Federal Student Loans: Challenges in Estimating Federal Subsidy  
Costs (29-SEP-05, GAO-05-874).					 
                                                                 
In fiscal year 2004, the federal government made or guaranteed	 
about $84 billion in loans for postsecondary education through	 
two loan programs--the Federal Family Education Loan Progam	 
(FFELP) and the Federal Direct Loan Program (FDLP). Under FFELP, 
private lenders fund the loans and the government guarantees them
a minimum yield and repayment if borrowers default. When the	 
interest rate paid by borrowers is lower than the guaranteed	 
minimum yield, the government pays lenders special allowance	 
payments (SAP). Under FDLP, the U.S. Treasury funds the loans	 
that are originated through participating schools. Under the	 
Federal Credit Reform Act (FCRA) of 1990 the government 	 
calculates, for purposes of the budget, the net cost of extending
or guaranteeing credit over the life of a loan, called a subsidy 
cost. Agencies generally update, or reestimate, subsidy costs	 
annually to include actual program results and adjust future	 
program estimates. GAO examined (1) whether reestimated subsidy  
costs have differed from original estimates for FFELP and FDLP	 
loans disbursed in fiscal years 1994 through 2004, (2) what	 
factors explain changes between reestimated and original subsidy 
rates--that is subsidy cost estimates per $100 disbursed; and (3)
which federal costs and revenues associated with the student loan
programs are not included in subsidy cost estimates.		 
-------------------------Indexing Terms------------------------- 
REPORTNUM:   GAO-05-874 					        
    ACCNO:   A38637						        
  TITLE:     Federal Student Loans: Challenges in Estimating Federal  
Subsidy Costs							 
     DATE:   09/29/2005 
  SUBJECT:   Aid for education					 
	     Cost analysis					 
	     Higher education					 
	     Interest rates					 
	     Student loans					 
	     Subsidies						 
	     Cost estimates					 
	     Dept. of Education Federal Family			 
	     Education Loan Program				 
                                                                 
	     Dept. of Education William D. Ford 		 
	     Direct Loan 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-05-874

United States Government Accountability Office

GAO

                       Report to Congressional Committees

September 2005

FEDERAL STUDENT LOANS

                 Challenges in Estimating Federal Subsidy Costs

GAO-05-874

September 2005

FEDERAL STUDENT LOANS

Challenges in Estimating Federal Subsidy Costs

[IMG]

What GAO Found

Both FFELP and FDLP subsidy cost reestimates have differed from original
estimates for loans made in fiscal years 1994 through 2004, reflecting the
challenges inherent in estimating the actual costs of loans made under
each of these federal loan programs. Reestimated subsidy costs for FFELP
loans were close to or lower than original estimates for loans made in
fiscal years 1994 to 2002, but higher than originally estimated for loans
made in fiscal years 2003 and 2004. FDLP reestimated subsidy costs were
generally similar to or higher than originally estimated for loans made in
fiscal years 1994 through 2004.

Differences between original and reestimated subsidy cost estimates per
$100 disbursed were, in part, due to market interest rates that were lower
than originally forecasted, greater than anticipated loan consolidation,
and the availability of additional data on student loans. Each of these
factors has affected reestimated subsidy costs for each loan program in a
different way. For example, interest rates fell to lower than expected
levels in 2001 and the condition persisted through 2004. For FFELP, lower
than expected interest rates have made the difference between the borrower
interest rate and lender yield smaller than expected resulting in lower
SAP paid to lenders, which in turn resulted in lower reestimated subsidy
cost estimates. For FDLP, lower than expected interest rates contributed
to higher reestimated subsidy costs because the government received
smaller interest payments from borrowers than originally anticipated and,
in some cases, the rate paid by student borrowers fell below the
government's fixed borrowing rate.

Certain federal costs and revenues associated with the student loan
programs, such as federal administrative expenses, some costs of risk
associated with lending money over time, and federal tax revenues
generated by both student loan programs, are not included in subsidy cost
estimates. For example, under current law, federal administrative expenses
are excluded from subsidy cost estimates. Moreover, both loan programs
generate federal tax revenues from private sector companies and investors
that are encompassed in the revenue portion of the budget but are not
included in subsidy cost calculations. Estimating the amount of federal
tax revenues generated by the loan programs would be difficult and was
beyond the scope of our review.

Education reviewed a draft copy of this report and did not have any
comments.

                 United States Government Accountability Office

Contents

     Letter                                                                 1 
                                    Results in Brief                        6 
                                       Background                           8 
                Reestimated Subsidy Costs Differed from Original Estimates 
                                                                       for 
                                   Both Loan Programs                      12 
               Recent Low Interest Rates, Increased Loan Consolidation,    
               and                                                         
                   Additional Data Contributed to Differences between      
                     Reestimated and Original Subsidy Cost Estimates       19 
                    Certain Federal Costs and Revenues Associated with the 
                                                                   Student 
                  Loan Programs Are Not Included in Subsidy Cost Estimates 29 
                                 Concluding Observations                   34 
                                     Agency Comments                       35 
Appendix I                Comparison of Fiscal Year 2006 FDLP and FFELP 
               Reestimated Subsidy Costs per $100 Disbursed,               
                                 by Loan Type and Cohort                   
Appendix II           GAO Contacts and Staff Acknowledgments            

Tables

Table 1: FFELP Reestimated and Original Subsidy Cost Estimates per $100
Disbursed, by Loan Type and Loan Cohort 15 Table 2: FDLP Reestimated and
Original Subsidy Cost Estimates Per $100 Disbursed, by Loan Type and Loan
Cohort 17 Table 3: Comparison of Reestimated and Original Subsidy Cost

Estimates for All Loans Disbursed between Fiscal Years

1994 and 2004 18 Table 4: OMB Interest Rate Projections for the 91-day
Treasury Bill as Shown in the 1999 President's Budget Compared to CBO's
Projections and Actual Interest Rates 20 Table 5: Comparison of Cost
Estimates per $100 Disbursed in

Fiscal Year 2006 with and without Administrative

Expenses 30

Figures

Figure 1: FFELP Cash Flows Used in Subsidy Cost Estimates 11

Figure 2: FDLP Cash Flows Used in Subsidy Cost Estimates 12 Figure 3:
Comparison of Total Reestimated FFELP Subsidy Costs to Original Estimates,
by Loan Cohort 13 Figure 4: Comparison of Total Reestimated FDLP Subsidy
Costs and Original Estimates, by Loan Cohort 16 Figure 5: Comparison of
Rates for 3-month Commercial Paper and 91-day Treasury Bill, 1997 to 2004
21 Figure 6: Comparison of Borrower Interest Rate to Lender Yield for
Loans Disbursed on or after January 1, 2000 22

Figure 7: Actual and Projected Borrower Interest Rate for a Stafford Loan
in Repayment Compared to the Discount Rate for the Fiscal Year 1999 Loan
Cohort 23

Figure 8: Consolidation Example: Borrower's Perspective 25 Figure 9:
Consolidation Example: Education's Perspective 26 Figure 10: Comparison of
Reestimated Subsidy Costs for

Subsidized Stafford Loans in FFELP and FDLP, by Loan Cohort 38

Figure 11: Comparison of Reestimated Subsidy Costs for Unsubsidized
Stafford Loans in FFELP and FDLP, by Loan Cohort 39

Figure 12: Comparison of Reestimated Subsidy Costs for PLUS Loans in FFELP
and FDLP, by Loan Cohort 40 Figure 13: Comparison of Reestimated Subsidy
Costs for Consolidation Loans in FFELP and FDLP, by Loan Cohort 41

Abbreviations

FCRA Federal Credit Reform Act of 1990
FDLP William D. Ford Direct Loan Program
FFELP Federal Family Education Loan Program
HEA Higher Education Act
OMB Office of Management and Budget
PLUS Parent Loans for Undergraduate Students
SAP special allowance payments

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 29, 2005

Congressional Committees

In fiscal year 2004, the federal government made or guaranteed about $84
billion in loans to assist students in paying for their postsecondary
education under title IV of the Higher Education Act (HEA), as amended.
Federal student loans are primarily administered through two programs: the
Federal Family Education Loan Program (FFELP) and the William D. Ford
Direct Loan Program (FDLP). The federal government's role in financing and
administering loans for these two programs differs significantly. Under
FFELP, private lenders, such as banks, fund the loans, and the federal
government guarantees FFELP lenders a minimum yield on the loans they make
and repayment if borrowers default. When the interest rate paid by
borrowers is lower than the minimum yield guaranteed to lenders, the
government pays lenders the difference-a subsidy called special allowance
payments (SAP). Additionally, statedesignated guaranty agencies receive
federal funding to perform a variety of administrative functions in FFELP
and also work with lenders and borrowers to prevent loan defaults and
collect on the loans after default. Under FDLP, the U.S. Treasury funds
the loans, which are originated through participating schools. The
Department of Education contracts with private-sector firms to provide
administrative functions for its student loan programs.

In general, which program a student uses to obtain a loan depends upon
which program the student's school has chosen to use. Both FFELP and FDLP
offer students and their parents the same types of loans to pay for
postsecondary education-Stafford subsidized, Stafford unsubsidized, Parent
Loans for Undergraduate Students (PLUS), and consolidation loans. The
interest rate borrowers pay on Stafford and PLUS loans is a variable rate
based on a statutory formula.1 Subsidized loans are awarded

1The formula for calculating borrower interest rates on Stafford loans
currently being disbursed is based on the 91-day Treasury bill (T-bill)
rate plus 1.7 percent while the borrower is in school, and plus 2.3
percent when the borrower is in repayment. Stafford rates are capped at
8.25 percent. The formula in effect for calculating interest rates on PLUS
loans currently being disbursed is based on the 91-day T-bill rate plus
3.1 percent, and the PLUS rates are capped at 9 percent. Borrower rates
are reset on July 1 each year, based on the T-bill rate from the last
Treasury auction conducted before June 1.

based on a student's financial need and the federal government pays the
interest on behalf of students while they are attending school and during
a brief grace period when the student first leaves school. Unsubsidized
and PLUS loans are available to borrowers regardless of financial need,
and borrowers are responsible for interest payments during the life of the
loan. Consolidation loans allow borrowers to combine multiple federal
student loans into a single loan with a fixed interest rate based on the
weighted average of the interest rates in effect on the loans being
consolidated.

In recent years, competition between FFELP and FDLP has been credited with
improving services provided for both schools and borrowers and enhancing
borrower benefits, but there has been ongoing debate about whether the
costs and benefits of one program outweigh those of the other. Assessing
and comparing the total costs and benefits of the two loan programs would
require consideration of, among other things, costs incurred by schools in
operating the loan programs, quality of services provided to schools and
borrowers, benefits to society and individuals from postsecondary
education, as well as federal costs and revenues generated by each loan
program-federal budgetary costs. With respect to the latter, the technical
nature of how the government accounts for the federal budgetary costs of
the two loan programs, and disagreement about whether estimates of subsidy
costs fully represent federal costs have made debate about comparing costs
of the loan programs challenging.

The Federal Credit Reform Act of 1990 (FCRA) significantly changed the way
that the federal government accounted for the budgetary costs of credit
programs, including FFELP and FDLP. Prior to FCRA, the government
calculated costs on a cash basis-whereby costs and revenues were recorded
when money was paid or received. On a cash basis, direct loans initially
appeared to be as expensive as grants because the budget did not recognize
the expected repayment of direct loans. Loan guarantees, on the other
hand, initially appeared to be cost free (or could even appear to make
money because of upfront fees paid by borrowers and lenders to the
government) because the budget did not recognize expected federal payments
to lenders as a consequence of loan defaults. Under FCRA, the government
calculates, for purposes of the budget, the net cost to the government of
extending or guaranteeing credit over the life of a loan-called the
subsidy cost. FCRA was enacted to require agencies to measure the lifetime
costs of a loan in a way that would permit better cost comparisons between
guaranteed and direct loans, and between credit and non-credit programs
(e.g., grants). Agencies are required to estimate the subsidy cost to the
government of a direct loan or a loan guarantee based on the net present
value of all estimated cash

flows, excluding administrative costs, when preparing their annual
budget.2 Agencies generally update or revise these estimates, called
reestimates, annually to take into account changes in interest rates and
other conditions. To provide Congress with information about federal costs
for the student loan programs to use as it considers reauthorization of
the HEA, we examined:

1. 	whether reestimated subsidy costs have differed from original
estimates for FFELP and FDLP loans disbursed in fiscal years 1994 through
2004;

2. 	what factors explain changes between reestimated and original subsidy
rates-that is, subsidy cost estimates per $100 disbursed; and

3. 	which federal costs and revenues associated with the student loan
programs are not included in subsidy cost estimates.

To determine whether subsidy cost estimates have changed over time, we
compared original subsidy cost estimates to the most recent estimates by
analyzing subsidy cost estimates and reestimates for loans made and
guaranteed in each fiscal year, called a loan cohort, from 1994-the first
fiscal year loans were disbursed through FDLP-to 2004.3 We collected and
analyzed this information for loan cohorts by loan type in both FFELP and
FDLP as presented in the Budget of the United States Government. On the
basis of our review of the documentation for these data, we determined
that the data were sufficiently reliable for the purpose of our
examination. To determine the factors that explain changes in subsidy cost
estimates, we reviewed documentation prepared by the Department of
Education (Education) for its financial statement audits and interviewed
Education officials. To determine which federal costs and revenues are not
included in the subsidy cost estimates, we reviewed the HEA and related
regulations, FCRA, Office of Management and Budget

2"Present value" is the worth of future streams of returns or costs for a
program in terms of money paid immediately. In calculating present value,
future amounts are converted into their "money now" equivalents using a
discount rate. For purposes of making subsidy cost estimates, the discount
rate is determined by the Office of Management and Budget (OMB) and is
generally the average annual interest rate for marketable zero-coupon U.S.
Treasury securities with the same maturity from the date of disbursement
as the cash flow being discounted.

3Original cost estimates for a given loan cohort are based on the estimate
that appeared in the Budget of the United States Government for the fiscal
year a loan was made or guaranteed. The most recent reestimates are based
on the fiscal year 2006 budget.

Results in Brief

(OMB) guidance, Education's financial statements and auditor reports, and
the federal budget. We also interviewed officials with Education, OMB, the
Congressional Budget Office, and an FFELP lender and reviewed studies
about estimating the costs of the student loan programs. Moreover, we
gathered data from Education on federal administrative costs in each loan
program and reviewed literature about discounting cash flows and
incorporating risk into cost estimates. We conducted our work in
accordance with generally accepted government auditing standards from
January 2005 to August 2005.

Both FFELP and FDLP subsidy cost reestimates differed from original
estimates for loans made in fiscal years 1994 through 2004, reflecting the
challenges inherent in estimating the costs of loans made under each of
the federal student loan programs. Differences in total subsidy cost
estimates were driven both by differences between expected and actual loan
volume, as well as changes in subsidy rates-that is, subsidy cost
estimates per $100 disbursed. Reestimated subsidy costs for FFELP loans
were close to or lower than original estimates for loans made in fiscal
years 1994 to 2002, but higher than originally estimated for loans made in
fiscal years 2003 and 2004. After controlling for loan volume, reestimated
FFELP subsidy costs per $100 of loans disbursed were often lower than
originally estimated. FDLP reestimated subsidy costs were generally
similar to or higher than originally estimated for loans made in fiscal
years 1994 through 2004 both at the aggregate level and after controlling
for loan volume. Moreover, while some original estimates of FDLP subsidy
costs per $100 of loans disbursed projected a net gain for the government,
subsequent reestimates project a smaller gain or even a net cost to
government, thus illustrating that originally anticipated increases in
federal revenues may not, in fact, ultimately materialize. Although FDLP
reestimated subsidy costs have been higher than originally expected, they
have generally remained lower than those of FFELP. According to Education
officials, FDLP subsidy cost estimates per $100 disbursed are lower than
those of FFELP because, even though long-term estimates of interest
subsidies to borrowers and default costs are roughly equivalent under both
programs, under FFELP there are large cash outflows in the form of special
allowance payments to lenders while under FDLP there are large cash
inflows, net of payments to Treasury, in the form of borrower interest
payments and no SAP paid to lenders.

Differences between original and reestimated subsidy cost estimates per
$100 disbursed were, in part, due to market interest rates that were lower
than originally forecasted by OMB, greater than anticipated loan

consolidation, and the availability of additional data on student loan
borrowers. For FFELP, lower than expected interest rates made the
difference between the borrower interest rate and lender yield smaller
than expected, resulting in lower SAP paid to lenders, which, in turn,
resulted in lower reestimated subsidy cost estimates. In the case of FDLP,
lower than expected interest rates have resulted in lower than expected
interest payments from borrowers to the government, thus leading to higher
reestimated subsidy costs. Additionally, higher than anticipated
consolidation loan volume, resulting, in part, from low interest rates,
contributed to differences between original and reestimated subsidy costs
for both programs. In FFELP, it contributed to lower reestimated subsidy
costs for the underlying loan cohorts repaid by consolidation loans,
because the length of time Education anticipated paying SAP to lenders was
shortened. Estimated subsidy costs for recently disbursed FFELP
consolidation loans, which reflect costs associated with default risk and
SAP to lenders, are, however, quite large in comparison to previous years.
In FDLP, greater than expected prepayment due to consolidation decreased
the anticipated interest payments on the underlying loans, which, in turn,
contributed to higher reestimated subsidy cost estimates of the underlying
loan cohorts. Furthermore, additional data for both FFELP and FDLP loans
have enabled Education to refine its cash flow model when it reestimated
subsidy costs. For example, according to Education officials, data on
FFELP and FDLP borrowers' use of deferment options that allow them to
delay making payments on a loan when they return to school or are
experiencing economic hardship only recently became available. Education
refined its model to explicitly include assumptions about borrowers' use
of deferment, which has improved its cash flow estimates.

Certain federal costs and revenues associated with the student loan
programs are not included in subsidy cost estimates, such as federal
administrative expenses, some costs of risk associated with lending money
over time, and federal tax revenues generated by both student loan
programs. Under current law, federal administrative expenses are excluded
from subsidy cost estimates. For the fiscal year 2005 loan cohort,
Education estimated that cost estimates for FDLP would increase by $1.45
and for FFELP by $0.69 per $100 in loans disbursed if federal
administrative expenses were included. The large difference is because the
federal government is primarily responsible for administering the FDLP
while in FFELP lenders and guaranty agencies perform administrative tasks.
In addition, subsidy cost estimates do not include all risk that the
government incurs by lending money over time. Subsidy cost estimates
factor in anticipated cash flows from the loans, which incorporate some

risks that the government incurs, such as credit risk represented by a
default rate-the rate at which the government expects borrowers not to pay
back their student loans. However, some risks are not explicitly included
in subsidy cost estimates, such as interest rate risk- unanticipated
fluctuations in the interest rate due to changes in the economy that cause
changes in the present value of the loans' cash flows. Lastly, both loan
programs generate federal tax revenues from privatesector companies and
investors that participate in the federal student loan programs. These
revenues are encompassed in the revenue portion of the budget but are not
included in subsidy cost calculations. Estimating the amount of federal
tax revenues generated by the loan programs would be challenging.
Calculations of total federal costs would be enhanced were these
additional costs and revenues considered, though doing so may require
complex methodologies and/or data that are not currently readily
available.

We provided Education with a copy of our draft report for review and
comment. Education reviewed the report and did not have any comments.

Background 	The federal government makes loans to students through
private- and public-sector lenders in the FFELP or directly to students
through FDLP. These two programs are among the largest of the federal
government's credit programs. At the end of 2004, there were about $245
billion in outstanding FFELP loans, about 20 percent of total federal
guaranteed loans outstanding, and $107 billion in outstanding FDLP loans,
about 43 percent of total federal direct loans outstanding.

Types of Federal Loans and Terms for Borrowers

Students and parents are able to borrow the same types of loans through
FFELP and FDLP, which include:

o  	Subsidized and Unsubsidized Stafford Loans-variable rate loans
available to students. The federal government pays the interest on behalf
of subsidized loan borrowers while the student is in school and during a
brief grace period when the student first leaves school.

o  	PLUS Loans-variable rate loans made to parents, on behalf of students.
The borrower pays all interest costs.

o  	Consolidation Loans-borrowers may combine multiple federal student
loans into a single loan. The interest rate is fixed based on the weighted
average of the interest rates in effect on the loans being consolidated.

Under either loan program borrowers are able to repay loans earlier than
required, with no penalty. The programs have several repayment options
available to borrowers. For Stafford and PLUS loans, the standard
repayment in both loan programs is a fixed amount per month for up to 10
years. Borrowers have other repayment options that allow them to extend
repayment for up to 30 years, gradually increase the monthly payment, or
base monthly payments on their adjusted gross income. The criteria for
some of the alternative repayment options are different in FFELP and FDLP.
For consolidation loans, the repayment terms depend on the loan amount.
Moreover, borrowers that graduate, leave school, or become a less than
half-time student are given a 6-month grace period before they must begin
to repay their Stafford or consolidation loans.4 All borrowers may
postpone repayment through deferment or forbearance if they meet certain
criteria and the loan is not in default. Deferment is allowed for
borrowers who remain in a postsecondary school at least halftime, a
graduate program, or have experienced economic hardship. For borrowers who
are temporarily unable to meet repayment obligations but are not eligible
for deferment, lenders may grant a temporary and limited time period in
which these borrowers do not need to repay their student loans, called
forbearance.

Developing Subsidy Cost Estimates

The FCRA guidance issued by OMB and accounting standards provide the
framework for the process Education uses to calculate subsidy costs for
student loans. Subsidy costs are calculated by estimating the federal
government's future cash flows for loans made or guaranteed in a
particular fiscal year, called a loan cohort. In estimating cash flows for
a loan cohort, Education must make assumptions about loan characteristics
and future borrower behavior, such as:

o  type and dollar amount of loans obligated or guaranteed, and

o  	how many borrowers will pay early, pay late, or default on their loans
and at what point in time.

Moreover, the model used to estimate future cash flows includes
assumptions about future interest rates. OMB provides Education with
interest rate assumptions that are used for the discount rate, borrower

4A less than half-time student takes one to five credits in a
postsecondary school. An unsubsidized Stafford borrower pays interest
while in the grace period but the government continues to pay the interest
for subsidized Stafford borrowers during this time.

interest rate, and lender yields. Education aggregates cash flows by loan
cohort, loan type, and risk category, which reflects the differences in
the likelihood of default. Education has five risk categories, which
include, in order of higher to lower risk of default: (1) students at
proprietary schools, (2) students at 2-year colleges, (3) freshman and
sophomores at 4-year colleges, (4) juniors and seniors at 4-year colleges,
and (5) students at graduate schools.

Federal Government's Role in Each Loan Program and Difference in Cash
Flows

Although the method for calculating the subsidy cost is the same for both
FFELP and FDLP, the federal government's role in each loan program differs
significantly, which, in turn, affects the type and timing of cash flows
in each program. In FFELP, private lenders, such as banks, fund the loans,
and the federal government guarantees lenders a statutorily specified
minimum yield that is tied to, and varies with, market financial
instruments. When the interest rate paid by borrowers is below that yield,
the federal government gives lenders subsidy payments, called SAP.
Moreover, the federal government, through state-designated guaranty
agencies, guarantees repayment of loans if borrowers default. Guaranty
agencies provide insurance to lenders for 98 percent of the unpaid
principal of defaulted loans. The federal government, in turn, pays
guaranty agencies 95 percent of their default claims.5 Guaranty agencies
also perform various administrative functions in the FFELP. As shown in
figure 1, under FFELP cash inflows to the federal government include fees
and other payments from lenders and outflows from the federal government
include SAP and default payments. FFELP cash flows are spread out over the
life of the loan.

5For loans disbursed on or after October 1, 1998.

Figure 1: FFELP Cash Flows Used in Subsidy Cost Estimates

Source: GAO analysis.

Under FDLP, the U.S. Treasury funds the loans, which are originated
through participating schools and contractors. Education's Office of
Federal Student Aid is responsible for delivering funds to schools
participating in FDLP, monitoring its contracts, and providing technical
assistance to schools. Education contracts with private-sector companies
to perform various administrative activities in FDLP, such as originating
and servicing loans, and collecting defaulted loans. As shown in figure 2,
FDLP cash inflows to the federal government are repayments of principal
and interest payments and outflows include loan disbursements to
borrowers. Because the federal government funds the loans, cash outflows
occur in the early years as loan disbursements are made. Cash inflows, in
the form of principal repayment and interest payments, occur in later
years as borrowers enter repayment.

Figure 2: FDLP Cash Flows Used in Subsidy Cost Estimates

Both FFELP and FDLP reestimated subsidy costs have differed from original
estimates for loans made in fiscal years 1994 through 2004, highlighting
the challenges in estimating the costs of federal student loans. FFELP
reestimated subsidy costs were similar to or lower than original estimates
for loans made in fiscal years 1994 to 2002, but higher than originally
estimated for loans made in fiscal years 2003 and 2004. In comparison,
FDLP reestimated subsidy costs were generally similar to or higher than
original estimates for loans made in fiscal years 1994 through

Cash inflows

Borrower principal Cash outflows repaymentsa

Borrower interest Loan disbursements payments to borrowersb

Payments to contractors

                             Source: GAO analysis.

aPrincipal repayments may be less than disbursements, reflecting defaults,
loan discharges, and loan forgiveness.

bLoan disbursements are decreased by the amount of origination fees
charged to borrowers.

Annually, agencies are generally required to update or "reestimate" loan
costs for differences in estimated loan performance, such as differences
between assumed and actual default rates, the actual program costs
recorded in the accounting records, and new forecasts of future economic
conditions, such as interest rates. Reestimates include all aspects of the
original cost estimate, including prepayments, defaults, delinquencies,
recoveries, and interest. Reestimates of the credit subsidy allow agency
management to compare the original budget estimates with actual program
results to identify variances from the original estimate, assess the
quality of the original estimate, and adjust future program estimates as
appropriate.

Reestimated Subsidy Costs Differed from Original Estimates for Both Loan
Programs

2004. Across all types of loans, FDLP subsidy costs per $100 of loans
disbursed were, for almost all loan cohorts, lower than those of FFELP.

FFELP Reestimated Subsidy Costs Were Generally Similar to or Lower Than
Original Estimates, with the Exception of Loans Disbursed in Fiscal Years
2003 and 2004

Reestimated subsidy costs for FFELP loans disbursed between fiscal years
1994 and 2002 were, in general, close to or lower than original estimates,
while reestimated subsidy costs for loans disbursed in 2003 and 2004 were
higher than originally expected, as shown in figure 3.

Figure 3: Comparison of Total Reestimated FFELP Subsidy Costs to Original
Estimates, by Loan Cohort

Total subsidy costs (nominal dollars in billions) 12

10

8

6

4

2

0 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 Loan cohort

Original subsidy cost estimate Reestimated subsidy cost Source: GAO
analysis of the Budget of the United States Government, fiscal years 1994
to 2006.

From fiscal years 1994 to 1999, reestimated subsidy costs for FFELP were
typically close to original estimates, while loans disbursed from fiscal
year 2000 to fiscal year 2002 had reestimated subsidy costs that were
lower than original estimates, ranging from $1.5 to $2.2 billion lower.
Reestimated subsidy costs for loans disbursed in fiscal years 2003 and
2004 were $2.7 and $3.6 billion higher than original estimates.
Differences between reestimated and original subsidy costs estimates for
the 2003 and 2004 loan cohorts were in part due to significant differences
between expected and actual loan volume. For example, Education originally
estimated about $40 billion in FFELP loans would be disbursed in 2003 when
actually $69 billion was disbursed that year. The large difference was
primarily due to a significantly higher volume of FFELP consolidation
loans than originally estimated and the relatively high subsidy costs per
$100 of these loans compared to consolidation loans made in previous
years.

After controlling for loan volume, FFELP reestimated subsidy costs per
$100 disbursed were generally close to or lower than original subsidy cost
estimates across loan types. As shown in table 1, for FFELP Stafford
unsubsidized and PLUS loans, reestimated subsidy costs per $100 disbursed
were lower for all loan cohorts than what was originally estimated-except
fiscal year 1999. For subsidized Stafford loans, about two-thirds of the
loan cohorts had lower reestimated subsidy costs per $100 disbursed.
Slightly over half of all consolidation loan cohorts had lower reestimated
subsidy costs per $100 disbursed than originally estimated.

    Table 1: FFELP Reestimated and Original Subsidy Cost Estimates per $100
  Disbursed, by Loan Type and Loan Cohort Loan cohort (fiscal year) Loan type
        Estimate 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004

Stafford Original $15.59 $15.31 $22.84 $22.75 $18.80 $16.41 $21.25 $23.53 $23.14
 $16.81 $17.64 subsidized Reestimated 19.62 20.59 19.94 19.98 19.03 18.21 16.16
                            13.40 12.39 14.18 15.41

  Stafford     Original   0.63  3.79 4.74 6.77  7.27 0.90  7.89 8.38  6.85 4.61   4.92 
unsubsidized  Reestimated -0.45 0.26 0.01 0.34  0.99 2.54  2.02 1.84  2.34 3.67   3.87 
    PLUS       Original   2.50  2.75 1.64 3.33  3.65 -1.91 6.32 5.38  3.92 4.67   3.26 
              Reestimated 0.36  0.73 0.98 0.99  1.00 1.89  1.40 1.55  1.62 1.87   1.60 
Consolidation  Original              7.86 0.06  0.75 -3.55 2.35 2.91  3.51 10.19 15.76 
              Reestimated -0.64 0.47 0.57 -0.17 2.12 1.15  0.80 -0.46 3.11 11.21 15.98 

FDLP Reestimated Subsidy Costs Were Generally Similar to or Higher Than
Original Estimates

Source: GAO analysis of the Budget of the United States Government fiscal
years 1994 to 2006.

Note: Original subsidy cost estimates are the rates that appeared in the
appendix to the budget for the associated fiscal year. Reestimated subsidy
cost estimates are the rates that appeared in table 8 of the credit
supplement of the fiscal year 2006 budget. Negative amounts represent
revenue to the government and occur when expected cash inflows exceed cash
outflows. Positive amounts represent a cost to the government. Cost
estimates for FFELP subsidized loans are considerably higher than those
for all other loan types because the government pays the borrowers'
interest costs during the student's in-school, grace, and deferment
periods.

Reestimated subsidy costs for FDLP loans were in general similar to or
higher than original estimates for loans disbursed between fiscal years
1994 and 2004. For FDLP loans disbursed between fiscal years 1994 and
1999, total reestimated subsidy costs were in general close to original
estimates, but there was one loan cohort that had higher reestimated
subsidy costs and another with much lower reestimated subsidy costs than
originally expected, as shown in figure 4.

Figure 4: Comparison of Total Reestimated FDLP Subsidy Costs and Original
Estimates, by Loan Cohort

Total subsidy costs (nominal dollars in billions)

1.5

1.0

0.5

0.0

-0.5

-1.0

-1.5

-2.0

1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004

Loan cohort

Original subsidy cost estimate

Reestimated subsidy cost

Source: GAO analysis of the Budget of the United States Government, fiscal
years 1994 to 2006.

In comparison, reestimated subsidy costs for FDLP loans disbursed between
fiscal years 2000 and 2004 were higher than original estimates. In some
cases original estimates projected a net gain for the government, but
subsequent reestimates project a smaller gain or even a net cost for the
government. For example, original subsidy cost estimates of the fiscal
year 2000 loan cohort projected a net gain of $930 million for the
government and reestimated subsidy costs project a net cost of $1.1
billion. Such swings in estimated subsidy costs illustrate that originally
anticipated federal revenues may not, in fact, ultimately materialize.
Differences between total reestimated and original subsidy cost estimates
were not driven by differences between original and actual loan volume,
but rather by changes in the subsidy rates-that is, subsidy costs per $100
disbursed.

FDLP reestimated subsidy costs per $100 disbursed were usually close to or
higher than original subsidy cost estimates across loan types. For
example, as shown in table 2, reestimated subsidy costs per $100 disbursed
for FDLP Stafford unsubsidized, and PLUS loans were, for almost all loan
cohorts, higher than original estimates. For Stafford

 subsidized and consolidation loans, slightly over half of the loan cohorts had
     reestimated subsidy costs that were higher than originally estimated.

     Table 2: FDLP Reestimated and Original Subsidy Cost Estimates Per $100
       Disbursed, by Loan Type and Loan Cohort Loan cohort (fiscal year)

 Loan type    Estimate     1994   1995   1996   1997   1998   1999   2000   2001   2002   2003   2004 
  Stafford    Original   $12.42 $14.45 $16.54 $10.38 $13.65 $13.79 $4.06  $8.03  $2.66   $4.97 -$0.12 
 subsidized  Reestimated  13.75  13.21  11.23  11.22  10.10 12.41  14.81  11.24    6.89   2.52   4.06 
  Stafford    Original    -2.44  -6.99  -1.66 -11.77  -6.93  -7.78 -16.38 -14.36 -22.20 -12.25 -14.91 
unsubsidized Reestimated  -6.50  -5.13  -6.00  -5.25  -5.13  -0.91   2.99   2.01 -0.94   -5.67  -5.70 
    PLUS      Original    -4.86  -3.71 -11.10  -9.36  -6.34  -9.49 -13.41 -12.36 -16.21  -9.80 -14.72 
             Reestimated   0.81  -1.30  -2.88  -2.54  -2.76  -0.95   2.21   1.45 -1.20   -4.80  -4.48 

  Consolidation Original -0.59 -6.59 -0.14 -0.61 -7.72 -3.92 -6.96 -3.75 1.13

Reestimated 2.23 0.77 1.58 -0.51 -2.97 1.62 -3.42 -4.29 -6.00 -1.75

FDLP Reestimated Subsidy Costs Were Lower Than FFELP Reestimated Subsidy
Costs

Source: GAO analysis of the Budget of the United States Government, fiscal
years 1994 to 2006.

Note: Original subsidy cost estimates are the rates that appeared in the
appendix to the budget for the associated fiscal year. Reestimated subsidy
cost estimates are the rates that appeared in table 7 of the credit
supplement of the fiscal year 2006 budget. Negative amounts represent
revenue to the government and occur when expected cash inflows exceed cash
outflows. Positive amounts represent a cost to the government. FDLP
Stafford subsidized loans will typically have positive subsidy costs
because the government does not collect interest payments from borrowers
during students' in-school, grace, or deferment periods.

For most Stafford unsubsidized and PLUS loan cohorts, and slightly over
half of consolidation loan cohorts, reestimated subsidy costs per $100
disbursed were higher than the original estimate, but still project a net
gain for the federal government. For example, Stafford unsubsidized loans
disbursed in fiscal year 1998 were originally estimated to have a net gain
of $6.93 for every $100 in loans disbursed. Reestimated subsidy costs show
that the projected net gain for these same loans is estimated to be $5.13
per $100 disbursed. Some loan cohorts that originally projected a net gain
for the federal government have reestimated subsidy costs with a net cost
to the government. For example, PLUS loans disbursed in fiscal year 2000
that were originally projected to have a net gain of $13.41 per $100
disbursed were subsequently reestimated to have a net cost of $2.21 per
$100 disbursed.

For all loans disbursed between fiscal years 1994 and 2004, FDLP
reestimated subsidy costs were lower than FFELP reestimated subsidy costs
in aggregate and after controlling for loan volume. Reestimated total
subsidy costs for FDLP loans were $2.5 billion compared to $36.6 billion
for FFELP loans, as shown in table 3 below.

Table 3: Comparison of Reestimated and Original Subsidy Cost Estimates for
All Loans Disbursed between Fiscal Years 1994 and 2004

              Original total Reestimated   Difference Total loan  Reestimated 
                     subsidy total         between               
    Loan     cost estimate a subsidy costb       cost volume     subsidy cost 
                                            estimates disbursed           per 
program        (billions)    (billions) (billions) (billions)         $100 
                                                                   disbursedc 
    FFELP              $35.2         $36.6       $1.4       $396        $9.20 
    FDLP               -$2.1          $2.5       $4.6       $150        $1.70 

Source: GAO analysis of the Budget of the United States Government, fiscal
years 1994 to 2006.

aOriginal subsidy cost estimates are the sum of the subsidy cost estimates
for each of the fiscal year 1994 to fiscal year 2004 cohorts, as presented
in the appendix to the budget for the associated fiscal year. Negative
amounts represent a gain to the government.

bReestimated subsidy costs are the total subsidy costs listed in the
credit supplement of the fiscal year 2006 budget.

cThe subsidy costs per $100 disbursed are based on the reestimated total
subsidy costs divided by the total loan volume disbursed. These subsidy
costs are not directly comparable to those reported in the budget; subsidy
costs shown are based on loans disbursed whereas under FCRA subsidy costs
are based on loans originated.

After controlling for loan volume and comparing reestimated subsidy costs
across the four types of loans-Stafford subsidized and unsubsidized, PLUS,
and consolidation-FDLP reestimated subsidy costs per $100 disbursed were
in general lower than FFELP reestimated subsidy costs per $100 disbursed.
(See app. I for comparisons of reestimated subsidy costs of FDLP and FFELP
loans, by loan type.) The difference between the reestimated subsidy cost
for FDLP and FFELP varied significantly and depended on the type of loan
and the year that the loan was disbursed. For example, reestimated subsidy
costs per $100 disbursed for FDLP subsidized Stafford loans disbursed in
fiscal year 2003 were $11.66 lower than for FFELP subsidized Stafford
loans, while the difference for the same loans disbursed in 2000 was $1.35
per $100 disbursed.

The primary reason for the difference in subsidy cost estimates between
FFELP and FDLP were differences in the structure of the programs rather
than the characteristics of the borrowers. According to Education
officials, estimates of long-term costs associated with subsidizing
borrowers' interest; canceling repayment of loans due to death,
disability, and bankruptcy; and defaulted loans are roughly equivalent in
both programs. However, under FFELP there are larger cash outflows in the
form of SAP to lenders than cash inflows of lender fees, while in FDLP
there are large cash inflow projections, net of interest payments to
Treasury, in the form of borrower interest payments and no SAP or guaranty
fees.

Recent Low Interest Rates, Increased Loan Consolidation, and Additional
Data Contributed to Differences between Reestimated and Original Subsidy
Cost Estimates

Differences between original and reestimated subsidy cost estimates per
$100 disbursed can be explained, in part, by lower than expected market
interest rates, greater than anticipated loan consolidation, and more data
on student loans incorporated into cash flow model. Differences between
actual and expected interest rates and rates of consolidations affected
reestimated subsidy costs for each loan program in a different way. For
example, lower than expected interest rates over the last several years
have resulted in lower reestimated subsidy cost estimates for FFELP and
higher reestimated subsidy costs for FDLP. Larger than expected volumes of
consolidation loans, which stemmed in part from low interest rates,
contributed to lower FFELP reestimated subsidy costs for the underlying
loan cohorts and higher FDLP reestimated subsidy cost estimates of the
underlying loan cohorts. Furthermore, the availability of additional data
for both FFELP and FDLP loans have enabled Education to refine its cash
flow model, which has also contributed to differences between reestimated
and original subsidy costs.

Interest Rates Lower Than Previously Forecasted Contributed to Differences
between Reestimated and Original Subsidy Cost Estimates

Interest rates fell to lower than expected levels in 2001 and persisted at
those levels through 2004, which affected subsidy cost estimates in both
FFELP and FDLP because estimates, especially for the FDLP, are highly
sensitive to changes between projected and actual interest rates. Cost
estimates for the loan programs are sensitive to such changes because
borrower interest rates in both FFELP and FDLP and the lender yield in the
FFELP, are variable rates. As a result, differences between projected and
actual interest rates can have a significant impact on estimates of cash
flows in both loan programs. OMB's interest rate projections made prior to
2001, as well as those by other government agencies and the private
sector, were considerably higher than actual interest rates for 2001 and
beyond. For example, as shown in table 4, actual interest rates from 2001
to 2003 were substantially lower than OMB's forecasts of interest rates
used in the budget for fiscal year 1999 and fluctuated slightly from year
to year. To the degree that such fluctuations were unanticipated, they
contributed to volatility in subsidy cost reestimates from year to year.

Table 4: OMB Interest Rate Projections for the 91-day Treasury Bill as Shown in
 the 1999 President's Budget Compared to CBO's Projections and Actual Interest
                                     Rates

                OMB's 1999 interest        CBO's January 1999 Actual interest 
           Year     rate projection interest rate projections            rate 
           1999                 4.9                       5.2            4.64 
           2000                 4.8                       4.8            5.82 
           2001                 4.7                       4.7             3.4 
           2002                 4.7                       4.7            1.61 
           2003                 4.7                       4.7            1.01 

Source: Budget of the United States Government, fiscal year 1999 and
Federal Reserve data.

For FFELP, lower than expected interest rates have resulted in lower than
expected SAP to lenders, which, in turn, resulted in lower reestimated
subsidy cost estimates. As interest rates decreased, the difference, or
spread, between the 3-month commercial paper (CP) and the 91-day Treasury
bill narrowed.6 For example, as can be seen in figure 5, the average rates
on the 91-day T-bill and the 3-month CP were 5.82 and 6.33, respectively,
in 2000, a difference of 0.51. However, in 2004 the difference between the
two rates was 0.15. The spread between commercial paper and Treasury bill
rates serves as the primary basis for SAP payments to the lenders, and, as
the spread narrowed, Education paid lower SAP, thus lowering reestimated
subsidy costs.

6The lender yield is calculated quarterly and for loans originated on or
after January 1, 2000, is the 3-month commercial paper rate plus a
supplement (for Stafford loans the supplement is 1.74 while the borrower
is in school or in a grace or deferment period and 2.34 otherwise). The
borrower interest rate is set annually, and for loans originated on or
after July 1, 1998, is the 91-day T-bill plus a supplement (for Stafford
loans the supplement is 1.70 while in school/grace/deferment and 2.30
otherwise).

Figure 5: Comparison of Rates for 3-month Commercial Paper and 91-day
Treasury Bill, 1997 to 2004

Percent

7

6

5

4

3

2

1

0 1997 1998 1999 2000 2001 2002 2003 2004

Year

91-day Treasury bill

3-Month commercial paper

Source: GAO analysis of Federal Reserve Bank Data.

Note: Rates for 3-month Commercial Paper and 91-day Treasury bill are
annual averages quoted on a discount basis and are not comparable to bond
yields.

The climate of declining interest rates not only narrowed the spread
between the T-bill rate and the CP rate and reduced SAP payments, it also
eliminated SAP payments for some loans because interest rates paid by
borrowers were higher than the guaranteed lender yield. Whether SAP is
paid on a loan can change during a year because borrower interest rates
are adjusted annually based on the final auction of T-bills before June 1
of each year while lender yields are adjusted each quarter. Thus in a
climate of declining interest rates, SAP on certain loans was eliminated
because the 3-month CP rate-on which the lender yield is based-fell, for a
particular quarter, below the annually adjusted borrower rate. SAP was
zero in 50 percent of the quarters for Stafford loans issued after January
1, 2000 through July 1, 2005. This is illustrated in figure 6, where one
can also see that the more recent climate of rising interest rates could
lead to increased SAP.

Figure 6: Comparison of Borrower Interest Rate to Lender Yield for Loans
Disbursed on or after January 1, 2000

Percent

9

8

7

6

5

4

3

2

1

0

2000 2001 2002 2003 2004 2005 Year (June 30)

Borrower interest rate

Lender yield

Source: GAO analysis of Department of Education data.

Note: Lender yield is for loans made on or after January 1, 2000, and the
borrower interest rate is for loans made on or after July 1, 1998, while
borrower is in repayment.

In contrast, lower than expected interest rates contributed to higher
reestimated FDLP subsidy costs. Under FDLP, the government had originally
anticipated larger interest payments from borrowers as they repaid their
loans because original subsidy cost estimates were based on forecasts that
did not anticipate the significant decline in interest rates. Lower than
expected interest rates thus resulted in lower than expected cash inflows
to the government and higher FDLP subsidy cost reestimates. For example,
using the numbers in table 4, one can see that original subsidy cost
estimates made for the 1999 loan cohort assumed that interest rates on the
91-day Treasury bill would be 4 times higher than they actually were when
some students would be entering repayment on loans they obtained in 1999.
Moreover, original estimates were based on the assumption that the
interest rate paid by borrowers on those loans would be higher than the
interest rate Education pays to Treasury for borrowing the funds to make
the loans. As can be seen in figure 7, the borrower interest rate fell
below the discount rate (rate paid to Treasury) in 2001.

Again, such a climate of lower than anticipated interest rates led to
higher reestimates of subsidy costs. As interest rates rise, the interest
paid by borrowers will increase-possibly to rates higher than the discount
rate.

Figure 7: Actual and Projected Borrower Interest Rate for a Stafford Loan
in Repayment Compared to the Discount Rate for the Fiscal Year 1999 Loan
Cohort

Percent

9

8

7

6

5

4

3

2

1

0 1999 2000 2001 2002 2003 2004 2005

Year

Discount rate for fiscal year 1999 loan cohort

Projected borrower rate

Actual borrower rate

Source: GAO analysis of Department of Education data.

Note: Borrower interest rate is for a Stafford loan disbursed in fiscal
year 1999 in repayment status. The projected borrower rate was calculated
using the Department of Education's 1999 projections of the 91-day
Treasury bill and adding 2.30 percentage points.

Lower than expected interest rates also affected the actual rate used to
discount cash flows for FFELP and FDLP subsidy cost estimates. When
subsidy cost estimates are first prepared for the budget, agencies use an
estimated discount rate. Education sets the actual discount rate when a
loan cohort is fully disbursed. Because subsidy cost estimates are
prepared prior to when a loan is disbursed, it is expected that
differences between the estimated and actual discount rate will contribute
to differences between reestimated and original subsidy cost estimates.
For example, the actual discount rate for loans disbursed in fiscal year
2002 was lower than originally estimated, which lowered reestimated
subsidy costs slightly in both FFELP and FDLP.

Higher Than Expected Consolidation Volume Was Another Factor Contributing
to Differences between Reestimated and Original Subsidy Cost Estimates

Higher than expected consolidation volume, which stemmed in part from low
interest rates, also affected reestimated subsidy costs. As we have
previously reported, the number of borrowers consolidating their loans has
increased substantially over the last several years.7 Consolidation
activity has been higher than expected in both loan programs since fiscal
year 1999. When borrowers consolidated their student loans and locked in
recent low interest rates, they effectively paid off the underlying loans-
Stafford subsidized and unsubsidized and PLUS-ahead of schedule and
started a new consolidation loan. With the new consolidation loans,
borrowers began new repayment periods that could be up to 30 years from
when the consolidation loans were made. Because Education calculates
subsidy costs for consolidation loans separately, it must adjust original
estimates of the underlying loans to reflect unanticipated prepayments.
Education considers the consolidation a new loan in the year that the loan
was disbursed. Figures 8 and 9 provide a simplified example of
consolidation from both the borrower's and Education's perspective.

7GAO, Student Loan Programs: As Federal Costs of Loan Consolidation Rise,
Other Options Should Be Examined, GAO-04-101 (Washington, D.C.: Oct. 31,
2003); and GAO, Student Loan Programs: Lower Interest Rates and Higher
Loan Volume Have Increased Federal Consolidation Loans, GAO-04-568T
(Washington, D.C.: Mar. 29, 2004).

            Figure 8: Consolidation Example: Borrower's Perspective

Figure 9: Consolidation Example: Education's Perspective

Case 1: Continuing the previous example, assume that the loans were made
through FFELP and have been in repayment since January 2002. Assume also
that Education expected the repayment to lenders on these underlying loans
to continue into January 2012, at which time the loans would be fully
repaid. In estimating the subsidy costs of the loans, Education made
assumptions about the likelihood that the borrower would default (based on
the type of school attended as well as the borrower's payment history
since 2002) and the amount of SAP that would be paid to lenders through
2012.

However, now that the underlying loans are paid in full, reestimated
subsidy costs for the 2000 and 1997 origination cohorts reflect the
unexpected changes, and would be lower for these three specific loans than
original estimates.

The new consolidation loan is part of the 2005 loan cohort. Education
makes assumptions about expected repayment and likelihood of paying SAP.
The borrower rate is fixed and relatively low compared to projected market
interest rates in the future. Because the yield that the government
guarantees the lender is variable and market interest rates are expected
to rise in the future, Education's estimated subsidy costs for the new
consolidation loan includes expectations that SAP payments to the lender
will be necessary and will increase in the future. The net effect may be
higher subsidy costs for the consolidation loan than those estimated for
the three underlying loans.

Case 2: Assume instead that the three Stafford loans were made through
FDLP (i.e., the two lender assumption from figure 8 does not apply and the
borrower was not motivated to consolidate to reduce the number of monthly
payments). Again assume that the loans have been in repayment since
January 2002 and that Education expected repayment on the underlying loans
to continue into January 2012.

The borrower's FDLP consolidation resulted in a shorter repayment period
to Education for the three Stafford underlying loans and less interest
payments than had been expected for the three loans. Thus, reestimated
subsidy costs would be higher (i.e., the inflow of interest payments from
the borrowers would be less) than original estimates for these three
underlying loans. If the interest rate paid by the borrower was less than
the rate paid by Education to borrow the funds to make the loans, then
prepaying a loan would not necessarily result in higher reestimated
subsidy costs.

The new consolidation loan is part of the 2005 loan cohort. The borrower
rate is fixed and relatively low compared to projected market interest
rates in the future. The net effect may be higher subsidy costs for the
consolidation loan than those estimated for the three underlying loans.

Consolidation activity has been particularly high for FFELP loans,
increasing from about $7 billion in fiscal year 2000 to $37 billion in
fiscal year 2004. Education had not anticipated such an increase in
consolidation loans, which contributed to lower reestimated subsidy costs
for the underlying loan cohorts. Under FFELP, consolidation loans
shortened the length of time Education anticipated paying SAP to lenders
and eliminated default risk on the underlying loans, thus lowering
reestimated subsidy costs. Estimated subsidy costs for recent
consolidation cohorts, which reflect costs associated with default risk
and SAP to lenders, are quite large in comparison to previous
consolidation loan cohorts. For example, reestimated subsidy costs per
$100 disbursed for consolidation loans made in 2003 were $11.21 and in
2004 were $15.98 compared to $3.11 for consolidation loans made in 2002.
The increase is due in part because borrowers locked in lower fixed
interest rates on their consolidation loans and the minimum yield
guaranteed to lenders is projected to be much higher than the fixed
interest rate paid by borrowers, thus requiring the government to pay
higher SAP than they would have on the 2002 loans.

Consolidation activity in FDLP also increased-from $5 billion in fiscal
year 2000 to $8 billion in fiscal year 2004. As borrowers consolidated
their loans, they repaid the underlying loans that shortened the length of
time Education had expected to receive interest payments on these loans.
According to Education, it had calculated that the interest payments from
borrowers would contribute positively to Education's cash flows because
expected interest rates that borrowers paid to Education were higher than
the rate Education paid to borrow the funds. However, greater than
expected prepayment due to consolidation decreased the anticipated
interest payments on the underlying loans, which in turn contributed to
higher reestimated subsidy cost estimates of the underlying loan cohorts.8
Moreover, as we reported in August 2004, large amounts of FDLP loans-
about $7.5 billion between 1998 and 2002-were consolidated into FFELP.9 As
a result, Education will not receive any of the future projected interest
payments on those loans that are now FFELP loans, which also contributed
to higher reestimated FDLP subsidy costs. Additionally, for

8If the interest rate paid by the borrower was less than the rate paid by
Education to borrow the funds to make the loans, then prepaying a loan
would not necessarily result in higher reestimated subsidy costs.

9GAO, Student Consolidation Loans: Further Analysis Could Lead to Enhanced
Default Assumptions for Budgetary Cost Estimates, GAO-04-843 (Washington,
D.C.: Aug. 20, 2004).

the FDLP loans consolidated into FFELP, the government may need to pay SAP
that it otherwise would not have had to pay.

Availability of Additional Data for Both FFELP and FDLP Loans Have Enabled
Education to Refine Its Cash Flow Model

More data for both FFELP and FDLP loans has allowed Education to make
refinements to its cash flow model, a result of changes made by Education
to address recommendations in our prior reports and by Education's
auditors.10 The addition of data about borrower behavior to the cash flow
model has also contributed to the differences between reestimated and
original subsidy costs. For example, Education officials reported that in
recent years, data on FFELP and FDLP borrowers' use of deferment options,
which allow them to delay making payments on a loan when they return to
school or are experiencing economic hardship, has become available. With
this data Education is able to explicitly include in its model the number
of students using deferment options and project the effect on cash flows
in both FFELP and FDLP, rather than implicitly including deferments in its
model through adjustments in the length of time a loan was expected to be
in repayment. According to Education officials, more FFELP borrowers than
they had predicted have used deferment options and, when this data was
incorporated into FFELP's cash flow model, it contributed to an increase
in reestimated FFELP subsidy costs of $5 billion in fiscal year 2003.
Education reported that deferment data will be added to the FDLP cash flow
model and will be reflected in reestimated subsidy costs in the fiscal
year 2007 Budget of the United States Government.

Education also noted that more data has become available in FDLP because
the program has been in existence for 10 years and in FFELP because of
improvements made by guaranty agencies. Previously, Education had based
its FDLP cash flow assumptions on FFELP data, but Education now has data
on when borrowers default or enter repayment based on FDLP borrowers.
According to Education, actual defaults in FDLP have not been much
different from the assumptions made using FFELP data because defaults are
best predicted by the borrower and the type of school attended rather than
from which loan program the student borrowed. According to Education
officials, guaranty agencies-that are

10GAO, Department of Education's Federal Direct Loan Programs: Status of
Recommendations to Improve Cost Estimates and Presentation of Update Cash
Flow Information, GAO-04-567R (Washington, D.C.:. Mar. 29, 2004); and GAO,
Department of Education: Key Aspects of the Federal Direct Loan Program's
Cost Estimates,

GAO-01-197 (Washington, D.C.: Jan. 12, 2001).

responsible for reporting on the status of a loan, i.e., in repayment,
deferred, defaulted, or in-school-have made changes in their data systems
and the quality checks on the data. As a result, Education has been better
able to estimate default rates, subsequent collections, and their effect
on cash flows in FFELP. In particular, Education noted that there have
been improvements in the data Education uses in estimating of collections
of defaulted loans in both FFELP and FDLP, which showed higher than
originally estimated collections and contributed to lower reestimated
subsidy costs.

Certain Federal Costs and Revenues Associated with the Student Loan
Programs Are Not Included in Subsidy Cost Estimates

Additional federal costs and revenues associated with the student loan
programs, such as federal administrative expenses, some costs of risk
associated with lending money over time, and federal tax revenues
generated by both student loan programs are not included in subsidy cost
estimates. These are important factors to consider when determining costs
of the student loan programs; however, they are difficult to measure.
Under current law, federal administrative expenses are excluded from
subsidy cost estimates. In addition, subsidy cost estimates do not
explicitly include all risk that the government incurs by lending money
over time. Moreover, both loan programs generate federal tax revenues that
are not included in subsidy cost calculations.

Federal Administrative Expenses, by Law, Are Not Included in Subsidy Cost
Estimates

Under FCRA, federal administrative expenses are excluded from subsidy cost
estimates. Federal administrative expenses for the student loan programs
have been accounted for in Education's budget on a cash basis-showing how
much money is allocated for administering all federal student aid programs
in one fiscal year. The federal government is primarily responsible for
administering the FDLP and, for the most part, Education has contracted
with private-sector companies to perform administrative tasks, such as
originating and servicing loans. In the FFELP, lenders and guaranty
agencies perform administrative functions. In addition to the SAP paid to
lenders to guarantee a minimum yield, which includes coverage of the
administrative expenses incurred, Education pays guaranty agencies account
maintenance fees for their administrative costs. In fiscal year 2006,
Education requested $939 million for administrative expenses for all
federal student loan and grant aid programs. Of this amount, $238 million
was for FFELP administrative expenses and $388 million was for FDLP
administrative expenses.

When FCRA was first passed there were concerns about whether agencies
could change existing accounting systems to estimate long term

administrative expenses for a loan program. Over the last few years,
Education's Office of Federal Student Aid has been developing a system
that allocates its administrative expenses to each student aid program in
a particular fiscal year so that management would have information that
could be used for decision making purposes. While developing the system,
Education officials reported that some administrative expenses are clearly
linked to either FFELP or FDLP-such as payments to originate or service
FDLP loans, and servicing defaulted FFELP loans. However, other
administrative expenses are incurred by both loan programs, such as
information systems used to process financial aid applications, thus
requiring Education to develop a systematic way to allocate such expenses
to FFELP or FDLP.

In the fiscal year 2006 budget, Education included, as supplementary
information, modified cost estimates that included estimated
administrative expenses. As shown in table 5, if administrative expenses
are included, subsidy cost estimates for loans disbursed in fiscal year
2006 would increase by $1.45 per $100 disbursed in FDLP and by $0.69 per
$100 disbursed in FFELP.

Table 5: Comparison of Cost Estimates per $100 Disbursed in Fiscal Year
2006 with and without Administrative Expenses

Subsidy cost per Modified cost per $100 disbursed $100 disbursed including
administrative expenses

                               FFELP 11.96 12.65

                                FDLP -0.53 0.92

Source: Fiscal year 2006 Budget of the United States Government appendix,
p. 371.

To produce cost estimates that included administrative expenses, Education
not only needed to know how much of an expense was allocated to FDLP or
FFELP, but also had to project how such costs might change in the future
and whether an expense was paid now or later. For example, servicing costs
for an FDLP loan while the borrower is in-school are paid in the first
years that a loan is disbursed and are lower than the same costs when a
borrower is in repayment that are typically paid several years later.
According to Education, determining the timing of the expense was
important because expenses in later years were discounted and, therefore,
cost less in present value terms than those made in the first year.
Moreover, Education officials acknowledged that there are limitations with
these estimates because they assumed that administration of student aid
programs would remain the same in the future. They reported that there is
the possibility that administration processes and

functions will change based on legislative or technological changes, but
it was not possible to develop assumptions that could be used in
estimating the effects of any such changes.

Current Subsidy Estimates Do Not Incorporate All Risk Associated with
Lending Money Over Time

While current subsidy cost estimates account for some risks- uncertainties
regarding future cash flows-they do not include all risks incurred when
lending money over time. Among the risks borne by any lender are credit
risk-the possibility that the loan will not be fully repaid-and interest
rate risk-unanticipated fluctuations in the interest rate due to changes
in the economy that cause changes in the present value of the loans' cash
flows.11 Some studies have commented that by not incorporating all risks
in subsidy cost estimates, the government does not present an accurate
picture of the costs of its credit programs, including both FFELP and
FDLP.12 Risk can be reflected in subsidy cost estimates in different ways.
For example, one way is to incorporate it in estimates of cash flows, and
another way is to adjust the discount rate to reflect the risk.

Currently, Education incorporates some risks into its FFELP and FDLP
subsidy cost estimate model by explicitly adjusting cash flow estimates.
For example, credit risk is explicitly incorporated into Education's
subsidy cost model. Cash flow estimates are adjusted to reflect the
likelihood that borrowers will default on their loans based primarily on
the type of school a borrower attends (e.g., 2-year college, graduate
school, etc.). Interest rate risk, however, is not explicitly incorporated
into Education's model. Interest rate fluctuations can affect estimates of
SAP and borrower interest payments as well as borrower behavior with
respect to loan prepayment and consolidation. Although Education uses
estimated prepayment rates in adjusting estimated FFELP and FDLP cash
flows, these estimates are based on historical averages rather than an

11For example, the risk of a borrower prepaying a loan is a form of
interest rate risk. As discussed previously, when interest rates declined
below previously forecasted levels, increasing numbers of borrowers
prepaid their variable rate student loans by obtaining a consolidation
loan and obtained a low fixed rate of interest for the life of the
consolidation loan in doing so.

12For example, Congressional Budget Office, Estimating the Value of
Subsidies for Federal Loans and Loan Guarantees (Washington, D.C.: August
2004); Lucas, Phaup, and Prasad, Valuing Federal Loans and Loan
Guarantees: The Effect of Risk (October 2003); Zimmerman and Miles,
"Substituting Direct Government Lending for Guaranteed Student Loans: How
Budget Rules Distorted Economic Decisionmaking," National Tax Journal,
December 1994.

econometric forecast of how interest rates might fluctuate in the future
and, thereby, influence borrowers' decisions to prepay or consolidate
their loans. Relying on historical averages-especially if such averages do
not reflect a variety of interest rate environments and stable loan terms
and borrower characteristics-may not reflect the tendency for prepayments
to increase or decrease at times when it is advantageous for borrowers.

CBO and others have suggested that, rather than adjusting cash flows, the
discount rate could be changed to incorporate certain types of risk, such
as interest rate risk, in estimating subsidy costs of federal credit
programs. Currently, subsidy cost estimates calculate the net present
value of the loans using the "risk-free" discount rate determined by OMB
in accordance with FCRA, which reflects the government's cost of borrowing
funds. The rate is known as risk-free because an investor buying a U.S.
Treasury instrument knows with certainty what cash flows will be received
and when they will be received and there is assumed to be no probability
of default on the investment. This risk-free discount rate tends to be
relatively low compared to interest rates used to discount cash flows in
private industry, where interest rates reflect the market's valuation of
transactions and incorporate considerations of various types of risk. In a
2004 report, CBO proposed, among other methods, using a risk-adjusted
discount rate, rather than the risk-free rate, to estimate subsidy costs
of federal credit programs.13 In the case of federal student loans, one
way to calculate a risk-adjusted discount rate would be to evaluate the
secondary market for student loans, where student loans are often sold to
banks or other investors. However, there are limitations to this approach
given numerous differences in private-sector versus public sector
assessments of risk. Notwithstanding this, the market price of the student
loans would reflect the market's valuation of the loans, because the
expected cash flows would have been discounted using a higher discount
rate that incorporates risks-such as interest rate risk-that are not
included in Education's subsidy cost model. The present value (price) of
loans being sold on the secondary market would tend to be lower than the
government's valuation of similar loans, i.e., loans with similar default
risk, loan amount, time to repayment, and other factors. This difference
in loan valuation could be helpful in determining a risk-adjusted discount
rate to use in calculating the cost to the government, although
determining an appropriate rate would be challenging.

13Congressional Budget Office, p. 7.

Incorporating interest rate risk would affect subsidy cost estimates for
both credit programs, FFELP and FDLP. Modeling interest rate risk more
systematically through the cash flow estimates would affect prepayment and
interest payment projections under FDLP, as well as SAP projections and
prepayment activities under FFELP. The extent to which subsidy cost
estimates would change for FFELP and FDLP would depend on the interest
rate scenarios forecasted and the subsequent effect on cash flows in each
program. However, using a risk-adjusted discount rate would have a greater
impact on the subsidy cost estimates of FDLP relative to FFELP.14 This
difference would result, in part, because of differences in the amount and
timing of cash flows: FDLP has large cash outlays early in a loan's life
and large cash inflows later, when loans are in repayment. Thus these late
cash inflows would be discounted at a higher rate and would have a smaller
present value than under the current discounting methodology. FFELP, on
the other hand, generates some cash inflows to the government early while
cash outflows occur later as loans default or when SAP payments, if any,
are made.

Private-Sector Activity in Both Student Loan Programs Generates Tax
Revenues for the Federal Government That Are Not Included in Subsidy Cost
Estimates

Both FFELP and FDLP generate federal tax revenues that are reflected in
the revenue portion of the budget but are not included in subsidy cost
calculations. Federal tax revenues are generated by a variety of sources,
including private-sector lenders that account for a majority of the
lenders that make or hold FFELP loans. Many of these lenders participate
actively in the multi-billion dollar financial services industry of
taxable and taxexempt bonds, asset-backed securities, and other debt
instruments and pay federal taxes on the income earned from these sources
as well as from their student loan business. In addition, other
private-sector companies that work with FFELP lenders and investors buying
student loan bonds and securities also generate federal tax revenues from
the income earned from their participation in FFELP. Moreover, to service
and collect defaulted FFELP loans, Education contracts with private-sector
companies that are another source of federal tax revenue.

Although FDLP is financed and primarily administered by the federal
government, Education contracts with private-sector companies for many key
administrative tasks, such as servicing loans while borrowers are in

14For purposes of making subsidy cost estimates for the budget, future
cash flows are converted, using a discount rate into their "money now"
equivalents to reflect the time value of money.

Concluding Observations

school, repayment, or default. In fiscal year 2004 Education reported that
it paid $321 million to private-sector contractors to service student
loans and perform other administrative tasks in the FDLP. These
private-sector contractors earn income from their participation in FDLP on
which they may pay federal taxes. Another source of tax revenue is income
tax paid by U.S. investors that hold Treasury securities used to finance
FDLP loans.

Estimating the dollar amount of federal tax revenues generated by private
sector entities and investors in FFELP and FDLP would be challenging. For
example, many lenders are large publicly traded financial services
companies with student loans being one portion of their business, making
it difficult to identify the tax revenue generated from their student loan
business. Moreover, to make an estimate of tax revenues would require
knowledge of each lender's profits from its student loan business and
applicable tax rates.

Significant reestimates of subsidy costs over the past 10 years illustrate
the challenges of estimating the lifetime costs of loans. As we have
shown, subsidy cost estimates and reestimates are sensitive to the
assumptions used in estimating these costs. The historically low interest
rates that persisted over the last several years were below levels
previously forecasted. Because cost estimates for FFELP and especially for
FDLP loans are sensitive to changes between projected and actual interest
rates, subsidy cost reestimates varied from original estimates. To the
extent that current assumptions correctly predict future loan performance
and interest rates, subsidy costs per $100 of FFELP loans made from fiscal
years 1994 to 2004 will be, in many cases, less costly than originally
anticipated. On the other hand, over the same time period, subsidy costs
per $100 of FDLP loans will in many cases be higher than originally
anticipated.

FDLP subsidy costs per $100 of loans disbursed have, in general, remained
lower than those of FFELP. Nonetheless, if current assumptions correctly
predict future loan performance and economic conditions, the originally
estimated gain to the government from FDLP loans made in fiscal years 1994
to 2004 will not materialize, and instead these loans will result in a net
cost to the government. In reality, however, subsidy cost estimates of
FFELP and FDLP loans made in fiscal years 1994 to 2004 will continue to
change as future reestimates incorporate actual experience and new
interest rate forecasts. Similarly, initial subsidy cost estimates for
loans made in the future will also change over the life of these loans and
at times be lower or higher than initially estimated, depending on the
extent to

which loan performance and interest rates differ from assumptions used to
develop initial estimates. Actual subsidy costs for a cohort of student
loans will remain unknown until all payments that will be made on such
loans have been collected.

Despite the fact that subsidy cost estimates will change from year to
year, estimates developed in accordance with FCRA more fully and
accurately present the expected long-term costs of federal student loans
than did the prior method of calculating costs based on single-year cash
flows to and from the government. As a result of FCRA, the budget is a
more useful tool for allocating resources among the myriad of competing
demands for federal dollars than it once was. Subsidy cost estimates, for
example, provide policymakers the means to more accurately evaluate the
long-term budgetary implications of potential legislative, regulatory, and
administrative reforms. At the same time, it is important for policymakers
to understand how credit reform subsidy cost estimates are developed and
to recognize that such estimates will change in the future. Decisions made
in the short-term on the basis of these estimates can have long-term
repercussions for the fiscal condition of the nation.

While subsidy cost estimates include many of the federal costs associated
with FFELP and FDLP loans, they do not capture all federal costs and
revenues associated with the loan programs. Consideration of all federal
costs and revenues of the loan programs would be an important component of
a broader assessment of the costs and benefits of the two programs.
Because federal administrative expenses-in accordance with FCRA-are
excluded from subsidy cost estimates, for example, these estimates can
underestimate the total lifetime costs of FFELP and FDLP loans. Other
costs and revenues are also not considered in subsidy costs estimates,
including interest rate risk inherent to lending programs, and federal tax
revenues generated by private-sector activity in both FFELP and FDLP.
Calculations of total federal costs would be enhanced were these
additional costs and revenues considered, though doing so may require
complex methodologies and/or data that are not currently readily
available.

Agency Comments 	We provided Education with a copy of our draft report for
review and comment. Education reviewed the report and had no comments.
Education noted that because the report did not include recommendations
for the Department, it was not providing a formal response to be included
in the report.

As agreed with your offices, unless you publicly announce its contents
earlier, we plan no further distribution of this report until 30 days from
its
date. At that time we will send copies of this report to the Secretary of
Education, appropriate congressional committees, and other interested
parties. We will also make copies available to others 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 staff have any questions about this report, please contact
me
at (202) 512-7215 or [email protected]. Contact points for our Offices of
Congressional Relations and Public Affairs may be found on the last page
of this report. Key contributors to the report are listed in appendix II.

Cornelia M. Ashby
Director, Education, Workforce,

and Income Security Issues

List of Congressional Committees

The Honorable Michael B. Enzi
Chairman
Committee on Health, Education, Labor, and Pensions
United States Senate

The Honorable Judd Gregg
Chairman
Committee on the Budget
United States Senate

The Honorable John A. Boehner
Chairman
Committee on Education and the Workforce
House of Representatives

The Honorable Tom Davis
Chairman
Committee on Government Reform
House of Representatives

The Honorable Pete Hoekstra
Chairman
Permanent Select Committee on Intelligence
House of Representatives

The Honorable Howard P. "Buck" McKeon
Chairman
Subcommittee on 21st Century Competitiveness
Committee on Education and the Workforce
House of Representatives

The Honorable Jim Nussle
Chairman
Committee on the Budget
House of Representatives

Appendix I: Comparison of Fiscal Year 2006 FDLP and FFELP Reestimated Subsidy
Costs per $100 Disbursed, by Loan Type and Cohort

Figure 10: Comparison of Reestimated Subsidy Costs for Subsidized Stafford
Loans in FFELP and FDLP, by Loan Cohort

Subsidy cost per $100 disbursed (nominal dollars) 25

20

15

10

5

0 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 Loan cohort

FFELP reestimated subsidy cost FDLP reestimated subsidy cost Source: GAO
analysis of the fiscal year 2006 Budget of the United States Government
credit supplement tables 7 and 8.

Appendix I: Comparison of Fiscal Year 2006 FDLP and FFELP Reestimated
Subsidy Costs per $100 Disbursed, by Loan Type and Cohort

Figure 11: Comparison of Reestimated Subsidy Costs for Unsubsidized
Stafford Loans in FFELP and FDLP, by Loan Cohort

Subsidy cost per $100 disbursed (nominal dollars)

6

4

2

0

-2

-4

-6

-8

1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004

Loan cohort

FFELP reestimated subsidy cost FDLP reestimated subsidy cost Source: GAO
analysis of the fiscal year 2006 Budget of the United States Government
credit supplement tables 7 and 8.

Appendix I: Comparison of Fiscal Year 2006 FDLP and FFELP Reestimated
Subsidy Costs per $100 Disbursed, by Loan Type and Cohort

Figure 12: Comparison of Reestimated Subsidy Costs for PLUS Loans in FFELP
and FDLP, by Loan Cohort

Subsidy cost per $100 disbursed (nominal dollars) 3

2

1

0

-1

-2

-3

-4

-5

-6

1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004

Loan Cohort

FFELP reestimated subsidy cost FDLP reestimated subsidy cost Source: GAO
analysis of the fiscal year 2006 Budget of the United States Government
credit supplement tables 7 and 8.

Appendix I: Comparison of Fiscal Year 2006 FDLP and FFELP Reestimated
Subsidy Costs per $100 Disbursed, by Loan Type and Cohort

Figure 13: Comparison of Reestimated Subsidy Costs for Consolidation Loans
in FFELP and FDLP, by Loan Cohort

Subsidy cost per $100 disbursed (nominal dollars)
20

15

10

5

0

-5

-10
1995 1996 1997 1998 1999 2000 2001 2002 2003 2004

Loan cohort

FFELP reestimated subsidy cost

FDLP reestimated subsidy cost

Source: GAO analysis of the fiscal year 2006 Budget of the United States
Government credit supplement tables 7 and 8.

Appendix II: GAO Contacts and Staff Acknowledgments

GAO Contact

Staff Acknowledgments

(130429)

Cornelia M. Ashby (202) 512-7215

The following individuals made important contributions to the report:
Jeff Appel, Assistant Director; Andrea Sykes, Analyst-in-Charge,
Nagla'a El-Hodiri, Jeffrey W. Weinstein, Christine Bonham, Marcia Carlsen,
Austin Kelly, Mitch Rachlis and Lauren Kennedy.

GAO's Mission

Obtaining Copies of GAO Reports and Testimony

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.

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, Contact:

Waste, and Abuse in Web site: www.gao.gov/fraudnet/fraudnet.htm

E-mail: [email protected] Programs Automated answering system: (800)
424-5454 or (202) 512-7470

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

                           PRINTED ON RECYCLED PAPER
*** End of document. ***