double indexing of the 1972 amendments. The 1977 amendments
corrected for the unintended growth in benefits from double indexing, and
replacement rates declined rapidly as a result. This pattern of increasing
and then declining benefit levels is known as the notch.17 Finally,
replacement rates have been considerably more stable since the 1977
amendments took effect, a fact that has helped to stabilize program costs.
(See app. II.)

16In this figure, replacement rates are the annual retired worker benefits
at age 65 divided by career-average earnings. Illustrative workers have
career-average earnings equal to about 45, 100, and 160 percent of Social
Security's Average Wage Index, respectively, for low, medium, and high
earners. These three cases have earnings patterns that reflect differences
by age in the probability of work and in average earnings levels. Taxable
maximum earners have earnings equal to the maximum earnings taxable under
OASDI in each year. Using illustrative workers holds other factors equal
that might also affect replacement rates. For example, using illustrative
workers filters out the effects of changes in the covered population or
changes in work and retirement patterns.

17See GAO, Social Security: GAO's Analysis of the Notch Issue,
GAO/T-HEHS-94-236 (Washington, D.C.: Sept. 16, 1994).

Figure 2: Social Security's Earnings Replacement Rates for Illustrative
Workers

Note: Replacement rates are the annual retired worker benefits at age 65
divided by career-average earnings. Illustrative workers have
career-average earnings equal to about 45, 100, and 160 percent of Social
Security's Average Wage Index, respectively, for low, medium, and high
earners. These three cases have earnings patterns that reflect differences
by age in the probability of work and in average earnings levels. Taxable
maximum earners have earnings equal to the maximum earnings taxable under
OASDI in each year. Variations in these illustrative replacement rates
result not only from program changes but also from short-term fluctuations
in the growth rate of wages, which helps determine the earnings histories
of the illustrative earners.

  Indexing Benefits to Prices for Existing Beneficiaries

After initial benefits have been set for the first year of entitlement,
benefits in subsequent years increase with a COLA designed to keep pace
with inflation and thereby help to maintain the purchasing power of those
benefits. The COLA is based on the consumer price index (CPI), in contrast
to the indexing of lifetime earnings and initial benefits, which are based
on the national average wage index. 18

18 Specifically, Social Security's COLAs are based on the consumer price
index for urban wage earners and clerical workers (CPI-W), as opposed to
the CPI series for all urban consumers (CPI-U).

  Indexing Maximum Taxable Earnings to Wages

The cap on taxable earnings increases each year to keep pace with changes
in average wages. As a result, in combination with a constant tax rate,
total program revenues tend to keep pace with wage growth and therefore
also with benefits to some degree. In 2006, the cap is set at $94,200. As
the distribution of earnings in the economy changes, the percentage of
total earnings that fall below the cap can also change. (See app. II.)

Table 1 summarizes the various indexing and automatic adjustment
approaches that affect most workers and beneficiaries under the current
program.

Table 1: Key Indexing Approaches under Current U.S. Social Security System

Approach              How it works            Comments                     
Benefit provisions    
Wage-indexing initial Before averaging        Maintains relative standards 
benefit calculation   workers' earnings over  of living across age groups  
                         their careers, AIME     (that is, replacement        
                         adjusts actual earnings rates), at time of           
                         using average wage      retirement.                  
                         index.                                               
                                                 Actuarial balance of the     
                         Bendpoints of PIA       program is relatively        
                         formula rise over time  insensitive to economic      
                         according to wage       fluctuations because benefit 
                         growth.                 levels and tax revenues are  
                                                 both linked to wages.        
                         Earnings replacement                                 
                         factors in PIA formula  Initial benefits keep pace   
                         remain constant.        with standard of living, as  
                                                 reflected by wage levels.    
Price-indexing        Benefits rise yearly    Purchasing power of benefits 
post-entitlement      according to rise in    remains constant over time,  
benefits              the CPI.                once benefits start.         
Tax provisions        
Wage-indexing maximum Earnings are only taxed Share of earnings not taxed  
taxable earnings      on the first $94,200    can change as income         
                         per year in 2006. Limit distribution changes.        
                         rises every year        
                         according to average    
                         wage growth.            
Constant tax rate     Earnings are taxed      Program revenue rises        
                         yearly at 12.4 percent  annually with the rise in    
                         (6.2 percent from       wages.                       
                         workers and 6.2 percent                              
                         from employers).        Constant tax rate maintains  
                                                 the same proportion of taxes 
                                                 for all workers earning less 
                                                 than maximum taxable         
                                                 earnings.                    

Source: GAO.

Various Reform Proposals Include Indexing Provisions

Various reform proposals have suggested changes to most of the indexing
features of the current Social Security system. Some proposals would use
alternative indexes for initial benefits in order to slow their growth.
Other proposals would take the same approach but would limit benefit
reductions on workers with lower earnings. Some propose modifying the COLA
in the belief that the CPI overstates the rate of inflation. Still others
propose indexing revenue provisions in new ways.

Changes to the indexing of Social Security's initial benefits could be
implemented by changing the indexing of lifetime earnings or the PIA
formula's bendpoints.19 However, they could also be implemented by
adjusting the PIA formula's replacement factors, even though these factors
are not now indexed. Under this approach, which is used in this report,
the replacement factors are typically multiplied by a number that reflects
the index being used. The replacement factors would be adjusted for each
year in which benefits start, beginning with some future year. So such
changes would not affect current beneficiaries. Indexing the replacement
factors would reduce benefits at the same proportional rate across income
levels, while changing the indexing of lifetime earnings or the bendpoints
could alter the distribution of benefits across income levels. Recent
reform proposals, as described by the Social Security Administration's
(SSA) Office of the Chief Actuary in its evaluations, generally implement
indexing changes as adjustments to the PIA formula's replacement factors.

Two indexing approaches-to reflect changes in the CPI or increasing
longevity-have been proposed as alternatives to the average wage index for
calculating initial benefits. Proponents of using CPI indexing for initial
benefit calculations generally offer the rationale that wage indexing has
never been fiscally sustainable and CPI indexing would slow the growth of
benefits to an affordable level while maintaining the purchasing power of
benefits. They say that maintaining the purchasing power of benefits
should be the program's goal, as opposed to maintaining relative standards
of living across age groups (that is, earnings replacement rates), which
the current benefit formula accomplishes. Proponents of longevity indexing
offer the rationale that increasing longevity is a key reason for the
system's long-term insolvency. Since people are living longer on average,
and are expected to continue to do so in the future, they will therefore
collect benefits for more years on average. Using an index that reflects
changes in life expectancy would maintain relatively comparable levels of
lifetime benefits across birth years and thereby promote intergenerational
equity. Also, longevity indexing could encourage people to work longer.

19Andrew G. Biggs, Jeffrey R. Brown, Glenn Springstead, "Alternative
Methods of Price Indexing Social Security: Implications for Benefits and
System Financing," National Bureau of Economic Research, Working Paper
11406 (2005).

Some indexing proposals accept the need to slow the growth of initial
benefits in general but seek to protect benefit levels for the lowest
earnings levels, consistent with the program's goal of helping ensure
income adequacy. Such proposals would modify how a new index would be
applied to the formula for initial benefits so that the formula is still
wage-indexed below a certain earnings level. As a result, they would
maintain benefits promised under the current program for those with
earnings below that level such as, for example, those in the bottom 30
percent of the earnings distribution. Such an approach has been called
progressive price indexing.

A few proposals would alter the COLA used to adjust benefits for current
retirees. Some proposals respond to methodological concerns that have been
raised about how the CPI is calculated and would adjust the COLA in the
interest of accuracy.20 In general, such changes would slightly slow the
growth of the program's benefit costs. However, other proposals call for
creating a new CPI for older Americans (CPI-E) specifically tailored to
reflect how inflation affects the elderly population and using the CPI-E
for computing Social Security's COLA.21 Depending on its construction,
such a change could increase the program's benefit costs.

Some proposals would index revenues in new ways. Some would apply a
longevity index to payroll tax rates, again focused on the fact that
increasing life expectancy is a primary source of the program's
insolvency. Proponents of indexing tax rates feel that benefits are
already fairly modest, so the adjustment for longevity should not come
entirely from benefit reductions. Other proposals would institute other
types of automatic revenue adjustments. Some would raise the maximum
taxable earnings level gradually until some percentage of total earnings
are covered and then maintain that percentage into the future. Implicitly,
such proposals reflect a desire to hold constant the percentage of
earnings subject to the payroll tax. Still another proposal would provide
for automatically increasing the tax rate when the ratio of trust fund
assets to annual program costs is projected to fall.

20For more information on the CPI and how it overstates the true rate of
inflation, see Advisory Commission to Study the Consumer Price Index,
"Toward a More Accurate Measure of the Cost of Living," Final Report to
the Senate Committee on Finance, Dec. 1996, which is known as the Boskin
Commission report. A variety of changes have been made to the CPI since
that report, including changes that in turn affect Social Security's COLA.
In addition, a new "chained" CPI reflects how consumers substitute one
product for another when their relative prices change. This new CPI is not
yet used by government agencies, but some reform proposals call for using
a variation of it in computing COLAs.

21For example, the elderly allot a larger proportion of their expenses to
medical care than the general population, which partially depends on
Medicare's coverage and premiums.

Table 2 summarizes the various indexing and automatic adjustment
approaches that reform proposals have contained.

Table 2: Summary of Indexes and Automatic Adjustments Proposed in United
States

                                                  Rationales offered by       
Provision             How it works             proponents                  
Provisions affecting initial benefit calculations for future beneficiaries
Longevity indexing    Proportionally reduces   Increasing longevity is a   
                         replacement factors in   key reason for the system's 
                         PIA formula to reflect   long-term solvency problem. 
                         adjustments for                                      
                         increasing life          People are living longer on 
                         expectancy.a             average and therefore       
                                                  collecting benefits for     
                                                  more years on average.      
                                                                              
                                                  Would maintain comparable   
                                                  levels of lifetime benefits 
                                                  across birth years and      
                                                  thereby promote             
                                                  intergenerational equity.   
Price indexingb       Proportionally reduces   "Wage-indexing ... has      
                         replacement factors in   never been fiscally         
                         PIA formula to reflect   sustainable."c              
                         changes in index.a                                   
                                                  Would slow the growth of    
                                                  benefits to an affordable   
                                                  level.                      
                                                                              
                                                  Would still maintain the    
                                                  purchasing power of         
                                                  benefits.                   
Progressive price     Proportionally reduces   Protects benefits for lower 
indexing              replacement factors in   earnings to help ensure     
                         PIA formula to reflect   income adequacy.            
                         changes in index.a       
                                                  
                         But no change to factors 
                         for earnings below a     
                         certain level.           
                                                  
                         Effectively adds new     
                         bendpoint between two    
                         current ones.            
Provisions affecting benefit COLAs for current and future beneficiaries
Revise COLA to        Use more accurate CPI in Greater accuracy.           
reflect more accurate determining COLA.        
calculation of CPI                             
Provisions affecting taxes for current and future workers
Longevity indexing of Proportionally increase  Increasing longevity is a   
payroll tax rates     payroll tax rate to      key reason for the system's 
                         reflect changes in       long-term solvency problem. 
                         index.                   
Increase payroll tax  Use trustees'            Ensure ongoing solvency.    
rates to ensure       intermediate projections 
maintaining ratio of  of trust fund ratios.    
trust fund assets to                           
program costs                                  
Increase maximum      Use recent data on       Promote intergenerational   
taxable earnings to   earnings distribution    equity by ensuring          
ensure a constant     with trustees'           consistent application of   
percentage of         intermediate projections payroll tax.                
aggregate earnings    of wage growth.          
are taxed                                      

Source: GAO.

aAverage indexed monthly earnings would be computed as under present law.

bAn implication of price indexing is that it would slow benefit growth to
a greater degree if wages grow faster than projected, even as Social
Security's financial situation would be improving.

cCommission to Strengthen Social Security. "Strengthening Social Security
and Creating Personal Wealth for All Americans: Report of the President's
Commission" p. 120, Washington, D.C.: Dec. 21, 2001.

    A Variety of Indexing Approaches Highlight International Reform Efforts

Faced with adverse demographic trends, many countries have enacted reforms
in recent years to improve the long-term fiscal sustainability of their
national pension systems. New indexing methods now appear in a variety of
forms around the world in earnings-related national pension systems.22 In
general, they seek to contain pension costs associated with population
aging. Some indexing methods affect both current and future retirees.

Retirement Indexing Approaches in Other Countries Generally Focus on Benefit
Reductions instead of Increased Contributions

A number of reforms have focused on methods that primarily adjust benefits
rather than taxes to address the fiscal solvency of national pension
systems. There are two main reasons for this. First, contribution rates
abroad are generally high already, making it politically difficult to
raise them much further. For example, while in the United States total
employer-employee Social Security contribution rates are 12.4 percent of
taxable earnings, they are above 16 percent in Belgium and France, more
than 18 percent in Sweden and Germany, above 25 percent in the Netherlands
and the Czech Republic, and over 30 percent in Italy.23 In fact, some
countries have stipulated a ceiling on employee contribution rates in
order to reassure the young-or current contributors-that the burden would
be shared among generations. For example, Japan settled, with the 2004
Reform Law, its pension premium rates for the next 100 years with an
increase of 0.35 percent per year until 2017, at which time premium levels
are to be fixed at 18.3 percent of covered wages. Similarly, Canada chose
to raise its combined employer-employee contribution rate more quickly
than previously scheduled, from 5.6 percent to 9.9 percent between 1997
and 2003, and maintain it there until the end of the 75-year projection
period.24 This increase is meant to help Canada's pension system build a
large reserve fund and spread the costs of financial sustainability across
generations.25 Germany's recent reforms set the workers' contribution rate
at 20 percent until 2020 and at 22 percent from 2020 to 2030. Second,
increasing employee contribution rates without significantly reducing
benefit levels will tend to make continued employment less attractive
compared to retirement. In the context of population aging and fiscally
stressed national pension systems facing many countries, reform measures
seek to do the opposite: encourage people to remain in the labor force
longer to enhance the fiscal solvency of pension programs. Contribution
rates that become too high are not likely to provide sufficient incentives
to continue work.

22In earnings-related public pension systems reviewed here, indexation
appears in different forms but in all cases affects the way in which
pension rights are accrued. For example, in notional defined contribution
systems such as in Sweden and Italy, workers earn a notional rate of
return on their contributions (based on their earnings), and indexation is
implicit in that notional return. In point systems such as in Germany,
workers earn pension points (also based on their earnings) that are
multiplied by a pension-point value at the time of retirement. There,
indexation is implicit in the value of the pension point.

23 It is important to note that the structure of public pension programs
differ across countries, and hence are not strictly comparable. For
example, contributions in some cases help finance maternity/paternity and
unemployment benefits in addition to old age benefits.

24 The total employer and employee contributions of 9.9 percent may appear
low, but the retirement pension benefits these generate are relatively
modest, replacing only 25 percent of average pensionable earnings.

Indexing Approaches Aim at Containing Costs

One commonly used means of reducing, or containing the growth of, promised
benefits involves changing the method used to compute initial benefits.
For example, France, Belgium, and South Korea now adjust past earnings in
line with price growth rather than wage growth to determine the initial
pension benefits of new retirees. In general, this shift to price
indexation tends to significantly lower benefits relative to earnings, as
over long periods prices tend to grow more slowly than wages.26 Because of
compounding, the effect of such a change is larger when benefits are based
on earnings over a long period than when they reflect only the last few
years of work, as in pension plans with benefits based on final salaries.
In fact, the OECD estimates that, in the case of a full-career worker with
45 years of earnings, price indexation can lead to benefits 40 percent
lower than with wage indexation.27 In contrast to full price indexing,
some nations use an index that is a mix of price growth and wage growth,
which tends to produce higher benefits than those calculated using price
indexation only, then adjust the relative weights of the two to cover
program costs. Finland, for example, changed its indexation of initial
benefits from 50 percent prices and 50 percent wages to 80 percent and 20
percent, respectively. Similarly, Portugal's index combines 75 percent
price growth and 25 percent wage growth.28

25 Canada's reserve fund is managed by an Investment Board that operates
independently from the government since the late 1990s and invests in both
foreign and domestic assets subject to some restrictions.

26 As in the United States in the 1970s and 1980s, prices at times grow
faster than wages; nonetheless, these periods remain exceptional.

27 A full-career worker is defined as one having earnings between the ages
of 20 and 65. The computation reflects the average effect, in OECD
countries, for a manufacturing worker with average earnings.

A few countries have moved away from wage indexing but without necessarily
adopting price indexation. Sweden, for instance, uses an index that
reflects per capita wage growth to compute initial benefits, provided the
system is in fiscal balance. However, when the system's obligations exceed
its assets, a "brake" is applied automatically that allows the indexation
to be temporarily abandoned.29 This automatic balancing mechanism (ABM)
ensures that the pension system remains financially stable.30 In Germany
and Japan, recent reforms changed benefit indexation from a gross-wage
base to a net-wage base-i.e., gross wages minus contributions. In Italy,
workers' benefit accounts rise in line with gross domestic product (GDP)
growth so both the changes in the size of the labor force and in
productivity dictate benefit levels.

Another approach countries have used is adding a longevity index to the
formula determining pension payments. In Sweden, Poland, and Italy, for
example, remaining life expectancy at the time of retirement inversely
affects benefit levels. Thus, as life spans gradually increase, successive
cohorts of retirees get smaller benefit payments unless they choose to
begin receiving them later in life than those who retired before them.
Also, people who retire earlier than their peers in a given cohort get
significantly lower benefits throughout their remaining life than those
who retire later. Longevity indexing helps ensure that improvements in
life expectancy do not strain the system financially.31

28 In most OECD countries, the formula used varies by either individual
earnings, age or length of service.

29 Using per capita wage growth, i.e., wage growth divided by the labor
force, as an index implies that when the labor force shrinks, per capita
wage growth goes up. As a result, benefits increase right when the number
of contributors gets smaller, creating an imbalance.

30 More precisely, the average-wage-growth indexation is reduced whenever
the Balance Ratio is less than 1, where Balance Ratio = (Contribution
Asset + Buffer Funds)/Pension Liability. The index then automatically
becomes average wage growth multiplied by the Balance Ratio, and remains
so as long as the Balance Ratio is less than 1. The Buffer Funds are a
collection of reserve funds to which part of pension contributions are
transferred. These are then invested in domestic and foreign assets with
the objective of achieving the highest possible returns. The Buffer Funds
play an important role in ensuring the financial stability of the pension
program insofar as high rates of return on these funds may partially or
fully compensate for any adverse demographic or economic developments.

Germany, on the other hand, now uses a sustainability factor that links
initial benefits to the system's dependency ratio-i.e., the number of
people drawing benefits relative to the number paying into the system.
This dependency ratio captures variations in fertility, longevity, and
immigration, and consequently makes the pension system self-stabilizing.
For example, higher fertility and immigration, which raise labor force
growth, will, other things equal, improve the dependency ratio, leading to
higher pension benefits, while higher longevity or life expectancy will
increase the dependency ratio, and hence cause benefits to decline.32

Indexing Approaches Affect Both Current and Future Beneficiaries

In some of the countries we studied, changes in indexing methods affect
both current and future retirees. In Japan, for example, post-retirement
benefits were indexed to wages net of taxes before 2000. However, reforms
enacted that year altered the formula by linking post-retirement benefits
to prices. As a result, retirees saw their subsequent benefits rise at a
much slower pace. The 2004 reforms reduced retirees' purchasing power
further by introducing a negative "automatic adjustment indexation" to the
formula. With this provision, post-retirement benefits increase in line
with prices minus the adjustment rate, currently fixed at 0.9 percent
until about 2023. This rate is the sum of two demographic factors: the
decline in the number of people contributing to the pension program
(projected at 0.6 percent) plus the increase in the number of years people
collect pensions (projected at 0.3 percent). This negative adjustment also
enters the formula determining the benefit of new recipients as past
earnings are indexed to net wages minus the same 0.9 percent adjustment
rate.

Sweden's ABM modifies both the retirement accounts of workers-or future
retirees-and the benefits paid to current pensioners. As explained
earlier, this mechanism is triggered whenever system assets fall short of
system liabilities. Moreover, post-retirement benefits in Sweden are
indexed each year to an economic factor equal to prices plus the average
rate of real wage increase minus 1.6 percent, which is the projected real
long-term growth in wages. As a result, if average real wages grow
annually at 1.6 percent, post-retirement benefits are adjusted for price
increases. On the other hand, if real wage growth falls below 1.6 percent,
benefits do not keep up with prices, leading to a decline in retiree
purchasing power.33

31 The longevity factor enters the formula determining initial benefits
for a given cohort and does not change for that cohort after the normal
retirement age. It ensures that the present value of benefits does not
increase with life expectancy across cohorts.

32 Changes in fertility or longevity are likely to affect the dependency
ratio in the long run, but little in the short run.

Germany's sustainability factor affects those already retired, as it is
included in the formula that adjusts their benefits each year. If, as
projected, the number of contributors falls relative to that of
pensioners, increasing the dependency ratio, all benefits are adjusted
downward, so all cohorts share the burden of adverse demographic trends.
This intergenerational burden sharing is also apparent in the indexation
of all benefits to net wages-wages minus contributions, which affect
workers and pensioners alike. Thus an increase in contributions,
everything else equal, lowers both initial benefits and benefits already
being paid.

Table 3 summarizes relevant characteristics of earnings-related public
pension programs in selected countries.

33 Benefits at the time of retirement are determined by remaining life
expectancy and a growth "norm" of 1.6 percent. Benefits are then adjusted
each year for inflation plus or minus deviations from this norm.

Table 3: Characteristics of Earnings-Related Public Pension
Programs-Selected Countries

                                            Years of                               
                                            individual                             
         Remaining  Normal     Average      earnings    Indexing of                
         life       retirement contribution considered  earnings for 
         expectancy age (early rate         in initial  calculating  Indexing of
         at age 65a retirement (percentage  benefit     initial      benefits in
         men-women  age)b      of earnings) calculation benefits     retirement
Belgium  15.8       65 (60)    16.36        Lifetime    Prices       100% prices   
                                            average                  
         19.7                                                        
                                                                     
         (2002)                                                      
Canada   17.4       65 (60)    9.9          Lifetime    Average      100% prices   
                                            average     earnings     
         20.8                               excluding                
                                            worst 15%                
                                            of years                 
Czech    13.9       Men: 63    28           Since 1985  Average      67% prices    
Republic                                    moving to   earnings     33% real wage 
         17.3       Women:                  30                       growth        
                    59-63                                            
                                                                     
                    (men: 60,                                        
                    women:                                           
                    56-60)c                                          
France   17.1       60         16.45        Lifetime    Prices       100% prices   
                                            average                  
         21.4                                                        
                                            (public                  
         (2002)                             employees:               
                                            best 20                  
                                            moving to                
                                            25)                      
Germany  16.1       65 (63)    19.5         Lifetime    Average net  100% wages    
                                            average     earnings     net of        
         19.6                                           (subject to  contributions 
                                                        demographic  (subject to   
                                                        adjustment)  demographic   
                                                                     adjustment)   
Italy    16.7       65 (60)    32.7         Lifetime    5-year       Between       
                                            average     moving       75-100%       
         20.7                                           average of   prices        
                                                        GDP growth   depending on  
         (2001)                                         (subject to  benefit level 
                                                        demographic  
                                                        adjustment)  
Japan    18.2       65 (60)    13.58        Lifetime    Average      100% prices   
                                            average     earnings     (subject to   
         23.3                                           (subject to  demographic   
                                                        demographic  adjustment)   
         (2004)                                         adjustment)  
Sweden   17.4       65 (61)    18.5         Lifetime    Average      100% prices   
                                            average     earnings     plus real     
         20.6                                           (subject to  wages less    
                                                        demographic  1.6% (subject 
         (2004)                                         and fiscal   to            
                                                        adjustments) demographic   
                                                                     and fiscal    
                                                                     adjustments)  
United   16.8       67 (62)    12.4         Best 35     Average      100% prices   
States                                                  earnings up  
         19.8                                           to age 60    

Source: GAO

aIn 2003 unless otherwise indicated.

b2002 data, including legislated changes.

cWomen's retirement ages (full and early) depend on the number of
children.

         Indexing Can Be Used to Achieve Desired Distributional Effect

In the U.S. Social Security program, indexing can have different effects
on the distribution of benefits and on the relationship between
contributions and benefits, depending on how it is applied to benefits or
taxes. There are a variety of proposals that would change the current
indexing of initial benefits, including a move to the CPI, to longevity or
mortality measures, or to the dependency ratio.34 When the index is
implemented through the benefit formula, each will have a proportional
effect, with constant percentage changes at all earnings levels, on the
distribution of benefits (i.e., the progressivity of the current system is
unchanged). However, indexing provisions can be modified to achieve other
distributional effects. For example, so-called progressive indexing
applies different indexes at different earnings levels in a manner that
seeks to protect the benefits of low-income workers. Indexing payroll tax
rates would also have distributional effects. Such changes maintain
existing benefit levels but affect equity measures like the ratio of
benefits to contributions across age cohorts, with younger cohorts having
lower ratios because they receive lower benefits relative to their
contributions. Finally, proposals that modify the indexing of COLAs for
existing beneficiaries have important and adverse distributional effects
for groups that have longer life expectancies, such as women and highly
educated workers, because such proposals would typically reduce future
benefits, and this effect compounds over time. In addition, disabled
worker beneficiaries, especially those who receive benefits for many
years, would also experience lower benefits.

Proposals to Index Initial Benefits Have a Proportional Effect on the
Distribution of Benefits

There are a variety of proposals that would change the current indexing of
initial benefits from the growth in average wages. These include a move to
a measure of the change in prices like the CPI, to longevity measures that
seek to capture the growth in population life expectancies, or to the
dependency ratio that measures changes in the number of retirees compared
to the workforce. We analyzed three indexing scenarios; the dependency
ratio index, which links the growth of initial benefits to changes in the
dependency ratio, the ratio of the number of retirees to workers; the CPI
index, which links the growth of initial benefits to changes in the CPI;
and the mortality index, which links the growth of initial benefits to
changes in life expectancy to maintain a constant life expectancy at the
normal retirement age.35 Figure 3 illustrates the projected distribution
of benefits for workers born in 1985 under three different indexing
scenarios36 (on the left side of the figure) and under a so-called benefit
reduction benchmark that reduces benefits just enough to achieve program
solvency over a 75-year projection period (on the far right).37 Median
benefits under the dependency ratio index and the CPI index are lower than
the median benefit for the benchmark; they reduce benefits more than is
needed to achieve 75-year solvency.38 In contrast, the mortality index has
a higher median benefit level than the benchmark, so without further
modifications, it would not achieve 75-year solvency.

34 Longevity and mortality are differing measures of life expectancy.

35 See appendix I for more information on these indexes.

36 We focused on workers born in 1985 because all prospective program
changes under all alternative policy scenarios would be almost fully
phased in for such workers.

37 The benefit reduction benchmark is a hypothetical benchmark policy
scenario that would achieve 75-year solvency by only reducing benefits.
For ease of modeling, the benefit reduction benchmark takes the form of
reductions in the benefit formula factors. Each formula factor is reduced
annually by subtracting a constant proportion of the factor's value under
current law, resulting in a constant percentage reduction of currently
promised benefits for everyone. See appendix I for more information about
the benefit reduction benchmark. Consistent with the Social Security
trustees' report, we use a 75-year projection period in assessing the
solvency of different indexing scenarios and our benchmarks. The 75-year
projection period has been standard practice for many years, although it
does not capture sustainability over longer time horizons. We believe it
is important to consider sustainability, and there are different ways to
do so, but this issue is outside the scope of this report.

38 While the level of solvency differs among these scenarios, the level of
benefits under each scenario is lower than promised benefits, and
replacement rates have declined in each scenario.

Figure 3: Indexing Changes with a Larger Proportional Reduction Have a
Greater Impact on the Distribution of Benefits, but Scaling to Achieve
75-Year Solvency Illustrates That the Proportional Effects Have Similar
Results

Note: Benefits are for all individuals in the GEMINI 1985 cohort sample in
2052 (the year the cohort reaches age 67). Scenarios are modeled using the
intermediate assumptions of the 2005 trustees' report. The dependency
ratio index links the growth of initial benefits to changes in the
dependency ratio, the ratio of the number of retirees to workers. The
dependency ratio index has a 197 percent improvement in 75-year solvency,
generating far more programmatic savings than is needed to achieve
solvency. The CPI index links the growth of initial benefits to changes in
the CPI. The CPI index has a 127 percent improvement in 75-year solvency,
generating more programmatic savings than is needed to achieve solvency.
The mortality index links the growth of initial benefits to changes in
life expectancy to maintain a constant life expectancy at the normal
retirement age. The mortality index has a 72 percent improvement in
75-year solvency, which does not generate enough programmatic savings to
be solvent. The benefit reduction benchmark is a hypothetical benchmark
policy scenario that would achieve 75-year solvency by only reducing
benefits. Thus, the benchmark has a 100 percent improvement in 75-year
solvency, being exactly solvent at the end of the 75-year period. The
scaled scenarios are adjusted to achieve a 75-year actuarial balance of
zero. While scaling allows comparisons across distributions over 75 years,
the different indexing scenarios are not identical in terms of
sustainability. For a more complete description of the indexing scenarios,
the benchmark, or the scaling, see appendix I.

Regardless of the index used to modify initial benefits, most proposals
apply the new index in a way that has proportional effects on the
distribution of benefits.39 Thus, benefits at all levels will be affected
by the same percentage reduction, for example, 5 percent, regardless of
earnings. The left half of figure 3 illustrates this proportionality in
terms of monthly benefits. While the level of benefits differs, the
distribution of benefits for each scenario has a similar structure.
However, the range of each distribution varies by the difference in the
size of the proportional reduction. A larger proportional reduction-the
dependency ratio index-will result in a distribution with a similar
structure, compared to promised benefits. However, each individual's
benefits are reduced by a constant percentage; therefore, the range of the
distribution, the difference between benefits in the 25th and 75th
percentile, would be smaller, compared to promised benefits. This
proportional reduction in benefits is also illustrated in figure 4, which
compares the currently scheduled or promised benefit formula with our
three alternative indexing scenarios. Under each scenario, the line
depicting scheduled benefits is lowered, by equal percentages at each AIME
amount, by the difference between the growth in covered wages and the new
index. Each indexing scenario maintains the shape of the current benefit
formula; thus the progressivity of the system is maintained, but the line
for each scenario is lower than scheduled benefits, which would affect the
adequacy of benefits.

39 The general application of these indexes is to multiply the PIA
formula's replacement factors by a factor that reflects the new index.
This is the approach taken by the Social Security actuaries and most
proposals. See appendix I.

Figure 4: Proportional Indexing Changes Would Maintain the Progressivity
of the Current Benefit Formula, but Could Reduce Adequacy (Initial
Benefits in 2050)

Note: The illustrated PIAs are for individuals who become eligible in
2050. The dependency ratio index links the growth of initial benefits to
changes in the dependency ratio, the ratio of the number of retirees to
workers. The CPI index links the growth of initial benefits to changes in
the CPI. The mortality index links the growth of initial benefits to
changes in life expectancy to maintain a constant life expectancy at the
normal retirement age. For more information on each index see appendix I.

The proportional effects of indexing are best illustrated by adjusting, or
scaling, each index to achieve comparable levels of solvency over 75
years.40 Thus, for those indexes that do not by themselves achieve
solvency, the benefit reductions are increased until solvency is achieved;
for those that are more than solvent, the benefit reductions are decreased
until solvency is achieved but not exceeded.

40 While scaling allows comparisons across distributions over 75 years,
the different indexing scenarios are not identical in terms of
sustainability.

The right half of figure 3 shows the distribution of monthly benefits for
each of the scaled indexing scenarios and the benchmark scenario. Once the
different indexing scenarios are scaled to achieve solvency, the
distribution of benefits for each scenario is almost identical in terms of
the level of benefits. Differences in the distributions deal with the
timing associated with implementing the changes. Scaling the indexing
scenarios also reveals that the shape of the distributions is the same.
The distributions of monthly benefits for the indexing scenarios are also
very similar to the distribution of benefits generated under the benefit
reduction benchmark. Therefore, changes to the benefit formula, applied
through the replacement factors, will have similar results regardless of
whether the change is an indexing change or a straight benefit reduction,
because of the proportional effect of the change.

Indexing Approaches Could Also Be Modified to Achieve Nonproportional Effects

Indexing could also be modified to achieve other distributional goals. For
example, so-called progressive indexing, or the use of different
indexes-such as prices and wages-at various earnings levels, has been
proposed as a way of changing the indexing while protecting the benefits
of low-income workers. Thus, under progressive price indexing, those
individuals with indexed lifetime earnings below a certain point would
still have their initial benefits adjusted by wage indexing; those
individuals with earnings above that level would be subject to a
combination of wage and price indexing on a sliding scale, with those
individuals with the highest lifetime earnings having their benefits
adjusted completely by price indexing.41

The effect that progressive price indexing would have on the benefit
formula can be seen in figure 5, where the CPI indexing scenario is
compared to a progressive CPI indexing scenario and to benefits promised
under the current program formula.42 Many lower-income individuals would
do better under the progressive application of the CPI index than under
the CPI indexing alone. However, a progressive application of CPI indexing
does not by itself achieve 75-year solvency, and further changes would be
necessary to do so. Figure 6 shows what happens to the benefit formula
when each of these indexing scenarios is scaled to achieve comparable
levels of solvency over 75 years. Under progressive price indexing, to
protect the benefits of low-income workers, the indexing to prices at
higher earnings levels begins to flatten out benefits, causing the line in
figure 6 to plateau. Thus, under this scenario, most individuals with
earnings above a certain level would receive about the same level of
benefits regardless of income-in the case of figure 6, a retiree with
average indexed monthly earnings of $2,000 would receive a similar benefit
level as someone with average indexed monthly earnings of $7,000. Since
progressive price indexing would change the shape of the benefit formula,
making it more progressive, it would reduce individual equity for higher
earners, as they would receive much lower benefits relative to their
contributions.

41 For more details on the progressive price indexing proposal, see
provision B7 of the August 10, 2005 Office of the Chief Actuary (OCACT)
memo at
http://www.ssab.gov/documents/advisoryboardmemo--2005tr--08102005.pdf ,
which was the basis for our analysis.

42 Progressive price indexing and progressive CPI indexing are two ways of
referring to the same proposal.

Figure 5: Lower-Income Individuals Would Fare Comparatively Better under
the Progressive Application of the CPI Index than under the CPI Index
Alone (Initial Benefits in 2050)

Note: The illustrated PIAs are for individuals who become eligible in
2050. The CPI index links the growth of initial benefits to changes in the
CPI. Progressive CPI indexing uses different indexes at various earnings
levels. Individuals with earnings below a certain point would have their
initial benefits adjusted by wage indexing; those individuals with
earnings above that level would be subject to a combination of wage and
price indexing on a sliding scale, with those individuals with the highest
earnings having their benefits adjusted completely by price indexing.
However, progressive CPI indexing does not achieve 75-year solvency, it
has only a 74 percent improvement in solvency. For more information on the
indexes see appendix I.

Figure 6: Scaling the Progressive Application of the CPI Index to Achieve
Equivalent Solvency Demonstrates That Most Individuals above a Certain
Point Would Receive About the Same Level of Benefits (Initial Benefits in
2050)

Note: The illustrated PIAs are for individuals who become eligible in
2050. The CPI index links the growth of initial benefits to changes in the
CPI. Progressive CPI indexing uses different indexes at various earnings
levels. Individuals with earnings below a certain point would have their
initial benefits adjusted by wage indexing; those individuals with
earnings above that level would be subject to a combination of wage and
price indexing on a sliding scale, with those individuals with the highest
earnings having their benefits adjusted completely by price indexing.
While scaling allows comparisons across distributions over 75 years, the
different indexing scenarios are not identical in terms of sustainability.
For more information on the indexes and the scaling, see appendix I.

While proposals that have suggested progressive indexing have focused on
using prices, any index can be adjusted to achieve the desired level of
progressivity, and the results will likely be similar. However, to the
extent that wages grow faster than the new index over a long period of
time, the benefit formula will eventually flatten out and all individuals
above a certain income level would receive the same level of benefits.

Indexing Applied to Taxes Would Have Adequacy and Equity Considerations

Indexing changes could also be applied to program financing. Under the
current structure of the system, one way this could be accomplished is by
indexing the Social Security payroll tax rate.43 As with indexing
benefits, the payroll tax rate could be indexed to any economic or
demographic variable. Under the tax scenarios presented, only the indexing
of taxes would change, so promised benefits would be maintained. However,
workers would be paying more in payroll taxes, which, like any tax change,
could affect work, saving, and investment decisions.

While benefit levels would be higher under tax increase scenarios, as
compared to benefit reduction scenarios, the timing of the tax changes
matters, just as it did with benefit changes. Since benefits would be
unchanged in the tax-increase-only scenarios, we use benefit-to-tax ratios
to compare the effects of different tax increase scenarios. Benefit-to-tax
ratios compare the present value of Social Security lifetime benefits with
the present value of lifetime Social Security taxes.44 The benefit-to-tax
ratio is an equity measure that focuses on whether, over their lifetimes,
beneficiaries can expect to receive a fair return on their contributions
or get their "money's worth" from the system. With benefits unchanged in
the tax increase scenarios, the benefit-to-tax ratios would vary across
scenarios because of differences in the timing of tax increases.45

To illustrate the effects of the timing of a change in tax rates, figure 7
shows the benefit-to-tax ratios, for four different birth cohorts, for two
tax increase scenarios: (1) the dependency ratio tax indexing scenario
scaled to achieve 75-year solvency and (2) our tax increase benchmark
scenario that increases taxes just enough to achieve program solvency over
a 75-year projection period.46 By raising payroll taxes once and
immediately, the tax increase benchmark would spread the tax burden more
evenly across generations. This is seen in figure 7, where the
benefit-to-tax ratios are fairly stable across cohorts for this
scenario.47 The dependency ratio tax indexing scenario would increase the
tax rate annually, in this case with changes in the dependency ratio.
Under this scenario, later cohorts would face a higher tax rate and thus
bear more of the tax burden, compared to earlier cohorts. This would
result in declining benefit-to-tax ratios across cohorts, with later
generations receiving relatively less compared to their contributions.

43 Under the current system, the maximum taxable earnings level, the level
at which earnings are subject to the Social Security payroll tax, is
indexed to the growth in wages, but the payroll tax rate itself is not
indexed. Some proposals have suggested changing the indexing of the
maximum taxable earnings level so that it maintains coverage of 90 percent
of all wages. Other proposals have not focused on the 90 percent goal, but
rather have suggested raising or completely eliminating the cap. With any
of these changes, several issues arise, most importantly whether the
benefit formula takes into account these higher earnings. These issues go
beyond the scope of our work, and thus we did not analyze changes to the
maximum taxable earnings level.

44 A value less than one, for example, indicates that benefits collected
fall short of taxes paid. The present value of benefits or taxes is the
equivalent value, at a point in time, of the entire stream of benefits the
individual receives or taxes the individual pays in his or her lifetime.

45 Changing benefits would also affect the benefit-to-tax ratios, which
would have adequacy and equity considerations.

46 The tax increase benchmark is a hypothetical benchmark policy scenario
that would achieve 75-year solvency by only increasing payroll taxes. It
raises payroll taxes once and immediately by the amount of Social
Security's actuarial deficit as a percentage of payroll (1.96 percentage
points divided evenly between employers and employees). It results in the
smallest ultimate tax rate that would achieve 75-year solvency and spreads
the tax burden evenly across generations. See appendix I for more
information about the tax increase benchmark.

47 Since the 1955 cohort reaches age 62 in 2017, the earliest age of
eligibility for retired worker benefits, members of this cohort will spend
fewer years contributing to the system at the higher tax rate than the
other cohorts. Thus, their benefit-to-tax ratios will be higher than those
for the other cohorts. Also, since lifetime benefits grow over time as
people live longer, the benefit-to-tax ratios for the tax increase
benchmark will begin to increase, as can be seen for the 2000 cohort.

Figure 7: A Onetime Payroll Tax Increase Would Spread the Tax Burden More
Evenly across Cohorts than Gradual Increases through an Index

Note: Benefit-to-tax ratios are the sum of the present value of family
benefits divided by the present value of family taxes summed for all
individuals in the cohort that survive until age 24. Scenarios are modeled
using the 2005 trustees' report intermediate assumptions. The dependency
ratio tax index links the growth in the payroll tax rate to changes in the
dependency ratio, the ratio of the number of retirees to workers. The
dependency ratio tax index has been adjusted-or "scaled"-to achieve a
75-year actuarial balance of zero. While scaling allows comparisons across
distributions over 75 years, the different indexing scenarios are not
identical in terms of sustainability. The tax increase benchmark is a
hypothetical benchmark policy scenario that would achieve 75-year solvency
by only increasing payroll taxes. For a more complete description of the
indexing scenario, the scaling, or the benchmark, see appendix I.

Revising COLA for Existing Beneficiaries Would Have Important Distributional
Implications for Multiple Subpopulations

Indexing changes can also be applied to the COLA used to adjust existing
benefits. Under the current structure of the program, benefits for
existing beneficiaries are adjusted annually in line with changes in the
CPI. The COLA helps to maintain the purchasing power of benefits for
current retirees. Some proposals, under the premise that the current CPI
overstates the rate of price inflation because of methodological issues
associated with how the CPI is calculated, would alter the COLA. Figure 8
shows the difference in benefit growth over time under the current COLA
and two alternatives: growing at rate of CPI minus 0.22 and growing at
rate of CPI minus 1.48 Changes to the COLA would also have adequacy
implications. After 20 years, benefits growing at the rate of the CPI
minus 0.22 would slow the growth of benefits by about 4 percent below the
level given by the current COLA and growing at the rate of the CPI minus 1
by about 17 percent. This slower benefit growth would improve the finances
of the system, but would also alter the distribution of benefits,
particularly for some subpopulations. Since changes to the COLA compound
over time, those most affected are those with longer life expectancies,
for example, women, as they would have the biggest decrease in lifetime
benefits as they tend to receive benefits over more years. In addition, as
education is correlated with greater life expectancy, highly educated
workers would also experience a significant benefit decrease. There could
also be a potentially large adverse effect on the benefits paid to
disabled beneficiaries, especially among those who become disabled at
younger ages and receive benefits for many years. These beneficiaries
could have a large decrease in lifetime benefits.49

48The 0.22 percentage point reduction in the growth of the CPI has been
proposed as a modification to the COLA to correct methodological issues
associated with how the CPI is calculated. Thus the COLA would be based on
a new CPI-W series that would reflect a "superlative" formula, of the type
currently used for the new chained CCPI-U. The 1 percentage point
reduction in the CPI is another possibility for slowing the growth of
benefits that has been analyzed by the Office of the Actuary at SSA.

49 Since the current benefit formula links the calculation of benefits for
all beneficiaries, any proposed changes would affect the benefits of
disabled workers as well as retirees. Proposals to reform Social Security
often modify the benefit formula without taking into account that the
circumstances facing disabled workers differ from those facing retired
workers. See the next section of this report for a discussion of this
issue, as well as GAO-03-310 , and GAO, Social Security Reform: Potential
Effects on SSA's Disability Programs and Beneficiaries, GAO-01-35
(Washington, D.C.: Jan. 24, 2001).

Figure 8: The Growth of $1,000 Benefit under the CPI and Two Alternatives
Illustrate That Those Beneficiaries Who Receive Benefits Longer Will Be
Affected the Most

Reducing the COLA would also have equity implications. Since the COLA is
applied to all beneficiaries, reductions in the COLA would lower the
return on contributions for all beneficiaries. However, the magnitude of
the effect will vary across subpopulations, similar to its effect on
adequacy. Those individuals who have the biggest decrease in lifetime
benefits will have the biggest decrease in individual equity. While these
individuals have a large decrease in equity, they would still receive
higher lifetime benefits since they live longer and collect benefits over
more years. Individuals with shorter life expectancies will experience a
decrease in equity, but they will fare comparably better than other groups
that live longer, since their lifetime benefits will decrease much less.
Therefore, men, African-Americans, low earners, and less educated
individuals would experience a much smaller decrease in equity compared to
their counterparts.

                    Key Considerations in Choosing an Index

Indexing raises other important considerations about the program's role,
the stability of the variables underlying the index, and the treatment of
Disability Insurance (DI) beneficiaries. The choice of the index implies
certain assumptions about the appropriate level of benefits and taxes for
the program. Thus, if the current indexing of initial benefits was changed
to price growth, there is an implication that the appropriate level of
benefits is one that maintains purchasing power over time rather than the
current approach that maintains a relative standard of living across age
groups (i.e., replacement rates). The solvency effects of an index are
predicated upon the relative stability and historical trends of the
underlying economic or demographic relationships implied by the index. For
example, the 1970s were a period of much instability, in which actual
inflation rates and earnings growth diverged markedly from past
experience, with the result that benefits unexpectedly grew much faster
than expected. Finally, since the benefit formulas for the Old-Age and
Survivors Insurance (OASI) and DI programs are linked, an important
consideration of any indexing proposal is its effect on the benefits
provided to disabled workers. Disabled worker beneficiaries typically
become entitled to benefits much sooner than retired workers and under
different eligibility criteria. As with other ways to change benefits, an
index that is designed to improve solvency by adjusting retirement
benefits may result in large reductions to disabled workers, who often
have fewer options to obtain additional income from other sources.

Choice of a Particular Index Implies Assumptions about the Appropriate Level of
Benefits and Taxes, Adequacy and Equity

The choice of an index suggests certain assumptions about the appropriate
level of benefits and the overall goal of the program.50 The current
indexing of initial benefits to wage growth implies that the appropriate
level of benefits is one that maintains replacement rates across birth
years. In turn, maintaining replacement rates implies a relative standard
of adequacy and an assumption that initial benefits should reflect the
prevailing standard of living at the time of retirement. In contrast,
changing the current indexing of initial benefits to price growth implies
that the appropriate level of benefits is one that maintains purchasing
power.51 In turn, maintaining purchasing power implies an absolute
standard of adequacy and an assumption that initial benefits should
reflect a fixed notion of adequacy regardless of improvements in the
standard of living. Also, any index that does not maintain purchasing
power results in workers born in one year receiving higher benefits than
workers with similar earnings born 1 year later.52 This would occur with
any benefit change that would reduce currently promised benefits more than
price indexing initial benefits would, since price indexing maintains the
purchasing power of initial benefits. In the case of longevity indexing,
if the growth of initial benefits were indexed to life expectancy, then
this implies that the increased costs of benefits that stem from
increasing life expectancy should be borne by all future beneficiaries,
even if society has become richer. Therefore, the desired outcome, in
terms of initial benefit levels at the time of retirement, should drive
the choice of an index.

50 Most proposals that change the indexing of initial benefits would
implement the new index through the benefit formula by multiplying the
replacement factors by the difference between the growth in wages and the
growth in the new index. In such instances, changing the indexing would
not likely pose any serious implementation issues from an agency
operational perspective.

The current indexing of existing benefits with the COLA implies that
maintaining the purchasing power of benefits for current retirees is the
appropriate level of benefits. Revising the COLA to reflect a more
accurate calculation of the CPI retains this assumption. However,
adjusting the COLA in a way that does not keep pace with the CPI would
change that assumption and imply a view that the costs of reform should be
shared by current as well as future retirees.

Similarly, on the revenue side, the program currently uses a constant tax
rate, which maintains the same proportion of taxes for all workers earning
less than the maximum taxable earnings level. Applying a life expectancy
index to payroll tax rates suggests that the appropriate level of taxes is
one that prefunds the additional retirement years increased life
expectancy will bestow on current workers, but also that the appropriate
level of benefits is one that maintains replacement rates, as benefits are
unchanged.

51 Purchasing power reflects the amount of goods and services individuals
can afford with a given level of benefits.

52 This is the so-called notch effect. Such a situation occurred
immediately after the 1977 amendments. Notches generate controversy and
confusion among beneficiaries because of inequities that result from them.
See GAO/T-HEHS-94-236 .

Stability of Economic or Demographic Relationships Underlying the Index Is a
Consideration

Indexing raises other considerations about the stability of the underlying
relationships between the economic and demographic variables captured by
the index. The choice of an index includes issues of risk and methodology.
Some indexes could be based on economic variables that are volatile,
introducing instability because the index generates wide swings in
benefits or taxes. In other cases, long-standing economic or demographic
relationships premised by the index could change, resulting in
unanticipated and unstable benefit or tax levels. While most indexes will
also pose methodological issues, these can become problematic to address
after the index has already been widely used, and the correction will have
implications for benefits or taxes. An example is the current measurement
limitations of the CPI. In other instances, the index may be based on
estimates about future trends in variables like mortality that could later
prove incorrect and erode public confidence in the system.

Some indexes are premised on the past behavior of economic or demographic
relationships. If these long-standing relationships diverge for a
significant period of time, they may result in unanticipated and unstable
benefit or tax levels. For example, the 1972 amendments that introduced
indexing into the Social Security program were premised on the belief that
over time, wage growth will generally substantially exceed price
inflation. However, for much of the 1970s, actual inflation rates and
earnings growth diverged markedly from past experience; price inflation
grew much faster than wages, with the result that benefits grew much
faster than anticipated. This development introduced major instability
into the program, which was unsustainable. Congress addressed this problem
when it passed the 1977 amendments.53 Moreover, even though the 1977
amendments succeeded in substantially stabilizing the replacement rates
for initial benefits, a solvency crisis required reforms just 6 years
later with the 1983 amendments. High inflation rates resulted in high
COLAs for existing benefits just as recession was depressing receipts from
the payroll taxes. The indexing of initial benefits under the 1977
amendments did not address the potential for such economic conditions to
affect COLAs or payroll tax receipts.

Many indexes have methodological issues associated with their calculation,
which can become problems over time. For example, the CPI has long been in
use by the Social Security program and other social welfare programs.
However, the CPI is not without its methodological problems. Some studies
have contended that the CPI overstates inflation for a number of reasons,
including that it does not account for how consumers can substitute one
good for another because the calculation assumes that consumers do not
change their buying patterns in response to price changes.54 Correcting
for this "substitution effect" would likely lower the CPI. Changing the
calculation in response to this concern might improve accuracy but is
controversial because it would also likely result in lower future benefits
and put more judgment into the calculation.

53 For more detail on the 1977 amendments, see appendix II.

Indexes that are constructed around assumptions about future experience
raise other methodological issues. An example is a mortality index, which
seeks to measure future changes in population deaths. Such a measure would
presumably capture an aspect of increased longevity or well-being in
retirement and could be viewed as a relevant determinant of program
benefits or taxes. Accuracy in this index would require forecasts of
future mortality based on assumptions of the main determinants influencing
future population deaths (i.e., medical advances, diet, income changes).
Such forecasts would require a clear consensus about these factors and how
to measure and forecast them. However, currently there is considerable
disagreement among researchers in terms of their beliefs about the
magnitude of mortality change in the future.55 In choosing an index, such
methodological issues would need to be carefully considered to maintain
public support and confidence.

54For more information on the CPI and how it overstates the true rate of
inflation, see Advisory Commission to Study the Consumer Price Index,
"Toward a More Accurate Measure of the Cost of Living," Final Report to
the Senate Committee on Finance, Dec. 1996; Congressional Budget Office,
"Is the Growth of the CPI a Biased Measure of Changes in the Cost of
Living?" (Washington, D.C., 1994). In recent years a variety of changes
have been made to the CPI, including changes that in turn affect Social
Security's COLA. In addition, a new "chained" CPI reflects how consumers
substitute one product for another when their relative prices change. This
new CPI is not yet used by government agencies, but some reform proposals
call for using a variation of it in computing COLAs.

55 See Ronald D. Lee and Lawrence R. Carter, "Modeling and Forecasting
U.S. Mortality," Journal of the American Statistical Association, Vol. 87,
No. 419, 1992; and Michael Sze, Stephen C. Goss, and Jose Gomez de Leon,
"Effect of Aging Population with Declining Mortality on Social Security
and NAFTA Countries," North American Actuarial Journal, Vol. 2, No. 4,
1998.

The Treatment of Disabled and Survivor Beneficiaries Poses Challenges When
Modifying the Indexing

Under the current structure of the U.S. Social Security system, the OASI
and DI programs share the same benefit formula. Thus, any changes that
affect retired workers will also affect survivors and disabled workers.
However, the circumstances facing these beneficiaries differ from those
facing retired workers. For example, the disabled worker's options for
alternative sources of income, especially earnings-related income, to
augment any reduction in benefits are likely to be more limited than are
those for the retired worker. Further, DI beneficiaries enter the program
at younger ages and may receive benefits for many years. As a result,
disabled beneficiaries could be subject to benefit changes for many years
more than those beneficiaries requiring benefits only in retirement.56

These differing circumstances among beneficiaries raise the issue of
whether any proposed indexing changes, or any other benefit changes,
should be applied to disabled worker and survivor beneficiaries, as well
as to retired worker beneficiaries.57 If disabled worker beneficiaries are
not subject to indexing changes applied to retirees, benefit levels for
disabled workers could ultimately be higher than those of retired workers.
This difference in benefit levels would occur because disabled workers
typically become entitled to benefits sooner than retired workers, and
thus any reductions in their replacement factors would be smaller. Such a
differential could increase the incentive for older workers to apply for
disability benefits as they near retirement age.

Excluding the disability program from indexing changes has implications
for solvency and raises implementation issues. If the indexing changes are
not applied to the disability program, even larger benefit reductions or
revenue increases would be needed to achieve fiscal solvency. Since the
OASI and DI programs share the same benefit formula, excluding disabled
worker beneficiaries from indexing changes might also necessitate the use
of two different benefit formulas or require a method to recalculate
benefits in order to maintain different indexing in each program. Such
changes could lead to confusion among the public about how the programs
operate, which may require significant additional public education.

56 For more information on the effects of reform on the DI program and
beneficiaries, see GAO-01-35 .

57 Some proposals have suggested reducing the disabled worker benefit only
at the time of conversion from DI to retired worker status, but only in
proportion to the percentage of their potential working years that
occurred in a nondisabled state.

                            Concluding Observations

Indexing has played an important role in the determination of Social
Security's benefits and revenues for over 30 years. As in other countries
seeking national pension system reform, recent proposals to modify the
role of indexing in Social Security have primarily focused on addressing
the program's long-term solvency problems. In theory, one index may be
better than another in keeping the program in financial balance on a
sustainable basis. However, such a conclusion would be based on
assumptions about the future behavior of various demographic and economic
variables, and those assumptions will always have considerable
uncertainty. Future demographic patterns and economic trends could emerge
that affect solvency in ways that have not been anticipated. So, while
indexing changes may reduce how often Congress needs to rebalance the
program's finances, there is no guarantee that the need will not arise
again.

Yet program reform, and the role of indexing in that reform, is about more
than solvency. Reforms also reflect implicit visions about the size,
scope, and purpose of the Social Security system. Indexing initial
benefits, existing benefits, tax rates, the maximum taxable earnings
level, or some other parameter or combination will have different
consequences for the level and distribution of benefits and taxes, within
and across generations and earnings levels. These questions relate to the
trade-off between income adequacy and benefit equity.

In the final analysis, indexing, like other individual reforms, comes down
to a few critical questions: What is to be accomplished or achieved, who
is to be affected, is it affordable and sustainable, and how will the
change be phased in over time? Although these issues are complex and
controversial, they are not unsolvable; they have been reconciled in the
past and can be reconciled now. Indexing can be part of a larger, more
comprehensive reform package that would include other elements whose
cumulative effect could achieve the desired balance between adequacy and
equity while also achieving solvency. The challenge is not whether
indexing should be part of any necessary reforms, but that necessary
action is taken soon to put Social Security back on a sound financial
footing.

                                Agency Comments

We provided a draft of this report to SSA and the Department of the
Treasury. SSA provided technical comments, which we have incorporated as
appropriate.

We are sending copies of this report to the Social Security Administration
and the Treasury Department, as well as other interested parties. Copies
will also be made available to others upon request. In addition, the
report will be available at no charge on the GAO Web site at
http://www.gao.gov . Please contact me at (202) 512-7215, if you have any
questions about this report. Other major contributors include Charles
Jeszeck, Michael Collins, Anna Bonelli, Charles Ford, Ken Stockbridge,
Seyda Wentworth, Joseph Applebaum, and Roger Thomas.

Barbara D. Bovbjerg Director, Education, Workforce, and Income Security
Issues

Appendix I: Methodology

                             Microsimulation Model

Description

Genuine Microsimulation of Social Security and Accounts (GEMINI) is a
microsimulation model developed by the Policy Simulation Group (PSG).
GEMINI simulates Social Security benefits and taxes for large
representative samples of people born in the same year. GEMINI simulates
all types of Social Security benefits, including retired worker, spouse,
survivor, and disability benefits. It can be used to model a variety of
Social Security reforms including the introduction of individual accounts.

GEMINI uses inputs from two other PSG models, the Social Security and
Accounts Simulator (SSASIM), which has been used in numerous GAO reports,
and the Pension Simulator (PENSIM), which has been developed for the
Department of Labor. GEMINI relies on SSASIM for economic and demographic
projections and relies on PENSIM for simulated life histories of large
representative samples of people born in the same year and their
spouses.58 Life histories include educational attainment, labor force
participation, earnings, job mobility, marriage, disability, childbirth,
retirement, and death. Life histories are validated against data from the
Survey of Income and Program Participation, the Current Population Survey,
Modeling Income in the Near Term (MINT3),59 and the Panel Study of Income
Dynamics. Additionally, any projected statistics (such as life expectancy,
employment patterns, and marital status at age 60) are, where possible,
consistent with intermediate cost projections from Social Security
Administration's Office of the Chief Actuary (OCACT). At their best, such
models can provide only very rough estimates of future incomes. However,
these estimates may be useful for comparing future incomes across
alternative policy scenarios and over time.

GEMINI can be operated as a free-standing model or it can operate as a
SSASIM add-on. When operating as an add-on, GEMINI is started
automatically by SSASIM for one of two purposes. GEMINI can enable the
SSASIM macro model to operate in the Overlapping Cohorts (OLC) mode or it
can enable the SSASIM micro model to operate in the Representative Cohort
Sample (RCS) mode. The SSASIM OLC mode requests GEMINI to produce samples
for each cohort born after 1934 in order to build up aggregate payroll tax
revenues and OASDI benefit expenditures for each calendar year, which are
used by SSASIM to calculate standard trust fund financial statistics. In
either mode, GEMINI operates with the same logic, but typically with
smaller cohort sample sizes in OLC mode than in the RCS or
stand-alone-model mode.

58 While these models use sample data, our report, like others using these
models, does not address the issue of sampling errors. The results of the
analysis reflect outcomes for individuals in the simulated populations and
do not attempt to estimate outcomes for an actual population.

59 MINT3 is a detailed microsimulation model developed jointly by the
Social Security Administration, the Brookings Institution, RAND, and the
Urban Institute to project the distribution of income in retirement for
the 1931 to 1960 birth cohorts.

For this report we used GEMINI to simulate Social Security benefits and
taxes primarily for 100,000 individuals born in 1985. Benefits and taxes
were simulated under our tax increase (promised benefits) and proportional
benefit reduction (funded benefits) benchmarks (described below) and
various indexation approaches.

Assumptions and Limitations

To facilitate our modeling analysis, we made a variety of assumptions
regarding economic and demographic trends. In choosing our assumptions, we
focused our analysis to illustrate relevant points about distributional
effects and hold equal as much as possible any variables that were either
not relevant to or would unduly complicate that focus. As a result of
these assumptions, as well as issues inherent in any modeling effort, our
analysis has some key limitations, especially relating to risk and changes
over time.

  2005 Social Security Trustees' Assumptions

The simulations are based on economic and demographic assumptions from the
2005 Social Security trustees' report.60 While the 2006 trustees' report
has been released, the assumptions have changed very little from the 2005
assumptions. We used trustees' intermediate assumptions for inflation,
real wage growth, mortality decline, immigration, labor force
participation, and interest rates.

  Distributional Effects Over Time

We simulated benefits for individuals born in 1955, 1970, 1985, and 2000.
However, the majority of our figures focus on individuals born in 1985
because all prospective indexing changes would be almost fully phased in
for these individuals. However, the distributional effects might change
over time. This is because each index phases in over time and reduces the
primary insurance amount (PIA) formula factors (or increases the Old-Age
and Survivors Insurance (OASI) and Disability Insurance (DI) taxes) at
different rates. For example, individuals in the 1955 cohort that survive
to age 65 do so in the year 2020, so the benefit reductions (or tax
increases), which we specify to begin sometime between 2006 and 2012,
depending on the scenario, have only been implemented for about 8 to 14
years. Additionally, members of the cohort that become disabled might
become disabled prior to the implementation of annual PIA reductions or
tax increases. Such issues become less pronounced with the younger
cohorts.

60 The Board of Trustees, Federal Old-Age and Survivors Insurance and
Disability Insurance Trust Funds, The 2005 Annual Report of the Board of
Trustees of the Federal Old-Age and Survivors Insurance and Disability
Insurance Trust Funds (Washington, D.C.: Mar. 23, 2005).

  Pre-retirement Mortality

To capture the distributional impact of pre-retirement mortality, we
calculated benefit-to-tax ratios and lifetime benefits for all sample
members who survived past age 24. However, our measure of well-being,
lifetime earnings, may not be the best way to assess the well-being of
those who die before retirement. Some high-wage workers are classified as
low lifetime earners simply because they did not live very long, and
consequently our analysis overstates the degree to which those who die
young are classified as low earners. As a result, our measures
underestimate the degree to which Social Security favors lower earners
under all of the scenarios we analyze.61

Description of Alternative Policy Scenarios

  CPI Indexing

To simulate consumer price indexing (CPI) indexing, which essentially
links the growth of initial benefits to changes in the CPI, we
successively modified the PIA formula replacement factors (90, 32, and 15)
beginning in 2012, reducing them successively by real wage growth in the
second prior year. This specification mimics provision B6 of the August
10, 2005 memorandum to SSA's Chief Actuary regarding the provision
requested by the Social Security Advisory Board (SSAB), which is an update
of provision 1 of Model 2 of the President's Commission to Strengthen
Social Security (CSSS).62 As noted in the CSSS solvency memorandum from
SSA's Chief Actuary, "[t]his provision would result in increasing benefit
levels for individuals with equivalent lifetime earnings across
generations (relative to the average wage level) at the rate of price
growth (increase in the CPI), rather than at the rate of growth in the
average wage level as in current law."

61 For benefit-to-tax ratios we followed the methodology followed in
GAO-04-747 ; see appendix I of this report for more detail.

62 See page 3 of
http://www.ssab.gov/documents/advisoryboardmemo-2005tr-08102005.pdf for
description of the provision. For the original provision from the
President's Commission to Strengthen Social Security, see page 4 of
http://www.ssa.gov/OACT/solvency/PresComm_20020131.pdf .

This provision as specified and scored by OCACT in the SSAB memo would
increase the size of the long-range OASDI actuarial balance (reduce the
actuarial deficit) by an estimated 2.38 percent of taxable payroll. Using
the overlapping cohort mode of SSASIM, we estimated this provision as
increasing the size of the long-range OASDI actuarial balance by 2.43
percent of taxable payroll, or 5 basis points more than the OCACT scoring.

  Mortality Indexing

To simulate mortality indexing, which links the growth of initial benefits
to changes in life expectancy to maintain a constant life expectancy at
the normal retirement age, we successively modified the PIA formula
replacement factors (90, 32, 15) beginning in 2009, reducing them annually
by multiplying them by 0.995. This specification mimics provision 1 of
Model 3 of CSSS.63 The CSSS solvency memorandum notes that the 0.995
successive reduction "reduces monthly benefit levels by an amount
equivalent to increasing the normal retirement age (NRA) for retired
workers by enough to maintain a constant life expectancy at NRA, for any
fixed age of benefit entitlement."64

This provision as specified and scored-using the intermediate assumptions
of the 2001 trustees' report-in the CSSS memo by SSA's Office of the Chief
Actuary would reduce the size of the long-range OASDI actuarial balance
(reduce the actuarial deficit) by an estimated 1.17 percent of taxable
payroll. Using the overlapping cohort mode of SSASIM and specifications,
which mimic the intermediate assumptions of the 2005 trustees' report, we
estimated this provision as increasing the size of the long-range OASDI
actuarial balance by 1.39 percent of taxable payroll, or 22 basis points
more than the earlier OCACT scoring.

63 For more information on provision 1 or Model 3, see page 8 of the CSSS
proposal at http://www.ssa.gov/OACT/solvency/PresComm_20020131.pdf .

64 We chose the CSSS specification because it was already scored and
readily available. Other constructions or interpretations of a mortality
index are certainly possible. For example, life expectancy at birth or
some other age could be used. Further, life expectancy could be defined as
period or cohort. A period life table represents the mortality conditions
at a specific point in time, whereas a cohort table depicts the mortality
conditions of a specific group of individuals born in the same year or
series of years.

  Dependency Indexing

Benefits

To simulate so-called dependency indexing of benefits, which links the
growth of initial benefits to changes in the dependency ratio, we
successively modified the PIA formula replacement factors (90, 32, and 15)
beginning in 2010, by reducing them annually by an index that follows the
inverse of the increase in the aged dependency ratio from 2 years prior.
For example, the reduction for 2010 is given by dividing the 2009 PIA
formula factors (90, 32, and 15) by 1.0098, which is rate of increase from
2007 to 2008.65

This provision as specified has not been scored by OCACT. Using the
overlapping cohort mode of SSASIM and specifications that mimic the
intermediate assumptions of the 2005 trustees' report, we estimated this
provision as increasing the size of the long-range OASDI actuarial balance
by 3.78 percent of taxable payroll.

Taxes

To simulate so-called dependency indexing of payroll taxes, which links
the growth of payroll taxes to changes in the dependency ratio, we
increased the initial OASI and DI tax rates (both employer and employee
combined) in 2009 by a cumulative index that increases annually by the
rate of increase in the aged dependency ratio from 2 years prior. For
example, the increase for 2010 is given by multiplying the 12.4 percent
tax rate (employer and employee combined-10.60 percent for OASI and 1.80
percent for DI) by 1.0098-the rate of increase from 2007 to 2008-to arrive
at a rate of 10.70 percent for OASI and 1.82 percent for DI. By 2050 the
cumulative index is 1.863, and the tax rates (employer and employee
combined) are 19.75 percent for OASI and 3.35 percent for DI.

This provision as specified has not been scored by OCACT. Using the
overlapping cohort mode of SSASIM and specifications that mimic the
intermediate assumptions of the 2005 trustees' report, we estimated this
provision as increasing the size of the long-range OASDI actuarial balance
by 6.98 percent of taxable payroll.

65 The aged dependency ratio is 0.204 and 0.206 under the intermediate
assumptions of the 2005 Trustees' report for 2007 and 2008, respectively.

  Scaling to Achieve Comparable Levels of Solvency over 75 Years

We modified the aforementioned CPI, mortality, and dependency indexes to
"scale" them to achieve comparable levels of solvency over a 75-year
period-the same actuarial period used by OCACT in trustees' reports and
solvency memorandums.66 To scale the proposals, we modified the PIAs (or
OASI and DI tax rates in the case of the aged dependency ratio tax
increase index) by a scaled factor equal to the inverse of the percentage
of solvency attained by the original, unscaled version of the proposal.
For each year in the 75-year period, the scaling factor is multiplied by
the percentage point difference between the unscaled PIA factors and the
factors prior to implementation of the proposal (i.e., 90, 32, and 15).67
The application of the so-called scaling factor to the PIA factors (or
OASDI tax rates) conveniently modifies the index in such a way that
75-year actuarial balance is 0.68

66 Though we do not present SSASIM or GEMINI results for the progressive
CPI index, we also scaled this proposal. For an OCACT scoring of this
proposal, which was the basis of our SSASIM OLC estimates for the scaled
and unscaled versions presented in the report, see provision B7 of the
August 10, 2005 OCACT memo at
http://www.ssab.gov/documents/advisoryboardmemo--2005tr--08102005.pdf. To
scale this scenario, we consulted with OCACT and only scaled the third and
fourth PIA formula factors, as these were the only factors reduced in the
original provision. This effectively sped up the rate of indexing so that
the benefit reductions were faster than pure price indexing across
generations of steady maximum earners. Additionally, we had to slightly
raise the scaling value for this scenario because the third and fourth
formula factors would need to have been of a negative value beginning in
2065. However, we censored negative values at zero and raised the scaling
factor by one percentage point to achieve a 75-year actuarial balance of 0
for the scaled version of progressive CPI index.

67 In the case of the aged dependency ratio tax increase index, the
increase from the initial OASI and DI tax rates is multiplied by the
scaling factor-again, represented by the inverse of the percentage of
solvency attained by the index.

68 Despite a similar 75-year actuarial balance across the indexes studied,
each index may have a unique balance in the 75th year because of the
unique timing of the benefit reductions (or tax increases) of each index.

Table 4: Application of So-called Scaling of PIA Factors for Aged
Dependency Ratio Benefit Reduction Index: An Example Using 2050 PIA
Formula Factors

                                          PIA formula PIA formula PIA formula 
                                            factor or   factor or   factor or 
Step Description                       calculation calculation calculation 
1    PIA formula factors in 2009              90.0        32.0        15.0 
2    PIA formula factors in 2050              48.3        17.2         8.1 
3    Difference between factors in                                         
        2050 and initial factors (i.e.,                           
        the factors in 2009 or prior)            41.7        14.8         6.9
4    Scaling factora                          .508        .508        .508 
5    Difference multiplied by scaling                                      
        factor (i.e., step 3 * step 4)           21.2         7.5         3.5
6    New, so-called "scaled" PIA                                           
        formula factors (i.e., step                               
        1-step 5)                                68.8        24.5        11.5

Source: GAO.

aThe aged dependency benefit reduction index increases the 75-year OASDI
actuarial balance by 3.78 percent of taxable payroll. This is 196.9
percent of 1.92 percent of taxable payroll, which is the amount required
to produce a 75-year actuarial balance of 0 under the intermediate
assumptions of the 2005 trustees' report. The inverse (i.e., 1/x) of 196.9
percent is 50.8 percent.

Data Reliability

To assess the reliability of simulated data from GEMINI, we reviewed PSG's
published validation checks, examined the data for reasonableness and
consistency, preformed sensitivity analysis, and compared our solvency
estimates, where applicable, with published results from the actuaries at
the Social Security Administration.

PSG has published a number of validation checks of its simulated life
histories. For example, simulated life expectancy is compared with
projections from the Social Security trustees; simulated benefits at age
62 are compared with administrative data from SSA; and simulated
educational attainment, labor force participation rates, and job tenure
are compared with values from the Current Population Survey. We found that
simulated statistics for the life histories were reasonably close to the
validation targets.

For sensitivity analysis, we simulated benefits and taxes for policy
scenarios under a number of alternative specifications, including limiting
the sample to those who survive to retirement. Our findings were
consistent across all specifications.

                           Benchmark Policy Scenarios

According to current projections of the Social Security trustees for the
next 75 years, revenues will not be adequate to pay full benefits as
defined by the current benefit formula. Therefore, estimating future
Social Security benefits should reflect that actuarial deficit and account
for the fact that some combination of benefit reductions and revenue
increases will be necessary to restore long-term solvency.

To illustrate a full range of possible outcomes, we developed hypothetical
benchmark policy scenarios that would achieve 75-year solvency either by
only increasing payroll taxes or by only reducing benefits.69 In
developing these benchmarks, we identified criteria to use to guide their
design and selection. Our tax-increase-only benchmark simulates "promised
benefits," or those benefits promised by the current benefit formula,
while our benefit-reduction-only benchmarks simulate "funded benefits," or
those benefits for which currently scheduled revenues are projected to be
sufficient. Under the latter policy scenarios, the benefit reductions
would be phased in between 2010 and 2040 to strike a balance between the
size of the incremental reductions each year and the size of the ultimate
reduction.

SSA actuaries scored our original 2001 benchmark policies and determined
the parameters for each that would achieve 75-year solvency.70 Table 5
summarizes our benchmark policy scenarios. For our benefit reduction
scenarios, the actuaries determined these parameters assuming that
disabled and survivor benefits would be reduced on the same basis as
retired worker and dependent benefits. If disabled and survivor benefits
were not reduced at all, reductions in other benefits would be greater
than shown in this analysis.

69These benchmarks were first developed for our report GAO-02-62 . We have
since used them in other studies, including GAO-03-310 ; GAO, Social
Security Reform: Analysis of a Trust Fund Exhaustion Scenario, GAO-03-907
(Washington, D.C.: July 29, 2003); GAO, Social Security and Minorities:
Earnings, Disability Incidence, and Mortality Are Key Factors That
Influence Taxes Paid and Benefits Received GAO-03-387 (Washington, D.C.:
Apr. 23, 2003); and GAO-04-747 .

70The Social Security actuaries provided these scorings for a previous
report and used assumptions from the 2001 trustees' report. The actuaries
did not believe it was necessary to provide new scorings using updated
assumptions for the purposes of our study, since the assumptions and the
estimates of actuarial balance on which they are based have changed little
from the 2001 report. In particular, they did not believe that the
differences in assumptions would materially affect the shape of the
distribution of benefits, which is the focus of our analysis. All
estimates related to the indexing scenarios and benchmark policy scenarios
were simulated using the SSASIM OLC mode.

Table 5: Summary of Benchmark Policy Scenarios

                                                                 Ultimate new 
                                                                      benefit 
Benchmark policy                                    Phase-in   reductionsa 
scenario          Description                       period       (percent) 
Tax increase only Increases payroll taxes in 2006   Immediate            0 
(promised         by amount necessary to achieve              
benefits)         75-year solvency (0.98 percent of           
                     payroll each for employees and              
                     employers)                                  
Proportional      Reduces benefit formula factors   2010-2040           25 
benefit reduction proportionally across all                   
(funded benefits) earnings levels                             

Source: GAO.

aThese benefit reduction amounts do not reflect the implicit reductions
resulting from the gradual increase in the full retirement age that has
already been enacted.

Criteria

According to our analysis, appropriate benchmark policies should ideally
be evaluated against the following criteria:

           1. Distributional neutrality: The benchmark should reflect the
           current system as closely as possible while still restoring
           solvency. In particular, it should try to reflect the goals and
           effects of the current system with respect to redistribution of
           income. However, there are many possible ways to interpret what
           this means, such as

                        a. producing a distribution of benefit levels with a
                        shape similar to the distribution under the current
                        benefit formula (as measured by coefficients of
                        variation, skewness, kurtosis, and so forth),
                        b. maintaining a proportional level of income
                        transfers in dollars,
                        c. maintaining proportional replacement rates, and
                        d. maintaining proportional rates of return.

           2. Demarcating upper and lower bounds: These would be the bounds
           within which the effects of alternative proposals would fall. For
           example, one benchmark would reflect restoring solvency solely by
           increasing payroll taxes and therefore maximizing benefit levels,
           while another would solely reduce benefits and therefore minimize
           payroll tax rates.
           3. Ability to model: The benchmark should lend itself to being
           modeled within the GEMINI model.
           4. Plausibility: The benchmark should serve as a reasonable
           alternative within the current debate; otherwise, the benchmark
           could be perceived as an invalid basis for comparison.
           5. Transparency: The benchmark should be readily explainable to
           the reader.

Tax-Increase-Only, or "Promised Benefits," Benchmark Policies

Our tax-increase-only benchmark would raise payroll taxes once and
immediately by the amount of Social Security's actuarial deficit as a
percentage of payroll. It results in the smallest ultimate tax rate of
those we considered and spreads the tax burden most evenly across
generations; this is the primary basis for our selection. The later that
taxes are increased, the higher the ultimate tax rate needed to achieve
solvency, and in turn the higher the tax burden on later taxpayers and
lower on earlier taxpayers. Still, any policy scenario that achieves
75-year solvency only by increasing revenues would have the same effect on
the adequacy of future benefits in that promised benefits would not be
reduced. Nevertheless, alternative approaches to increasing revenues could
have very different effects on individual equity.

Benefit-Reduction-Only, or "Funded Benefits," Benchmark Policies

We developed alternative benefit reduction benchmarks for our analysis.
For ease of modeling, all benefit reduction benchmarks take the form of
reductions in the benefit formula factors; they differ in the relative
size of those reductions across the three factors, which are 90, 32, and
15 percent under the current formula. Each benchmark has three dimensions
of specification: scope, phase-in period, and the factor changes
themselves.

  Scope

For our analysis, we apply benefit reductions in our benchmarks very
generally to all types of benefits, including disability and survivors'
benefits as well as old-age benefits. Our objective is to find policies
that achieve solvency while reflecting the distributional effects of the
current program as closely as possible. Therefore, it would not be
appropriate to reduce some benefits and not others. If disabled and
survivor benefits were not reduced at all, reductions in other benefits
would be deeper than shown in this analysis.

  Phase-in Period

We selected a phase-in period that begins with those becoming initially
entitled in 2010 and continues for 30 years. We chose this phase-in period
to achieve a balance between two competing objectives: (1) minimizing the
size of the ultimate benefit reduction and (2) minimizing the size of each
year's incremental reduction to avoid "notches," or unduly large
incremental reductions. Notches create marked inequities between
beneficiaries close in age to each other. Later birth cohorts are
generally agreed to experience lower rates of return on their
contributions already under the current system. Therefore, minimizing the
size of the ultimate benefit reduction would also minimize further
reductions in rates of return for later cohorts. The smaller each year's
reduction, the longer it will take for benefit reductions to achieve
solvency, and in turn the greater the eventual reductions will have to be.
However, the smallest possible ultimate reduction would be achieved by
reducing benefits immediately for all new retirees by 13 percent; this
would create a notch.

In addition, we feel it is appropriate to delay the first year of the
benefit reductions for a few years because those within a few years of
retirement would not have adequate time to adjust their retirement
planning if the reductions applied immediately. The Maintain Tax Rates
(MTR) benchmark in the 1994-1996 Advisory Council report also provided for
a similar delay.71

Finally, the timing of any policy changes in a benchmark scenario should
be consistent with the proposals against which the benchmark is compared.
The analysis of any proposal assumes that the proposal is enacted, usually
within a few years. Consistency requires that any benchmark also assumes
enactment of the benchmark policy in the same time frame. Some analysts
have suggested using a benchmark scenario in which Congress does not act
at all and the trust funds become exhausted.72 However, such a benchmark
assumes that no action is taken while the proposals against which it is
compared assume that action is taken, which is inconsistent. It also seems
unlikely that a policy enacted over the next few years would wait to
reduce benefits until the trust funds are exhausted; such a policy would
result in a sudden, large benefit reduction and create substantial
inequities across generations.

71Advisory Council on Social Security. Report of the 1994-1996 Advisory
Council on Social Security, Vols. 1 and 2. Washington, D.C.: Jan. 1997.

72See GAO-03-907 , in which we analyzed such a policy scenario under a
congressional request.

  Defining the PIA Formula Factor Reductions

When workers retire, become disabled, or die, Social Security uses their
lifetime earnings records to determine each worker's PIA, on which the
initial benefit and auxiliary benefits are based. The PIA is the result of
two elements-the Average Indexed Monthly Earnings (AIME) and the benefit
formula. The AIME is determined by taking the lifetime earnings record,
indexing it, and taking the average of the highest 35 years of indexed
wages.73 To determine the PIA, the AIME is then applied to a step-like
formula, shown here for 2006.

PIA = 90% (AIME1 =< $656)

+ 32% (AIME2 > $656 and =< $3955)

+ 15% (AIME3 > $3955)

where AIMEi is the applicable portion of AIME.

All of our benefit-reduction benchmarks are variations of changes in PIA
formula factors.

Proportional reduction: Each formula factor is reduced annually by
subtracting a constant proportion of that factor's value under current
law, resulting in a constant percentage reduction of currently promised
benefits for everyone. That is,

Fit+1 = Fi t - (Fi2006 x)

where

Fit represents the three PIA formula factors in year t and

x = constant proportional formula factor reduction.

The value of x is calculated to achieve 75-year solvency, given the chosen
phase-in period and scope of reductions.

73 The highest 35 years of salary are used in the calculation of a retired
worker benefit. The disabled worker benefit is calculated using the number
of years between the age of entitlement and age 21, divided by 5.

The formula for this reduction specifies that the proportional reduction
is always taken as a proportion of the current law factors rather than the
factors for each preceding year. This maintains a constant rate of benefit
reduction from year to year. In contrast, taking the reduction as a
proportion of each preceding year's factors implies a decelerating of the
benefit reduction over time because each preceding year's factors gets
smaller with each reduction. To achieve the same level of 75-year
solvency, this would require a greater proportional reduction in earlier
years because of the smaller reductions in later years.

The proportional reduction hits lower earners harder than higher earners
because the constant x percent of the higher formula factors results in a
larger percentage reduction over the lower earnings segments of the
formula. For example, in a year when the cumulative size of the
proportional reduction has reached 10 percent, the 90 percent factor would
then have been reduced by 9 percentage points, the 32 percent factor by
3.2 percentage points, and the 15 percent factor by 1.5 percentage points.
As a result, earnings in the first segment of the benefit formula would be
replaced at 9 percentage points less than the current formula, while
earnings in the third segment of the formula would be replaced at only 1.5
percentage points less than the current formula.74

Table 6 summarizes the features of our benchmarks.

74Other analyses have addressed the concern about the effect of the
proportional reduction on low earners by modifying that offset to apply
only to the 32 and 15 percent formula factors. The MTR policy in the 1994
to 1996 Advisory Council report used this approach, which in turn was
based on the individual account (IA) proposal in that report. However, the
MTR policy also reflected other changes in addition to PIA formula
changes.

Table 6: Summary of Benchmark Policy Scenario Parameters

                                Annual PIA factor             
                              reduction (percentage     Ultimate PIA factor
                                     point)              (2040) (percent)
                                  90      32      15       90      32      15 
Benchmark       Phase-in  percent percent percent  percent percent percent 
policy scenario period     factor  factor  factor   factor  factor  factor 
Tax increase                                                               
only (promised                                                     
benefits)            2006       0      00       0    90.00   32.00   15.00
Proportional                                                               
benefit                                                            
reduction                                                          
(funded                                                            
benefits)       2010-2040    0.74    0.26    0.12    67.07   23.85   11.18

Source: GAO's analysis as scored by SSA actuaries.

Note: Annual PIA factor reductions rounded to the nearest hundredth of a
percent.

Appendix II: Background on Development of Social Security's Indexing
Approach

Social Security did not originally use indexing to automatically adjust
benefit and tax provisions; only ad hoc changes were made. The 1972
amendments provided for automatic indexing of benefits and taxes for the
first time, but the indexing approach for benefits was flawed, introducing
potential instability in benefit costs. The 1977 amendments addressed
those issues, resulting in the basic framework for indexing benefits still
in use today.

                    Program Did Not Use Indexing until 1970s

Before the 1970s, the Social Security program did not use indexing to
adjust benefits or taxes automatically. For both new and existing
beneficiaries, benefit rates increased only when Congress voted to raise
them. The same was true for the tax rate and the cap on the amount of
workers' earnings that were subject to the payroll tax. Under the 1972
amendments to the Social Security Act, benefits and taxes were indexed for
the first time, and revisions in the 1977 amendments created the basic
framework still in use today.

Ad Hoc Benefit and Tax Changes Had Sporadic Effects

Until 1950, Congress legislated no changes to the benefit formula of any
kind. As a result, average inflation-adjusted benefits for retired workers
fell by 32 percent between 1940 and 1949. Under the 1950 amendments to the
Social Security Act, these benefits increased 67 percent in 1 year.
Afterward, until 1972, periodic amendments made various ad hoc adjustments
to benefit levels. Economic prosperity and regular trust fund surpluses
facilitated gradual growth of benefit levels through these ad hoc
adjustments.75 In light of the steady growth of benefit levels, the 1972
amendments instituted automatic adjustments to constrain the growth of
benefits as well as to ensure that they kept pace with inflation. Table 7
summarizes the history of benefit increases before 1972. It illustrates
that between 1940 and 1971, average benefits for all current beneficiaries
tripled while prices nearly doubled and wages more than quintupled.76 Some
benefit increases were faster and some were slower than wages increases.

75Until the 1970s, trust fund projections were routinely exceeded at least
in part as a result of actuarial methods that assumed no growth in average
earnings.

76These estimates of average benefit increases include both existing and
initial benefits.

Table 7: Percentage Increases in OASI Benefits, Prices, and Wages, by
Effective Date of OASI Change, 1950-1971

                                          Increase in                
                  Increase in OASI      consumer price    Increase in average
                      benefit                index               wages
                                Since      Since   Since       Since    Since 
Date of        Since prior January      prior January       prior  January 
changea         amendmentb    1940  amendment    1940   amendment     1940 
September                                                                  
1950                 81.3c    81.3      75.5c    75.5      148.8c    148.8
September                                                                  
1952                  14.1   106.9        9.3    91.8        12.5    179.9
September                                                                  
1954                  13.3   134.3        0.5    92.8         7.7    201.5
January                                                                    
1959 (1958)            7.7   152.4        7.9   108.0        19.4    259.9
January                                                                    
1965                   7.7   171.9        7.9   124.5        22.3    340.2
February                                                                   
1968 (1967)           14.2   210.5        9.3   145.4        18.0    419.4
January                                                                    
1970 (1969)           15.6   258.9       10.8   171.8        12.2    482.8
January                                                                    
1971                  10.4   296.2        5.2   185.9         5.3    513.7

Source: Martha Derthick, Policymaking for Social Security, The Brookings
Institution, Washington, D.C., 1979, p. 276. Reprinted with permission of
the Brookings Institution Press, and GAO analysis.

aYear of enactment, if different from year in which change took effect, is
in parentheses.

bAverage increases for current beneficiaries, that is, people who were on
the rolls. At the same time, increases approximately equal to these were
promised by statutory formula, to active workers.

cPercentage increase since January 1940, when OASI benefits were first
paid.

On the revenue side, payroll tax rates have never been indexed. However,
Social Security's revenue also depends on the maximum amount of workers'
earnings that are subject to the payroll tax. This cap is technically
known as the contribution and benefit base because it limits the earnings
level used to compute benefits as well as taxes.77 Just as with benefits,
the maximum taxable earnings level did not change until the 1950
amendments even as price and earning levels were increasing. From 1940 to
1950, the inflation-adjusted value of the cap fell by over 40 percent.
Also, until the 1972 amendments, adjustments to the maximum taxable
earnings level were made on an ad hoc basis. With the enactment of the
1972 amendments, the maximum taxable earnings level increased
automatically based on increases in average earnings. Figure 9 shows the
inflation-adjusted values for the maximum taxable earnings level before
automatic adjustments took effect in 1975.78 Figure 10 shows that as a
result of the fluctuations in the maximum taxable earnings level, the
proportion of earnings subject to the payroll tax varied widely before
indexing, ranging from 71 to 93 percent.

77The contribution and benefit base reflects the program's role of only
providing for a floor of protection.

78In 2006, the maximum taxable earnings cap is set at $94,200.

Figure 9: Inflation-Adjusted Values of the Maximum Taxable Earnings Level
before Automatic Adjustments, 1937-1975

Note: The maximum taxable earnings level is the level at which earnings
are subject to the payroll tax.

Figure 10: Percentage of Total Covered Earnings below Social Security's
Maximum Taxable Earnings Level, 1937-2005

     Indexing in 1972 Amendments Built on Previous Ad Hoc Benefit Increases

The 1972 amendments, in effect, provided for indexing initial benefits
twice for new beneficiaries. The indexing changed the benefit formula in
the same way that previous ad hoc increases had done.

Approach Used for Ad Hoc Benefit Increases

Before the 1972 amendments, benefits were computed essentially by applying
different replacement factors to different portions of a worker's
earnings. For example, under the 1958 amendments, a workers' PIA79 would
equal

58.85 percent of first $110 of average monthly wages plus 21.40 percent of
next $290,

where the 58.85 and 21.40 percents are the replacement factors that
determine how much of a worker's earnings will be replaced by the Social
Security benefit.80 Subsequent amendments increased benefits by
effectively increasing the replacement factors. For example, the 1965
amendments increased benefits by 7 percent for a given average monthly
wage by increasing the replacement factors by 7 percent to 62.97 from
58.85 and to 22.9 percent from 21.4.81 The automatic adjustments under the
1972 amendments increased these same replacement factors according to
changes in the CPI. These changes in the benefit computation applied
equally to both new and existing beneficiaries.82

To illustrate how the benefit formula worked, take, for example, a worker
with an average monthly wage of $200 who became entitled in 1959 (when the
1958 amendments first took effect). The PIA for this worker would be

58.85 percent of $110 plus 21.4 percent of the average monthly wage over
$110, that is, $200-110 = $90, which equals $64.74 + $19.26 = 84.00.

79 The PIA is the monthly amount payable to a retired worker who begins to
receive benefits at normal retirement age or (generally) to a disabled
worker. This is also the amount used as a base for computing all types of
benefits payable on the basis of one individual's earnings record.

80 The declining replacement factors for higher levels of earnings made
the formula progressive.

81 When the maximum taxable earnings level increased, a new replacement
factor would be applied to the newly covered portion of earnings. For
example, the 1965 amendments increased the maximum taxable earnings from
$4,800 to $6,600. Accordingly, the benefit formula added a new component,
with a replacement factor of 21.4 percent for the next $150 of average
monthly wages.

82 The fact that benefits were changed for both new and current
beneficiaries using the same computations came to be known as "coupling"
of benefit increases.

When the 1965 amendments took effect, this same beneficiary would have the
PIA recalculated using the new formula. Assuming no new wages, the average
monthly wage would still be $200, and the new PIA would be

62.97 percent of $110 plus 22.9 percent of the average monthly wage over
$110, that is, $200-110 = $90, which equals $69.27 + $20.61 = 89.88, which
is 7 percent greater than the previous $84.00.

Now consider the example of a new beneficiary, who became entitled in 1965
(when the 1965 amendments first became effective). For the purposes of
this illustration, to reflect wage growth, assume this worker had an
average monthly wage of $240.00, or 20 percent more than our previous
worker who became entitled in 1959. For this new beneficiary, the PIA in
1965 would be $99.04, which, as a result of the wage growth, is much more
than 7 percent higher than the initial benefit for the worker in 1959.

1972 Amendments Introduced Indexing

The 1972 amendments provided for automatic indexing of benefits and taxes
for the first time. The indexing approach for benefits was flawed and
raised issues that the 1977 amendments addressed; these issues help
explain the basic framework for indexing benefits still in use today. In
particular, the indexing approach in the 1972 amendments resulted in (1)
double-indexing benefits to inflation for new beneficiaries though not for
existing ones and (2) a form of bracket creep that slowed benefit growth
as earnings increased over time. Within a few years, the problems raised
by the double indexing under the 1972 amendments became apparent, with
benefits growing far faster than anticipated.

Under the 1972 amendments, indexing the replacement factors in the benefit
formula to inflation had the effect of indexing twice for new
beneficiaries. First, the increase in the replacement factors themselves
reflected changes in the price level. Second, the benefit calculations
were based on earnings levels, which were higher for each new group of
beneficiaries, partially as a result of inflation.83 Thus, benefit levels
grew for each new year's group of beneficiaries because both the benefit
formula reflected inflation and their higher average wages reflected
inflation. For existing beneficiaries who had stopped working, the average
earnings used to compute their benefits did not change, so growth in
earnings levels did not affect their benefits and double indexing did not
occur. Once the double indexing for new beneficiaries was understood, the
need became clear to index benefits differently for new and existing
beneficiaries, which was referred to as "decoupling" benefits.

83Part of the growth in wages reflects inflation. Wage growth makes the
average monthly earnings for a new year's group of beneficiaries higher on
average than for the previous year's group.

The effect of double indexing on replacement rates could be offset by a
type of "bracket creep" in the benefit formula, depending on the relative
values of wage and price growth over time. Bracket creep resulted from the
progressive benefit formula, which provided lower replacement rates for
higher earners than for lower earners. As each year passed and average
earnings of new beneficiaries grew, more and more earnings would be
replaced at the lower rate used for the upper bracket, making replacement
rates fall on average, all else being equal.

      Indexing Approach Introduced Potential Instability in Benefit Costs

The combination of double indexing and bracket creep implied in the 1972
amendments introduced a potential instability in Social Security benefit
costs. Price growth determined the effects of double indexing, and wage
growth determined the effects of bracket creep. The extent to which
bracket creep offset the effects of double indexing depended on the
relative values of price growth and wage growth, which could vary
considerably. Had wage and price growth followed the historical pattern at
the time, benefits would not have grown faster than expected and
replacement rates would not have risen; the inflation effect and the
bracket creep effect would have balanced out. However, during the 1970s,
actual rates of inflation and earnings growth diverged markedly from past
experience (see fig. 11), with the result that benefit costs grew far
faster than revenues.

Figure 11: Changes in Average Wage Index and Consumer Price Index,
1951-1985

In contrast, an indexing approach that stabilized replacement rates would
help to stabilize program costs. To illustrate this, annual benefit costs
can be expressed as a fraction of the total taxable payroll in a given
year, that is, total covered earnings.84 In turn, this can be shown to
relate closely to replacement rates.

84 In a pay-as-you-go system, the payroll tax would equal annual benefit
costs as a percentage of payroll.

While not precisely a replacement rate, the second term on the last line
above-the ratio of the average benefit to average taxable earnings-is
closely related to the replacement rates provided under the program. While
replacement rates are now relatively stable after the 1977 amendments, it
is the first term on the last line above-the ratio of beneficiaries to
workers-that has been increasing and placing strains on the system's
finances. The inverse of this is the ratio of covered workers to
beneficiaries. While 3.3 workers support each Social Security beneficiary
today, only 2 workers are expected to be supporting each beneficiary by
2040. (See fig. 12.)

Figure 12: Social Security Workers per Beneficiary

Note: This is based on the intermediate assumptions of the 2006 Social
Security trustees' report.

Related GAO Products

Social Security Reform: Answers to Key Questions. GAO-05-193SP .
Washington, D.C.: May 2005.

Options for Social Security Reform. GAO-05-649R . Washington, D.C.: May 6,
2005.

Social Security Reform: Early Action Would Be Prudent. GAO-05-397T .
Washington, D.C.: Mar. 9, 2005.

Social Security: Distribution of Benefits and Taxes Relative to Earnings
Level. GAO-04-747. Washington, D.C.: June 15, 2004.

Social Security Reform: Analysis of a Trust Fund Exhaustion Scenario.
GAO-03-907 . Washington, D.C.: July 29, 2003.

Social Security Reform: Analysis of Reform Models Developed by the
President's Commission to Strengthen Social Security. GAO-03-310 .
Washington, D.C.: Jan. 15, 2003.

Social Security: Program's Role in Helping Ensure Income Adequacy.
GAO-02-62. Washington, D.C.: Nov. 30, 2001.

Social Security Reform: Potential Effects on SSA's Disability Programs and
Beneficiaries. GAO-01-35. Washington, D.C.: Jan. 24, 2001.

Social Security: Evaluating Reform Proposals. GAO/AIMD/HEHS-00-29.
Washington, D.C.: Nov. 4, 1999.

Social Security: Issues in Comparing Rates of Return with Market
Investments. GAO/HEHS-99-110. Washington, D.C.: Aug. 5, 1999.

Social Security: Criteria for Evaluating Social Security Reform Proposals.
GAO/T-HEHS-99-94. Washington, D.C.: Mar. 25, 1999.

Social Security: Different Approaches for Addressing Program Solvency.
GAO/HEHS-98-33. Washington, D.C.: July 22, 1998.

Social Security: Restoring Long-Term Solvency Will Require Difficult
Choices. GAO/T-HEHS-98-95. Washington, D.C.: Feb. 10, 1998.

(130509)

GAO's Mission

The Government Accountability Office, the audit, evaluation and
investigative arm of Congress, exists to support Congress in meeting its
constitutional responsibilities and to help improve the performance and
accountability of the federal government for the American people. GAO
examines the use of public funds; evaluates federal programs and policies;
and provides analyses, recommendations, and other assistance to help
Congress make informed oversight, policy, and funding decisions. GAO's
commitment to good government is reflected in its core values of
accountability, integrity, and reliability.

Obtaining Copies of GAO Reports and Testimony

The fastest and easiest way to obtain copies of GAO documents at no cost
is through GAO's Web site ( www.gao.gov ). Each weekday, GAO posts newly
released reports, testimony, and correspondence on its Web site. To have
GAO e-mail you a list of newly posted products every afternoon, go to
www.gao.gov and select "Subscribe to Updates."

Order by Mail or Phone

The first copy of each printed report is free. Additional copies are $2
each. A check or money order should be made out to the Superintendent of
Documents. GAO also accepts VISA and Mastercard. Orders for 100 or more
copies mailed to a single address are discounted 25 percent. Orders should
be sent to:

U.S. Government Accountability Office 441 G Street NW, Room LM Washington,
D.C. 20548

To order by Phone: Voice: (202) 512-6000 TDD: (202) 512-2537 Fax: (202)
512-6061

To Report Fraud, Waste, and Abuse in Federal Programs

Contact:

Web site: www.gao.gov/fraudnet/fraudnet.htm E-mail: [email protected]
Automated answering system: (800) 424-5454 or (202) 512-7470

Congressional Relations

Gloria Jarmon, Managing Director, [email protected] (202) 512-4400 U.S.
Government Accountability Office, 441 G Street NW, Room 7125 Washington,
D.C. 20548

Public Affairs

Paul Anderson, Managing Director, [email protected] (202) 512-4800 U.S.
Government Accountability Office, 441 G Street NW, Room 7149 Washington,
D.C. 20548

www.gao.gov/cgi-bin/getrpt? GAO-06-804 .

To view the full product, including the scope

and methodology, click on the link above.

For more information, contact Barbara Bovbjerg at (202) 512-7215 or
[email protected].

Highlights of GAO-06-804 , a report to congressional addressees

September 2006

SOCIAL SECURITY REFORM

Implications of Different Indexing Choices

The financing shortfall currently facing the Social Security program is
significant. Without remedial action, program trust funds will be
exhausted in 2040. Many recent reform proposals have included
modifications of the indexing currently used in the Social Security
program. Indexing is a way to link the growth of benefits and/or revenues
to changes in an economic or demographic variable.

Given the recent attention focused on indexing, this report examines (1)
the current use of indexing in the Social Security program and how reform
proposals might modify that use, (2) the experiences of other developed
nations that have modified indexing, (3) the effects of modifying the
indexing on the distribution of benefits, and (4) the key considerations
associated with modifying the indexing. To illustrate the effects of
different forms of indexing on the distribution of benefits, we calculated
benefit levels for a sample of workers born in 1985, using a
microsimulation model. We have prepared this report under the Comptroller
General's statutory authority to conduct evaluations on his own initiative
as part of a continued effort to assist Congress in addressing the
challenges facing Social Security. We provided a draft of this report to
SSA and the Department of the Treasury. SSA provided technical comments,
which we have incorporated as appropriate.

Indexing currently plays a key role in determining Social Security's
benefits and revenues, and is a central element of many proposals to
reform the program. The current indexing provisions that affect most
workers and beneficiaries relate to (1) benefit calculations for new
beneficiaries, (2) the annual cost-of-living adjustment (COLA) for
existing beneficiaries, and (3) the cap on taxable earnings. Some reform
proposals would slow benefit growth by indexing the initial benefit
formula to changes in prices or life expectancy rather than wages. Some
would revise the COLA under the premise that it currently overstates
inflation, and some would increase the cap on taxable earnings.

National pension reforms in other countries have used indexing in various
ways. In countries with high contribution rates that need to address
solvency issues, recent changes have generally focused on reducing
benefits. Although most Organisation for Economic Co-operation and
Development (OECD) countries compute retirement benefits using wage
indexing, some have moved to price indexing, or a mix of both. Some
countries reflect improvements in life expectancy in computing initial
benefits. Reforms in other countries that include indexing changes
sometimes affect both current and future retirees.

Indexing can have various distributional effects on benefits and revenues.
Changing the indexing of initial benefits through the benefit formula
typically results in the same percentage change in benefits across income
levels regardless of the index used. However, indexing can also be
designed to maintain benefits for lower earners while reducing or slowing
the growth of benefits for higher earners. Indexing payroll tax rates
would maintain scheduled benefit levels but reduce the ratio of benefits
to contributions for younger cohorts. Finally, the effect of modifying the
COLA would be greater the longer people collect benefits.

Indexing raises considerations about the program's role, the treatment of
disabled workers, and other issues. For example, indexing initial benefits
to prices instead of wages implies that benefit levels should maintain
purchasing power rather than maintain relative standards of living across
age groups (i.e., replacement rates). Also, as with other ways to change
benefits, changing the indexing of the benefit formula to improve solvency
could also result in benefit reductions for disabled workers as well as
retirees.
*** End of document. ***