GAO-16-41, CREDIT REFORM: Current Method to Estimate Credit Subsidy Costs Is More Appropriate for Budget Estimates Than a Fair Value Approach

CREDIT REFORM

Current Method to

Estimate Credit

Subsidy Costs Is

More Appropriate for

Budget Estimates

Than a Fair Value

Approach

Report to the Ranking Member,

Subcommittee on Financial Services

and General Government, Committee

on Appropriations, U.S. Senate

January 2016

GAO-16-41

United States Government Accountability Office

Highlights of GAO-16-41, a report to the

Ranking Member, Subcommittee on Financial

Services and General Government,

Committee on Appropriations,

U.S. Senate

January 2016

CREDIT REFORM

Current Method to Estimate Credit Subsidy Costs Is

More Appropriate for Budget Estimates

Than a Fair

Value App

roach

What GAO Found

The Federal Credit Reform Act of 1990 (FCRA) requires agencies to estimate the

cost to the government of extending or guaranteeing credit. This cost, referred to

as subsidy cost, equals the net present value of estimated cash flows from the

government (e.g., loan disbursements and claim payments to lenders) minus

estimated cash flows to the government (e.g., loan repayments, interest

payments, fees, and recoveries on defaulted loans) over the life of the loan,

excluding administrative costs. Discount rates that reflect the federal

government’s cost of financing are used to determine the net present value of

estimated cash flows. Agencies generally update—or reestimate—subsidy costs

annually to reflect both actual loan performance and changes in expected future

loan performance.

Based on GAO’s analyses of credit program reestimates for direct loans and loan

guarantees obligated or committed from fiscal years 2001 through 2014 and

considering various factors to identify trends, GAO did not identify any overall

consistent trends in under- or overestimates of subsidy costs across federal

credit programs government-wide. Overall, both direct loan and loan guarantee

programs government-wide underestimated costs by $3.1 billion and

$39.0 billion, respectively, over the 14-year period. These amounts represent

less than 1 percent of the amounts disbursed or guaranteed during the period.

Annual reestimates fluctuated from year to year, indicating both under- and

overestimates of subsidy costs. Further, significant lifetime reestimates could

generally be explained by specific events affecting a few large programs. For

example, the Department of Housing and Urban Development’s Mutual Mortgage

Insurance Program reported underestimating costs over this period because of a

variety of factors, including long-term housing prices and interest rate changes

stemming from the mortgage and financial crises in the late 2000s.

Fluctuations in Direct Loan and Loan Guarantee Programs’ Annual Net Reestimates, 2006-

2014

Why GAO Did This Study

Federal direct loans and loan

guarantees outstanding have nearly

doubled from $1.5 trillion at the end of

fiscal year 2008 to $2.9 trillion at the

end of fiscal year 2014. For the past

several years, concerns have been

raised by some experts both in and out

of the federal government that FCRA

may understate credit program subsidy

costs. Some of these experts have

suggested that FCRA be modified with

an approach—referred to as the fair

value approach—to include certain

market risk not currently considered

under FCRA.

GAO was asked to examine the

budgetary treatment of the cost of

federal credit programs. This report

addresses (1) whether trends exist in

subsidy cost reestimates and what

factors, if any, help explain any

significant trends in reestimates and

(2) the implications of using the fair

value approach to estimate subsidy

costs in the budget and whether GAO

believes such concepts should be

incorporated into subsidy cost

estimates for the budget.

GAO analyzed reestimate data from

fiscal years 2001 to 2014 as reported

in the President’s Budgets and

conducted interviews with 30 experts.

What GAO Recommends

GAO supports maintaining the current

FCRA method for estimating credit

subsidy cost for the budget and

therefore is not making any

recommendations. The Congressional

Budget Office and the Office of

Management and Budget provided

technical comments on a draft of this

report, which have been incorporated

as appropriate.

View GAO-16-41. For more information,

contact Cheryl E. Clark at (202) 512-9377 or

clarkce@gao.gov, Susan J. Irving at (202)

512-6806 or [email protected], or Susan Offutt

at (202) 512-3763 or [email protected].

Highlights of GAO-16-41 (Continued)

United States Government Accountability Office

While subsidy cost estimates under the fair value approach may provide useful information to decision makers for

evaluating the costs against the benefits of credit programs, GAO does not support the use of the fair value approach to

estimate subsidy costs for the budget. Proponents of the fair value approach have asserted that beyond the cash flows

associated with a direct loan or loan guarantee, costs are imposed on taxpayers who would, in a similarly risky private

market transaction, require compensation for bearing the risk associated with making the loan or guarantee. Taxpayers as

investors with diversified portfolios would still demand compensation, or a premium, for bearing the risk that the

macroeconomy—the national or global economy—may falter. This risk—referred to as aggregate risk (a portion of overall

market risk)—arises from the possibility of significant economic downturns, when even a well-diversified portfolio of

financial investments will decrease in value. To incorporate the cost of bearing aggregate risk into subsidy cost estimates

for the budget, the fair value approach adds an aggregate risk premium to the discount rate used in FCRA calculations,

which is based on interest rates of Treasury securities. Including the aggregate risk premium incorporates a noncash cost

into the subsidy cost estimate. The actual cash flows to and from the federal government associated with a credit program

are the same under the fair value approach and FCRA. The debate over the fair value approach rests on whether the cost

associated with aggregate risk should be considered in the subsidy cost estimates for the budget of the federal

government.

Differences between Market and Treasury Interest Rates

Reflecting a different concern, some proponents of the fair value approach cited as motivation the perceived overreliance

on federal credit programs as a policy tool and the desire to correct any bias toward underestimates of costs under FCRA.

Raising the subsidy cost would likely result in fewer loans being made. In contrast, some proponents of FCRA stated that

any overreliance on credit programs should be addressed as a policy decision, and that to the extent that agencies were

underestimating subsidy costs under FCRA, improvements in the subsidy estimation process should be pursued.

The additional market risk recognized under the fair value approach does not reflect additional cash costs beyond those

already recognized by FCRA. The introduction of market risk into subsidy costs under the fair value approach would

(1) be inconsistent with long-standing federal budgeting practices primarily based on cash outlays; (2) be inconsistent with

the budgetary treatment of similarly risky programs; (3) introduce transparency and verification issues with respect to

inclusion of a noncash cost in budget totals; and (4) involve significant implementation issues, such as the need for

additional agency resources. Consequently, GAO does not support the use of the fair value approach to estimate subsidy

costs for the budget and believes the current FCRA methodology is more appropriate for this purpose as it represents the

best estimate of the direct cost to the government and is consistent with current budgetary practices.

Page i GAO-16-41 Credit Subsidy Cost Estimates

Letter 1

Background 4

o Overall Consistent Trends Identified in Under- or

Overestimates of Subsidy Costs, and Various Economic and

Portfolio Changes Caused Certain Significant Reestimates 16

Fair Value Subsidy Cost Estimates May Be Useful, but They Are

Not Consistent with Long-standing Federal Budgeting

Practices, Involve Significant Implementation Challenges, and

Should Not Be Recognized in Budget Costs 33

Concluding Observations 53

Agency Comments 53

Appendix I Objectives, Scope, and Methodology 55

Appendix II Analysis of Reestimates 59

Appendix III Fair Value Approach Implementation Considerations 66

Appendix IV GAO Contacts and Staff Acknowledgments 73

Related GAO Products 74

Tables

Table 1: Direct Loan Programs Lifetime Upward and Downward

Reestimates, 2001-2014 18

Table 2: Loan Guarantee Programs Lifetime Upward and

Downward Reestimates, 2001-2014 20

Table 3: Discretionary and Mandatory Credit Program

Reestimates, 2001-2014 23

Table 4: Direct Loan Programs’ Lifetime Reestimates by Agency,

2001-2014 60

Table 5: Loan Guarantee Programs’ Lifetime Reestimates by

Agency, 2001-2014 61

Contents

Page ii GAO-16-41 Credit Subsidy Cost Estimates

Table 6: Lifetime Upward and Downward Reestimates by Type of

Direct Loan Program, 2001-2014 63

Table 7: Lifetime Upward and Downward Reestimates by Type of

Loan Guarantee Program, 2001-2014 63

Figures

Figure 1: Federal Direct Loans and Loan Guarantees Outstanding,

2001-2014 5

Figure 2: Agencies with the Highest Direct Loans and Loan

Guarantees Outstanding as of Fiscal Year 2014 and

Types of Credit Programs 6

Figure 3: Calculation of Subsidy Costs for Direct Loans and Loan

Guarantees 8

Figure 4: Federal Credit Reform Act of 1990 Program and

Financing Account Transactions for Direct Loans and

Loan Guarantees 11

Figure 5: Effect of Discount Rates on the Value of Loan

Repayments and Subsidy Costs for a Direct Loan 12

Figure 6: Direct Loan Programs Representing a Significant

Percentage of Lifetime Upward and Downward

Reestimates, 2001-2014 19

Figure 7: Loan Guarantee Programs Representing a Significant

Percentage of Lifetime Upward and Downward

Reestimates, 2001-2014 21

Figure 8: Fluctuations in Direct Loan and Loan Guarantee

Programs’ Annual Net Reestimates, 2006-2014 22

Figure 9: Primary Drivers of the Largest Overall Upward and

Downward Lifetime Reestimates for Direct Loans and

Loan Guarantees by Purpose of Program, 2001-2014 25

Figure 10: 2008 Direct Loan and Loan Guarantee Programs’

Reestimates during the Financial Crisis, 2006-2014 32

Figure 11: Aggregate Risk 35

Figure 12: Differences between Market and Treasury Interest

Rates 37

Figure 13: Fair Value Costs versus Cash Cost over Time 42

Figure 14: Percentage Point Differences between Direct Loan and

Loan Guarantee Programs’ Original and Most Recent

Reestimated Subsidy Rates, 2001-2014 65

Page iii GAO-16-41 Credit Subsidy Cost Estimates

Abbreviations

CBO Congressional Budget Office

DOE Department of Energy

Ex-Im Export-Import Bank

FASAB Federal Accounting Standards Advisory Board

FCRA Federal Credit Reform Act of 1990

FCS Federal Credit Supplement

FFEL Federal Family Education Loan

GAAP generally accepted accounting principles

GSE Government Sponsored Enterprises

HUD Department of Housing and Urban Development

MBS Mortgage Backed Securities

MMI Mutual Mortgage Insurance

OMB Office of Management and Budget

SBA Small Business Administration

SBIC Small Business Investment Company

TARP Troubled Asset Relief Program

Treasury Department of the Treasury

USDA Department of Agriculture

VA Department of Veterans Affairs

This is a work of the U.S. government and is not subject to copyright protection in the

United States. The published product 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.

Page 1 GAO-16-41 Credit Subsidy Cost Estimates

441 G St. N.W.

Washington, DC 20548

January 29, 2016

The Honorable Christopher A. Coons

Ranking Member

Subcommittee on Financial Services and General Government

Committee on Appropriations

United States Senate

Dear Senator Coons:

The federal government uses credit programs that extend direct loans

and loan guarantees as tools to support specific social and public policy

objectives, such as those for housing, education, and small businesses.

Twenty-five years ago, the enactment of the Federal Credit Reform Act of

1990 (FCRA) changed the method used to budget for the cost of federal

credit programs.

Before fiscal year 1992, when FCRA took effect, the

cost of credit programs was recorded in the budget on a cash basis (the

expected amount of cash paid out minus the cash received in a given

year). As a result, the budget cost associated with a loan guarantee was

not recorded until a default occurred, which may have been many years

after the guarantee was made. Further, direct loans appeared to cost the

same as grants because the total amount of a loan was recorded as a

cost when the loan was made and loan repayments were not recorded

until the year received. Under FCRA, the budget records the federal

government’s estimated net long-term cost—referred to as the subsidy

cost—in the year the direct loan or loan guarantee is made. Agencies

generally update—or reestimate—these subsidy costs annually to reflect

both actual loan performance and changes in expected future loan

performance, which could be based on economic changes.

For the past several years, concerns have been raised by experts both in

and out of the federal government that subsidy costs may be

underestimated under FCRA procedures. Some of these experts have

suggested that FCRA be modified to include an approach—referred to as

the fair value approach in this report—that would account for certain

Pub. L. No. 101-508, § 13201(a), 104 Stat. 1388, 1388-609 (Nov. 5, 1990), classified, as

amended, at 2 U.S.C. §§ 661-661f.

Letter

Page 2 GAO-16-41 Credit Subsidy Cost Estimates

market risk not currently considered in FCRA subsidy cost estimates.

Specifically, taxpayers as investors with diversified portfolios would still

demand compensation, or a premium, for bearing the risk that the

macroeconomy—the national or global economy—may falter. This risk—

referred to as aggregate risk (a portion of overall market risk)—arises

from the possibility of significant economic downturns, when even a well-

diversified portfolio of financial investments will decrease in value. The

fair value approach would increase initial subsidy cost estimates for direct

loan and loan guarantee programs because of the added market risk. As

a result, because of the higher estimated initial subsidy cost, less federal

credit would be available, assuming the same level of spending was

provided for in the budget. Other experts, both in and out of the federal

government, did not agree that the fair value approach would be

beneficial in estimating credit subsidy costs for the budget.

Since the 2008 financial crisis, the amount of federal credit outstanding,

consisting of direct loans and loan guarantees, has nearly doubled from

$1.5 trillion at the end of fiscal year 2008 to $2.9 trillion at the end of fiscal

year 2014. In light of this growing portfolio of outstanding direct loans and

loan guarantees, as well as concerns about underestimates of subsidy

costs and the suggestions to modify FCRA, you asked us to review

issues related to the budgetary treatment of the cost of federal credit

programs. Our objectives were to determine (1) the extent to which trends

exist in the size and direction of subsidy cost reestimates across, or

within, federal credit programs and, based on this analysis of reestimates,

what factors, if any, help explain any significant trends in reestimates and

(2) the implications of using subsidy cost estimates developed under the

fair value approach in the budget and whether we believe such concepts

should be incorporated into subsidy cost estimates for the budget.

Also

at your request, we will issue a follow-up report addressing the factors

agencies should consider when developing subsidy cost estimates and to

what extent selected agencies are using those factors.

Market risk is the potential for loss resulting from movements in market prices, including

interest rates, commodity and stock prices, and foreign exchange rates.

For purposes of this report, “budget” includes the development and consideration of the

President’s Budget; congressional budget resolutions, allocations, and appropriations; and

compliance with budget controls.

Page 3 GAO-16-41 Credit Subsidy Cost Estimates

To analyze trends in subsidy cost reestimates, we used direct loan and

loan guarantee reestimate data for fiscal years 2001 through 2014, which

are reported in the fiscal years 2003 through 2016 President’s Budgets as

presented in the Federal Credit Supplement.

We categorized these

programs by seven loan purposes (e.g., housing or education) and by the

type of budget spending. We examined reestimate trends by agency,

program, purpose, cohort fiscal year (the fiscal year of obligation for direct

loans or commitment for loan guarantees), and the type of budget

spending (discretionary or mandatory).

We focused on identifying trends

in lifetime reestimate amounts,

annual reestimate amounts,

and a

comparison of original subsidy rates to reestimated subsidy rates. We

encountered some limitations with the data reported in the President’s

Budgets, which we individually evaluated based on professional

judgment. For each data limitation, we adjusted the data, identified

alternative data calculation methods when available, or determined that

the data were reliable for our purposes. Our conclusions about possible

trends were based on our professional judgment in assessing the data

and not based on a statistical analysis.

To evaluate implications of using

subsidy cost estimates developed under the fair value approach for the

budget, we reviewed literature related to FCRA, the purpose and uses of

The Federal Credit Supplement is supplementary material issued along with the federal

budget. It provides summary information about federal direct loan and loan guarantee

programs subject to FCRA. It includes information related to estimated (1) original subsidy

rates, obligations, commitments, and average loan sizes for future cohorts; (2) certain loan

terms and assumptions underlying original subsidy rates; and (3) annual and lifetime

reestimates, as well as loan disbursements to date for existing cohorts. (A cohort refers to

the fiscal year of obligation for direct loans or commitment for loan guarantees.)

Funding for discretionary spending programs is provided in appropriations acts.

Mandatory programs are those programs whose funding is provided for in laws other than

appropriations acts. In general, mandatory credit programs are entitlement programs for

which the amount of funding depends on eligibility and benefits rules contained in law.

The term lifetime reestimate refers to total reestimates reported for a cohort of loans

since inception of the cohort. Lifetime reestimates are presented by program, agency, or

type of direct loan or loan guarantee program for cohorts over a given period of time. For

example, if an agency recorded annual reestimates of $100, $50, and $20 over a 3-year

period for a cohort, the lifetime reestimate would be $170 ($100 + $50 + $20).

The term annual reestimates refers to reestimates of a cohort of loans or a group of

cohorts for a given fiscal year.

We did not use statistical analysis to estimate trends because, in our professional

judgment, the quantity of data (e.g., number of years and number of programs) was

insufficient to provide a credible statistical result.

Page 4 GAO-16-41 Credit Subsidy Cost Estimates

the federal budget, and the fair value approach. We also conducted

semistructured interviews with 30 individuals with expertise in one or

more of the following: FCRA; financial economics; and federal budgeting,

auditing, and accounting. See appendix I for additional details on our

scope and methodology.

We conducted this performance audit from June 2014 to January 2016 in

accordance with generally accepted government auditing standards.

Those standards require that we plan and perform the audit to obtain

sufficient, appropriate evidence to provide a reasonable basis for our

findings and conclusions based on our audit objectives. We believe that

the evidence obtained provides a reasonable basis for our findings and

conclusions based on our audit objectives.

Federal credit programs provide assistance to borrowers through two

principal methods: direct loans and loan guarantees. Direct loans are a

disbursement of funds by the government to a nonfederal borrower under

a contract that requires the repayment of such funds with or without

interest. Loan guarantees are any guarantees, insurance, or other

pledges with respect to the payment of all or a part of the principal or

interest on any debt obligation of a nonfederal borrower to a nonfederal

lender.

As shown in figure 1, since the 2008 financial crisis, the amount

of federal direct loans and loan guarantees outstanding has nearly

doubled from $1.5 trillion at the end of fiscal year 2008 to $2.9 trillion at

the end of fiscal year 2014.

Loan guarantees do not include the insurance of deposits, shares, or other withdrawable

accounts in financial institutions. Further, the credit and insurance activities of the Federal

Deposit Insurance Corporation, National Credit Union Administration, Pension Benefit

Guarantee Corporation, National Flood Insurance Program, National Insurance

Development Fund, Crop Insurance, and Tennessee Valley Authority are statutorily

exempted from FCRA and are therefore not considered “credit programs” for budgetary

purposes. 2. U.S.C. § 661e(a).

The amount for direct loans is the total face value outstanding and for loan guarantees

is the principal amount outstanding of the underlying loans that are guaranteed by the

federal government.

Background

Page 5 GAO-16-41 Credit Subsidy Cost Estimates

Figure 1: Federal Direct Loans and Loan Guarantees Outstanding, 2001-2014

As of September 30, 2014, 21 federal agencies reported that they had

direct loans or loan guarantees outstanding. As shown in figure 2, the

Department of Education (Education) (largely through its direct student

loan program) and the Department of Housing and Urban Development

(HUD) (largely through its single-family mortgage guarantee program)

held the largest share of federal direct loans and loan guarantees

outstanding, respectively.

Page 6 GAO-16-41 Credit Subsidy Cost Estimates

Figure 2: Agencies with the Highest Direct Loans and Loan Guarantees Outstanding as of Fiscal Year 2014 and Types of

Credit Programs

Page 7 GAO-16-41 Credit Subsidy Cost Estimates

Prior to the enactment of FCRA, credit programs—like most other federal

programs—were recorded in budgetary accounts on a cash basis (the

expected amount of cash paid out minus the cash received in a given

year). Because a loan guarantee does not require a cash outlay at the

time the guarantee is issued, guarantees initially appeared to be of no

cost to the federal budget; conversely, because the entire amount of a

direct loan is disbursed and recognized as a budget cost when the loan is

made, the cost of direct loans was recorded the same as grants in the

federal budget. Both were inaccurate and provided policymakers with

distorted information for comparing credit programs to noncredit programs

and to each other. This created a bias in favor of loan guarantees over

direct loans because loan guarantees appeared to be less expensive than

direct loans regardless of the actual lifetime cost to the government.

FCRA was enacted with the intent of improving the accuracy of the cost

of federal credit programs reported in the budget by requiring agencies to

measure the government’s net long-term cost of federal credit programs

to permit better cost comparisons both among credit programs and

between credit and noncredit programs. The policies enacted under

FCRA, sometimes referred to simply as credit reform, recognized that the

actual cost of a direct loan or loan guarantee was not captured by its cash

flows in any one year, but rather is the net present value—worth in terms

of money paid immediately—of its cash flows over the life of the loan.

Therefore, FCRA specified an approach using estimates of expected

cash flows, including future loan repayments and defaults as elements of

the cost to be recorded in the budget.

FCRA requires agencies to estimate the cost to the government of

extending or guaranteeing credit. This cost, referred to as subsidy cost,

equals the net present value of estimated cash flows from the

government minus estimated cash flows to the government over the life of

the loan and excluding administrative costs. This approach puts direct

loans and loan guarantees on an equal footing in terms of cost in the

federal budget and permits the costs of credit programs to be compared

with each other and with the costs of noncredit programs. Figure 3

The present value of a stream of future returns or costs is its worth in terms of money

paid immediately. In calculating present value under FCRA, prevailing interest rates on

Treasury securities provide the basis for converting future amounts into their “money now”

equivalents.

The Federal Credit Reform

Act of 1990

Subsidy Cost Estimates

and Reestimates

Page 8 GAO-16-41 Credit Subsidy Cost Estimates

illustrates the types of cash flows that affect the subsidy cost for direct

loans and loan guarantees. If the present value of estimated cash

outflows exceeds cash inflows, there is a subsidy cost. If the present

value of estimated cash inflows exceeds cash outflows, there is a

negative subsidy cost, referred to as subsidy income.

Figure 3: Calculation of Subsidy Costs for Direct Loans and Loan Guarantees

Page 9 GAO-16-41 Credit Subsidy Cost Estimates

To calculate subsidy costs, agencies begin by estimating the expected

cash outflows and inflows over the life of the loans for each cohort of

direct loans obligated or loan guarantees committed in the cohort year.

Agencies use historical information and various assumptions, including

the probabilities of default, borrower prepayments, or recoveries, and the

projected timing of these events, to make informed predictions about

expected future cash flows. These expected cash flows are then

discounted to the point of loan disbursement to determine the net present

value. A credit subsidy rate—the cost per dollar of credit assistance—is

then determined by dividing the subsidy cost by direct loan obligations or

loan guarantee commitments estimated to be made in that year.

The subsidy cost represents the net present value cost of making or

guaranteeing new loans and is included in the President’s Budget. FCRA

requires that agencies have budget authority to cover a program’s

subsidy cost to the government in advance—before new direct loan

obligations are incurred and new loan guarantee commitments are made.

The subsidy cost of credit programs may be categorized as either

discretionary or mandatory. Discretionary spending is provided for

through annual appropriations acts, typically in definite amounts.

Mandatory spending, however, is provided for in laws other than

appropriations acts. Often, these mandatory programs are entitlement

programs wherein the Congress controls spending indirectly by defining

eligibility and setting payment rules, with indefinite budget authority

provided by law to pay for the entitlements of all qualifying beneficiaries.

The data used for budgetary subsidy cost estimates are generally

updated—or reestimated—annually after the end of the fiscal year to

reflect actual loan performance and to incorporate any changes in

assumptions about future loan performance. Reestimates that increase

subsidy costs are referred to as upward reestimates (an agency would

need additional funds), while reestimates that decrease subsidy costs are

referred to as downward reestimates (an agency would return funds).

Regardless of whether the credit programs are discretionary or

mandatory, agencies do not need to request additional appropriations to

cover upward reestimates because FCRA provides permanent indefinite

Cohort refers to the fiscal year of obligation for direct loan obligations or loan guarantee

commitments of a program.

Page 10 GAO-16-41 Credit Subsidy Cost Estimates

budget authority for this purpose.

Accordingly, an upward reestimate

does not use up room under any discretionary spending caps and a

downward reestimate does not “free up” room under such caps.

FCRA established a special budgetary accounting system to record the

cash flows and budget information necessary to implement credit reform.

Direct loans and loan guarantees made on or after October 1, 1991—the

effective date of credit reform—use program and financing accounts to

handle credit transactions (see fig. 4).

The program account is included

in budget totals, receives appropriations for the subsidy cost of a credit

program, and records the budget authority and outlays for these costs.

The program account is used to pay the associated subsidy cost to the

financing account when a direct or guaranteed loan is disbursed. The

financing account, which is nonbudgetary,

is used to (1) collect the

subsidy cost from the program account, (2) borrow from the Department

of the Treasury (Treasury) to provide financing for loan disbursements,

and (3) record the cash flows between the government and the borrower

or lender associated with direct loans or loan guarantees over the life of

the loan. These cash flows include loan disbursements, default payments

to lenders, loan repayments, interest payments, recoveries on defaulted

loans, and fee collections.

Permanent indefinite budget authority is available for obligation and expenditure without

fiscal year limitation and is not limited to a specified amount or ceiling.

A liquidating account was also established to handle credit transactions on a cash basis

for direct loans and loan guarantees made before the implementation of FCRA.

Funding for administrative costs of credit programs varies by program, depending on the

language of the relevant authorizing statutes and appropriations acts. In some cases,

administrative costs are provided for as part of appropriations made to the program

account, while other programs charge their administrative costs to other accounts, such as

an agency’s general lump-sum appropriation.

Transactions from nonbudgetary accounts do not belong in the budget because they do

not represent net budget authority or outlays, but rather are a means of financing. This

contrasts with “off-budget,” which refers to activities that are budgetary in nature but are

required by law to be excluded from budget totals.

Program and Financing

Accounts under FCRA

Page 11 GAO-16-41 Credit Subsidy Cost Estimates

Figure 4: Federal Credit Reform Act of 1990 Program and Financing Account Transactions for Direct Loans and Loan

Guarantees

General Fund accounts in the U.S. Treasury hold all federal money not allocated by law to any other

fund account.

When calculating the net present value of expected cash flows, FCRA

requires that discount rates be based on interest rates of marketable U.S.

Treasury securities with similar maturities as the cash flows being

discounted. In addition, FCRA requires that the rate of interest charged

on financing account transactions with Treasury be the same as the final

discount rate used to calculate the net present value of cash flows when

Discount Rates under

FCRA

Page 12 GAO-16-41 Credit Subsidy Cost Estimates

estimating the subsidy cost of a credit program.

Figure 5 illustrates how

the discount rate would affect the value of future loan repayments and the

subsidy cost for a direct loan.

Figure 5: Effect of Discount Rates on the Value of Loan Repayments and Subsidy Costs for a Direct Loan

Note: This figure illustrates a simplified example of an interest-free loan and assumes that it performs

as required with no defaults

For loans made or guaranteed in fiscal year 2001 and thereafter, the discount rate is

based on interest rates on marketable zero-coupon Treasury securities with similar

maturities from the date of disbursement as the cash flow. For loans made or guaranteed

before fiscal year 2001, the discount rate is based on a disbursement-weighted average of

interest rates for marketable Treasury securities with similar maturities as the loans or

loan guarantees.

Page 13 GAO-16-41 Credit Subsidy Cost Estimates

The final discount rate for a cohort of direct loans or loan guarantees is

determined based on interest rates of Treasury securities prevailing

during the period when the loans are disbursed. Specifically, once a

cohort of loans is substantially disbursed (at least 90 percent), the final

discount rate for that cohort is determined, and this rate is used to

calculate reestimates of subsidy costs and for financing account interest

calculations. Using the same rate for reestimates and interest on the

financing account ensures that the financing account will break even over

time as it uses its collections to repay its Treasury borrowing. As a result,

the discount rate reflects the federal government’s actual borrowing cost

and incorporates into the subsidy cost calculation an agency’s cost of

financing its lending. For loan guarantee programs, which may or may not

borrow from Treasury, the financing account receives the subsidy cost

from the program account and holds these funds to serve as a reserve

against future loan guarantee defaults or other costs. FCRA requires that

these funds—referred to as uninvested funds—earn interest from

Treasury at the same rate as the discount rate used to calculate the

present value when estimating the subsidy cost. The Office of

Management and Budget (OMB) provides tools for agencies to use to

discount estimated cash flows and calculate interest on financing account

balances.

For several years, the Congressional Budget Office (CBO) and others

have raised the concern that FCRA subsidy cost estimates recorded in

the budget underestimate the costs of federal credit programs and do not

completely reflect the costs imposed on taxpayers. Some have suggested

revising FCRA to include in the subsidy cost an additional estimated cost

related to certain market risk. This suggestion has been referred to as the

fair value approach and centers around the debate that beyond the cash

flows associated with the direct loan or loan guarantee, which are

recognized under FCRA, costs are imposed on taxpayers who would, in a

similarly risky private market transaction, require compensation for

bearing the aggregate risk associated with making the loan.

However, it

is important to note that this cost to the taxpayer does not represent a

monetary cost to the government as it is not a cash outlay by Treasury.

See Deborah Lucas and Marvin Phaup, “Reforming Credit Reform,” Public Budgeting &

Finance (Winter 2008).

Fair Value Approach and

the Current Basis of the

Budget

Page 14 GAO-16-41 Credit Subsidy Cost Estimates

Instead, it can be thought of as representing a noncash social cost of

federal lending.

CBO has provided information on fair value estimates to aid

congressional decision making regarding the value of resources being

devoted to federal credit programs for several years. In a 2004 report,

CBO initially laid out the conceptual foundation for the fair value approach

and subsequently issued several more reports more fully examining the

implications of using the fair value approach in the decision-making and

the budget process.

In a 2014 testimony, CBO compared and

contrasted the use of the fair value approach versus FCRA in informing

decisions and in budget formulation and execution, laying out advantages

and disadvantages of each method.

At present, CBO provides

congressional budget committees with cost estimates for credit programs

using both the fair value approach and the FCRA methodology.

The term fair value in the context of suggested revisions to FCRA subsidy

costs is similar in concept to its use in private-sector generally accepted

accounting principles (GAAP) for financial accounting; however, there are

differences in how the two would be applied.

The fair value approach

discussed in this report relates to budgeting for federal credit programs,

while the term “fair value” under private-sector GAAP applies to the

valuation of certain assets and liabilities of private-sector entities and is

defined as the price that would be received to sell an asset or paid to

transfer a liability in an orderly transaction between market participants. In

a general sense, the “fair value” label for federal budgeting purposes

For example, Congressional Budget Office, Fair-Value Accounting for Federal Credit

Programs (Washington, D.C.: March 2012); Accounting for FHA’s Single-Family Mortgage

Insurance Program on a Fair-Value Basis (Washington, D.C.: May 2011); and Estimating

the Value of Subsidies for Federal Loans and Loan Guarantees (Washington, D.C.:

August 2004).

Congressional Budget Office, Testimony on Estimates of the Cost of the Credit

Programs of the Export-Import Bank (Washington, D.C.: June 2014).

CBO regularly reports a variety of cost estimates to the Congress related to both current

government programs and pending legislation. See, e.g., 2 U.S.C. §§ 602, 653.

“Generally accepted accounting principles” has a specific meaning for accountants and

auditors. Private-sector GAAP, established by the Financial Accounting Standards Board,

provides guidance that businesses follow in preparing their general purpose financial

statements, which provide users such as investors and creditors with useful information

that allows them to assess a business’s ongoing financial performance.

Page 15 GAO-16-41 Credit Subsidy Cost Estimates

suggests a parallel between federal and private-sector notions of

appropriate compensation for making loans. The label “fair value” reflects

the perspective that this compensation—the risk premium—that

taxpayers demand can be found by observing what they would require to

make similarly risky investments in private financial markets. That view

provides the analogue to the situation contemplated in private-sector

GAAP fair value.

The federal budget has for decades represented a system of cash

accounts. FCRA, therefore, was designed to put credit programs on the

same plane as noncredit programs: FCRA reflects only the cost

associated with estimated cash flows between the government and the

borrower or lender, such as repayment of principal, payments of interest

and fees, and claim payments to lenders. All federal programs, including

those for spending programs or tax provisions whose cash costs appear

in the budget, also have associated noncash costs and benefits that

accrue to society at large. However, these noncash costs and benefits

are not included in the budget. Accordingly, budget decisions demand

trade-offs beyond just the cash cost; such decisions need to also consider

noncash costs and benefits. For example, a project supported by federal

spending could reduce unemployment—while at the same time result in

detrimental environmental effects. Because the federal budget was not

envisioned to present a complete picture of the total costs and benefits to

society of government programs and policies, it should not be used as the

sole rationale for federal activities. Specifically, OMB guidance in Circular

No. A-94, Guidelines and Discount Rates for Benefit-Cost Analysis of

Federal Programs, directs agencies to perform program evaluations,

including consideration of social costs and not just the costs to the federal

government. The compensation cost to taxpayers of making risky

investments, grants, or loans would be a noncash social cost considered

in performing benefit-cost analysis.

A fuller examination of the fair value approach, including the nature of the

social, noncash costs and the arguments for and against their inclusion in

the budget, is presented later in this report.

Page 16 GAO-16-41 Credit Subsidy Cost Estimates

We analyzed direct loan and loan guarantee programs’ annual and

lifetime reestimates from fiscal years 2001 through 2014

by cohort,

program, agency, and purpose to identify any trends.

We also

considered whether a program’s funding was discretionary or mandatory.

Our analysis considered many factors, including magnitude of

reestimates, consistency in reestimate patterns over time, as well as the

number of programs with either upward or downward reestimates. Based

on our analyses of these factors related to reestimate data for the fiscal

years 2001 through 2014 cohorts, we did not identify any overall

consistent trends in under- or overestimates of subsidy costs across

federal credit programs government-wide.

However, there were some

programs with significant upward or downward reestimates, which could

generally be explained by specific events. For example, the 2008 financial

crisis likely contributed to upward annual reestimates for housing, small

business, and agriculture loan guarantee programs because borrowers’

ability to repay weakened with rising unemployment and falling market

revenue. Unless otherwise noted, the amounts discussed in this section

are based on information reported by OMB in the President’s Budgets.

See appendix II for additional detailed reestimate trend information.

Based on our analysis of reestimates for fiscal years 2001 through 2014

cohorts, we found that through fiscal year 2014, both direct loan and loan

guarantee programs had net lifetime upward reestimates, meaning that

original subsidy costs were underestimated. However, in concluding

about overall trends, we considered many factors, including magnitude of

reestimates, consistency in reestimate patterns over time, as well as the

number of programs with either upward or downward reestimates. Based

on our analyses of these factors related to reestimate data for the fiscal

years 2001 through 2014 cohorts, we did not identify any overall

Reestimates are generally prepared at the end of fiscal years. Therefore, our analysis of

reestimates from fiscal years 2001 through 2014 indicates that we evaluated reestimates

prepared after the end of those fiscal years. Further, our analysis included the fiscal years

2001 through 2014 cohorts.

Unless otherwise noted, we have excluded the Troubled Asset Relief Program (TARP)

from this analysis because the size of the TARP lifetime downward reestimate is

significantly larger than all the reestimates of other credit programs combined, and its

inclusion would make any trend analysis less meaningful.

Our conclusions about trends were based on our professional judgment in assessing the

data, and not based on a statistical analysis.

No Overall Consistent

Trends Identified in

Under- or

Overestimates of

Subsidy Costs, and

Various Economic

and Portfolio

Changes Caused

Certain Significant

Reestimates

No Clear Overall Trends in

Subsidy Cost Reestimates

across Federal Credit

Programs Government-

Wide

Page 17 GAO-16-41 Credit Subsidy Cost Estimates

consistent trends in under- or overestimates of subsidy costs across

federal credit programs government-wide.

Reestimates provide information to improve the estimation process and

yield insights into program performance. For example, upward

reestimates could indicate that borrowers are delinquent in repaying

loans, and as a result, loan performance is worse than initially expected.

Reestimates take into account actual loan performance and expected

changes in future loan performance, which can both vary because of

portfolio changes (e.g., changes in the credit-worthiness of borrowers)

and changes in the economy—external events that affect borrower

behavior. Reestimates of subsidy costs are to be expected as loan

cohorts age and as the economy deviates from the paths assumed when

subsidy costs were originally estimated. When estimating or reestimating

subsidy costs, agencies must respond to changes in their credit program

portfolios and the economy. While forecasting future loan performance

can be challenging, the goal is to avoid consistent bias, reflected in either

recurring upward or downward reestimates. Consistent bias in forecasting

loan performance would result in subsidy cost estimates that provide

misleading information about program performance. At the same time, a

period of recurring overestimates or underestimates does not necessarily

mean that the original estimates were biased. A particularly challenging

period for forecasting loan performance was the onset of the 2008

financial crisis. A key question in evaluating subsidy cost estimates is

whether agencies were able to foresee deteriorating performance in credit

program portfolios.

Government-wide, a total of 101 direct loan programs reported

reestimates for the fiscal years 2001 through 2014 cohorts. As shown in

table 1, 42 programs had lifetime upward reestimates and 59 programs

had lifetime downward reestimates. Overall, the reestimates for these

programs totaled a net lifetime upward reestimate of $3.1 billion, meaning

that the original subsidy costs of the cohorts obligated during this period

were underestimated by $3.1 billion, which is less than 1 percent of the

amount of loans disbursed by the cohorts in our review. Based on data

reported in the President’s Budget, the estimated lifetime subsidy income,

meaning negative subsidy cost, of the fiscal years 2001 through 2014

cohorts of direct loan programs totaled about $91 billion, after considering

Page 18 GAO-16-41 Credit Subsidy Cost Estimates

the effects of lifetime reestimates.

Therefore, the net lifetime upward

reestimate of direct loan programs’ fiscal years 2001 through 2014

cohorts, represents about 3 percent of the lifetime subsidy income.

Table 1: Direct Loan Programs Lifetime Upward and Downward Reestimates, 2001-

2014

Dollars in billions

Programs

Reestimates

Lifetime upward reestimates

$32.6

Lifetime downward reestimates

(29.5)

Overall net lifetime upward reestimate

101

$3.1

Source: GAO analysis of President’s Budgets. | GAO-16-41

Further, as shown in figure 6, four direct loan programs accounted for 88

percent of the lifetime upward reestimate of $32.6 billion, with Education’s

Direct Student Loan Program accounting for 71 percent.

Similarly, four

direct loan programs accounted for 88 percent of the lifetime downward

reestimates of $29.5 billion, with Treasury’s Government Sponsored

Enterprises (GSE) Mortgage Backed Securities (MBS) Purchase Program

accounting for 55 percent.

This was calculated by multiplying the most recently reported original subsidy rate and

disbursements for a cohort and adding the most recently reported lifetime reestimates for

that cohort.

Education’s Direct Student Loan Program provides financing to students or their parents

to help students obtain postsecondary education. This program is currently the largest

federal direct loan program with $694 billion outstanding as of September 30, 2014.

Disbursements were $130 billion and $134 billion during fiscal years 2013 and 2014,

respectively.

Treasury established the GSE MBS Purchase Program in response to the 2008 financial

crisis to provide stability and prevent disruption in the financial markets.

Page 19 GAO-16-41 Credit Subsidy Cost Estimates

Figure 6: Direct Loan Programs Representing a Significant Percentage of Lifetime Upward and Downward Reestimates, 2001-

2014

Government-wide, a total of 76 loan guarantee programs reported

reestimates for the fiscal years 2001 through 2014 cohorts. As shown in

table 2, 35 loan guarantee programs had lifetime upward reestimates and

41 programs had lifetime downward reestimates. Overall, the reestimates

for these programs totaled a net lifetime upward reestimate of $39 billion,

which is less than 1 percent of the amount of loans guaranteed by the

cohorts in our review. Based on data reported in the President’s Budget,

the estimated lifetime subsidy cost of the fiscal years 2001 through 2014

cohorts of loan guarantee programs totaled about $26 billion, after

considering the effects of lifetime reestimates.

Therefore, for the fiscal

This was calculated by multiplying the most recently reported original subsidy rate and

disbursements for a cohort and adding the most recently reported lifetime reestimates for

that cohort.

Page 20 GAO-16-41 Credit Subsidy Cost Estimates

years 2001 through 2014 cohorts, loan guarantee programs initially

estimated subsidy income of $13 billion.

Table 2: Loan Guarantee Programs Lifetime Upward and Downward Reestimates,

2001-2014

Dollars in billions

Programs

Reestimates

Lifetime upward reestimates

$104.5

Lifetime downward reestimates

(65.5)

Overall net lifetime upward reestimate

$39.0

Source: GAO analysis of President’s Budgets. | GAO-16-41

Further, as shown in figure 7, five loan guarantee programs accounted for

91 percent of the $104.5 billion lifetime upward reestimates, with HUD’s

Mutual Mortgage Insurance (MMI) Fund accounting for 65 percent.

Similarly, three loan guarantee programs accounted for 95 percent of the

$65.5 billion lifetime downward reestimates, with Education’s Federal

Family Education Loan (FFEL) program accounting for 91 percent.

HUD’s MMI Fund provides insurance to lenders for single-family homeowners and is

supported by insurance premiums paid by borrowers. This is the largest federal loan

guarantee program, with $1,169 billion of outstanding guarantees as of September 30,

2014. New loan guarantees of $240 billion and $135 billion were approved during fiscal

years 2013 and 2014, respectively.

Education’s FFEL program was the loan guarantee complement to the Direct Student

Loan Program. The Health Care and Education Reconciliation Act of 2010, Pub. L. No.

111-152, § 2201, 124 Stat. 1029, 1074 (Mar. 30, 2010), terminated Education’s authority

to guarantee new loans under the FFEL program after June 30, 2010.

Page 21 GAO-16-41 Credit Subsidy Cost Estimates

Figure 7: Loan Guarantee Programs Representing a Significant Percentage of Lifetime Upward and Downward Reestimates,

2001-2014

In concluding about overall trends, we considered many factors, including

magnitude of reestimates, consistency in reestimate patterns over time,

and the number of programs with either upward or downward

reestimates. Based on our analyses of these factors related to reestimate

data for the fiscal years 2001 through 2014 cohorts, we did not identify

any overall consistent trends in under- or overestimates of subsidy costs

across federal credit programs government-wide. Specifically, although

both direct loan and loan guarantee programs government-wide had

overall lifetime upward reestimates for the cohorts over the 14-year period

of our analysis, the reestimates fluctuated significantly from year to year.

For example, as shown in figure 8, from fiscal years 2006 through 2014,

direct loan programs recorded overall net upward reestimates for 4 years

and overall net downward reestimates for 5 years. Also during this period,

loan guarantee programs recorded overall net upward reestimates for 5

Page 22 GAO-16-41 Credit Subsidy Cost Estimates

years and overall net downward reestimates for 4 years. The large

upward reestimates in fiscal year 2012 for loan guarantees and in fiscal

year 2014 for direct loans were primarily attributed to HUD’s MMI Fund

and Education’s Direct Student Loan Program, respectively, which are

discussed later.

Figure 8: Fluctuations in Direct Loan and Loan Guarantee Programs’ Annual Net Reestimates, 2006-2014

We also analyzed reestimates over time based on whether programs are

discretionary or mandatory. Under FCRA, as previously discussed,

appropriations for initial subsidy costs of discretionary programs are

typically limited to a definite amount. Mandatory credit programs are

funded by the operation of their authorizing statutes, typically in indefinite

amounts. However, both discretionary and mandatory programs

automatically receive permanent indefinite budget authority for the cost of

upward reestimates. Thus, agencies with discretionary credit programs

theoretically could have an incentive to initially underestimate subsidy

costs so that they could make more loans or loan guarantees within a

given appropriation level. If this incentive did not exist and absent any

overriding economic trend, one could expect reestimates to lower the

Page 23 GAO-16-41 Credit Subsidy Cost Estimates

original subsidy cost estimates as often as they raise it, and the patterns

would be similar for discretionary and mandatory programs.

Overall, we did not identify any consistent trends that would indicate

whether for some discretionary programs there was an effort to benefit

from initially underestimating subsidy costs. As shown in table 3,

generally the same proportion of discretionary and mandatory credit

programs had upward and downward reestimates. The one pattern we

observed was in relation to the magnitude of loan guarantee reestimates.

Specifically, regarding the dollar amount of reestimates, about 95 percent

of the upward reestimates were from discretionary loan guarantee

programs, while about 93 percent of the downward reestimates were from

mandatory loan guarantee programs. These trends are primarily driven by

two programs—upward reestimates for HUD’s discretionary MMI Fund

and downward reestimates for Education’s mandatory FFEL program.

The reestimates for these programs, which can generally be explained by

economic events and portfolio changes, are described in more detail in

the following section.

Table 3: Discretionary and Mandatory Credit Program Reestimates, 2001-2014

Dollars in billions

Direct loan programs

Loan guarantee programs

Programs

Percentage

of programs

Reestimates

Programs

Percentage

of programs

Reestimates

Upward reestimates

Discretionary

34%

$7.0

43%

$99.8

Mandatory

25.6

4.7

Downward reestimates

Discretionary

42%

(5.3)

46%

(4.3)

Mandatory

16%

(24.2)

(61.2)

Total

101

$3.1

$39.0

Source: GAO analysis of President’s Budgets. | GAO-16-41

Page 24 GAO-16-41 Credit Subsidy Cost Estimates

Our analysis of reestimates trend data excluded activity from Treasury’s

Troubled Asset Relief Program (TARP).

TARP is an example of a

federal program that implemented a budgeting process similar to the fair

value approach. TARP was required by law to estimate the budgetary

cost of purchases and guarantees of troubled assets in accordance with

FCRA, except that such costs were required to be calculated by adjusting

the discount rate for market risk. These cost estimates were reestimated

annually, at fiscal year-end, to reflect revised assumptions for market risk,

asset performance, and other key variables and economic factors.

Interest earned or paid on financing account transactions with Treasury

was based on FCRA discount rates (not adjusted for market risk). As a

result, as program funds were repaid, reestimates reflected reduced costs

because of improvements in the economy and also because the noncash

cost that was considered in the market-risk-adjusted discount rates was

not reflected in the actual federal cash flows. Overall, TARP has had

lifetime downward reestimates of nearly $177 billion through fiscal year

2014. If the FCRA methodology had been used to initially estimate the

subsidy cost of TARP, the lifetime downward reestimate would have been

significantly less—meaning that the overestimate of initial subsidy costs

would have been less. As of September 30, 2014, TARP has reported an

estimated lifetime subsidy cost of $54.6 million for its direct loan,

investment, and guarantee programs.

We identified a few programs with significant upward or downward

lifetime reestimates for the fiscal years 2001 through 2014 cohorts, which

could generally be explained by portfolio or economic changes, as well as

revisions in estimation methodologies.

We analyzed estimates of fiscal years 2001 through 2014 cohorts based

on the purposes of the programs. Specifically, the types of direct loan

programs with the largest lifetime reestimates were in the education,

international, and other categories, and the types of loan guarantee

programs with the largest lifetime reestimates were in the housing,

TARP was authorized by the Emergency Economic Stabilization Act of 2008, which

gave the Secretary of the Treasury broad and flexible authority to establish TARP to

purchase and insure mortgages and other troubled assets. Among other things, this

permitted the Secretary to inject capital into banks and other commercial companies by

taking equity positions in those entities to help stabilize the financial markets. Pub. L. No.

110-343, div. A, 122 Stat 3765 (Oct. 3, 2008), classified in part, as amended, at 12 U.S.C.

§§ 5201-5261.

Significant Upward and

Downward Reestimates

Could Generally Be

Explained by Specific

Events Affecting a Few

Large Credit Programs

Page 25 GAO-16-41 Credit Subsidy Cost Estimates

education, and small business and agriculture categories. (See fig. 9.)

These reestimates could generally be explained by specific events

affecting a few large programs.

Figure 9: Primary Drivers of the Largest Overall Upward and Downward Lifetime Reestimates for Direct Loans and Loan

Guarantees by Purpose of Program, 2001-2014

The education-related category had the largest lifetime upward reestimate

for direct loans and the biggest swings in annual reestimates.

For

example, in fiscal year 2010, this category had a net downward

reestimate of $11.4 billion and, in fiscal year 2014, a net upward

reestimate of $18.7 billion, primarily driven by Education’s Direct Student

Loan Program. The reported estimated lifetime subsidy income for the

Direct Student Loan Program’s fiscal years 2001 through 2014 cohorts of

direct loans totaled about $73.3 billion, after considering a lifetime upward

reestimate of about $23.1 billion. The primary cost driver of the Direct

Student Loan Program is the difference between borrowers’ interest rates

on their loans and Education’s cost of borrowing to finance its lending,

The “education-related” direct loan category includes programs operated by the

Department of Education, as well as a program operated by the Department of Veterans

Affairs that provides financing to veterans with service-connected disabilities to obtain

vocational rehabilitation.

Education-Related Direct Loan

Programs

Page 26 GAO-16-41 Credit Subsidy Cost Estimates

represented by the discount rate in the subsidy cost calculation.

Consistent with the process to develop the President’s Budget, Education

prepares its initial subsidy cost estimates in advance of knowing the

actual borrower interest rates on its lending and interest rates to finance

its lending. Further, interest rate changes are impossible to predict with

certainty. As a result, changes in interest rates, which are outside the

control of Education, can have a significant effect on subsidy cost

reestimates for the program.

According to Education, this program experienced generally upward

reestimates over the years mainly driven by discount rate changes;

revised assumptions related to income-driven repayment plans

and

public service loan forgiveness;

and rising borrower default rates.

After

fiscal year 2014, the Direct Student Loan Program reported an upward

reestimate of $21.0 billion, which, according to Education officials, was

primarily related to newly implemented income-driven repayment plans

for borrowers.

This reestimate had a significant effect not only on the

reestimates in the education category but also on direct loan program

reestimates government-wide. The size of the Direct Student Loan

Under the Direct Student Loan Program, if borrowers’ outstanding federal student loan

debt is higher than their annual income or if it represents a significant portion of their

annual income, borrowers may want to repay their federal student loans under one of the

income-driven repayment plans available to borrowers, which are designed to make

student loan debt more manageable by reducing the monthly payment amount.

Under the Public Service Loan Forgiveness program, certain borrowers employed full-

time by certain public service organizations may have their remaining balances of Federal

Direct Student Loans forgiven after making 120 qualifying payments while in public service

employment. Qualifying payments must be made after October 1, 2007, under a qualifying

repayment plan; the first forgiveness of loan balances won’t be granted until October 1,

2017. Borrowers may request that Education certify their employment and loans in

advance for eligibility purposes.

However, in September 2015, Education released its updated 3-year default rate for the

fiscal year 2012 cohort. The cohort default rate for the fiscal year 2012 cohort is based on

the number of students who entered repayment in fiscal year 2012 and defaulted on or

before the end of fiscal year 2015 (a 3-year window) divided by the number of students

who entered repayment in fiscal year 2012. Based on these data for the fiscal year 2012

cohort, the 3-year default rate has decreased since the fiscal year 2010 cohort.

Education officials told us that the reestimates it reported after fiscal year 2014 (included

in the 2016 President’s Budget) included an error. The reestimate should have been an

upward reestimate of $19 billion. If this error had been identified and corrected prior to the

preparation of the President’s Budget, the lifetime reestimate for the fiscal years 2001

through 2014 cohorts of direct loan programs would have been an underestimate, or a

lifetime upward reestimate, of $1.1 billion rather than $3.1 billion.

Page 27 GAO-16-41 Credit Subsidy Cost Estimates

Program also contributes to the magnitude of its reestimates. For

example, for the fiscal years 2001 through 2014 cohorts, Education has

reported loan disbursements totaling about $825 billion. With this loan

volume, a 1 percentage point change in the subsidy rate for all cohorts

would result in an $8.25 billion reestimate. In comparison, a 1 percentage

point change in the subsidy rate of a loan program with less loan volume

would have a smaller reestimate amount.

The lifetime upward reestimates for the international category were

primarily driven by the Export-Import Bank’s (Ex-Im) Direct Loan

Program. The reported estimated lifetime subsidy cost for the fiscal years

2001 through 2014 cohorts of Ex-Im’s Direct Loan Program totaled about

$1.1 billion, after considering a lifetime upward reestimate of $2.8 billion.

Ex-Im reported that upward reestimates were due to estimation modeling

changes made in fiscal year 2012. Specifically, Ex-Im uses a loss

estimation model to estimate subsidy costs. The model calculated loss

rates based on historical data, including default and loss history of prior

loan transactions, as well as variables that can be used to predict defaults

and losses, such as transaction amount and length, obligor type, product

type, and risk rating. According to Ex-Im, in 2012, it added qualitative

factors (e.g., minimum loss rate; global economic risk; and region,

industry, and aircraft portfolio obligor concentration risk) to account for

specific risks associated with the agency’s current portfolio. Ex-Im added

qualitative factors to the estimation methodology to better and more

accurately measure default risk. When Ex-Im added qualitative factors,

there was also an increase in direct loan exposure because of the

financial crisis, which magnified the impact of the qualitative factors. Ex-

Im’s direct loan exposure increased from $4.5 billion in 2008 to

$32.7 billion in 2014.

The downward reestimates in the direct loan other category were

primarily driven by Treasury’s GSE MBS Purchase Program. The

reported lifetime subsidy income of this program totaled about

$21.8 billion, after considering a lifetime downward reestimate of

$16.1 billion. Under this program, Treasury purchased $226 billion of

GSE MBS from the secondary market from September 2008 through

December 2009. As a result of scheduled principal repayments, the

Ex-Im’s Direct Loan Program provides financing for international buyers of U.S. goods

and services.

International Direct Loan

Programs

Other Direct Loan Programs

Page 28 GAO-16-41 Credit Subsidy Cost Estimates

portfolio declined to $136 billion as of March 2011, when Treasury began

the orderly disposition of the remaining MBS portfolio. Treasury

completed the orderly disposition of this portfolio in March 2012.

Following the principal repayments and sales, significant downward

reestimates were recorded in fiscal years 2009 and 2011. These

downward reestimates were the result of higher-than-projected proceeds

from MBS repayments and sales because of improving market conditions

following the 2008 financial crisis.

The housing category had the largest lifetime upward reestimates for loan

guarantee programs and experienced significant annual upward

reestimates from fiscal years 2008 through 2012, ranging from $8.5 billion

to $28.1 billion during this period. These reestimates were primarily driven

by HUD’s MMI Fund. The reported estimated lifetime subsidy cost for the

fiscal years 2001 through 2014 cohorts of MMI loan guarantees totaled

about $18.7 billion, after considering a lifetime upward reestimate of

$75.3 billion.

A variety of economic, portfolio, and market changes

contributed to upward reestimates for MMI’s forward mortgages, which

accounted for $68.2 billion of the MMI Fund’s lifetime upward reestimate.

Chief among these changes affecting the cost were downward

adjustments to long-term housing price and interest rate assumptions

stemming from the mortgage and financial crises in the late 2000s.

Housing prices are an important variable in estimating the MMI Fund’s

subsidy costs because of their influence on the probability of mortgage

default and the severity of losses in the event of default.

However, long-

term housing price forecasts are inherently uncertain. In addition, the

budget process requires these assumptions to be made more than a year

in advance. HUD officials told us that the assumptions used to make the

original credit subsidy estimates were more optimistic than the revised

The MMI Fund insures both single-family forward and reverse mortgages. With forward

mortgages, the borrower’s monthly loan payments to the lender add to the borrower’s

home equity and decrease the loan balance. Reverse mortgages permit persons 62 years

and older to convert their home equity into cash advances. With reverse mortgages, the

borrower receives payments from the lender. The lender adds the principal and interest to

the loan balance, reducing the homeowner’s equity. In this report, we focus on MMI-

insured single-family forward mortgages.

For example, falling house prices reduce home equity. In general, lower levels of home

equity as a percentage of home value are associated with relatively poorer loan

performance because homeowners with negative equity may find it difficult to sell or

refinance the property to avoid foreclosure. They may also have incentives to stop making

mortgage payments to minimize their financial losses.

Housing Loan Guarantee

Programs

Page 29 GAO-16-41 Credit Subsidy Cost Estimates

assumptions used to make the reestimates, which contributed to upward

reestimates. Another key reason for the MMI Fund’s upward reestimates

was a downward adjustment in mortgage interest rate assumptions.

Mortgage interest rates are an important variable in estimating the fund’s

subsidy costs because of their influence on both mortgage prepayments

and defaults.

A number of portfolio and mortgage market developments associated with

higher-than-expected losses also contributed to the MMI Fund’s upward

reestimates. For example, growth in the proportion of MMI-insured

mortgages with seller-funded down-payment assistance and competition

from private mortgage institutions in the low down-payment mortgage

market contributed to upward reestimates that began in the early 2000s.

MMI-insured loans with seller-funded down-payment assistance grew

from about 6 percent of the MMI Fund’s business in 2000 to about 30

percent in the mid-2000s. Unlike other key mortgage industry participants,

HUD allowed borrowers to obtain down-payment assistance from

nonprofits that operated programs supported partly by financial

contributions and service fees from participating property sellers. The

seller-funded down-payment assistance loans performed worse than

comparable loans with down-payment assistance from other sources or

with no assistance, possibly because these homebuyers had less equity

in the transactions and thus may have been more likely to default. For

example, the MMI Fund cost estimates reflected an estimated subsidy

rate of positive 6.35 percent for seller-funded down-payment assistance

loans, compared with negative 0.01 percent for non-seller-funded down-

payment assistance.

Similar to Education’s Direct Student Loan Program, the size of the MMI

Fund also contributes to the magnitude of its reestimates. Because the

amount of loan guarantees provided is so large, a minor change in the

subsidy rate during the reestimate process can result in a significant

For example, declining mortgage rates cause early mortgage terminations because of

borrower refinancing. The refinancings can be expected to reduce the MMI Fund’s

expected premium income (thereby increasing estimated subsidy costs) in part because

not all borrowers refinance into new MMI-insured mortgages. Additionally, borrowers who

do not refinance (and are therefore paying interest rates higher than current market rates)

have greater incentive to default because the present values of their mortgages are higher

than their outstanding mortgage balances. As a result, the MMI Fund’s projected losses

for these borrowers can be expected to increase, resulting in higher estimated subsidy

costs.

Page 30 GAO-16-41 Credit Subsidy Cost Estimates

reestimate in terms of dollars. For example, for the fiscal years 2001

through 2014 cohorts, HUD has reported loan guarantees totaling about

$2.2 trillion under the forward mortgage issuance program. With this

volume of loan guarantees, a 1 percentage point change in the subsidy

rate for all cohorts would result in a $22 billion reestimate.

The education-related category had the largest lifetime downward

reestimates for loan guarantee programs, driven by Education’s FFEL

program.

The reported estimated lifetime subsidy income of the FFEL

program totaled about $8.4 billion for the cohorts since fiscal year 2001,

after considering the effects of lifetime downward reestimates of

$59.6 billion. The significant downward reestimates were attributable to

many factors; chief among them were interest rate changes.

Programmatic changes, such as Education’s ability to purchase

outstanding guaranteed loans from lenders,

were reflected as

modifications under FCRA.

In addition, the program was terminated by

statute in 2010, prohibiting new loan guarantees after June of that year.

The small business and agriculture loan guarantee category had lifetime

upward reestimates of $7.4 billion, which were primarily driven by

reestimates from the Small Business Administration’s (SBA) 7(a) Loan

Guarantee and 504 Certified Development Company Loan Guarantee

Programs.

The reported estimated lifetime subsidy cost for the fiscal

The “education-related” loan guarantee category includes a program operated by the

Department of Education, as well as a program initiated by the Department of Health and

Human Services that provided financing for medical students.

The Ensuring Continued Access to Student Loans Act of 2008, Pub. L. No. 110-227, § 7

(May 7, 2008), classified as amended at 20 U.S.C. § 1087i-1, authorized the Secretary of

Education to purchase or enter into forward commitments to purchase outstanding FFEL

program loans from private lenders. This authority was to expire on July 1, 2009; however,

a 2008 amendment to the statute extended this authority through July 1, 2010.

A modification is a government action that (1) differs from actions assumed in the

baseline estimate of cash flows and (2) changes the estimated cost of an outstanding

direct loan (or direct loan obligation) or an outstanding loan guarantee (or loan guarantee

commitment). Any cost associated with a modification is budgeted for separately from a

reestimate.

SBA’s 7(a) Loan Guarantee Program is SBA’s most common loan guarantee program; it

provides financial help for businesses, and its special requirements of eligibility are based

on specific aspects of each business and its principals. The 504 Loan Guarantee Program

provides financial help to business applicants meeting certain eligibility requirements, and

loan funds are to be used for major fixed assets, such as equipment or real estate.

Education-Related Loan

Guarantee Programs

Small Business and Agriculture

Loan Guarantee Programs

Page 31 GAO-16-41 Credit Subsidy Cost Estimates

years 2001 through 2014 cohorts of 7(a) Loan Guarantee and 504

Certified Development Company Loan Guarantee Programs’ loan

guarantees totaled about $5.0 billion and $3.9 billion, after the effect of

lifetime upward reestimates of $4.1 billion and $3.8 billion, respectively.

These programs experienced higher-than-normal reestimates for fiscal

years 2009 and 2010. SBA reported that the cause of these reestimates

was the downturn in the economy and an unprecedented continuing

period of high unemployment, which increased default claim payments

and the uncertain effects these events would have on future default claim

payments. In addition, SBA reported that worse-than-expected

performance in loans originated during fiscal years 2005 through 2008

also contributed to the upward reestimates.

To further analyze reestimates, we evaluated annual reestimates for

direct loan and loan guarantee programs from fiscal years 2006 to 2014

to determine whether the 2008 financial crisis had an effect on

reestimates.

We found that the negative economic outlook caused by

the economic downturn during the 2008 financial crisis likely increased

the projected cost of housing, small business, and agriculture loan

guarantee programs, as evidenced by upward reestimates during the

financial crisis. For fiscal years 2008 through 2012, the housing loan

guarantee programs’ annual net upward reestimates ranged from

$8.5 billion to $28.1 billion, while the small business and agriculture

related loan guarantee programs’ annual net upward reestimates ranged

from $1.0 billion to $4.3 billion. These reestimates for both categories of

programs were substantially larger than the reestimates before and after

this period. In addition, rising default rates associated with Education’s

Direct Student Loan Program could likely also be, in part, attributed to the

financial crisis. However, this program had upward reestimates before

and after the financial crisis. (See fig. 10.)

We started this trend analysis with fiscal year 2006 to include a few years prior to the

2008 financial crisis.

Reestimates during the

Financial Crisis

Page 32 GAO-16-41 Credit Subsidy Cost Estimates

Figure 10: 2008 Direct Loan and Loan Guarantee Programs’ Reestimates during the Financial Crisis, 2006-2014

Page 33 GAO-16-41 Credit Subsidy Cost Estimates

The fair value approach to estimate subsidy costs, supported by some

experts we interviewed, would add market risk to what is recognized

under FCRA. While credit program subsidy cost estimates under the fair

value approach may provide useful information to decision makers for

evaluating the costs against the benefits of credit programs, the additional

market risk recognized under the fair value approach does not reflect

additional cash costs beyond those recognized by FCRA. The

introduction of market risk into credit program subsidy costs under the fair

value approach would be inconsistent with long-standing federal

budgeting practices and presents several implementation issues. More

specifically, the fair value approach would (1) add noncash costs into the

budgeting process, which is based on cash costs; (2) be inconsistent with

the budget treatment of similarly risky programs; (3) lack transparency

with respect to inclusion of a noncash cost in budget totals; (4) involve

significant implementation issues; and (5) be complicated by comparisons

to GAAP fair value. Further, as discussed in the prior section, we did not

identify consistent trends in overall under- or overestimates of subsidy

costs across federal credit programs government-wide. Consequently, we

do not support the fair value approach to estimate subsidy costs for the

budget and believe the current FCRA methodology is more appropriate,

as it represents the best estimate of the direct cost to the government and

is consistent with current budgetary practices .Other experts we

interviewed, as well as OMB, opposed the fair value approach.

The fair value approach to estimate credit program costs may be useful

for evaluating the costs against the benefits of credit programs in that it

accounts for market risks beyond those already recognized under FCRA.

CBO and others have reported that subsidy cost estimates prepared

under FCRA requirements do not completely reflect the costs imposed on

taxpayers by federal credit programs, thus prompting proposals to modify

FCRA subsidy cost estimates based on the fair value approach. More

specifically, proposals calling for the adoption of the fair value approach

are motivated by the idea that the social costs to taxpayers of making

loans should be recognized. The social cost of making a loan includes not

only the estimated cash cost to the government for making the loan and

for providing for the possibility of default—both of which are currently

recognized under FCRA requirements—but also the noncash cost to

taxpayers for bearing the risk associated with the loan—which is not

currently recognized under FCRA requirements. Because such social

costs are not currently being accounted for, proponents of the fair value

approach asserted that budget decisions may be distorted by an incentive

to overly rely on risky credit programs as a policy tool because credit

Fair Value Subsidy

Cost Estimates May

Be Useful, but They

Are Not Consistent

with Long-standing

Federal Budgeting

Practices, Involve

Significant

Implementation

Challenges, and

Should Not Be

Recognized in Budget

Costs

Fiscal Year 2014 Subsidy Costs

For fiscal year 2014, the estimated subsidy

cost for new loans and guarantees across all

credit programs was an estimated subsidy

income of $35.7 billion. This represented

about .01 percent of estimated federal outlays

and receipts for the year.

Source: GAO analysis of fiscal year 2016 President’s Budget.

| GAO-16-41

Fair Value Approach Cost

Estimates May Be Useful

to Decision Makers for

Evaluating the Costs

against the Benefits of

Credit Programs

Page 34 GAO-16-41 Credit Subsidy Cost Estimates

programs look relatively less expensive than other forms of federal

assistance (e.g., grants).

Proponents of the fair value approach have generally viewed the

government as a pass-through to a collection of taxpayers who bear the

cost of risk, recognizing that the government must ultimately either reduce

spending or turn to taxation to recoup the cost of a defaulted loan. In this

sense, taxpayers could be considered as analogous to investors in a

private corporation who must contribute more resources if the corporation

experiences losses. In private financial markets, taxpayers as investors

with diversified portfolios would still demand compensation for bearing the

risk that the macroeconomy—the national or global economy as a

whole—may falter. As shown in figure 11, this risk—referred to as

aggregate risk (which represents a portion of overall market risk)—arises

from the possibility of significant economic downturns, when even a well-

diversified portfolio of financial investments will experience a reduction in

value. In such bad economic times, loan payments are likely to fall just

when returns are worth most to investors. Compensation for taking on this

aggregate risk—the aggregate risk premium—is over and above that

demanded to account for the time value of money (i.e., the notion that a

dollar today is worth more than a dollar in the future because it can earn

interest). The fair value approach is based on the premise that the

subsidy cost of credit programs should include this aggregate risk

premium due to taxpayers, even if it is not a cash cost to Treasury.

Aggregate Risk Premium

In private financial markets, taxpayers as

investors with diversified portfolios would still

demand compensation for bearing the risk

that the macroeconomy—the national or

global economy as a whole—may falter. This

risk—referred to as aggregate risk (which

represents a portion of overall market risk)—

arises from the possibility of significant

economic downturns, when even a well-

diversified portfolio of financial investments

will experience a reduction in value.

Source: GAO. | GAO-16-41

Page 35 GAO-16-41 Credit Subsidy Cost Estimates

Figure 11: Aggregate Risk

While proponents of the fair value approach suggested several methods

to estimate the cost of credit programs under the fair value approach, we

are focusing on the most commonly discussed method for use in federal

budgeting, which calls for adjusting the discount rate to account for the

aggregate risk premium.

Some proponents of the fair value approach

also viewed adjusting the discount rate as the most straightforward

method to derive fair value approach measurements. Under this method

The adjustment to the discount rate can be gleaned from market data on the spreads on

securities of comparable risk and maturity. Two other methods suggested to estimate

subsidy costs under the fair value approach are (1) market prices of similar products

offered by private companies adjusted to account for borrower and product differences

and (2) an options pricing model, which is a type of model that many private-sector entities

use to evaluate guarantees.

Page 36 GAO-16-41 Credit Subsidy Cost Estimates

to implement the fair value approach, the difference between FCRA and

fair value approach subsidy cost estimates arises from the choice of

discount rates. The estimates of expected cash flows associated with the

loan program, including repayments, interest, or the net amount lost

through defaults or other loan performance, are the same in both

approaches. However, the difference in discount rates means that those

expected cash flows will have different present values under the fair value

approach and FCRA.

To incorporate the cost of bearing this aggregate risk into budget costs for

loans, the fair value approach adds an aggregate risk premium to the

discount rate used in FCRA calculations, which is determined based on

interest rates on Treasury securities. The Treasury rate is considered to

be default-free because investors are protected against default by the

government’s ability to cover its debts by raising tax revenue or by cutting

other federal spending.

Adding the aggregate risk premium to the

discount rate incorporates a noncash social cost into the subsidy cost

estimate. When the expected future cash flows (i.e., payments of interest

and repayments of principal) associated with the loan are discounted

using a discount rate above the Treasury rate, the resulting present value

of these future payments to the government is lower than it would be if

discounting were done using the Treasury rate. A lower present value for

these future payments to the government translates into a higher subsidy

cost to the government when the loan is made.

As shown in figure 12,

the aggregate risk premium is one factor that may explain differences

between market interest rates and Treasury interest rates. The debate

over the fair value approach rests on whether the cost associated with

aggregate risk should be considered in the subsidy cost of credit

programs for the budget of the federal government.

Treasury securities are also often called riskless. However, investors are exposed to

interest rate risk when they purchase Treasury securities.

Applying the fair value approach to a loan guarantee program also results in a higher

estimated cost to the government. For loan guarantees, the effect of a higher aggregate

risk is to increase the cost to the government of future guarantee default payments above

their estimated value using default-free or “riskless” Treasury rates. The aggregate risk

increases the value of the guarantee because guarantee payments are most likely to be

made when the economy is depressed and resources are scarce. Mathematically, to

approximate this higher cost requires the use of discount rates that are effectively lower

than default-free Treasury rates.

Page 37 GAO-16-41 Credit Subsidy Cost Estimates

Figure 12: Differences between Market and Treasury Interest Rates

The budget, as presented under FCRA, does not recognize an aggregate

risk premium, and the cost of the government making loans is lower than

it would be under the fair value approach. Some proponents of the fair

value approach have stated that the disconnect of subsidy cost estimates

under FCRA from market valuation is at odds with the presentation of

other costs in the budget that do reflect values in private market

transactions (e.g., purchase of office supplies or wages paid). Market

prices are key benchmarks in assessing the costs to society of using its

resources in one project versus another. In this context, the budget under

FCRA does not reflect the social costs of making loans, which fair value

approach proponents believed should be considered in making budget

decisions that require trade-offs in the presence of a federal budget

constraint.

The social benefits and costs of extending credit is an important

consideration in decision making about funding levels. As stated in OMB

Circular No. A-94, federal programs should be evaluated based on

Page 38 GAO-16-41 Credit Subsidy Cost Estimates

comprehensive estimates of the expected benefits and costs to society

and not just the monetary benefits and costs to the government.

Therefore, for credit programs, both cash costs to the government and

social costs imposed on taxpayers could be weighed against benefits to

borrowers and to the public at large. These social costs would include the

cost of bearing risk that the government has transferred to taxpayers,

represented by the aggregate risk premium. OMB Circular No. A-129,

Policies for Federal Credit Programs and Non-Tax Receivables, requires

agencies to provide benefit-cost analyses as part of biennial program

reviews. These reviews are also intended to take account of changes in

program risks and costs. In this context, the fair value approach

aggregate risk premium could be classified as a cost imposed by a

program and included in total societal costs along with the cash costs to

the government. It is in this setting that several proponents of FCRA

believed, and we concur, that the fair value approach would be most

appropriately considered by decision makers, rather than in budget

estimates.

Reflecting a different concern, some proponents of the fair value

approach to budgeting for federal credit programs, cited as motivation the

perceived over-reliance on federal credit as a policy tool and the desire to

correct any bias toward underestimated costs under FCRA. Consistent

with this view, some proponents of the fair value approach generally

supported the approach for initial subsidy cost estimates that are used by

congressional decision makers to determine loan levels to approve. After

loan levels are approved, these proponents of the fair value approach

generally expressed no preferences for the methodology used to

determine subsidy cost estimates for existing credit program portfolios.

The effects of the 2008 financial crisis on parts of the federal credit

portfolio demonstrated the sensitivity of loan performance to conditions in

the macroeconomy. Subsidy cost estimates of loan guarantee programs

related to housing and small business and agriculture, in particular, had

not anticipated the possibility of large losses associated with the crisis in

the mortgage market and extreme weakness in market demand for goods

and services experienced during the 2008 financial crisis. Repayment of

student loans also declined significantly during this period. Some

proponents of the fair value approach told us that adopting the fair value

approach would help address concerns about the over-reliance on federal

credit programs by raising the subsidy cost of credit programs, likely

Page 39 GAO-16-41 Credit Subsidy Cost Estimates

resulting in fewer loans being made.

Also, according to some

proponents of the fair value approach, raising the cost of credit programs

would weigh against any bias in underestimating costs. In contrast,

proponents of the FCRA methodology stated that to the extent that

agencies were underestimating subsidy costs under FCRA, this would be

more appropriately addressed through improvements in the subsidy

estimation process rather than application of the fair value approach.

Going forward, there are some reasons to expect that agencies’

estimates of expected loan losses will better reflect the possibility of a rise

in defaults in a macroeconomic downturn. Typically, agencies project

future losses based on historical experience. Before the 2008 financial

crisis, program experience generally did not include the effect of a

significant economic downturn. Therefore, expected loss projections

made prior to 2008 would have likely underestimated the actual losses.

Now, the depressing effects of the 2008 financial crisis across the

government’s credit portfolio are included in each program’s historical

data. Although agencies can project individual loan programs’ expected

losses with the downturn included in historical data, the possibility of

extreme losses occurring simultaneously across loan programs would not

be reflected in the loss estimates of any one program.

For that, the

government would have to evaluate performance across its entire credit

portfolio, trying to assess the potential size of total losses in a bad

economy. However, our analysis of credit program reestimates over the

course of the 2008 financial crisis revealed that only parts of the portfolio

experienced significant increases in reestimates, implying some degree of

diversification against the risk of aggregate losses. While performance of

Under the fair value approach, while credit programs would have higher initial subsidy

rates, this may not affect the amount of loans that are made by programs with indefinite

budget authority, such as Education’s Direct Student Loan Program, because, under such

programs, loans are made based on the eligibility of applicants and not budgetary

spending limits.

Considering that the federal government’s entire portfolio comprises many different

types of loans (e.g., housing, disaster, and alternative energy), overall losses will be

higher the more loans of each type default. Some loan types experience losses under

similar circumstances, for example, mortgage and small business loans default during

recession, which would likely raise overall losses during a downturn. However, disaster or

alternative energy loan performance is not likely related to a recession, and if these loan

types do not experience defaults in downturns, then the government’s overall losses

would be limited to those that do. Projections of losses for a given loan type consider only

that type’s default history and not the possibility that other loan types might perform

similarly.

Page 40 GAO-16-41 Credit Subsidy Cost Estimates

parts of the federal credit portfolio does track with financial market

outcomes, performance of others, such as disaster loans, may not.

While fair value approach estimates may provide useful information for

evaluating the costs against the benefits of credit programs, the additional

market risk recognized under that approach do not reflect additional

estimated cash costs beyond those recognized by FCRA. The concept of

the fair value approach to recognize the aggregate risk premium due to

taxpayers beyond the FCRA-reflected risks (based on expected cash

outlays) evolved from the consideration of investor behavior in private

financial markets. Subsidy cost estimates prepared under the fair value

approach would be inconsistent with long-standing budgeting practices

and involve significant implementation challenges. More specifically, the

fair value approach would

• add noncash social costs into the budget process, which is primarily

based on cash costs to Treasury;

• result in inconsistency with the budgetary treatment of similarly risky

programs;

• result in less transparency and cause verification issues related to

noncash costs in budget totals;

• involve significant implementation issues; and

• be complicated by comparisons between the fair value approach and

private-sector GAAP fair value.

Because of these issues, we do not support the fair value approach to

estimate credit program subsidy costs for the budget and believe that the

current FCRA methodology is more appropriate for this purpose. In

addition, OMB has reported that it opposes the fair value approach

because it would make the budgeting process less transparent and less

accurate.

Deployment of the fair value approach in the federal budget would

provide a perspective on the cost of aggregate risk that is not currently

represented in the budget for any type of federal program and would add

noncash costs to primarily cash totals found in budget accounts. As

previously discussed, the fair value approach would add the

Office of Management and Budget, Statement from OMB Spokesman Steve Posner on

H.R. 1871 and H.R. 1872 (Apr. 7, 2014).

Fair Value Approach

Subsidy Cost Estimates

Are Not Consistent with

Long-standing Federal

Budgeting Practices,

Involve Significant

Implementation

Challenges, and Should

Not Be Included in the

Budget

Noncash Social Cost versus

Cash Budget Cost to Treasury

Page 41 GAO-16-41 Credit Subsidy Cost Estimates

compensation cost to taxpayers of bearing aggregate risk to the FCRA

subsidy cost estimate, thereby explicitly introducing noncash social costs

into what is fundamentally a cash budgeting system. Proponents of both

the fair value approach and FCRA recognized the aggregate risk

premium as representing a social cost reflecting the ultimate transfer of

the risk of making loans from the government to taxpayers. However,

these social costs do not represent a cash cost to the federal government

and would not be recognized under FCRA estimates because no money

is disbursed or received by Treasury as a result of the different cost

calculation and its inclusion in budget figures. As shown in figure 13, the

initial subsidy cost would be higher under the fair value approach. As the

loan is repaid in the future, the agency would reestimate the subsidy cost,

creating a difference in the timing of the subsidy cost recognition

compared to that under the FCRA methodology. If the loan performed as

expected, the resulting reestimates under the fair value approach would

be downward, in essence amortizing the aggregate risk premium initially

estimated. In contrast, under the FCRA methodology, if the loan

performed as expected, there would be no reestimate. If the loan

performed worse than expected, under both the fair value approach and

FCRA, the agency would update the estimated cash flows and record a

reestimate. However, the resulting final lifetime subsidy cost under the fair

value approach or FCRA would be the same, after completing

reestimates, because the cash flows to and from the government are the

same. The cash flows between the government and the borrower are not

affected by the different discount rates used under the fair value approach

or FCRA.

Page 42 GAO-16-41 Credit Subsidy Cost Estimates

Figure 13: Fair Value Costs versus Cash Cost over Time

Note: This figure illustrates a simplified example of an interest-free loan and assumes that it performs

as required with no defaults.

Net cash flows are lower than actual returns because of financing costs associated with providing a

loan to a borrower.

Page 43 GAO-16-41 Credit Subsidy Cost Estimates

The concepts that currently underlie the federal budget date to the 1967

Report of the President’s Commission on Budget Concepts.

The

commission identified the major purposes of the budget, which included

processes to propose particular programs to advance policy goals and to

propose total expenditures and revenues intended to promote stability

and growth in the macroeconomy. In practice, OMB has observed that the

budget provides “the means for the President and the Congress to decide

how much money to spend, what to spend it on, and how to raise the

money they have decided to spend.” In other words, the budget supports

decisions about which programs to fund and at what level given the

presence of an overall budget constraint (in that government spending

must be supported by tax revenues or debt). Government programs’ fiscal

effects on the macroeconomy are associated with financial inflows and

outflows of Treasury.

The inclusion of social costs in the budget would depart from the concept

of cash expenditures and revenues that currently underlies the budget.

Some proponents of the fair value approach stated that in practice, the

budget already includes noncash costs. Specifically, FCRA introduced

this apparent anomaly by presenting the current and future costs of loan

making in terms of today’s dollars, meaning their worth today, so that the

costs may be compared to the outlays made today when the loan is

disbursed. The subsidy cost of the loan is the budget authority the

Congress appropriates and is the net present value of outflows and

inflows from the government over the life of the loan. Putting credit

program costs in present value terms differs from the current cash basis

used in other programs’ budget accounts to reflect the time value of

money. FCRA could be called modified cash budgeting and serves to

represent the lifetime cost of the loan in today’s dollars. This is consistent

with the method used to compute the fiscal gap and discounts cash flows

in and out of Treasury at the default-free Treasury rate of borrowing.

FCRA does not include in subsidy costs any estimates of costs that do

Report of the President’s Commission on Budget Concepts (Washington, D.C.: October

1967). The commission was appointed by the President on March 3, 1967. It was formed

as a bipartisan commission to conduct a thorough and objective review of budget

concepts and to address long-standing questions about the budget presentation and the

treatment of individual accounts within the budget.

The fiscal gap is the amount of spending reductions or tax increases on average over a

75-year period to keep debt held by the public as a share of gross domestic product from

exceeding its current level.

Excerpts from the Report of the

President’s Commission on Budget

Concepts, October 1967

“In deciding whether it is possible to develop a

unified budget, one must distinguish between

competing budget concepts, which cause

confusion, and complementary budget

concepts, which actually aid in understanding

the scope and economic impact of the

Government.”

“The work of the Congress and the executive

branch should be facilitated by budget

concepts in which all the different major

purposes come to focus in a comprehensive

unified budget, and public understanding of

the budget and usefulness of budget

information should be furthered.”

“The budget must serve simultaneously as an

aid in decisions about both the efficient

allocation of resources among competing

claims and economic stabilization and

growth.”

“The budget totals must be readily useful for

analysis of the impact of the Federal budget

on the economy. …To be able to do this in the

simplest possible fashion, rules for calculating

budget receipts and expenditures should lead

to a measure of surplus and deficit which is

useful for analyzing the economic impact of

the budget.”

Source: Report of the President’s Commission on Budget

Concepts, October 1967. | GAO-16-41

Page 44 GAO-16-41 Credit Subsidy Cost Estimates

not represent cash flows to or from the government and will not eventually

be paid to or from Treasury.

The adoption of the fair value approach would, all else equal, initially

increase subsidy cost estimates and the federal budget deficit by

including the cost of bearing aggregate risk.

The higher initial estimated

subsidy cost under the fair value approach would increase the budget

authority required for a given volume of loans compared to that required

under FCRA, thereby raising the federal budget deficit through the effect

on outlays. The federal budget deficit has significance in policy making

and in analyzing the economy, and experts who opposed the fair value

approach cautioned against the inclusion in the budget of what they

termed phantom costs of aggregate risk under the fair value approach.

Ultimately though, as noted above, net cash outlays under both the fair

value approach and FCRA are the same, so any upward bias in initial

subsidy cost estimates introduced by the fair value approach would be

eliminated through downward reestimates as receipts from loan

repayments are recognized.

Under the fair value approach, if actual cash flows were as initially

estimated, there would be an additional amount in the financing account

that represents the compensation to taxpayers for bearing risk, which is a

social cost and not a cash cost to Treasury. Social costs do not produce

fiscal effects insofar as they are not cash receipts or disbursements from

Treasury. Consequently, to avoid misrepresenting the fiscal effect of

credit programs on the federal budget deficit, some proponents of the fair

value approach suggested recording a revenue to offset this noncash

social cost to clear the additional amount in the financing account and

correct deficit balances.

Several experts we spoke with disagreed with the fair value approach of

reflecting social costs or the compensation to taxpayers for bearing risks

associated with credit programs in subsidy cost estimates for the budget.

These experts, who favored the current FCRA approach, rejected the

characterization by some financial economists of the government’s being

a pass-through to a collection of taxpayers, no different than investors in

a private corporation. Unlike private corporations, the federal government

The budget deficit is the amount by which the government’s budget outlays exceed its

budget receipts for a given period, usually a fiscal year.

Page 45 GAO-16-41 Credit Subsidy Cost Estimates

is enduring and can avoid insolvency through exercise of its power to levy

taxes and can more easily access credit through the Treasury securities

market. These proponents of FCRA noted that the government has a

lower cost of borrowing than a private financial institution, and as an

entity, the government has the ability to more readily collect on loans

through tax refund offsets and garnishment of wages.

In addition, some

federal loans, such as student loans, may not be discharged through

bankruptcy. Therefore, these experts claimed, the cost of providing a

credit program is less expensive for the government than the private

sector. The more fundamental disagreement over the fair value approach

concerns whether the government’s true cost of capital is represented by

the “riskless” rate at which it borrows or whether that cost should also

include compensation to taxpayers for bearing aggregate risk. This

debate in the financial economics literature about the government’s cost

of capital dates back several decades and has not been settled.

If the fair value approach were deployed for the government’s portfolio of

direct loans and loan guarantees, inconsistency of treatment across

budgetary accounts would have to be considered. Potential

inconsistencies arise from (1) the comparability of loan program budget

totals that include social costs with noncredit program cash costs that do

not include social costs, (2) the exclusion of social costs associated with

costs of programs that present risks similar to those of loans, and (3) the

treatment of the presence of risk and uncertainty in federal activities other

than loan making. These inconsistencies could distort the decisions the

budget is intended to support by making loans appear relatively more

expensive than other programs because of the inclusion of noncash

social costs.

The federal budget is primarily recorded on a cash and obligations

basis.

The inclusion of social cost in the form of the fair value approach

risk premium raises the relative costs of loans compared to other

For example, the federal government has the ability to garnish wages without a court

order, while other creditors must first obtain a court order to garnish wages.

Kenneth J. Arrow and Robert C. Lind, “Uncertainty and the Evaluation of Public

Investment Decisions,” American Economic Review, vol. 60, no. 3, (1970).

An obligation is a definite commitment that creates a legal liability of the government for

the payment of goods and services ordered or received. Payment can be made

immediately or in the future.

Inconsistency with the

Budgetary Treatment of

Similarly Risky Programs

Page 46 GAO-16-41 Credit Subsidy Cost Estimates

programs budgeted based on cash. However, many other federal

programs impose social costs in addition to incurring cash outlays. One

example would be the costs of environmental damage resulting from

federally funded construction projects. Such social costs may not be

priced in the market as an aggregate risk premium might be, but they are

real costs to society in terms of losses in environmental quality and

services. If the fair value approach were to be adopted for budgeting

credit program costs, maintaining consistency in the budget would require

recognition of the social costs of all federal programs across the board,

presenting a daunting analytical problem, at a minimum.

The introduction of aggregate risk compensation in the budget would

create another inconsistency in the budget’s portrayal of the costs of

uncertain outcomes. Under the fair value approach, an agency would—as

it does now under FCRA—make an estimate of expected loan

performance, including repayments, prepayments, defaults, and

recoveries. The expected cash flows would be associated with average

losses over the life of the loan cohort, implying equal weight on both lower

and higher levels of loss. With the fair value approach, the addition of the

aggregate risk premium implicitly weights the worse outcome more

heavily by adding a cost associated with the possibility that losses are

greater than expected. Elsewhere, the budget presents as point estimates

the expected value of uncertain spending outcomes with the prospects of

better and worse outcomes equally weighted. For example, the

President’s Budget includes funding requests for the costs of fighting

forest fires. That request represents the expected value of the funding

level required based on the probability of a good outcome (i.e., few fires)

and a bad outcome (i.e., an active fire season). To the extent that this risk

cannot be diversified—that is, offset by risk elsewhere in government

spending—then, as with the fair value approach, a risk premium would be

added. Weighting bad outcomes more heavily, as with the fair value

approach, would thus be inconsistent with the treatment of risk and

uncertainty in all other budget estimates.

Cost estimates prepared under the fair value approach in the budget

would introduce a number of transparency and verification issues.

Regarding transparency in the presentation of the budget, it may not be

obvious that credit program costs included an aggregate risk premium, so

the lack of comparability with other programs’ cash budgeting would not

be clear. Moreover, as previously discussed, some proponents of the fair

value approach believed that the use of fair value would address the

perceived over-reliance on federal credit and correct for any bias in

underestimates of costs. However, as also previously discussed, our

Transparency and Verification

Issues

Page 47 GAO-16-41 Credit Subsidy Cost Estimates

analysis of reestimates for the fiscal years 2001 through 2014 cohorts of

direct loans and loan guarantees did not identify any overall consistent

trends in under- or overestimates of subsidy costs across federal credit

programs government-wide. Further, some proponents of FCRA asserted

that any problem with subsidy cost estimates will not be solved by

revising a budget estimate via the fair value approach but by engaging

decision makers in discussions about the types and amounts of federal

credit that should be provided in the budget. For example, one expert

stated that concerns about credit program losses could be addressed by

tightening eligibility criteria or reducing appropriations. In addition,

proponents of FCRA felt that any inaccuracy in estimating subsidy costs

should more appropriately be tackled by strengthening the data and

methods agencies use in forecasting expected cash flows.

Using the fair value approach may also mask information about direct

loan and loan guarantee performance. Under FCRA, the annual

reestimates may provide useful information to analyze actual loan

performance and to evaluate the quality of the estimation process as,

year by year, actual loan performance is recorded and expectations of

future loan performance are updated. Under the fair value approach, such

information may not be transparent because the effects of the adjusted

discount rate would also be included.

Further, under the fair value approach, the accuracy of the estimate of the

aggregate risk premium is not transparent and verifiable, which is an

important consideration given its cost-increasing effect. A primary

challenge in the fair value approach to budgeting is identifying the

aggregate risk premium applicable to any particular program. If the credit

program has a close counterpart in private credit markets, referred to as a

comparable, the difficulty is reduced though not eliminated.

For

example, private credit markets could likely provide a proxy for HUD’s

MMI Fund. However, there is likely not a private market comparable for

Using a private market comparable is a valuation technique in which a recently sold

asset is used to help determine the value of a similar asset.

Page 48 GAO-16-41 Credit Subsidy Cost Estimates

SBA’s Disaster Loan Program.

In the absence of a market comparable,

the analytical task would be significant and subjective, and in any event,

resulting estimates could be subject to criticism and controversy that

would be hard to resolve.

Moreover, even when a market comparable

exists, the task of extracting the aggregate risk premium from the

difference between the default-free Treasury rate and the observed

market rate would still be subjective. This complication arises because

other kinds of costs and risks that are not applicable to government

lending may be compensated in the market rate, as was illustrated in

figure 12. Arriving at a verifiable and auditable estimate of the aggregate

risk premium for each credit program is essential if there is to be

confidence in decisions that depend on comparing their relative costs.

Some proponents of the fair value approach suggested that TARP

illustrates successful implementation of the fair value approach. TARP

was required to use market-risk adjusted discount rates when determining

subsidy costs for the budget. For investment programs, TARP’s

valuations were based on market prices of securities for comparable

institutions and not solely the aggregate risk. In addition, because the

intention of TARP was to liquidate the assets when economically feasible

rather than hold them to maturity, the use of market prices was an

appropriate measure of expected cash proceeds to be received from

liquidation when the assets were sold. Therefore, TARP’s use of market-

risk adjusted discount rates may not be directly comparable to other credit

programs. Under FCRA and in accordance with OMB Circular No. A-11,

Under the Disaster Loan Program, SBA provides low-interest disaster loans to

businesses of all sizes, private nonprofit organizations, homeowners, and renters. SBA

disaster loans can be used to repair or replace the following items damaged or destroyed

in a declared disaster: real estate, personal property, machinery and equipment, and

inventory and business assets. By law, borrower interest rates depend on whether each

applicant has credit available elsewhere. For applicants unable to obtain credit elsewhere,

the interest rate will not exceed 4 percent. For those who can obtain credit elsewhere, the

interest rate will not exceed 8 percent. Borrower interest rates are determined for each

declared disaster. SBA will not decline a loan for lack of a particular amount of collateral.

In the absence of comparable market prices, analysts must resort to alternative

valuation methods, such as adjusting the discount rate for risk or applying options pricing.

For example, in implementing the market-risk adjustment required for TARP, Treasury

adjusted the discount rate using techniques applied to bank stock valuation and bank

reserves. As proponents of the fair value approach have pointed out, making such

estimates is complicated, although they have asserted that the approximations thus

obtained are at least closer to market valuations than under FCRA, which makes no

adjustment for market risk.

Page 49 GAO-16-41 Credit Subsidy Cost Estimates

Preparation, Submission, and Execution of the Budget, if an agency were

to plan to sell loan assets, the cash flows developed to estimate subsidy

costs would include assumptions related to expected sale proceeds,

which could be based on market information.

Several experts we spoke with stated that there would be significant

challenges to implementation of the fair value approach. (See app. III for

further details on these implementation challenges.) Several experts,

including proponents of both the fair value approach and FCRA, stated

that agencies and OMB do not currently have sufficient resources to

implement the fair value approach for subsidy cost estimates in the

budget. Some proponents of the fair value approach suggested that

private-sector expertise could be recruited to address the challenges of

identifying the appropriate aggregate risk premium for a program.

However, other experts stated that for programs that lack a market

comparable, private-sector expertise may or may not be more suited to

address the challenge of identifying the aggregate risk premium.

Conceptual and practical implementation difficulties would also confront

agency analysts. For example, a decision would need to be made about

whether the aggregate risk premium for an existing cohort should be

updated during the reestimate process. Updating the aggregate risk

premium could add significant volatility to subsidy cost estimates, making

it more difficult to understand reasons for reestimates (i.e., poor estimates

or worse-than-expected performance). However, not updating the

aggregate risk premium would result in cost estimates based on out-of-

date or noncurrent estimates of risks. Further, to the extent that the

financial accounting for credit programs continues to be consistent with

the budgeting for credit programs, not updating the aggregate risk

premium would be difficult for an agency to support during a financial

statement audit because the auditor would expect credit program cost

estimates to be based on current risk information. In addition, because

fair value approach subsidy cost estimates would include noncash costs

that would, if actual cash flows were as initially estimated, eventually

result in downward reestimates, decisions would need to be made on

whether to recognize these downward reestimates each year or as a one-

time adjustment. If the fair value approach were used, decisions would

also need to be made regarding the interest rate used to calculate interest

income and expense amounts on financing account balances so that the

financing account will break even over time as it uses its collections to

repay its Treasury borrowing.

Additional Significant

Implementation Issues

Page 50 GAO-16-41 Credit Subsidy Cost Estimates

Other implementation challenges would relate to programs that are

currently required to set borrower fees or interest rates to produce

subsidy-neutral cost estimates, such as the Department of Energy’s Loan

Guarantee Program,

or even negative-subsidy cost estimates to cover

administrative costs or to generate reserves, such as HUD’s MMI Fund

loan guarantee program. Decisions would also be needed related to

whether any increases to discretionary spending caps would be made if

the fair value approach were implemented.

Lastly, because the financial

accounting for credit programs currently mirrors FCRA and budgetary

accounting, the Federal Accounting Standards Advisory Board would

need to determine whether accounting standards would need to be

revised for any changes made related to the use of the fair value

approach for the budget.

Consideration of the fair value approach has been complicated by the

introduction of financial economics concepts and by the use of the term

“fair value” in both FCRA and private-sector GAAP contexts. Financial

economics aims to explain the behavior of private market participants

when confronted with risk and uncertainty in making investment

decisions. This requires an understanding of the alternatives available for

managing risks and of their respective costs. Application of these private-

The Department of Energy’s (DOE) Loan Guarantee Program was established by Title

XVII of the Energy Policy Act of 2005 to encourage early commercial use of new or

significantly improved technologies in energy projects. The act—specifically section

1703—originally authorized DOE to guarantee loans for energy projects that (1) use new

or significantly improved technologies as compared with commercial technologies already

in service in the United States and (2) avoid, reduce, or sequester emissions of air

pollutants or man-made greenhouse gases. For loan guarantees approved under this

authority, if no appropriation is provided to cover the subsidy cost of the program, DOE

must collect payments from borrowers to offset the initial credit subsidy costs associated

with their loan. See Pub. L. No. 109-58, tit. XVII, 119 Stat. 594, 1117 (Aug. 8, 2005),

classified as amended at 42 U.S.C. §§ 16511-16516.

Under the Balanced Budget and Emergency Deficit Control Act of 1985, as amended,

federal program funds may be sequestered in the event that aggregate discretionary

appropriations do not conform to previously established spending caps. See 2 U.S.C. §

901-907d.

The Federal Accounting Standards Advisory Board (FASAB) was established by the

Secretary of the Treasury, the Director of the Office of Management and Budget, and the

Comptroller General of the United States to develop accounting standards and principles

for the federal government. The American Institute of Certified Public Accountants

recognized FASAB as the board that promulgates GAAP for federal entities. FASAB has

nine members, three federal (selected by its sponsoring agencies) and six public or

nonfederal members.

Comparisons between the Fair

Value Ap

proach and Private-

Sector GAAP Fair Value

Page 51 GAO-16-41 Credit Subsidy Cost Estimates

sector concepts to a government setting is an unfamiliar exercise to many

in the public sector. Further, awareness of the private-sector GAAP

guidance on the use of fair value for measuring private financial

institutions’ assets may mistakenly suggest relevance to federal

budgeting. The label “fair value” as it relates to federal credit program

subsidy cost proposals reflects the perspective that the risk premium

taxpayers demand can be found by observing what they would require to

make similarly risky investments in private financial markets. That view

provides the analogue to the situation contemplated in private-sector

GAAP uses of fair value, wherein the valuation of certain assets and

liabilities of private-sector entities is based on the price that would be

received to sell an asset or paid to transfer a liability in an orderly

transaction between market participants. For those who advocate the

application of private-sector accounting principles to the public sector,

employing private-sector GAAP in estimating the costs of federal credit

programs is attractive. However, any parallels between the public and

private-sector settings are not straightforward. Consequently, the

suggestion to apply private-sector GAAP to federal budgeting for credit

programs creates the potential for misrepresentation of the costs of

government programs.

While the concepts behind the fair value approach and private-sector

GAAP fair value are similar, there are differences in how the two would be

applied. First, under private-sector GAAP, fair value is defined as the

price that would be received to sell an asset or paid to transfer a liability in

an orderly transaction between market participants. In contrast, because

private-sector GAAP fair value—which is based on market prices—

includes costs that may not be relevant to the federal government, such

as liquidity costs,

the fair value approach extracts aggregate risk from

overall market risk to recognize the risk applicable to federal credit

programs in the view of a taxpayer. Consequently, the fair value approach

does not reflect the price that would be received to sell an asset or paid to

transfer a liability in an orderly transaction between market participants

because the aggregate risk premium recognized under the fair value

approach represents only a portion of the risks and other factors that are

reflected in market prices.

Liquidity cost represents compensation to investors for holding an asset that may be

more difficult to sell quickly than Treasury securities of a corresponding maturity.

Federal Generally Accepted Accounting

Principles (U.S. GAAP)

U.S. GAAP is established by the Federal

Accounting Standards Advisory Board

(FASAB).

The primary accounting standard for federal

credit programs is Statement of Federal

Financial Accounting Standards No. 2,

Accounting for Direct Loans and Loan

Guarantees, which became effective in fiscal

year 1994.

When FASAB established this accounting

standard, it stated that it recognized the value

of having financial accounting support the

budget. It also endorsed the logic underlying

the Federal Credit Reform Act of 1990

(FCRA) and recommended that accounting

standards for credit programs be consistent

with budgeting under FCRA.

Source: GAO analysis of U.S. GAAP. | GAO-16-41

Page 52 GAO-16-41 Credit Subsidy Cost Estimates

Second, under private-sector GAAP, generally loans that a bank holds for

investment are recorded at amortized cost, net of an impairment

allowance for estimated credit losses.

Such loans typically comprise the

bulk of assets held by banks insured by the Federal Deposit Insurance

Corporation. However, loans originated with the intent to sell in the

secondary market to government-sponsored entities and other investors

are measured at the lower of cost or fair value, unless the institution has

elected to account for the loans at fair value under the fair value option.

The private-sector GAAP fair value measurement provides information on

the current value of the asset that is available for sale. In contrast, the fair

value approach would be applied to all credit programs, regardless of

whether the government plans to sell assets or hold them to maturity.

Currently, Treasury is the only federal agency routinely selling some of its

credit programs’ assets.

However, under FCRA and in accordance with

OMB Circular No. A-11, Preparation, Submission, and Execution of the

Budget, if an agency were to plan to sell loan assets, the cash flows

developed to estimate subsidy costs would include assumptions related

to expected sale proceeds, which could be based on market information.

Essentially, amortized cost is outstanding principal adjusted for any charge-offs,

deferred fees or costs, and unamortized discount or premium.

TARP was required by law to estimate the budgetary cost of purchases and guarantees

of troubled assets in accordance with FCRA, except that such costs were required to be

calculated by adjusting the discount rate for market risk. In addition, for financial

accounting purposes, TARP’s investment programs were measured based on private-

sector GAAP. However, the intention of TARP was to liquidate the assets when

economically feasible rather than hold them to maturity, and private-sector GAAP fair

value was an appropriate measure of expected proceeds to be received from liquidation

for financial accounting purposes. Therefore, TARP’s use of private-sector GAAP fair

value is not directly comparable to its use for other federal credit programs.

Private-Sector Generally Accepted

Accounting Principles (GAAP) Fair Value

To increase consistency and comparability in

GAAP fair value measurements and related

disclosures, GAAP requires entities to

disclose fair value information based on a

hierarchy of inputs used to determine fair

values. The fair value hierarchy includes three

levels of inputs:

• The highest priority—or level 1—refers to

fair values derived from direct market

prices for the specific asset (e.g., quoted

prices of common stock).

• Level 2 refers to fair values derived from

market observable data, other than a

direct market quote, for a similar asset

from the same entity.

• Level 3—the lowest priority—refers to fair

values derived from unobservable inputs,

which are primarily derived from

management’s best estimate of how a

market participant would assess the risk

inherent in the asset. These unobservable

inputs are used because there is little to

no direct market activity.

What percentage of large domestic

financial institutions’ loan portfolios are

measured using GAAP fair value?

Financial institution

Percentage of loan

portfolio measured

at GAAP fair value

JP Morgan Chase

Bank of America

Wells Fargo

Citigroup

U.S. Bancorp

As of the end of 2013, the five largest (based

on total assets) domestic private-sector

financial institutions used GAAP fair value for

approximately 2 percent or less of their loan

portfolios and these fair values were classified

as level 2 or level 3.

Source: GAO analysis of private-sector GAAP and financial

institutions’ financial statements and note disclosures as of

the end of 2013. | GAO-16-41

Page 53 GAO-16-41 Credit Subsidy Cost Estimates

GAO believes the current FCRA methodology is more appropriate to

estimate credit program subsidy costs for the budget. The construction,

use, and interpretation of the federal budget as a system of primarily cash

accounts have been the norm for decades. Subsidy cost estimates

prepared under the fair value approach would not be consistent with

federal budgeting practices of recognizing expected cash outlays.

Further, the application of the fair value approach to only credit programs,

the loss in transparency and in the ability to verify the noncash cost, and

significant implementation issues argue against change from the FCRA

approach. In addition, to the extent that agencies were consistently

underestimating subsidy costs under FCRA, this kind of bias would be

more appropriately addressed through improvements in the subsidy

estimation process, rather than application of the fair value approach.

However, based on our analyses of reestimate data for the fiscal years

2001 through 2014 cohorts, we did not identify any consistent trends in

overall under- or overestimates of subsidy costs across federal credit

programs government-wide. Although the fair value approach may be

useful to decision makers for evaluating the costs against the benefits of

federal credit programs in that it accounts for market risks beyond those

risks already recognized under FCRA, and the provision of this type of

information is consistent with OMB’s guidance in OMB Circular No. A-129

requiring biennial reviews of credit programs, it is not appropriate for use

in estimating credit subsidy costs for the budget.

We provided a draft of this report to CBO and OMB for their review and

comment. Both CBO and OMB provided technical comments, which we

have incorporated as appropriate.

As agreed with your office, unless you publicly announce the contents of

this report earlier, we plan no further distribution until 30 days from the

report date. At that time, we will send copies to other interested parties. In

addition, the report will be available at no charge on the GAO website at

http://www.gao.gov.

Concluding

Observations

Agency Comments

Page 54 GAO-16-41 Credit Subsidy Cost Estimates

If you or your staff have any questions about this report, please contact

Cheryl E. Clark at (202) 512-9377 or [email protected], Susan J. Irving at

(202) 512-6806 or [email protected], or Susan Offutt at (202) 512-3763 or

offutts@gao.gov. Contact points for our Offices of Congressional

Relations and Public Affairs may be found on the last page of this report.

GAO staff members who made key contributions to this report are listed

in appendix IV.

Sincerely yours,

Cheryl E. Clark

Director

Financial Management and Assurance

Susan J. Irving, PhD

Director for Federal Budget Analysis

Strategic Issues

Susan Offutt, PhD

Chief Economist

Applied Research and Methods

Appendix I: Objectives, Scope, and

Methodology

Page 55 GAO-16-41 Credit Subsidy Cost Estimates

To determine whether there were similarities in the size and direction of

reestimates across, or within, federal credit programs and, based on this

analysis of reestimates, to identify factors, if any, that help explain any

significant trends in reestimates, we examined reestimate data reported

for all direct loan and loan guarantees by cohort for fiscal years 2001

through 2014. We chose loan cohorts beginning in fiscal year 2001 to

ensure that our analysis covered a full business cycle. Reestimate data

were obtained from the President’s Budgets for fiscal years 2003 through

2016. Specifically, we used tables 7 and 8 reestimate data from the

Federal Credit Supplement (FCS) to the Budget of the U.S. Government

for fiscal years 2003 through 2016. The FCS for fiscal year 2003 is the

first year in which data for the fiscal year 2001 cohorts were reported.

Because of the 2-year lag between the timing of the loan cohort

disbursement and the reporting in the FCS, the fiscal year 2014 cohort of

direct loans and loan guarantees (from the fiscal year 2016 FCS) is the

most recent cohort in our data.

The reestimate data were in spreadsheets that reproduce the data as

they appear in the FCS. In spreadsheets, the first column contains five

different fields: (1) the agency name, (2) the bureau name, (3) the

program name, (4) the risk category name, and (5) the cohort year. We

developed a computer algorithm to distinguish among these fields and

identified the unique agency, bureau, and risk category combination as a

“program.” We retained this information for each cohort of direct loans

and loan guarantees.

We reviewed the agency, bureau, and risk category data manually to look

for inconsistencies such as spelling and formatting differences that could

affect the number of distinct direct loan and loan guarantee programs

identified by our computer algorithm. We found that some names were

reported differently over time. For example, the risk category Farm

Storage Facility Loan Program reported in the fiscal year 2003 to 2008

FCSs is the same as the risk category Farm Storage Facility Loans

reported in the fiscal year 2009 to 2016 FCSs. In another example, the

Indian Land Acquisition risk category, reported in fiscal year 2003 to 2008

FCSs, was called Indian Tribe Land Acquisition in subsequent years’

FCSs. In cases such as these, we standardized the program names to

facilitate our analysis of credit programs over time. We cleaned the data

to remove footnotes, standardized the coding of missing values, and

reformatted numeric values to permit mathematical calculations, as

appropriate. For example, numbers saved as text characters were

multiplied by 1 to obtain values that could be added, subtracted,

multiplied, or divided.

Appendix I: Objectives, Scope, and

Methodology

Appendix I: Objectives, Scope, and

Methodology

Page 56 GAO-16-41 Credit Subsidy Cost Estimates

We supplemented data from the FCS with information on whether

programs were mandatory or discretionary, which we obtained from

tables 1 and 2 of the FCS or based on research of the program’s

budgetary funding type. We also used professional judgment to classify

the primary purpose of each program as one of seven possible

categories: (1) education-related; (2) energy, transportation, and

infrastructure; (3) housing; (4) international; (5) small business and

agriculture; (6) other; and (7) the Troubled Asset Relief Program (TARP).

Because TARP was a unique, one-time program implemented in

extraordinary circumstances, and because the amount of its lifetime

downward reestimate was so large in comparison to all other credit

programs, we excluded TARP from our analysis.

From the spreadsheets, we created a single file of cohort-level direct loan

data and a single file of cohort-level loan guarantee data. These data

were the basis for our examination of trends in reestimates across

programs and over time. We examined patterns in direct loan and loan

guarantee programs by agency, by program, by purpose, by fiscal year

the direct loan or loan guarantee was approved, and by program funding

type (mandatory and discretionary). Our analysis focused on lifetime

reestimate amounts, annual reestimate amounts, and a comparison of

most recently reported reestimated and original subsidy rates for each

program. For each year of the FCS, we calculated program-level lifetime

reestimate amounts as the sum of the lifetime reestimate amounts

reported for each cohort in the program. We examined the direction of the

most recent lifetime reestimate amounts by purpose and by agency. We

identified programs with relatively large lifetime reestimate amounts and

examined their influence on upward and downward reestimates

government-wide. For each program, we calculated the annual reestimate

amount as the annual change in the program-level lifetime reestimate

amounts. Our conclusions about possible trends were based on our

professional judgment in assessing the data, and not based on a

statistical analysis.

We also examined reestimated subsidy rates for each program relative to

the most recently reported original subsidy rate. For each FCS, we

calculated these program-level relative reestimated rates as a weighted

We did not use statistical analysis to estimate trends because, in our professional

judgment, the quantity of data (e.g., number of years and number of programs) was

insufficient to provide a credible statistical result.

Appendix I: Objectives, Scope, and

Methodology

Page 57 GAO-16-41 Credit Subsidy Cost Estimates

average of the cohort-level data. In these calculations, differences

between the reestimated rate and the most recently reported original

subsidy rate were weighted by the size of each cohort, as measured by

the cohort’s reported total disbursements to date.

In assessing data reliability, we encountered some data limitations, which

we individually evaluated based on professional judgment. For each data

limitation, we adjusted the data, identified alternative data calculation

methods when available, or determined that the data were reliable for our

purposes. For example, not all programs reported data in the FCS each

year. In addition, one program did not report reestimate information by

cohort. Instead, all reestimates were combined and reported with the

most recent cohort. We also noted some discrepancies between the

lifetime reestimate amount reported for each cohort and the running total

of the annual reestimate amounts that had been reported since the cohort

of direct loans or loan guarantees was first disbursed.

These

discrepancies tended to be less prevalent for later cohort years because

of quality controls implemented by the Office of Management and Budget

(OMB) in fiscal year 2008.

To evaluate implications of using subsidy cost estimates developed under

the fair value approach for the budget and to determine whether we

believe such concepts should be incorporated into subsidy cost estimates

for the budget, we reviewed literature related to the Federal Credit

Reform Act of 1990 (FCRA), the budget, and the fair value approach. This

literature included GAO reports, Congressional Budget Office reports,

Congressional Research Service reports, OMB guidance and budget

information, the 1967 Report of the President’s Commission on Budget

Concepts, legislation, position papers, and academic reports. We

identified a diverse group of individuals or organizations with expertise in

FCRA, the fair value approach, financial economics, and federal

budgeting through our review of relevant reports in these issue areas,

based on internal discussions with GAO staff familiar with experts in this

area, and based on recommendations from experts interviewed. We

considered expert experience, work history, and published work relevant

to our engagement to help ensure expertise qualifications and to achieve

a balanced review. We conducted semistructured interviews with 30

To adjust for this limitation, rather than using the reported annual reestimate amount, we

calculated the annual reestimate amounts based on differences in the reported lifetime

reestimate amounts.

Appendix I: Objectives, Scope, and

Methodology

Page 58 GAO-16-41 Credit Subsidy Cost Estimates

experts to achieve a variety of expertise and viewpoints. The views

expressed by the experts should be interpreted in the context of the

following qualification. Although we were able to secure the participation

of a balanced, highly qualified group of individuals, we could not interview

all of the individuals with expertise in relevant fields because of the need

to limit the number of interviews conducted.

We conducted this performance audit from June 2014 to January 2016 in

accordance with generally accepted government auditing standards.

Those standards require that we plan and perform the audit to obtain

sufficient, appropriate evidence to provide a reasonable basis for our

findings and conclusions based on our audit objectives. We believe that

the evidence obtained provides a reasonable basis for our findings and

conclusions based on our audit objectives.

Appendix II: Analysis of Reestimates

Page 59 GAO-16-41 Credit Subsidy Cost Estimates

We analyzed direct loan and loan guarantee programs’ annual and

lifetime reestimates by cohort, program, and agency purpose from fiscal

years 2001 through 2014 to identify any trends. We also considered the

purpose of the credit programs based on the following categories:

(1) education-related; (2) energy, transportation, and infrastructure;

(3) housing; (4) international; (5) small business and agriculture; and

(6) other. We excluded the Department of the Treasury’s (Treasury)

Troubled Asset Relief Program (TARP) from our analysis because TARP

was a unique, one-time program implemented in extraordinary

circumstances. Further, the size of TARP downward reestimates dwarfs

the reestimates of all other credit programs, and its inclusion would make

any trend analysis less meaningful. The amounts discussed in this

appendix are based on information reported in the President’s Budgets.

We analyzed the reestimates for fiscal years 2001 through 2014 by

agency. Table 4 illustrates the agencies with direct loan programs, the

total loan disbursements from these programs during the period, the

number of direct loan programs at each agency, the number of programs

that had either upward or downward lifetime reestimates, and the amount

of the upward or downward reestimates. For example, the Department of

Education (Education) had seven direct loan programs, with one program

comprising lifetime upward reestimates of $23.1 billion and six programs

comprising lifetime downward reestimates of $7.7 billion. Overall,

Education’s direct loan programs represented a net lifetime upward

reestimate of $15.4 billion. In contrast, Treasury had five direct loan

programs, with two programs comprising lifetime upward reestimates of

$2.4 billion and three programs comprising lifetime downward reestimates

of $16.4 billion. Overall, Treasury’s direct loan programs represented a

net lifetime downward reestimate of $14.0 billion.

Appendix II: Analysis of Reestimates

Reestimates by Agency

Appendix II: Analysis of Reestimates

Page 60 GAO-16-41 Credit Subsidy Cost Estimates

Table 4: Direct Loan Programs’ Lifetime Reestimates by Agency, 2001-2014

Dollars in millions

Agency Programs

Programs

with lifetime

upward

reestimate

Programs

with lifetime

downward

reestimate

Loan amounts

disbursed

Lifetime

upward

reestimate

Lifetime

downward

reestimate

Net lifetime

reestimate

Department of

Education

$940,574 $23,067

$(7,670)

$15,397

Export-Import Bank

29,329

2,843

Small Business

Administration

15,114 1,078

(13)

1,065

International

programs

8,859 531

531

Department of

Veterans Affairs 6

4,545 72

(13)

Department of Health

and Human Services 4

1,374 44

(9)

Overseas Private

Investment

Corporation

2,829 20

Federal

Communications

Commission

3 0

(1)

Department of the

Interior 2

29 1

(6)

(5)

Department of Housing

and Urban

Development

83 0

(10)

Department of State

(15)

Department of

Commerce 7

756 5

(43)

(38)

Department of

Transportation 3

6,910 39

(110)

(71)

Department of

Agriculture 45

114,008 1,944

(2,020)

(76)

Department of

Homeland Security 2

1,196 47

(259)

(212)

Department of Defense

1,464

(245)

Department of Energy

17,444

484

(2,678)

(2,194)

Department of the

Treasury

244,956 2,401

(16,382)

(13,981)

Total

101

$1,389,499

$32,576

$(29,474)

$3,102

Source: GAO analysis of President’s Budgets. | GAO-16-41

Appendix II: Analysis of Reestimates

Page 61 GAO-16-41 Credit Subsidy Cost Estimates

International programs include International Monetary Programs and International Security

Assistance.

Department of the Treasury excludes the Troubled Asset Relief Program reestimates.

Table 5 summarizes the lifetime reestimates for agencies with loan

guarantee programs for fiscal years 2001 through 2014. As shown, the

Department of Housing and Urban Development and Education

accounted for the largest net lifetime upward and downward reestimates,

respectively.

Table 5: Loan Guarantee Programs’ Lifetime Reestimates by Agency, 2001-2014

Dollars in millions

Agency Programs

Programs

with lifetime

upward

reestimate

Programs

with lifetime

downward

reestimate

Value of loan

amounts

guaranteed

Lifetime

upward

reestimate

Lifetime

downward

reestimate

Net lifetime

reestimate

Department of

Housing and Urban

Development

4 4

$7,171,143

$85,958 $(2,000)

$83,958

Small Business

Administration 19

10 9

312,263

10,170 (1,025)

9,145

Department of

Veterans Affairs 2

1 1

860,273

4,795 (105)

4,690

Department of

Agriculture 21

12 9

235,542

3,162 (1,851)

1,311

U.S. Agency for

International

Development

3 4

11,360

426 (40)

386

Department of the

Treasury

1 0

1,627

22 0

Department of

Energy 1

1 0

3,740

2 0

Department of

Defense 2

1 1

702

14 (16)

(2)

Department of the

Interior 2

0 2

901

0 (21)

(21)

Department of

Transportation

1 1

2,503

0 (27)

(27)

Department of

Health and Human

Services

0 3

578

0 (40)

(40)

Department of

Commerce 2

0 2

406

0 (102)

(102)

Export-Import Bank

166,910

(338)

Appendix II: Analysis of Reestimates

Page 62 GAO-16-41 Credit Subsidy Cost Estimates

Dollars in millions

Agency Programs

Programs

with lifetime

upward

reestimate

Programs

with lifetime

downward

reestimate

Value of loan

amounts

guaranteed

Lifetime

upward

reestimate

Lifetime

downward

reestimate

Net lifetime

reestimate

Overseas Private

Investment

Corporation

1 3

12,684

21 (415)

(394)

Department of

Education 1

0 1

700,143

0 (59,552)

(59,552)

Total

$9,480,775

$104,570

$(65,532)

$39,038

Source: GAO analysis of President’s Budgets. | GAO-16-41

This amount includes the total amount of single- and multifamily loan guarantees issued by the

Department of Housing and Urban Development (HUD) over the 14-year period. If a loan guaranteed

under a HUD program is refinanced and the new loan is guaranteed under a HUD program, both

guaranteed loan amounts would be included in this total. In addition, this amount includes guarantees

issued by the Government National Mortgage Association (Ginnie Mae). Ginnie Mae guarantees the

timely payment of principal and interest on securities issued by financial institutions and backed by

pools of federally insured or guaranteed mortgage loans. Therefore, this amount would include the

initial value of the security guaranteed by Ginnie Mae, as well as the initial value of the underlying

federally guaranteed loan.

Department of the Treasury excludes the Troubled Asset Relief Program reestimates.

The Health Education Assistance Loan program is listed under the Department of Health and Human

Services in this table. In 2014, this program was transferred to the Department of Education. The

value of loan amounts guaranteed under this program was $450 million, and the program had

$36 million in downward reestimates.

We also analyzed the reestimates for fiscal years 2001 through 2014

based on the program purpose. Table 6 summarizes direct loan program

reestimates by purpose. Direct loan programs related to education and

international lending experienced net lifetime upward reestimates. The

other category, which represents programs that do not fit in the other

categories, including Treasury’s Government Sponsored Enterprises

Mortgage Backed Securities Purchase Program and the Small Business

Administration’s (SBA) Disaster Loan Program experienced net lifetime

downward reestimates.

Reestimates by Program

Purpose

Appendix II: Analysis of Reestimates

Page 63 GAO-16-41 Credit Subsidy Cost Estimates

Table 6: Lifetime Upward and Downward Reestimates by Type of Direct Loan Program, 2001-2014

Dollars in millions

Type of

direct loan program

Programs

with lifetime

upward

reestimate

Programs

with lifetime

downward

reestimate

Loan amounts

disbursed

Lifetime

upward

reestimate

Lifetime

downward

reestimate

Net lifetime

reestimate

Education-related

$940,607

$23,067

$(7,670)

$15,397

International

41,402

3,393

(89)

3,304

Housing

29,634

478

(398)

Small business and

agriculture 13 1

12 20,264

(1,024)

(1,000)

Energy, transportation,

and infrastructure 25 13

12 94,462

2,038

(3,577)

(1,539)

Other

263,130

3,576

(16,716)

(13,140)

Total

101

$1,389,499

$32,576

$(29,474)

$3,102

Source: GAO analysis of President’s Budgets. | GAO-16-41

As shown in table 7, loan guarantee programs related to housing and

small business and agriculture experienced net lifetime upward

reestimates, while loan guarantee programs related to education

experienced net lifetime downward reestimates.

Table 7: Lifetime Upward and Downward Reestimates by Type of Loan Guarantee Program, 2001-2014

Dollars in millions

Type of

loan guarantee

program

Programs

with lifetime

upward

reestimate

Programs

with lifetime

downward

reestimate

Value of loan

amounts

guaranteed

Lifetime

upward

reestimate

Lifetime

downward

reestimate

Net lifetime

reestimate

Housing

$8,164,739

$93,580

$(2,121)

$91,459

Small business and

agriculture 27

400,131

10,183

(2,826)

7,357

Energy, transportation,

and infrastructure 14

21,681

325

(180)

145

Other

2,677

(25)

International

190,954

447

(792)

(345)

Education-related

700,593

(59,588)

Total

$9,480,775

$104,570

$(65,532)

$39,038

Source: GAO analysis of President’s Budgets. | GAO-16-41

Appendix II: Analysis of Reestimates

Page 64 GAO-16-41 Credit Subsidy Cost Estimates

We also compared the most recently reported original and reestimated

subsidy rates for direct loan and loan guarantee programs.

To calculate

the difference between the original and reestimated subsidy rates for

each program, the difference for each cohort of direct loans or loan

guarantees was weighted based on reported disbursements for direct

loans or guaranteed loan amounts disbursed for loan guarantees. As

shown in figure 14, we found that for most of the programs, reestimated

subsidy rates were within 10 percentage points of the original subsidy

rates. For example, if the calculated original subsidy rate for a program

was 12 percent, with a 10 percentage point difference, the calculated

reestimated subsidy rate would fall between 2 percent and 22 percent.

However, as indicated in figure 14, there were a few programs that had

larger differences between the most recently reported original and

reestimated subsidy rates. For example, the Department of Energy’s

Advanced Technology Vehicles Manufacturing direct loan program

reported an overall lifetime downward reestimate of $2.7 billion mostly

attributable to a significant drop in the credit subsidy cost estimate for one

loan, as a result of a significantly improved credit rating, reducing the

original subsidy rate by 41.2 percentage points, from 45.3 percent to a

reestimated rate of 4.1 percent.

As another example, SBA’s Small

Business Investment Company (SBIC) participating securities loan

guarantee program reported an overall lifetime upward reestimate of

$1.7 billion, primarily because of reestimates for fiscal years 2009 and

2010. SBA reported that these reestimates were due to the downturn in

the economy during the financial crisis that resulted in lower-than-

projected recoveries.

For SBIC, the calculated original subsidy rate of

0.3 percent increased to a calculated reestimated rate of 24.7 percent.

We used the most recently reported reestimated rate as it represents the most current

estimate of the subsidy rate updated for the actual program performance. The difference

between the most recent reestimated and the most recently published original rates

indicates how accurate the initial cost estimate is based on the most recently available

information.

The Advanced Technology Vehicles Manufacturing loan guarantee program supports the

development of advanced technology vehicles and associated components in the United

States.

The SBIC participating securities loan guarantee program facilitates the flow of long-term

capital to America’s small businesses.

Differences between

Original and Reestimated

Subsidy Rates

Appendix II: Analysis of Reestimates

Page 65 GAO-16-41 Credit Subsidy Cost Estimates

Figure 14: Percentage Point Differences between Direct Loan and Loan Guarantee Programs’ Original and Most Recent

Reestimated Subsidy Rates, 2001-2014

Appendix III: Fair Value Approach

Implementation Considerations

Page 66 GAO-16-41 Credit Subsidy Cost Estimates

Based on our review of literature and interviews with individuals

knowledgeable about the Federal Credit Reform Act (FCRA), the federal

budget, and the fair value approach, we identified a number of issues that

would need to be considered if the fair value approach were implemented

for credit program subsidy cost estimates in the federal budget.

The primary task in the fair value approach to budgeting for credit

programs is identifying the risk premium applicable to any particular

program. Generally, most proponents of the fair value approach whom we

interviewed suggested that private market information could be used to

determine the risk premium. If the type of federal lending has a close

counterpart in private credit markets, the difficulty of this task is reduced,

but not eliminated. However, most proponents of the FCRA methodology

suggested that it would be difficult to find a private market comparable for

federal credit programs because the federal government tends to

intervene in either inefficient or nonexistent markets. In the absence of a

private market comparable, the analytical task of arriving at an

appropriate aggregate risk premium would be significant and subjective.

Moreover, even when a market comparable exists, the task of extracting

the aggregate risk premium from the difference between the default-free

Treasury interest rate and the observed market rate would be subjective.

This complication arises because some of the costs and risks that may be

compensated in the market rate might be significant to an individual

investor but not relevant to the federal government, such as a more or

less favorable tax treatment or liquidity. Overall, the experts we

interviewed could not provide clear insight into how aggregate risk

premium applicable to the federal government could be extracted from

the market rate. Arriving at a verifiable and auditable estimate of the

aggregate risk premium for each credit program is essential if there is to

be confidence in decisions that depend on comparing their relative costs.

Under FCRA, when calculating subsidy cost estimates, all agencies use

the same methodology to determine discount rates, which are based on

interest rates of Treasury securities. Under the fair value approach,

discount rates would be derived from a market comparable, if available,

and each agency could use different methodologies to determine the

aggregate risk premium. Of the experts we interviewed, proponents of the

FCRA methodology generally believed that because of the addition of the

risk premium in the fair value approach discount rate, fair value approach

estimates would be more subjective than FCRA estimates, in part

because more assumptions would be involved. Some proponents of the

Appendix III: Fair Value Approach

Implementation Considerations

Determining the Aggregate

Risk Premium

Subjectivity

Appendix III: Fair Value Approach

Implementation Considerations

Page 67 GAO-16-41 Credit Subsidy Cost Estimates

fair value approach, on the other hand, believed that fair value approach

cost estimates would be as, or less, subjective than FCRA estimates

because market information would inform the cost estimates. If the fair

value approach resulted in more subjective estimates, this could create a

wider variety of estimates, reducing transparency and consistency across

programs. This inconsistency would be even greater if each agency

estimating subsidy costs used a different methodology to determine the

aggregate risk premium applicable to a program. Some experts we

interviewed suggested, however, that the Office of Management and

Budget (OMB) or the Department of the Treasury (Treasury) should be

responsible for developing the fair value approach aggregate risk

premium to add on to the discount rate for each credit program, which

could reduce inconsistencies. Other experts suggested that OMB could

develop guidance for agencies on appropriate methods for determining

the aggregate risk premium to help ensure consistent and appropriate

application of the fair value approach across programs.

Most experts we interviewed believed that implementing the fair value

approach would require additional resources for agencies directly

involved in making the subsidy cost estimates and for entities overseeing

the fair value approach estimation process, such as OMB. The types of

additional resources that would be needed include additional technical

training for current employees, hiring of new staff, use of private-sector

experts, and funds to cover additional administrative costs associated

with development of fair value approach methodologies and estimates.

Some experts stated that the resources used to implement the fair value

approach would be better used by agencies to either (1) improve their

estimates of expected cash flows used in both the FCRA or fair value

approach subsidy cost estimates or (2) gain a more thorough

understanding of, and reporting on, the benefits of credit programs.

Under FCRA, the discount rate is locked—meaning that the discount rate

is determined and cannot be changed—after a cohort of loans is

substantially disbursed. This helps ensure that the reasons for

reestimates are limited to changes in estimation methodologies or actual

loan performance, rather than fluctuations in interest rates, which are

outside the control of agencies. Based on our interviews, there were

differing views as to whether the discount rate should be locked or should

fluctuate under the fair value approach. Under the fair value approach, if

the discount rate was allowed to fluctuate, reestimates could fluctuate

significantly year to year based on the market. This volatility in the

Resources

Locking the Discount Rate

Appendix III: Fair Value Approach

Implementation Considerations

Page 68 GAO-16-41 Credit Subsidy Cost Estimates

reestimates would be caused by market conditions, which are outside the

control of agencies. In addition, it would be more difficult to determine

whether reestimates were due to estimates of expected cash flows that

differed from actual loan performance or changes in the aggregate risk

premium. Under the fair value approach, the accuracy of the estimate of

the aggregate risk premium cannot be objectively assessed. For example,

under FCRA agencies perform comparisons of estimated and actual loan

performance to assess the estimation process. Because the aggregate

risk premium reflects a noncash cost, there is no method that can be

used to verify the reasonableness of the aggregate risk premium

estimate. In addition, updating the risk premium increases the

administrative activities and would require additional resources.

Conversely, under the fair value approach if the discount rate were locked

in order to avoid this volatility in the reestimates and to reduce workload

burdens, then the resulting reestimates would be based on out-of-date

risk information. The financial accounting for credit programs currently

mirrors FCRA and budgetary accounting. If the financial accounting were

to be revised to incorporate any changes to implement fair value

approach subsidy cost estimates for the budget, then not updating the

aggregate risk premium would result in cost estimates based on

noncurrent risks. This would be difficult for an agency to support during a

financial statement audit because auditors would expect estimates to be

based on currently available risk information.

As previously discussed, the financial accounting for credit programs is

consistent with the budgeting for credit programs. Auditing credit program

balances as part of an agency’s financial statement audit includes

determining whether direct loans and loan guarantees outstanding are

properly reported in the agency’s financial statements and footnotes. This

auditing goal would not change if the fair value approach were adopted

for financial reporting purposes. One expert noted that with the fair value

approach, external auditors would need to focus on assessing whether

agencies have identified an appropriate private market comparable to

identify the aggregate risk premium, or if the agency claims there was no

market comparable, the auditor will need to verify the reasonableness of

that claim. Under the fair value approach, each agency could identify its

own methodologies to identify a private market comparable and to derive

the aggregate risk premium to add to the discount rate for a particular

program. However, the current cash budgeting process allows federal

credit program officials and auditors to see the outflows and inflows of

accounts and permits the comparison of actual cash flows to estimated

cash flows with the goal of verifying the reasonableness of subsidy cost

Auditability

Appendix III: Fair Value Approach

Implementation Considerations

Page 69 GAO-16-41 Credit Subsidy Cost Estimates

estimates and improving estimates going forward. With the fair value

approach, there would be no mechanism to verify the reasonableness of

the aggregate risk premium because it does not correlate to any cash

flow of the program. Depending on how well agencies were able to

(1) find a private-sector comparable, (2) establish methodologies to derive

the aggregate risk premium, and (3) adequately document their decision-

making process and resulting aggregate risk premiums, this could create

financial statement auditability issues for agencies.

Under the fair value approach, the aggregate risk premium included in

discount rates would be derived from a market comparable, if available. If

no market comparable is available, this task would be more complicated

and assumptions would need to be made. Some experts we spoke with

believed that market information could vary more widely than interest

rates on Treasury securities, resulting in greater volatility in original

subsidy estimates year to year. In addition, depending on how the fair

value approach is implemented during the reestimate process, if the

aggregate risk premium is updated for annual reestimates, many experts

we interviewed believed that reestimates would be more volatile from

year to year. If the fair value approach resulted in more volatile subsidy

estimates and reestimates, this could lead to greater swings in the deficit

than are currently experienced under FCRA.

Under the fair value approach, the financing account would include the

subsidy cost associated with the noncash cost reflected in the aggregate

risk premium. Therefore, when a loan is fully repaid, and if actual cash

flows were as initially estimated, there would be an additional amount in

the financing account that represents a social cost and not a cash cost to

Treasury. Social costs do not produce fiscal effects insofar as they are

not cash receipts or disbursements from Treasury. Some proponents of

the FCRA methodology referred to this as a phantom cost, which could

cause swings in the federal deficit depending on how the social cost is

cleared from the financing account. This could (1) occur immediately after

the loans are disbursed, (2) be amortized over the life of the loan cohort

through the reestimate process, or (3) be cleared by a closing reestimate

after the entire loan cohort was repaid. To avoid misrepresenting the

fiscal effect on the federal deficit, the social cost in the financing account

would have to be liquidated by means of an offsetting “phantom” receipt

immediately after the loans are disbursed. If the social cost is amortized

over the life of a cohort or cleared at the end of the cohort, this would be

reflected in downward reestimates for programs. Several of the

Volatility

Timing of Recognition of

the Fair Value Approach

Downward Reestimate

Appendix III: Fair Value Approach

Implementation Considerations

Page 70 GAO-16-41 Credit Subsidy Cost Estimates

proponents of the fair value approach whom we interviewed were

indifferent as to when this downward reestimate took place as it would be

outside the congressional decision-making process.

Under FCRA, the discount rate is also used as the interest rate to

calculate interest income and expense on financing account balances so

that the financing account will break even over time as it uses its

collections to repay its Treasury borrowing. Under the fair value

approach, the financing account interest rate could continue to be based

on the FCRA-defined discount rate based on interest rates on Treasury

securities, as was done with Troubled Asset Relief Program, or it could be

based on discount rates including the aggregate risk premium. If the

financing account interest rate is based on the FCRA discount rate, then

recording the annual interest income and expense on the financing

account helps to amortize the additional amount in the financing account

that represents the noncash social cost. If the interest rate includes the

aggregate risk premium, this would in essence monetize the social cost

but would result in interest payments between agencies and Treasury to

be based on rates not tied to Treasury borrowing costs. Based on our

interviews, most experts did not have a conclusive answer or did not

provide an answer as to what the financing account interest rate should

be under the fair value approach. Many proponents of the fair value

approach did not have an opinion on which rate should be used because

the financing account transactions are outside the congressional

decision-making process.

Some federal credit programs, such as the Department of Energy’s Loan

Guarantee Program, have statutory mandates to charge borrowers fees

at a level that will result in an initial subsidy cost of zero. Other programs,

such as the Department of Housing and Urban Development’s Mutual

Mortgage Insurance Fund loan guarantee program, have statutory

mandates to charge fees at a level to help establish reserves for

unexpected future losses. For programs with these types of requirements,

agencies estimate expected loan performance cash flows and then

determine fees that are necessary to generate the required subsidy rate.

Under the fair value approach, while the expected loan performance cash

flows would be the same, because the discount rate includes the

aggregate risk premium, the cost appears higher. As a result, agencies

would need to charge borrowers higher fees to cover the initially

estimated noncash social costs of a program. However, these noncash

social costs would then be offset, or cleared, during the reestimate

Financing Account Interest

Rate

Programs with Specific

Subsidy Rate

Requirements

Appendix III: Fair Value Approach

Implementation Considerations

Page 71 GAO-16-41 Credit Subsidy Cost Estimates

process, meaning that borrowers would be charged more than is needed

to cover the initially estimated cash costs of the program. These higher

fees could affect borrower demand for the credit program. The experts we

interviewed stated that this was a congressional decision on how fees

should be charged. The experts also believed that borrower fees should

not be returned if the estimated cost of a program turned out to be lower

than originally estimated in part because of the administrative burden and

the fact that if costs were underestimated, the government would not

request more fees from borrowers.

Recent concurrent budget resolutions have included spending cap limits

on discretionary spending, which are enforceable during the

congressional budget process. One expert pointed out that when FCRA

was implemented, there was an adjustment to the spending caps then in

place to reflect this “change in concept.” If the fair value approach were

implemented and were considered a “change in concept,” the Budget

Control Act of 2011 provided that the President’s budget could include

adjustments to discretionary spending limits, subject to consultation with

the Senate and House, to reflect this conceptual change.

Because of the

expected higher costs under the fair value approach, some experts

speculated that these caps would likely be raised.

The Federal Accounting Standards Advisory Board (FASAB)

established

the primary accounting standard for federal credit programs, Statement of

Federal Financial Accounting Standards No. 2, Accounting for Direct

Loans and Loan Guarantees, which became effective in fiscal year 1994.

When this standard was issued, FASAB stated that it recognized the

value of having financial accounting support the budget. It also endorsed

the logic underlying FCRA and recommended that accounting standards

for credit be consistent with budgeting under FCRA. FASAB also stated

that as more experience is gained, some modifications in budgetary

requirements may be needed and that so long as the modifications are

Pub. L. No. 112-25, § 101 (Aug. 2, 2011) (amending the Balanced Budget and

Emergency Deficit Control Act of 1985, Pub. L. No. 99-187, § 251(b)(1) (Dec. 12, 1985)),

classified as amended at 2 U.S.C. § 901(b)(1).

FASAB was created by OMB, Treasury, and GAO to develop accounting standards for

the federal government. These accounting standards are considered generally accepted

accounting principles for federal entities.

Spending Caps

Consistency between

Federal Budget and

Financial Accounting

Standards

Appendix III: Fair Value Approach

Implementation Considerations

Page 72 GAO-16-41 Credit Subsidy Cost Estimates

made on a credit reform basis and do not materially affect the basic

recognition and measurement principles embodied in the accounting

standards, it intended that accounting practices for direct loans and loan

guarantees should change as needed to remain consistent with the

budget. If the fair value approach were implemented for the budget,

FASAB would need to determine if that approach is consistent with the

basic recognition and measurement principles embodied in the

accounting standards. Having the accounting standards for credit

programs mirror the budgeting for credit programs means that the

financial statement audit can help provide assurance that the agencies’

budgeted amounts are reasonable.

Appendix IV: GAO Contacts and Staff

Acknowledgments

Page 73 GAO-16-41 Credit Subsidy Cost Estimates

Cheryl E. Clark, (202) 512-9377 or clarkce@gao.gov

Susan J. Irving, (202) 512-6806 or irvings@gao.gov

Susan Offutt, (202) 512-3763 or [email protected]

In addition to the contacts named above, Marcia Carlsen (Assistant

Director), Carol Henn (Assistant Director), Ardith Spence (Assistant

Director), Maria Belaval, Mark Cheung, David Chrisinger, Oliver Culley,

Robert Dacey, Francine DelVecchio, Melissa Emrey-Arras, Natasha

Guerra, Cole Haase, Debra Hoffman, Wilfred Holloway, Karen Jarzynka-

Hernandez, Jason Kelly, Jason Kirwan, Dragan Matic, Rebecca Perkins,

Michael Reed, Oliver Richard, Mathew Scirè, Monasha Thompson, Frank

Todisco, Matthew Ward, and Steven Westley made key contributions to

this report.

Appendix IV: GAO Contacts and Staff

Acknowledgments

GAO Contacts

Staff

Acknowledgments

Related GAO Products

Page 74 GAO-16-41 Credit Subsidy Cost Estimates

Rural Housing Service: Progress on GAO Recommendations and

Preliminary Observations on Loan Guarantee Risk Management.

GAO-15-625T. Washington, D.C.: May 19, 2015.

DOE Loan Programs: Current Estimated Net Costs Include $2.2 Billion in

Credit Subsidy, Plus Administrative Expenses. GAO-15-438. Washington,

D.C.: April 27, 2015.

Telecommunications: USDA Should Evaluate the Performance of the

Rural Broadband Loan Program. GAO-14-471. Washington, D.C.:

May 22, 2014.

DOE Loan Programs: DOE Should Fully Develop Its Loan Monitoring

Function and Evaluate Its Effectiveness. GAO-14-367. Washington, D.C.:

May 1, 2014.

Federal Student Loans: Better Oversight Could Improve Defaulted Loan

Rehabilitation. GAO-14-256. Washington, D.C.: March 6, 2014.

Small Business Administration: Actions Needed to Ensure Planned

Improvements Address Key Requirements of the Development Company

(504) Loan Program. GAO-14-233. Washington, D.C.: March 6, 2014.

Federal Student Loans: Borrower Interest Rates Cannot Be Set in

Advance to Precisely and Consistently Balance Federal Revenues and

Costs. GAO-14-234. Washington, D.C.: January 31, 2014.

FHA Mortgage Insurance: Applicability of Industry Requirements Is

Limited, but Certain Features Could Enhance Oversight. GAO-13-722.

Washington, D.C.: September 9, 2013.

COMPETES Reauthorization Act: Federal Loan Guarantees for

Innovative Technologies in Manufacturing. GAO-13-717R. Washington,

D.C.: July 24, 2013.

Department of Energy: Status of Loan Programs. GAO-13-331R.

Washington, D.C.: March 15, 2013.

Department of Energy: Advanced Technology Vehicle Loan Program

Implementation Is Under Way, but Enhanced Technical Oversight and

Performance Measures Are Needed. GAO-11-145. Washington, D.C.:

February 28, 2011.

Related GAO Products

Page 75 GAO-16-41 Credit Subsidy Cost Estimates

Department of Energy: Further Actions Are Needed to Improve DOE’s

Ability to Evaluate and Implement the Loan Guarantee Program.

GAO-10-627. Washington, D.C.: July 12, 2010.

Small Business Administration: Additional Measures Needed to Assess

7(a) Loan Program’s Performance. GAO-07-769. Washington, D.C.:

July 13, 2007.

Mortgage Financing: FHA’s $7 Billion Reestimate Reflects Higher Claims

and Changing Loan and Performance Estimates. GAO-05-875.

Washington, D.C.: September 2, 2005.

SBA Disaster Loan Program: Accounting Anomalies Resolved but

Additional Steps Would Improve Long-Term Reliability of Cost Estimates.

GAO-05-409. Washington, D.C.: April 14, 2005.

Mortgage Financing: Actions Needed to Help FHA Manage Risks from

New Mortgage Loan Products. GAO-05-194. Washington, D.C.:

February 11, 2005.

Export-Import Bank: OMB’s Method for Estimating Bank’s Loss Rates

Involves Challenges and Lacks Transparency. GAO-04-531. Washington,

D.C.: September 30, 2004.

Small Business Administration: Model for 7(a) Program Subsidy Had

Reasonable Equations, but Inadequate Documentation Hampered

External Reviews. GAO-04-9. Washington, D.C.: March 31, 2004.

Department of Education’s Federal Direct Loan Program: Status of

Recommendations to Improve Cost Estimates and Presentation of

Updated Cash Flow Information. GAO-04-567R. Washington, D.C.:

March 29, 2004.

Student Loan Programs: As Federal Costs of Loan Consolidation Rise,

Other Options Should Be Examined. GAO-04-101. Washington, D.C.:

October 31, 2003.

Maritime Administration: Weaknesses Identified in Management of the

Title XI Loan Guarantee Program. GAO-03-657. Washington, D.C.:

June 30, 2003.

Related GAO Products

Page 76 GAO-16-41 Credit Subsidy Cost Estimates

Small Business Administration: Accounting Anomalies and Limited

Operational Data Make Results of Loan Sales Uncertain. GAO-03-87.

Washington, D.C.: January 3, 2003.

Small Business Administration: Section 7(a) General Business Loans

Credit Subsidy Estimates. GAO-01-1095R. Washington, D.C.: August 21,

2001.

Department of Education: Key Aspects of the Federal Direct Loan

Program’s Cost Estimates. GAO-01-197. Washington, D.C.: January 12,

2001.

Department of Veterans Affairs: Credit Costs and Risks of Proposed VA

Small Business Loan Guarantee Program. GAO/GGD-00-158.

Washington, D.C.: June 30, 2000.

Credit Reform: Key Credit Agencies Had Difficulty Making Reasonable

Loan Program Cost Estimates. GAO/AIMD-99-31. Washington, D.C.:

January 29, 1999.

Credit Reform: Greater Effort Needed to Overcome Persistent Cost

Estimation Problems. GAO/AIMD-98-14. Washington, D.C.: March 30,

1998.

Budget Issues: Budgetary Treatment of Federal Credit Programs.

GAO/AFMD-89-42. Washington, D.C.: April 10, 1989.

(191043)

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 website (http://www.gao.gov). Each weekday

afternoon, GAO posts on its website newly released reports, testimony,

and correspondence. To have GAO e-mail you a list of newly posted

products, go to http://www.gao.gov and select “E-mail Updates.”

The price of each GAO publication reflects GAO’s actual cost of

production and distribution and depends on the number of pages in the

publication and whether the publication is printed in color or black and

white. Pricing and ordering information is posted on GAO’s website,

http://www.gao.gov/ordering.htm.

Place orders by calling (202) 512-6000, toll free (866) 801-7077, or

TDD (202) 512-2537.

Orders may be paid for using American Express, Discover Card,

MasterCard, Visa, check, or money order. Call for additional information.

Connect with GAO on Facebook, Flickr, Twitter, and YouTube.

Subscribe to our RSS Feeds or E-mail Updates.

Listen to our Podcasts and read The Watchblog.

Visit GAO on the web at www.gao.gov.

Contact:

Website: http://www.gao.gov/fraudnet/fraudnet.htm

E-mail: fraudnet@gao.gov

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

Katherine Siggerud, Managing Director, [email protected], (202) 512-

4400, U.S. Government Accountability Office, 441 G Street NW, Room

7125, Washington, DC 20548

Chuck Young, Managing Director, youngc1@gao.gov, (202) 512-4800

U.S. Government Accountability Office, 441 G Street NW, Room 7149

Washington, DC 20548

GAO’s Mission

Obtaining Copies of

GAO Reports and

Testimony

Order by Phone

Connect with GAO

To Report Fraud,

Waste, and Abuse in

Federal Programs

Congressional

Relations

Public Affairs

Please Print on Recycled Paper.