-------------------------Indexing Terms------------------------- 
REPORTNUM:   GAO-03-993		

TITLE:     Social Security: Proposed Totalization Agreement with 
Mexico Presents Unique Challenges

DATE:   09/30/2003 
				                                                                         
----------------------------------------------------------------- 

llllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllntries, and eliminate dual
social security taxes that employers and their employees pay when they
operate and reside in countries with parallel social security systems.
Because Mexicans are believed to represent a large share of the millions
of unauthorized workers present in the United States, a totalization
agreement with Mexico has raised concerns that they would become newly
eligible for social security benefits. To shed light on the possible
impacts, you asked GAO to (1) describe the Social Security
Administration's (SSA) processes for developing the agreement with Mexico,
(2) explain how the agreement might affect the payment of benefits to
Mexican citizens, and (3) assess the cost estimate for such an agreement.

GAO recommends that SSA (1) establish a formal process to identify and
assess risks of proposed agreements, (2) make future reports to the
Congress on these agreements more consistent and informative, and (3) work
with the Office of the Chief Actuary to improve the cost estimates for
agreements. SSA disagreed that additional processes were needed to assess
risks, but it agreed that cost estimates should be more consistent and
that it should regularly re-examine the accuracy of its estimates.

www.gao.gov/cgi-bin/getrpt?GAO-03-993.

To view the full product, including the scope and methodology, click on
the link above. For more information, contact Barbara D. Bovbjerg, (202)
512-7215, [email protected].

September 2003

SOCIAL SECURITY

Proposed Totalization Agreement with Mexico Presents Unique Challenges

SSA has no written policies or procedures it follows when entering into
totalization agreements, and the actions it took to assess the integrity
and compatibility of Mexico's social security system were limited and
neither transparent nor well-documented. SSA followed the same procedures
for the proposed Mexican agreement that it used in all prior agreements.
SSA officials told GAO that they briefly toured Mexican facilities,
observed how its automated systems functioned, and identified the type of
data maintained on Mexican workers. However, SSA provided no information
showing that it assessed the reliability of Mexican earnings data and the
internal controls used to ensure the integrity of information that SSA
will rely on to pay social security benefits.

The proposed agreement will likely increase the number of unauthorized
Mexican workers and family members eligible for social security benefits.
Mexican workers who ordinarily could not receive social security
retirement benefits because they lack the required 40 coverage credits for
U.S. earnings could qualify for partial social security benefits with as
few as 6 coverage credits. In addition, under the proposed agreement, more
family members of covered Mexican workers would become newly entitled
because the agreements usually waive rules that prevent payments to
noncitizens' dependents and survivors living outside the United States.

The cost of such an agreement is highly uncertain. In March 2003, the
Office of the Chief Actuary estimated that the cost of the Mexican
agreement would be $78 million in the first year and would grow to $650
million (in constant 2002 dollars) in 2050. The actuarial cost estimate
assumes the initial number of newly eligible Mexican beneficiaries is
equivalent to the 50,000 beneficiaries living in Mexico today and would
grow sixfold over time. However, this proxy figure does not directly
consider the estimated millions of current and former unauthorized workers
and family members from Mexico and appears small in comparison with those
estimates. The estimate also inherently assumes that the behavior of
Mexican citizens would not change and does not recognize that an agreement
would create an additional incentive for unauthorized workers to enter the
United States to work and maintain documentation to claim their earnings
under a false identity. Although the actuarial estimate indicates that the
agreement would not generate a measurable long-term impact on the
actuarial balance of the trust funds, a subsequent sensitivity analysis
performed at GAO's request shows that a measurable impact would occur with
an increase of more than 25 percent in the estimate of initial, new
beneficiaries. For prior agreements, error rates associated with
estimating the expected number of new beneficiaries have frequently
exceeded 25 percent, even in cases where uncertainties about the number of
unauthorized workers were less prevalent. Because of the significant
number of unauthorized Mexican workers in the United States, the estimated
cost of the proposed totalization agreement is even more uncertain than in
prior agreements.

Contents

  Letter

Results in Brief
Background
SSA's Process for Developing Agreements Is Not Thorough or Well-

Documented Totalization Agreements Will Increase Benefit Payments to

Mexican Citizens Poor Data Undermine the Reliability of SSA's Cost
Estimate Conclusions Recommendations Agency Comments and Our Evaluation 1

2 3

6

7

9 14 15 16

Appendix I 	Comparison of Totalized and Minimum Social Security Benefits

Appendix II Comments from the Social Security Administration

  Tables

Table 1: Existing Totalization Agreements between the United

States and Other Countries and Year of Effective Date of

the Original Agreements 5 Table 2: Precision of OCACT's Cost Estimates for
11 Prior Totalization Agreements 14 Table 3: Monthly Social Security
Benefits Payable in 2003 at

Different Earnings and Coverage Levels under a

Totalization Agreement and Compared to the Minimum

Benefit Payable 20

Abbreviations

INS Immigration and Naturalization Service
NAFTA North American Free Trade Agreement
OCACT Office of the Chief Actuary
SSA Social Security Administration

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 General Accounting Office Washington, DC 20548

September 30, 2003

The Honorable F. James Sensenbrenner, Jr.
Chairman
Committee on the Judiciary
House of Representatives

The Honorable E. Clay Shaw, Jr.
Chairman
Subcommittee on Social Security
Committee on Ways and Means
House of Representatives

Totalization agreements foster international commerce and protect
benefits for persons who have worked in foreign countries in two ways.
First, the agreements eliminate dual social security taxes that
multinational employers and their employees must pay when they operate
and reside in countries with parallel social security programs. Second,
the
agreements help to fill gaps in benefit protection for persons who have
worked in different countries for portions of their careers. Since 1977,
the
United States has entered into 20 totalization agreements.

Over the last year, the United States has been negotiating a totalization
agreement with Mexico that has received considerable attention among
the media and others regarding its potential impacts. Because Mexicans
represent a large share of the millions of unauthorized workers present in
the United States, a totalization agreement with Mexico has raised
concerns that many such workers would become newly eligible for social
security benefits at a time when long-term trust fund solvency is
threatened. To shed light on the possible impacts of such an agreement,
you asked us to (1) describe the Social Security Administration's (SSA)
processes for developing the proposed agreement with Mexico, (2) explain
how the agreement might affect the payment of social security benefits to
Mexican citizens, and (3) assess SSA's cost estimates for such an
agreement.

To address these objectives, we reviewed existing totalization agreements
and the laws governing them; interviewed and obtained key
documentation from SSA, Department of State, and Mexican Embassy
personnel; and reviewed a range of demographic data and estimates
addressing Mexican immigration. We also examined SSA's actuarial cost

  Results in Brief

estimates and supporting documentation for the proposed Mexican agreement.
We conducted our work between January and August 2003, in accordance with
generally accepted government auditing standards.

SSA has no written policies or procedures outlining the specific steps it
follows when entering into totalization agreements, and the actions it
took to assess the integrity and compatibility of Mexico's social security
system were limited and neither transparent nor well-documented. SSA said
the process it used to develop the proposed totalization agreement with
Mexico was the same as for prior totalization agreements. SSA officials
told us that they briefly toured Mexican facilities, observed how their
automated systems functioned, and identified the type of data maintained
on Mexican workers. However, SSA provided no information showing that it
assessed the reliability of Mexican earnings data and the internal
controls Mexico uses to ensure the integrity of information that SSA will
rely on to pay social security benefits.

The proposed agreement will increase the number of Mexican workers and
family members eligible for social security benefits. Mexican workers who
ordinarily could not receive benefits because they lack the required 40
coverage credits for U.S. earnings could qualify for partial Social
Security benefits with as few as 6 coverage credits. In addition, under
the proposed agreement, more family members of covered Mexican workers
would also become newly entitled because of the waiver of rules that
prevent payment to noncitizens' dependents and survivors living outside
the United States.

The cost of a totalization agreement with Mexico is highly uncertain. In
March 2003, the Office of the Chief Actuary (OCACT) estimated that the
cost of the Mexican agreement would be $78 million in the first year of
the agreement and would grow to $650 million (in constant 2002 dollars) in
2050. SSA's actuarial cost estimate assumes the initial number of newly
eligible Mexican beneficiaries is equivalent to the 50,000 beneficiaries
living in Mexico today and would grow sixfold over time. However, this
proxy figure does not directly consider the estimated millions of current
and former unauthorized workers and family members from Mexico and appears
small in comparison with those estimates. Although the actuarial estimate
indicates that the agreement would not generate a measurable impact on the
long-range actuarial balance of the trust funds, an increase of more than
25 percent in the estimate of initial, new beneficiaries would generate a
measurable impact. For prior agreements, error rates associated with
estimating the expected number of new beneficiaries have

frequently exceeded 25 percent, even in cases where uncertainties about
the number of unauthorized workers were less prevalent. Because of the
significant number of unauthorized Mexican workers in the United States,
the estimated cost of the proposed totalization agreement is even more
uncertain than for the prior agreements.

This report recommends that SSA establish formal processes for entering
into totalization agreements that include mechanisms to assess the risks
associated with such agreements and to document the range of analyses SSA
conducts. The report also recommends that reports of proposed agreements
be enhanced to make them more consistent and informative and that SSA
establish a regular process to reassess the accuracy of its actuarial
estimates. SSA and the OCACT commented on this report. SSA said that the
report did not sufficiently discuss the benefits of totalization
agreements and that its current process for evaluating whether to enter
into negotiations for totalization agreements was sufficient to identify
and assess risks. Our report specifically notes that such agreements
foster international commerce, protect benefits for persons who have
worked in foreign countries, eliminate dual social security taxes, and
foster enhanced diplomatic relations. With regard to SSA's current
processes, we could find no specific references to SSA examining data
reliability and program integrity. We are hopeful that SSA will conduct
such examinations of the Mexican Social Security system before submitting
a proposed agreement to the Congress for its review. OCACT generally
agreed with our recommendations and noted that they are consistent with
current practices. OCACT, however, took exception to the implication of
our statement that its estimated cost was more likely to be understated
than overstated. Our intent was not to imply that the OCACT estimate was
biased. Accordingly, we have revised our report to state the very large
difference between estimated and potential beneficiaries underscores the
uncertainty of the estimate, and the potential costs of an agreement could
be higher than OCACT projects. The full text of SSA's and OCACT's comments
appears in appendix II. The State Department was also provided a copy of
the draft report for review and advised us that it had no comments.

Background 	SSA administers the Old Age, Survivors, and Disability
Insurance programs under Title II of the Social Security Act. About 96
percent of the nation's work force is in social security-covered
employment and pays tax on its

annual earnings. When workers pay social security taxes, they earn
coverage credits, and 40 credits-equal to at least 10 years of work-
entitle them to social security benefits when they reach retirement age.1

In 1977, the Congress authorized the President to enter into totalization
agreements with other countries. These bilateral agreements are intended
to accomplish three purposes. First, they eliminate dual social security
coverage and taxes that multinational employers and employees encounter
when they operate and their workers temporarily reside and work for the
corporation, usually no more than 5 years, in a foreign country with its
own social security program. Under the agreements, U.S. employers and
their workers sent temporarily abroad would benefit by paying only U.S.
social security taxes, and foreign businesses and their workers would
benefit by paying only social security taxes to their home country.
Second, the agreements provide benefit protection to workers who have
divided their careers between the United States and a foreign country, but
lack enough coverage under either social security system to qualify for
benefits, despite paying taxes into both systems. Totalization agreements
allow such workers to combine (totalize) work credits earned in both
countries to meet minimum benefit qualification requirements. Third, most
totalization agreements improve the portability of social security
benefits by removing rules that suspend benefits to noncitizens who live
outside the benefit-paying country.

By law, proposed agreements are sent to the Congress, which has 60
legislative days to review them. The agreements become effective unless
either House of the Congress adopts a resolution of disapproval. Table 1
shows agreements in effect and the years they became effective.

1Different requirements govern the number of coverage credits necessary to
receive disability and survivors benefits for workers who become disabled
or die with relatively short work careers.

Table 1: Existing Totalization Agreements between the United States and
Other Countries and Year of Effective Date of the Original Agreements

                                 Countries Year

                                   Italy 1978

                                  Germany 1979

                                Switzerland 1980

                                  Belgium 1984

                                  Norway 1984

                                  Canada 1984

                              United Kingdom 1985

                                  Sweden 1987

                                   Spain 1988

                                  France 1988

                                 Portugal 1989

                                Netherlands 1990

                                  Austria 1991

                                  Finland 1992

                                  Ireland 1993

                                Luxembourg 1993

                                  Greece 1994

South Korea 2001

Chile 2001

Australia 2002 Source: SSA.

To qualify for totalized U.S. social security benefits, a worker must have
at least 6 but no more than 39 U.S. coverage credits. Benefit amounts are
based on the portion of time a foreign citizen worked in the United
States, and thus, are almost always lower than full social security
benefits. The average monthly, totalized social security benefit at the
end of 2001 was $162, compared with the average nontotalized monthly
social security benefit of $825. In 2001, SSA paid about $173 million
under totalization agreements to about 89,000 persons, including their
dependents. (Appendix I compares the amount of U.S. totalized benefits for
different coverage credits and earnings levels with a minimum benefit that
would be paid to a worker with 40 credits.)

Under U.S. law, immigrants may not work in the United States unless
specifically authorized. Nevertheless, immigrants often do work without
authorization and pay social security taxes. Under the Social Security
Act,

  SSA's Process for Developing Agreements Is Not Thorough or Well-Documented

all earnings from covered employment in the United States count towards
earning social security benefits, regardless of the lawful presence of the
worker, his or her citizenship status, or country of residence. Immigrants
become entitled to benefits from unauthorized work if they can prove that
the earnings and related contributions belong to them. However, they
cannot collect such benefits unless they are either legally present in the
United States or living in a country where SSA is authorized to pay them
their benefits. Mexico is such a country.

A lack of transparency in SSA's processes, and the limited nature of its
review of Mexico's program, cause us to question the extent to which SSA
will be positioned to respond to potential program risks should a
totalization agreement with Mexico take place. SSA officials told us that
the process used to develop the proposed totalization agreement with
Mexico was the same as for prior agreements with other countries. The
process-which is not specified by law or outlined in written policies and
procedures-is informal, and the steps SSA takes when entering into
agreements are neither transparent nor well-documented.

Current law does not prescribe how SSA should select potential agreement
countries. According to SSA, interest in a Mexican agreement dates back
more than 20 years. SSA officials noted that increased business
interaction between the two countries due to the North American Free Trade
Agreement (NAFTA) was a factor in the renewed negotiations. In addition,
because there is a totalization agreement with Canada, our other NAFTA
partner, SSA believed that equity concerns required consideration of an
agreement with Mexico. In February 2002, SSA sought clearance from the
Department of State to begin such negotiations.

The law also does not specify which elements of other countries' social
security systems must be evaluated during totalization agreement
negotiations. SSA officials met with Mexican officials to exchange
narrative information on their respective programs. Senior SSA officials
also visited Mexico for 2 days in August 2002. During their visit, these
officials told us that they toured social security facilities, observed
how Mexico's automated social security systems functioned, and identified
the type of data maintained on Mexican workers. SSA took no technical
staff on this visit to assess system controls or data integrity processes.
In effect, SSA only briefly observed the operations of the Mexican social
security program. Moreover, SSA did not document its efforts or perform
any additional analyses then, or at a later time, to assess the integrity
of Mexico's social security data and the controls over that data. In
particular,

SSA officials provided no evidence that they examined key elements of
Mexico' s program, such as its controls over the posting of earnings and
its processes for obtaining key birth and death information for Mexican
citizens. Nor did SSA evaluate how access to Mexican data and records is
controlled and monitored to prevent unauthorized use or whether internal
and external audit functions exist to evaluate operations.

Because all totalization agreements represent a financial commitment with
implications for social security tax revenues and benefit outlays, a
reasonable level of due diligence and analysis is necessary to help
federal managers identify issues that could affect benefit payment
accuracy or expose the nation's system to undue risk. Our Internal Control
Management and Evaluation Tool provides a risk assessment framework to
help federal managers mitigate fraud, waste, abuse, and mismanagement in
public programs, such as social security. A key component of this
framework is the identification of internal and external risks that could
impede the achievement of objectives at both the entity and program
levels. Identified risks should then be analyzed for their potential
effect and an approach devised to mitigate them.

SSA did not conduct these types of analyses in previous agreements or in
the case of the proposed Mexican agreement, despite documented concerns
among Mexican government officials and others regarding the integrity of
Mexico's records, such as those for birth, death, and marriage, as well as
its controls over assigning unique identification numbers to workers for
benefit purposes. Such information will likely play a role in SSA's
ability to accurately determine Mexican workers' initial and continuing
eligibility for benefits under a totalization agreement.

A totalization agreement with Mexico will increase the number of Mexican
citizens who will be paid U.S. social security benefits in two ways.
First, the agreement will make it easier for Mexican workers to qualify
for benefits. Second, it will remove some nonpayment restrictions that
affect benefit payments to non-U.S. citizens' family members residing in
another country, thus providing U.S. social security benefits to more
survivors and dependents of entitled Mexican workers.

Under current law, a worker must earn sufficient coverage credits to
qualify for benefits under the U.S. Social Security program. For example,
a worker who was born in 1929 or later generally needs 40 coverage credits
to be insured for retirement benefits. Credits are based on a worker's
annual earnings in social security-covered employment. At most, 4 credits
can be earned per year so that it takes at least 10 years of covered
earnings

  Totalization Agreements Will Increase Benefit Payments to Mexican Citizens

in the United States for a worker to accumulate the necessary 40 credits
and become insured for retirement benefits.

Currently, social security credits are earned by anyone who has worked in
covered employment in the United States. This is true even if the person
was unauthorized to work when he or she earned coverage credits. For
example, noncitizens, including Mexicans, who are at least 62 years old
and lawfully present in the United States, will receive retirement
benefits today as long as they meet the coverage credit threshold. Even
Mexican citizens who are not lawfully present in this country can receive
social security benefits earned through unauthorized employment if they
later return to live in Mexico. Similarly, under current law, noncitizen
dependents and survivors can also receive social security benefits under
some circumstances.

Totalization agreements generally expand benefits to both authorized and
unauthorized workers and create new groups of beneficiaries. This would be
the case for a totalization agreement with Mexico if it follows the same
pattern as all prior totalization agreements. Mexican citizens with fewer
than 40 coverage credits will be permitted to combine their annual
earnings under their home country's social security program with their
annual earnings under the U.S. Social Security program to meet the
40-credit requirement.2 In addition, more family members of covered
workers will qualify for dependent and survivor benefits. Totalization
agreements generally override Social Security Act provisions that prohibit
benefit payments to noncitizens' dependents and survivors who reside
outside the United States for more than 6 months, unless they can prove
that they lived in the United States for 5 years in a close family
relationship with the covered worker. If a totalization agreement with
Mexico is structured like others already in force, the 5-year rule for
dependents and survivors will be waived.

However, it is important to understand that not all unauthorized Mexican
citizens who have worked in the United States will receive totalization
benefits. Some will have earned at least 40 coverage credits and can
receive social security benefits without a totalization agreement. Still
others may have worked under false identities and may not be able to

2Under an agreement, U.S. citizens will also be able to receive totalized
Mexican benefits. The amount of time needed to qualify for Mexican social
security benefits is about 9.6 years under the former pay-as-you-go plan
that closed in July 1997 and 24 years under the defined contribution plan
that replaced it.

  Poor Data Undermine the Reliability of SSA's Cost Estimate

prove that they have the necessary coverage credits to be entitled to
benefits. Others still may not accumulate sufficient credits under the
Mexican social security system to totalize with their U.S. social security
coverage.

The cost of a totalization agreement with Mexico is highly uncertain. In
March 2003, the Office of the Chief Actuary estimated that the cost of the
Mexican agreement would be $78 million in the first year and would grow
$650 million (in constant 2002 dollars) in 2050. SSA's actuarial cost
estimate assumes the initial number of newly eligible Mexican
beneficiaries was equivalent to the 50,000 beneficiaries living in Mexico
today and would grow sixfold over time. However, this proxy figure is not
directly related to the estimated millions of current and former
unauthorized workers and their family members from Mexico and appears
small in comparison to those estimates. Furthermore, even if the baseline
estimate is used, a sensitivity analysis performed by OCACT shows that an
increase of more than 25 percent-or 13,000 new beneficiaries-would produce
a measurable impact on the long-range actuarial balance of the trust
funds. Our review of cost estimates for prior totalization agreements
shows that the actual number of beneficiaries has frequently been
underestimated and far exceeded the original actuarial estimates.

    Actuarial Estimates Are Based on Varied Data Sources

OCACT develops estimates of expected costs of totalization agreements by
analyzing pertinent data from prior agreements, work visas issued, foreign
corporations operating in the United States, and U.S. Census data. Because
of extensive unauthorized immigration from Mexico, OCACT concluded that
U.S. Census data, that would typically be used to estimate the number of
new beneficiaries under an agreement, were not reliable.

Instead, OCACT used the number of fully insured beneficiaries-U.S.
citizens and others living in Mexico-currently receiving U.S. social
security benefits as a proxy for the number of Mexican citizens who would
initially receive totalized benefits. The principal basis for this
assumption was a 1997 study of Mexican immigration patterns conducted by a
private nonprofit organization.3 This study indicated that the percentage
of Mexican immigrants who returned to Mexico after more than 10 years and,

3Belinda I. Reyes, Dynamics of Immigration: Return Migration to Western
Mexico, Public Policy Institute of California, January 1997.

therefore, could qualify for benefits is roughly equal to the percentage
that returned after staying 2 to 9 years and would not have the required
credits. Thus, OCACT assumed that the potential totalized initial new
beneficiaries would be equivalent to the 50,000 persons currently
receiving benefits in Mexico.

For the proposed Mexican agreement, both a short-term (covering the first
8 years of the agreement) and a long-term (covering 75 years) cost
estimate were developed.4 The estimated cost to the Social Security Trust
Funds would be about $78 million in the first year of the agreement. For
the long-term cost estimate, OCACT projected that the number of
beneficiaries would ultimately increase sixfold to 300,000 over a 45-year
period after the agreement took effect and equal about $650 million (in
constant 2002 dollars) in 2050. However, the actuarial analysis notes that
the methodology was indirect and involved considerable uncertainty.

As a rough check on the reasonableness of using current beneficiaries in
Mexico for its cost estimate, OCACT analyzed totalized beneficiary data
for Canadian citizens because Canada, like Mexico, is a NAFTA trading
partner and shares a large contiguous border. After determining the ratio
of Canadians receiving totalized versus fully insured benefits, OCACT
applied this ratio to the number of Mexican-born U.S. social security
beneficiaries and found that about 37,000 beneficiaries would be expected
under the agreement initially, if the Canadian experience proves
predictive of the Mexican outcome. According to OCACT, this comparison
increased its confidence that the assumed 50,000 new beneficiaries under
the agreement was within a reasonable range.

Estimated Cost of Mexican Limited data about unauthorized workers make any
estimate of the Agreement Is Highly expected costs of a Mexican
totalization agreement highly uncertain. A Uncertain significant variable
of any totalization agreement cost estimate is the

identification of the number of potential beneficiaries. Estimates of the

number of unauthorized Mexican immigrants living in the United States

4For prior agreements with other countries, the OCACT developed only
short-term estimates covering periods ranging from 1 to 5 years because it
was determined that the number of expected beneficiaries were too few to
have a measurable cost impact on the long-range actuarial balance of the
trust funds.

vary5 The federal government's estimate was published in January 2003 and
comes from the former Immigration and Naturalization Service (INS)6 INS
estimated that, as of January 2000, about 5 million, or 69 percent of all
unauthorized immigrants in the United States, were from Mexico. INS's
estimate also indicated that this figure was expected to increase by about
240,000 persons annually.

The INS estimate, however, does not include unauthorized Mexican workers
and family members who no longer live in the United States and could also
conceivably benefit from a totalization agreement. Economic disparity
between the United States and Mexico has fostered longstanding immigration
from Mexico to the United States dating back many decades. Various studies
also show that fewer than a third of Mexican immigrants stay more than 10
years in the United States, the minimum amount of time needed to qualify
for social security retirement benefits.7 For cost analysis purposes,
little is known about the population of former immigrants who have
returned to Mexico in terms of their age, work history, dependents, and
social security coverage. These factors increase the inherent uncertainty
of any long-range forecasts with regard to Mexico. It is under this
backdrop that OCACT set about developing an estimate of the costs of the
potential totalization agreement.

We have several concerns about OCACT's estimate of the number of expected
beneficiaries and cost of an agreement with Mexico. First, the use of the
50,000 fully insured beneficiaries receiving benefits in Mexico as a proxy
for individuals who might initially benefit from an agreement, does not
directly consider the estimated millions of unauthorized Mexican
immigrants in the United States and Mexico who are not fully insured and
might receive totalized benefits. Furthermore, despite the availability of
key data about earnings, work histories, years of employment, and
dependents for the 50,000 fully insured beneficiaries, OCACT did not
analyze this population to determine whether they represented a good

5For example, the Pew Hispanic Center estimated that there are between 3.4
and 5.7 million unauthorized Mexican citizens in the United States and the
Urban Institute has estimated that there are more than 4 million.

6In March 2003, INS functions were transferred to the Department of
Homeland Security. Responsibility for deriving these estimates now lies
with the Under Secretary Management, Office of Immigration Statistics.

7Reyes (1997), p. 13 lists several studies that document the temporary and
circular nature of Mexican migration to the United States.

proxy for individuals likely to qualify for totalized benefits. The cost
estimate also inherently assumes that the behavior of Mexican citizens
would not change after a totalization agreement goes into effect. Under
totalization, unauthorized workers would have an additional incentive to
enter the United States to work and to maintain the appropriate
documentation necessary to claim their earnings under a false identity.
Thus, a large number of Mexican citizens have likely earned some social
security coverage credits through both authorized and unauthorized work to
meet the 40-credit threshold requirement and are not directly accounted
for in SSA's estimate.

Second, SSA's reasonableness check using Canadian data faces similar
questions. While Mexico and Canada are NAFTA partners and share a common
border with the United States, there is a dramatic difference in the
extent of unauthorized immigration from these two countries and, in our
view, the Canadian experience is not a good predictor of experience under
an agreement with Mexico. Recent INS data show that Mexican citizens
account for about 69 percent of unauthorized U.S. immigrants, whereas
Canadian citizens account for less than 1 percent, and all other
totalization agreement countries combined account for less than 3 percent.
It is this population of unauthorized immigrants that makes estimating the
cost of a totalization agreement with Mexico particularly problematic.

Finally, even though SSA's actuarial analysis increases the number of
beneficiaries sixfold over time, the expected 300,000 beneficiaries in
2050 represents only about 6 percent of the estimated number of
unauthorized Mexicans in the United States today, and thus appears
relatively low. Although it would be unreasonable to expect all
unauthorized Mexicans in the United States to qualify for totalized
benefits, the very large difference between estimated and potential
beneficiaries underscores the uncertainty of the estimate and the
potential costs of an agreement could be higher than OCACT projects.

Indeed, it would take only a relatively small increase in new
beneficiaries from the original actuarial assumption of 50,000 initial new
beneficiaries to have a measurable impact on the long-range actuarial
balance of the trust funds. OCACT has estimated that the agreement would
not generate a measurable impact on the long-range actuarial balance.
However, a subsequent sensitivity analysis performed at our request shows
that a measurable impact on the long-range actuarial balance of the trust
funds will occur if the baseline figure is underestimated by more than 25
percent-just 13,000 additional beneficiaries above the estimated 50,000
new beneficiaries.

Our analysis of past actuarial estimates of expected beneficiaries under
totalization agreements shows that exceeding the 25 percent threshold has
not been unusual, even in agreements where uncertainty about the number of
unauthorized workers is substantially less.8 Our review of prior estimates
shows that OCACT frequently either overestimated or underestimated the
number of expected beneficiaries, usually by more than 25 percent (see
table 2). In fact, where underestimates occurred, the differences were
huge, involving several orders of magnitude. However, it is important to
note that the number of estimated beneficiaries for prior agreements is
substantially smaller than for the proposed Mexican agreement. Therefore,
the differences in actual beneficiaries from estimated beneficiaries have
a higher proportional impact. Furthermore, OCACT has not underestimated
the number of expected beneficiaries for the agreements we analyzed since
the 1991 agreement with Austria. Nevertheless, the numerous uncertainties
and data gaps associated with the Mexican agreement elevate the risks
associated with any cost estimate.

8OCACT staff told us that it would be best to look at precision of past
estimates by comparing the estimated number of beneficiaries for the last
year of the estimate with actual data for that same year. We were able to
make this comparison for 11 countries.

     Table 2: Precision of OCACT's Cost Estimates for 11 Prior Totalization
                        Agreements Beneficiaries Country

Effective year of agreement Estimated Actual

                Percent actual beneficiaries is greater/(less) than estimated
                                                                beneficiaries

                United Kingdom        1985    3,500          2,084       (40) 
                        Sweden        1987     100         211     
                         Spain        1988     300         377     
                        France        1988     200         968     
                      Portugal        1989     100         701     
                   Netherlands        1990     100         310     
                       Austria        1991     100         314     
                       Finland        1992     100         38            (62) 
                    Luxembourg        1993     40          12            (70) 
                       Ireland        1993    1,100        515           (53) 
                        Greece        1994    1,000        918            (8) 

Conclusions

Source: GAO analysis.

Note: Actual data were not available for years prior to 1987 so
comparisons for six earlier agreements could not be made. Also, comparison
could not be made for the three recent agreements.

Totalization agreements between the United States and other countries
often foster enhanced diplomatic relations and provide mutually beneficial
business, tax, and other incentives to employers and employees affected by
these agreements. At the same time, the agreements impose a financial cost
to both countries' social security programs. SSA's processes for entering
into these agreements have been informal and have not included specific
steps to assess and mitigate potential risks. Regardless of the country
under consideration, sound management practices dictate that SSA managers
have a risk management process in place to ensure that the interests of
the United States and the Social Security Trust Funds are protected.

Most totalization agreements have been with countries that are
geographically distant to the United States, have developed economies, and
represent only a fraction of the estimated unauthorized immigrants in the
United States. Still, all agreements include some level of uncertainty and
require due diligence on SSA's part to alleviate those uncertainties. An
agreement with Mexico, however, presents unique and difficult challenges
for SSA because so little is known about the size, work history, earnings,

and dependents of the unauthorized Mexican population. Furthermore, a
common border and economic disparity between the United States and Mexico
have fostered significant and longstanding unauthorized immigration into
the United States, making an agreement with Mexico potentially far more
costly than any other. Thus, for the Mexican agreement, additional
analyses to assess risks and costs may be called for.

A revised approach for entering into totalization agreements with all
countries would enhance the quality of information provided to the
Congress, which is tasked with reviewing these vital long-term
commitments. A more thorough prospective analysis will also provide a
better basis for determining whether agreements under consideration meet
the mutual economic and business needs of all parties. Finally, current
solvency issues require the Congress to think carefully about future trust
fund commitments resulting from totalization agreements. Having more
timely and complete information on the benefits, costs, and risks
associated with each agreement can only serve to better inform their
decisions.

                                Recommendations

In light of the potential impact of totalization agreements on the Social
Security Trust Funds, we recommend that the Commissioner of Social
Security

o  	establish a formal process to identify and assess the major risks
associated with entering into agreements with other countries. Such a
process should include mechanisms to assess the integrity of a country's
retirement data and records, as well as a means for documenting the range
of analyses conducted by SSA;

o  	enhance future reports to the Congress for proposed totalization
agreements with other countries by making them more consistent and
informative. Such reports should include consistent time periods for
estimating both the short- and long-term effects on the trust fund and, as
appropriate, include data on how alternative assumptions or sensitivity
analyses could affect costs and potential beneficiaries; and

o  	work with the Office of the Chief Actuary to establish a regular
process that examines original projected costs and beneficiaries affected
versus what actually transpired over time and use this information, as
appropriate, to adjust future estimating methods for totalization
agreements.

  Agency Comments
  and Our Evaluation

We obtained written comments on a draft of this report from the
Commissioner of SSA, as well as OCACT. The full texts of these comments
are reproduced in appendix II. We made limited changes to the report as
appropriate. The State Department was also provided a copy of the draft
report for review and advised us that it had no comments.

SSA said that the report did not sufficiently discuss the benefits of
totalization agreements to U.S. workers and employers and disagreed with
our recommendation that the agency establish a formal process to identify
and assess the major risks associated with entering into agreements with
other countries. The agency noted that its current informal process for
evaluating whether to enter into negotiations for totalization agreements
was sufficient to identify and assess risks.

Regarding the potential benefits of totalization agreements, our report
specifically notes that such agreements foster international commerce,
protect benefits for persons who have worked in foreign countries, and
eliminate dual social security taxes for multinational employers and
employees. Our concluding remarks also note that totalization agreements
often foster enhanced diplomatic relations between participating
countries. However, these agreements also have costs to the U.S. social
security system, and we continue to believe that SSA should take steps to
assess and mitigate risk during the negotiation process rather than after
an agreement is signed.

SSA also noted that it has specific criteria it follows when deciding
whether to enter into totalization agreements with other countries and
that the agency received detailed information on Mexico's social security
system during its 2-day visit to Mexico City. In reviewing SSA's criteria,
we could find no specific reference to data reliability and program
integrity as a factor in negotiations. Further, our review of the
activities surrounding SSA's visit to Mexico and the limited documentation
SSA received from Mexican social security officials shows that data
integrity issues and systems controls were not sufficiently examined. In
its comments, SSA notes that it is currently in the process of scheduling
additional visits to Mexican facilities outside of Mexico City and will
utilize SSA technical staff to further examine Mexico's social security
system. We are hopeful that-prior to submitting a proposed agreement with
Mexico-SSA will take additional steps to assess key data it will rely on
to determine Mexican worker's initial and continuing eligibility for U.S.
totalized benefits and that it will sufficiently document its efforts.
Enhancing its due diligence efforts and formalizing this process to
include all future totalization agreements would further improve SSA's
risk assessment efforts.

OCACT generally agreed with our recommendations that cost estimates for
future totalization agreements should be more consistent and informative
and that such agreements should be regularly analyzed to examine the
differences between original projections and actual experience as an aid
to making better estimates. OCACT noted that, consistent with the
U.S./Mexican totalization agreement, all future potential agreements would
include both long-range (75 year) and short-range (10 year) cost
projections. OCACT also noted that regularly examining the differences
between original projections and actual experience for future totalization
agreements made sense and was consistent with current practice. Although
we could find no evidence during our review that such analyses had
occurred on a systematic basis, we are pleased to hear that such analyses
are now being done and are hopeful that OCACT will both complete them in
the future and document and make available the results.

Both SSA and OCACT disagreed with our analysis and conclusions regarding
the estimates of the potential cost of a totalization agreement with
Mexico, as well as our statement that any difference between estimated and
actual costs will be on the high side. OCACT noted that, given the
relative uncertainty of the data, this outcome is possible, but that our
statement inaccurately implied that there was evidence that OCACT
estimates are more likely to be understated than overstated. OCACT went on
to note that a number of factors suggest that OCACT's estimate of 50,000
new beneficiaries, which will increase sixfold to 300,000 by 2050, could
indeed be too high.

Our intent was not to imply that OCACT's estimate was biased. Thus, we
have revised our report to state that, given the large disparity between
the estimated beneficiaries and the large number of undocumented Mexican
workers, the potential cost of an agreement could be higher than OCACT
projects. However, we continue to believe that a totalization agreement
with Mexico is both qualitatively and quantitatively different than any
other agreement signed to date, especially regarding estimating the
potential impact of millions of unauthorized workers and their families.
Thus, in assessing the risks of a totalization agreement with Mexico, we
believe it is important to discuss the potentially significant impact that
any underestimate of beneficiaries could have on the Social Security Trust
Funds. As table 2 shows, error rates associated with SSA's estimates of
potential beneficiaries under prior agreements have often been
substantial, even in cases where uncertainties about the number of
unauthorized workers were less prevalent. OCACT's comment that "taken as a
whole" its estimate of initial beneficiaries differs from actual initial
beneficiaries by only 3 percent is misleading because it nets
overestimates

against underestimates. OCACT prepares estimates of initial beneficiaries
for each proposed agreement with an individual country. Thus, any
comparison of estimated to actual initial beneficiaries should be on a
country-by-country basis, rather than by aggregating the error rates for
all agreements.

Finally, in response to our concern that the OCACT's original baseline
estimate of 50,000 first-year totalization beneficiaries did not directly
consider millions of current and former unauthorized Mexican workers,
OCACT said that this estimate was based on the best available data.
OCACT's comments also included excerpted text from the original estimate
in order to illustrate the analyses and assumptions that supported using
the 50,000 individuals already receiving Old-Age, Survivors, and
Disability Insurance benefits in Mexico as a proxy for potential
totalization beneficiaries. We acknowledge the data limitations facing
OCACT as well as its good faith effort to reasonably estimate the costs of
a totalization agreement with Mexico. However, based on our audit
work-which involved a thorough review of the full text of the actuarial
estimate, numerous in-depth interviews with OCACT officials to discuss
issues of concern, and regular consultation with our own Chief Actuary-it
seems reasonable to examine all sources of data and address the estimates
of unauthorized Mexican immigrants directly to provide a more complete
picture of possible outcomes from an agreement with Mexico. We continue to
believe that, given the magnitude of the proposed Mexican agreement
relative to other totalization agreements, it is not unreasonable to
expect that OCACT should develop and use a variety of approaches to
estimate potential costs and perhaps develop a range of cost estimates
based on those data sources and alternative assumptions. Such efforts
would better serve the information needs of the Congress in the event that
an agreement is ultimately submitted for its review.

We are sending copies of this report to the House and Senate committees
with oversight responsibilities for the Social Security Administration. We
will also make copies available to other interested parties upon request.
In addition, the report will be available at no charge on GAO's Web site
at http://www.gao.gov. If you or your offices have any questions
concerning this report, please call me or Daniel Bertoni, Assistant
Director, on

(202) 512-7215. Other major contributors to this report are Patrick
Dibattista, Gerard Grant, Daniel Schwimer, William Staab, and Paul Wright.

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

Appendix I: Comparison of Totalized and Minimum Social Security Benefits

Table 3: Monthly Social Security Benefits Payable in 2003 at Different
Earnings and Coverage Levels under a Totalization Agreement and Compared
to the Minimum Benefit Payable

                                 Monthly totalized social    
                                 security                    
                                 retirement benefit for the  
                                 credits                     
                                           earned            
                                                               Monthly social 
                                                             security benefit 
                 Social security                              with 40 credits 
                        earnings                             
                                 8 credits 20 credits 36                      
                          levela credits                               earned

                  Low earnings     $39.00   $99.00        $178.00     $296.00 
              Average earnings     $65.00   $163.00       $294.00     $561.00 
                 High earnings     $85.00   $212.00       $382.00     $702.00 
              Maximum earnings     $94.00   $237.00       $427.00     $899.00 

Source: Office of International Programs, SSA.

aA low earnings level equals 45 percent below the national average wages
for each year. An average earnings level equals the national average wages
for each year. A high earnings level equals 60 percent above the national
average wages for each year. A maximum earnings level equals the maximum
taxable amount of covered earnings for each year.

Page 20 GAO-03-993 Social Secur

Appendix II: Comments from the Social Security Administration

Appendix II: Comments from the Social Security Administration

Appendix II: Comments from the Social Security Administration

Appendix II: Comments from the Social Security Administration

Appendix II: Comments from the Social Security Administration

Appendix II: Comments from the Social Security Administration

Appendix II: Comments from the Social Security Administration

Appendix II: Comments from the Social Security Administration

Appendix II: Comments from the Social Security Administration

Appendix II: Comments from the Social Security Administration

  GAO's Mission

Obtaining Copies of GAO Reports and Testimony

The General Accounting 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.

The fastest and easiest way to obtain copies of GAO documents at no cost
is through the Internet. GAO's Web site (www.gao.gov) contains abstracts
and full-text files of current reports and testimony and an expanding
archive of older products. The Web site features a search engine to help
you locate documents using key words and phrases. You can print these
documents in their entirety, including charts and other graphics.

Each day, GAO issues a list of newly released reports, testimony, and
correspondence. GAO posts this list, known as "Today's Reports," on its
Web site daily. The list contains links to the full-text document files.
To have GAO e-mail this list to you every afternoon, go to www.gao.gov and
select "Subscribe to e-mail alerts" under the "Order GAO Products"
heading.

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. General Accounting 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

Contact:

To Report Fraud, Web site: www.gao.gov/fraudnet/fraudnet.htm

  Waste, and Abuse in E-mail: [email protected]

Federal Programs Automated answering system: (800) 424-5454 or (202)
512-7470

Jeff Nelligan, Managing Director, [email protected] (202) 512-4800

Public Affairs 	U.S. General Accounting Office, 441 G Street NW, Room 7149
Washington, D.C. 20548
*** End of document. ***