Budget Issues: Budgeting for Federal Insurance Programs (Chapter Report,
09/30/97, GAO/AIMD-97-16).

Pursuant to a congressional request, GAO reviewed the budget treatment
of federal insurance programs to assess whether the current cash-based
budget provides complete information on the government's cost and
whether accrual concepts could be used to improve budgeting for these
programs, focusing on: (1) potential approaches for using accrual
concepts in the budget for insurance programs; (2) trade-offs among
different approaches, including the current cash-based budget treatment;
and (3) potential implementation issues such as cost estimation.

GAO noted that: (1) the cash-based budget does not adequately reflect
the government's cost or the economic impact of federal insurance
programs because generally costs are recognized when claims are paid
rather than when the commitment is made; (2) in any particular year, the
cost of the government's insurance commitments may be understated or
overstated because the time between the receipt of program collections,
the occurrence of an insured event, and the final payment of a claim can
extend over many budget periods; (3) decision-making is best informed if
the government recognizes the costs of its commitments at the time it
makes them; (4) for insurance programs, accrual-based budgeting which
would recognize the expected long-term cost of the insurance commitment
at the time the insurance is extended offers the potential to overcome a
number of the deficiencies of cash-based budgeting by improving cost
recognition; (5) in most cases, the risk-assumed approach to accrual
would be analogous to a premium rate-setting process in that it looks at
the long-term expected cost of an insurance commitment at the time the
insurance commitment is extended; (6) in practical terms, however,
attempts to improve cost recognition occur on a continuum since
insurance programs and insurable events vary significantly; (7) the
challenges involved in bringing accrual-based estimates into the budget
are significant and dictate beginning with an informational and analytic
step; (8) development of models to generate reasonably reliable
risk-assumed estimates is made difficult by the nature of the risks
insured by the government, frequent program modifications, and the
sufficiency of data on potential losses; (9) the potential benefits of
accrual-based budgeting for federal insurance programs warrant continued
effort in the development of risk-assumed cost estimates; (10)
supplemental reporting of risk-assumed estimates in the budget as they
are developed over a number of years would help policymakers understand
the extent and nature of the estimation uncertainty and evaluate whether
a more comprehensive accrual-based budgeting approach should be adopted;
(11) supplemental reporting of risk-assumed estimates in the budget
would parallel the new accounting treatment required under accounting
standards developed by the Federal Accounting Standards Advisory Board
(FASAB); and (12) in requiring the disclosure of risk-assumed estimates
as supplemental information to agency financial statements, FASAB
recognized the usefulness of these estimates to better inform budget
decisions.

--------------------------- Indexing Terms -----------------------------

 REPORTNUM:  AIMD-97-16
     TITLE:  Budget Issues: Budgeting for Federal Insurance Programs
      DATE:  09/30/97
   SUBJECT:  Insurance
             Future budget projections
             Budget administration
             Accrual basis accounting
             Presidential budgets
             Cash basis accounting
             Budget authority
             Budget obligations
             Federal agency accounting systems
IDENTIFIER:  VA Service-Disabled Veterans Life Insurance Program
             FAA Aviation War Risk Insurance Program
             Navy Maritime War Risk Insurance Program
             FEMA National Flood Insurance Program
             Federal Employees Group Life Insurance Program
             Savings Association Insurance Fund
             PHS National Vaccine Injury Compensation Program
             VA Veterans Mortgage Life Insurance Program
             PBGC Pension Plan Insurance Program
             SAIF
             Bank Insurance Fund
             BIF
             Federal Crop Insurance Program
             NCUA Share Insurance Fund
             OPIC Political Risk Insurance Program
             
******************************************************************
** This file contains an ASCII representation of the text of a  **
** GAO report.  Delineations within the text indicating chapter **
** titles, headings, and bullets are preserved.  Major          **
** divisions and subdivisions of the text, such as Chapters,    **
** Sections, and Appendixes, are identified by double and       **
** single lines.  The numbers on the right end of these lines   **
** indicate the position of each of the subsections in the      **
** document outline.  These numbers do NOT correspond with the  **
** page numbers of the printed product.                         **
**                                                              **
** No attempt has been made to display graphic images, although **
** figure captions are reproduced.  Tables are included, but    **
** may not resemble those in the printed version.               **
**                                                              **
** Please see the PDF (Portable Document Format) file, when     **
** available, for a complete electronic file of the printed     **
** document's contents.                                         **
**                                                              **
** A printed copy of this report may be obtained from the GAO   **
** Document Distribution Center.  For further details, please   **
** send an e-mail message to:                                   **
**                                                              **
**                                            **
**                                                              **
** with the message 'info' in the body.                         **
******************************************************************


Cover
================================================================ COVER


Report to the Chairman, Committee on the Budget, House of
Representatives

September 1997

BUDGET ISSUES - BUDGETING FOR
FEDERAL INSURANCE PROGRAMS

GAO/AIMD-97-16

Budgeting for Federal Insurance Programs

(935218)


Abbreviations
=============================================================== ABBREV

  AID - Agency for International Development
  BEA - Budget Enforcement Act of 1990
  BFE - base flood elevation
  BIF - Bank Insurance Fund
  CAMEL - capital adequacy, asset quality, management practices,
     earnings, and liquidity
  CBO - Congressional Budget Office
  CCC - Commodity Credit Corporation
  CEC - Current Exposure to Claims
  CFO - Chief Financial Officer
  CRP - Conservation Reserve Program
  CUSIF - Credit Union Share Insurance Fund
  DELV - damage by elevation
  DOD - Department of Defense
  DOJ - Department of Justice
  DOT - Department of Transportation
  ERISA - Employee Retirement Income Security Act of 1974
  FAA - Federal Aviation Administration
  FASAB - Federal Accounting Standards Advisory Board
  FASB - Financial Accounting Standards Board
  FCIC - Federal Crop Insurance Corporation
  FDA - Food and Drug Administration
  FDIC - Federal Deposit Insurance Corporation
  FDICIA - FDIC Improvement Act of 1991
  FEGLI - Federal Employees' Group Life Insurance
  FEMA - Federal Emergency Management Agency
  FFB - Federal Financing Bank
  FIA - Federal Insurance Administration
  FICO - Financing Corporation
  FIRM - flood insurance rate map
  FIRREA - Financial Institutions Reform, Recovery, and Enforcement
     Act of 1989
  FRS - Federal Reserve System
  FSLIC - Federal Savings and Loan Insurance Corporation
  GATT - General Agreement on Tariffs and Trade
  GPRA - Government Performance and Results Act of 1993
  HHS - Department of Health and Human Services
  HRSA - Health Resources and Services Administration
  HUD - Housing and Urban Development
  IBNR - incurred but not yet reported
  MARAD - Maritime Administration
  NAP - Noninsured Assistance Program
  NCUA - National Credit Union Administration
  NFIP - National Flood Insurance Program
  OBRA'90 - Omnibus Budget Reconciliation Act of 1990
  OCC - Office of the Comptroller of the Currency
  OMB - Office of Management and Budget
  OPIC - Overseas Private Investment Corporation
  OPM - Office of Personnel Management
  OTS - Office of Thrift Supervision
  PAYGO - pay-as-you-go
  PBGC - Pension Benefit Guaranty Corporation
  PELV - probability of elevation
  PIMS - Pension Insurance Modeling System
  RIS - Retirement and Insurance Service
  RMA - Risk Management Agency
  RPA - Retirement Protection Act of 1994
  RTC - Resolution Trust Corporation
  SAIF - Savings Association Insurance Fund
  SBA - Small Business Administration
  SDVI - Service-Disabled Veterans Insurance
  SFAS - Statement of Financial Accounting Standards
  USDA - Department of Agriculture
  VA - Department of Veterans Affairs
  VBA - Veterans Benefit Administration
  VICP - Vaccine Injury Compensation Program
  VMLI - Veterans Mortgage Life Insurance

Letter
=============================================================== LETTER


B-275316

September 30, 1997

The Honorable John R.  Kasich
Chairman
Committee on the Budget
House of Representatives

Dear Mr.  Chairman: 

This report responds to your request that we review the budget
treatment of federal insurance programs to assess whether the current
cash-based budget provides complete information on the government's
cost and whether accrual concepts could be used to improve budgeting
for these programs.  As requested, we (1) identified potential
approaches for using accrual concepts in the budget for insurance
programs, (2) highlighted trade-offs among different approaches,
including the current cash-based budget treatment, and (3) discussed
potential implementation issues such as cost estimation.  We have
included a matter for congressional consideration and are making a
recommendation to the Director of the Office of Management and Budget
to develop and evaluate risk-assumed cost estimation methods for
federal insurance programs. 

We are sending copies of this report to the Ranking Minority Member
of your Committee, the Director of the Office of Management and
Budget, and interested congressional committees.  We will also make
copies available to others upon request.  Major contributors to this
report are listed in appendix VI. 

If you have any questions concerning this report, please call me on
(202) 512-9142. 

Sincerely yours,

Susan J.  Irving
Associate Director, Budget Issues


EXECUTIVE SUMMARY
============================================================ Chapter 0


   PURPOSE
---------------------------------------------------------- Chapter 0:1

For most federal programs, the cash basis of the federal budget
provides adequate information on absolute and comparative costs on
which to base decisions.  However, there are some programs--including
federal insurance programs--in which the cash consequences of current
decisions may not be seen for a number of years.  For these programs,
cash-based budgeting may provide not only incomplete but also
misleading information as to their cost.  Concern about improving the
information available to policymakers about the costs of various
commitments has led to questions about whether budgeting for these
programs should move toward an accrual basis under which the net
present value of the expected cost of the risk assumed by the
government would be recognized at the time the commitment is
extended. 

The Chairman of the House Committee on the Budget asked GAO to review
the treatment of federal insurance programs in the budget.  He
requested that GAO assess whether the current cash-based reporting
provides complete information about these programs and whether
accrual concepts--similar to those used for loans and loan guarantees
under the Credit Reform Act of 1990--could be used to improve
budgeting for these programs.  He has stated that making the cost of
these programs more visible will facilitate budget decision-making in
a time when difficult funding trade-offs must be made.  This report
(1) examines the shortcomings of cash-based budgeting for insurance
programs, (2) identifies how accrual-based budgeting could improve
the recognition of insurance program costs and their economic impact,
(3) examines approaches that could be used to incorporate
accrual-based cost information in the budget, and (4) identifies
implementation issues that can be anticipated in changing to
accrual-based budgeting for these programs. 


   BACKGROUND
---------------------------------------------------------- Chapter 0:2

The federal budget is the primary financial document of the
government.  The Congress and the American people rely on it to frame
their understanding of significant choices about the role of the
federal government and to provide them with the information necessary
to make informed decisions about individual programs and the
collective fiscal policy of the nation.  Historically, government
outlays and receipts have been reported on a cash basis, i.e.,
receipts are recorded when received and expenditures are recorded
when paid, without regard to the period in which the taxes or fees
were assessed or the costs incurred.  Although this has the advantage
of reflecting the cash borrowing needs of the government, over the
years, analysts and researchers have raised concerns that cash-based
budgeting does not adequately reflect either the cost of some
programs--such as federal credit or insurance--in which cash flows to
and from the government can span many budget periods or the timing of
their impact on economic behavior.\1

These concerns led in 1990 to changes in the budgetary treatment of
credit programs.  Budgeting for these programs is now done on an
accrual basis:  the net present value of the estimated cost to the
federal government over the entire life of a loan or loan guarantee
is recognized in the budget at the time the credit is extended.  The
same concerns also led in 1992 to a proposal to change the budget
treatment of deposit, pension, and other insurance programs. 
Although both GAO and CBO found problems with the specific proposal,
which was not adopted, both agencies endorsed further exploration of
accrual-based budgeting for insurance.\2

The shortcomings of the budget's cash-based reporting for insurance
programs became vividly apparent in the aftermath of the savings and
loan crisis.  During the 1980s, as hundreds of institutions became
insolvent and the government's insurance liabilities mounted, the
cash-based budget failed to provide timely information on the rising
cost of deposit insurance.  Although GAO and some industry analysts
raised concerns about the rapidly rising deposit insurance costs that
were accruing to the government, corrective action was delayed and
the government's ultimate cost increased.  The cash-based budget
provided little incentive to address the growing problem because it
did not recognize the costs until institutions were closed and
depositors paid.  This delayed budget recognition obscured the
program's, as well as the government's, underlying fiscal condition
and limited the budget process as a means for the Congress to assess
the problem.  These shortcomings of the cash-based budget led some
analysts to suggest that the earlier recognition of costs under an
accrual-based budgeting approach might have prompted quicker action
to address the growing deposit insurance commitments and thus limited
the government's ultimate cost. 

The magnitude of federal insurance commitments shown in
table 1--approximately $5 trillion in fiscal year 1995--and the risk
for significant future costs make consideration of how best to
provide adequate information on them important.  While more than half
of this
$5 trillion represented insured deposits at financial institutions,
the federal government also insures individuals and firms against a
variety of risks ranging from natural disasters under the flood and
crop insurance programs to employer bankruptcies under the pension
insurance program.  Other programs include life insurance for
veterans and federal employees, political risk insurance for overseas
investment, and programs covering vaccine injuries and war risks. 



                                Table 1
                
                Major Federal Insurance Programs, Fiscal
                               Year 1995

                         (Dollars in billions)

Program                                                     Face value
--------------------------------------------------  ------------------
Bank Deposit Insurance                                          $1,919
Private Pension Insurance                                          853
Savings Association Deposit Insurance                              709
Veterans Life Insurance                                            490
Federal Employees' Group Life Insurance                            353
National Flood Insurance                                           325
National Credit Union Share Insurance                              266
Federal Crop Insurance                                              26
Political Risk Insurance                                            21
Maritime War-Risk Insurance                                          2
Aviation War-Risk Insurance                                          2
Vaccine Injury Compensation                                          1
======================================================================
Total insurance in force                                        $4,967
----------------------------------------------------------------------
Source:  Budget of the United States Government, Fiscal Year 1997 and
the Office of Management and Budget. 


--------------------
\1 The cash basis adequately measures the amount and timing of
government borrowing, but for some programs, such as credit and
insurance, it misstates the size and timing of the impact of the
government's spending on private economic behavior. 

\2 For additional detail, see Accrual Budgeting (GAO/AFMD-92-49R,
February 28, 1992). 


   RESULTS IN BRIEF
---------------------------------------------------------- Chapter 0:3

The cash-based budget, which focuses on annual cash flows, does not
adequately reflect the government's cost or the economic impact of
federal insurance programs because generally costs are recognized
when claims are paid rather than when the commitment is made.  In any
particular year, the cost of the government's insurance commitments
may be understated or overstated because the time between the receipt
of program collections, the occurrence of an insured event, and the
final payment of a claim can extend over many budget periods.  In
addition, since it is generally the issuance of insurance rather than
the payment of the claim that affects economic behavior, the
cash-based budget may not accurately measure the timing and magnitude
of an insurance program's impact on economic behavior. 

As a general principle, decision-making is best informed if the
government recognizes the costs of its commitments at the time it
makes them.  For most programs, cash-based budgeting accomplishes
this.  However, for insurance programs, accrual-based budgeting,
which would recognize the expected long-term cost of the insurance
commitment at the time the insurance is extended, offers the
potential to overcome a number of the deficiencies of cash-based
budgeting by improving cost recognition.  In concept, recognition in
the budget of the risk assumed by the government would permit
policymakers to consider these costs in relation to other funding
demands and would improve the measurement of a program's impact on
private economic behavior.  In most cases, the risk-assumed approach
to accrual would be analogous to a premium rate-setting process in
that it looks at the long-term expected cost of an insurance
commitment at the time the insurance commitment is extended.  The
risk assumed by the government is essentially that portion of a full
risk-based premium not charged to the insured. 

In practical terms, however, attempts to improve cost recognition
occur on a continuum since insurance programs and insurable events
vary significantly.  For example, the extent of the improvement in
information in moving from cash-based to accrual-based information
would vary across programs depending on (1) the size and length of
the government's commitment, (2) the nature of the insured risks, and
(3) the extent to which costs are currently captured in the budget. 
The diversity of federal insurance programs also implies that the
period used for estimating risk assumed, the complexity of the
models, and the policy responses to this new information will vary. 

The challenges involved in bringing accrual-based estimates into the
budget are significant and dictate beginning with an informational
and analytic step.  Development of models to generate reasonably
reliable risk-assumed estimates is made difficult by the nature of
the risks insured by the government, frequent program modifications,
and the sufficiency of data on potential losses.  For some programs,
the development of risk-assumed estimates will require refining and
adapting available risk assessment models while, for other programs,
new methodologies may have to be developed.  The degree of difficulty
in developing estimates and the uncertainty surrounding these
estimates will likely be greatest for programs--such as deposit and
pension insurance--that require modeling complex interactions between
highly uncertain macroeconomic variables and human behavior.  Even
after years of research, significant debate and estimation disparity
exists in the modeling for these programs. 

Despite these challenges, the potential benefits of accrual-based
budgeting for federal insurance programs warrant continued effort in
the development of risk-assumed cost estimates.  Supplemental
reporting of risk-assumed estimates in the budget as they are
developed over a number of years would help policymakers understand
the extent and nature of the estimation uncertainty and evaluate
whether a more comprehensive accrual-based budgeting approach should
be adopted.  In evaluating these estimates for use in the budget, the
focus should not be on whether the estimates are exactly correct but
rather on the improvements in the quality of budget information they
provide to policymakers.  Any shift in the way a program's costs are
reflected in the budget has significant implications for
beneficiaries and taxpayers alike.  Better information about the
costs of commitments will permit more informed deliberations about
the appropriate design of insurance programs and about possible
responses to changes in program costs.  Given the large stakes
involved, it will also be important that the cost estimates be
perceived as unbiased and generally reliable. 

Supplemental reporting of risk-assumed estimates in the budget would
parallel the new accounting treatment required under accounting
standards developed by the Federal Accounting Standards Advisory
Board (FASAB).  In requiring the disclosure of risk-assumed estimates
as supplemental information to agency financial statements, FASAB
recognized the usefulness of these estimates to better inform budget
decisions.  FASAB also recognized the difficulty of preparing
reliable risk-assumed estimates and therefore did not require their
recognition on the financial statements as a liability.  In the
interim, as work on the development of risk-assumed estimates takes
place, the claims liability reported in agency financial statements
provide policymakers with useful information on insurance program
losses that are both probable and can be reasonably estimated as a
result of events that have occurred as of a given reporting date. 
This information should be considered during budget decisions. 
However, as FASAB recognized, the risk-assumed concept would in most
cases go further than the financial statement liability recognition
standard since the latter does not reflect losses inherent in the
government's commitment at the time the insurance is extended. 


   GAO'S ANALYSIS
---------------------------------------------------------- Chapter 0:4


      CASH-BASED BUDGETING FOR
      FEDERAL INSURANCE PROGRAMS
      GENERALLY PROVIDES
      INCOMPLETE INFORMATION FOR
      DECISION-MAKING
-------------------------------------------------------- Chapter 0:4.1

Although the cash-based budget may most accurately measure the
government's borrowing needs, for federal insurance it generally
provides incomplete or misleading information for resource allocation
and fiscal policy decisions.  The annual net cash flows currently
reported in the budget may obscure the government's cost for
insurance programs because premium collections are not matched with
the expected costs of insurance commitments.  This occurs for several
reasons that vary from program to program depending upon the
characteristics of the risk insured and the structure of the program. 
The mismatch is most obvious for programs in which the government's
commitment extends for many years into the future, such as for life
insurance and pension guarantees.  For example, from 1981 through
1992 accrued losses of the Pension Benefit Guaranty Corporation
(PBGC) ballooned its accumulated deficit as reported in its financial
statements from $190 million to $2.4 billion while the cash-based
budget reported positive net cash flows every year.\3 Thus, PBGC
appeared financially sound in the cash-based budget despite its
deteriorating condition. 

Even for programs in which the insurance commitment is short term,
cash-based reporting may not be adequate because some risks insured
by the government--e.g., flood and crop damage--result in losses that
although predictable are nevertheless variable on an annual basis. 
In order to accumulate reserves to pay claims in high loss years,
these programs need more funds than they pay out in other years. 
Because of their sporadic nature, these risks need to be pooled\4
over many years.  The annual net cash flows for any single budget
year will not accurately reflect the government's cost for operating
such programs on a continuing basis.  For example, from 1986 through
1995 the cash-based budget reported premium income exceeding claim
payments in 6 of the 10 years for the flood insurance program.  This
made the program appear in good financial shape, even though a
significant portion of the policies receive an unfunded subsidy and
the program has not been able to build sufficient reserves to cover
expected future high loss years.\5

The cash-based budget also generally falls short as a gauge of the
economic impact of federal insurance programs.  Although discerning
the economic impact of insurance programs is difficult, in general,
private economic behavior is affected when insurance is provided
rather than when claims are paid.  For example, the government may
influence the overseas investment decisions of U.S.  corporations
when it extends political risk insurance and the planting decisions
of farmers when it insures their crops. 


--------------------
\3 Liability for pension benefit payments is recorded in the
financial statements based on events that have occurred or are
probable to occur and can be reasonably estimated. 

\4 Pooling risk refers to the spreading of risk among a large number
of insureds in order to reduce the cost of bearing the risk. 

\5 Flood Insurance:  Financial Resources May Not Be Sufficient to
Meet Future Expected Losses (GAO/RCED-94-80, March 21, 1994). 


      ACCRUAL-BASED BUDGETING HAS
      THE POTENTIAL TO IMPROVE
      BUDGET INFORMATION AND
      INCENTIVES FOR FEDERAL
      INSURANCE PROGRAMS
-------------------------------------------------------- Chapter 0:4.2

The use of accrual-based budgeting for federal insurance programs has
the potential to overcome a number of the deficiencies of cash-based
budgeting.  Two characteristics of federal insurance programs support
the use of accrual-based budgeting:  (1) the government's commitment
to cover future losses that may occur beyond the current budget
period and (2) the difficulty of estimating and pooling insured risk
on an annual basis. 

As is true of the treatment of loan guarantees under credit reform,
accrual-based budgeting for insurance would recognize the estimated
cost of the government's insurance commitments when they are made
rather than when the cash consequences occur.  Conceptually,
therefore, accrual-based budgeting using risk-assumed cost estimates
would improve both the opportunities and incentives to control the
government's insurance costs.  Policymakers might be encouraged to
examine the underlying benefits and structure of insurance programs
before coverage is provided--i.e., when a new program is proposed--or
before large losses accumulate in existing coverage.  They would also
be able to make more accurate cost comparisons because risk-assumed
cost estimates reflect the government's subsidy cost--the difference
between expected claims and program income--regardless of when cash
flows occur.  Budgeting for the government's expected cost of the
risk it assumes would take into account the need to pool risks over
time and accumulate reserves for future high loss years.  Lastly, the
earlier recognition of the government's cost afforded by risk-assumed
estimates would more closely coincide with the economic impact of
federal insurance programs, which generally occurs when insurance
coverage is provided and the risk to the insured is lowered. 

In practical application, a risk-assumed approach to accrual-based
budgeting may vary across programs.  The diversity in federal
insurance programs means that the period used for estimating risk
assumed may differ due to the length of the government's commitment,
the nature of the risk insured, and the ability to estimate the risk
inherent in the insurance provided.  For example, in considering the
appropriate method to measure risk it is worth noting the difference
between uncertain and predictable but variable events.  The
occurrence of floods is measurable over the long term but predicting
the timing of their occurrence by more than a few days or hours is
considered impossible.  Thus, while flood losses may be variable on
an annual basis, it is possible to measure the long-term expected
risk of flooding.  However, determining the risk assumed for other
programs may be more difficult because they insure events that are
not only highly variable but also highly uncertain over the long
term.  For example, estimating future deposit and pension insurance
costs would require assessing the long-term solvency of private
firms, which is dependent on highly uncertain and volatile economic,
financial, and behavioral variables. 

In addition, the degree to which accrual-based budgeting would change
the information and incentives provided to decisionmakers for
insurance programs will vary based on the characteristics of
individual programs and the specific approach taken.  For example,
the limitations of cash-based budgeting and the improvements achieved
by shifting to accrual-based budgeting are most pronounced for the
two largest programs--deposit and pension insurance.  The size of
these programs in relation to total federal spending and, therefore,
their potential to distort resource allocation and fiscal policy
choices have been central to the argument for accrual-based budgeting
for federal insurance programs.  The failure of the cash-based budget
to adequately signal policymakers about the mounting losses from the
savings and loan crisis showed that such potentially large
misstatements of cost may have serious consequences for aggregate
budget and fiscal policy.  For other insurance programs, the
implications of cash-based budgeting for aggregate budget and fiscal
policy may not be as great, but accrual-based reporting could still
improve cost information. 


      INCORPORATING ACCRUAL-BASED
      INFORMATION IN THE BUDGET
      PRESENTS ESTIMATION AND
      IMPLEMENTATION CHALLENGES
-------------------------------------------------------- Chapter 0:4.3

The effective implementation of accrual-based budgeting for federal
insurance programs will depend on the ability to generate reasonable,
unbiased estimates of the risk assumed by the federal government. 
Although the risk-assumed concept itself is relatively
straightforward (the recognition of the difference between the full
risk premium and the actual premiums charged for the insurance at the
time coverage is extended), how to implement it for the wide range of
federal insurance programs raises complex issues, such as the
appropriate period over which to estimate risk.  In addition,
substantial effort will be required to improve available risk
assessment models and, in some cases, develop new methodologies.  The
extent of these difficulties varies significantly across the
programs.  While, in some cases, generating risk-assumed estimates
may not be problematic, in other cases, the difficulties faced may be
considerably more challenging than those currently faced for some
loan programs under credit reform. 

Estimating the risks inherent in most federal insurance programs is
difficult for a number of reasons.  Many federal insurance programs
cover complex, case-specific, or catastrophic risks that the private
sector has historically been unwilling or unable to cover.  As a
result, the development and acceptance of risk assessment
methodologies for individual insurance programs vary considerably. 
Lack of sufficient historical data for some federal insurance
programs also constrains risk assessment.  While private insurers
generally rely on historical data on losses and claim costs to assess
risk, data on the occurrence of insured events over sufficiently long
periods under similar conditions are generally not available for
federal insurance programs.  For some programs, such as the war-risk
programs, insured events are extremely rare.  For others, such as
crop and flood insurance, the variation in possible outcomes is
large, requiring several decades of data to adequately estimate
risks.  Frequent program modifications as well as fundamental changes
in the activities insured further reduce the predictive value of
available data and complicate risk estimation. 

Because insurance program costs are dependent on many economic,
behavioral, and environmental variables that cannot be known with
certainty in advance, there will always be uncertainty in reported
accrual-based estimates.  This will be true even as models are
developed and improved.  It will be important for policymakers to
understand the extent and nature of this uncertainty and to have
assurance that the estimates are unbiased.  In addition, as is true
for loan programs under credit reform, budgeting for federal
insurance programs would be more complex under an accrual-based
budgeting approach. 

In most cases, use of risk-assumed estimates in budgeting for federal
insurance programs, would be more forward looking than the liability
recognition standards traditionally used to prepare financial
statements.  For programs with short duration policies, such as crop
and flood insurance, the use of financial statement liability
recognition standards may not yield information very different from
what is currently reported on a cash basis in the budget.  For other
programs with long-term commitments, such as pension and life
insurance, the use of financial statement liability recognition
standards would improve the information available in the budget
compared to the cash basis.  However, accrual-based budgeting using
traditional financial statement liability recognition standards, in
most cases, would not provide recognition of the risks inherent in
the government's commitment at the time insurance is extended and,
thus, would not be as useful for budgeting as the risk-assumed
concept.  Nevertheless, until risk-assumed estimates are fully
developed, and the new accounting standards developed by FASAB are
implemented, insurance programs' financial statements, which are
included in the budget appendix, provide policymakers with valuable
information on insured events (losses) that are probable and
measurable as of a given date and should be considered in budget
discussions. 

Although the characteristics of the risk assumed by the government
under the various federal insurance programs make risk estimation
difficult, continued research and development of estimation
techniques could improve information on and increase attention given
to the cost of the government's commitments.  For example, the Office
of Management and Budget's effort to develop methodologies to
estimate the future costs of pension guarantees has helped focus
attention on the risk assumed by the government for this program. 


      SUPPLEMENTAL REPORTING OF
      RISK-ASSUMED ESTIMATES IN
      THE BUDGET WOULD HELP
-------------------------------------------------------- Chapter 0:4.4

Although the potential for risk-assumed accrual-based budgeting for
federal insurance programs to address the shortcomings of the current
cash-based approach argues for its implementation, the analytic and
implementation issues involved argue for beginning with supplemental
information.  Supplemental reporting of these estimates in the budget
as they are developed over a number of years could help policymakers
understand the extent and nature of the estimation uncertainty and
evaluate the desirability and feasibility of adopting a more
comprehensive accrual-based budgeting approach. 

Once several years of data have been reported as supplemental
information in the budget, these estimates should be evaluated to
determine their reliability.  In evaluating these estimates, the
focus should not be on whether the estimates are exactly correct but
rather on how they improve the quality of the information and
incentives provided to policymakers in the budget.  If the
risk-assumed estimates develop sufficiently so that their use in the
budget will not introduce an unacceptable level of uncertainty,
policymakers could consider whether to move beyond supplemental
information and incorporate risk-assumed, accrual-based estimates
into budget authority.  Beyond that, full integration of
accrual-based estimates into budget authority, outlays, and the
deficit could follow if it seemed appropriate and helpful. 

Supplemental reporting of risk-assumed cost estimates in the budget
would allow time to

  -- develop and refine estimation methodologies,

  -- assess the reliability of the risk-assumed estimates,

  -- gain experience and confidence in the risk-assumed cost
     measures,

  -- evaluate the feasibility of a more comprehensive accrual-based
     budgeting approach, and

  -- formulate cost-effective reporting procedures and requirements. 

During this period, policymakers should continue to draw on
information provided in audited financial statements.  As noted
above, financial statements provide earlier recognition of accruing
liabilities than does the cash-based budget for insurance
commitments.  Where applicable, agency efforts to comply with
accounting standards recently developed by FASAB, which require
disclosure of the risk-assumed estimates as supplemental information
to agency financial statements, could facilitate the reporting of
risk-assumed estimates in the budget.  In addition, the ongoing
efforts of various interagency working groups to identify ways to
comply with credit reform at the lowest possible cost, improve and
standardize audit requirements, and use credit reform data and
concepts for internal management purposes may be helpful in
addressing challenges faced in implementing accrual-based budgeting
for federal insurance programs. 


   MATTER FOR CONGRESSIONAL
   CONSIDERATION
---------------------------------------------------------- Chapter 0:5

The Congress may wish to consider encouraging the development and
subsequent reporting of annual risk-assumed cost estimates in
conjunction with the cash-based estimates for all federal insurance
programs in the President's budget.  The Congress may also wish to
consider periodically overseeing and assessing the reliability and
usefulness of these estimates, making adjustments, and determining
whether to move toward a more comprehensive accrual-based budgeting
approach for insurance programs. 


   RECOMMENDATION
---------------------------------------------------------- Chapter 0:6

GAO recommends that the Director of the Office of Management and
Budget develop risk-assumed cost estimation methods for federal
insurance programs and encourage similar efforts at agencies with
insurance programs.  As they become available, the risk-assumed
estimates should be reported annually in a standardized format for
all insurance programs as supplemental information along with the
cash-based estimates.  A description of the estimation methodologies
used and significant assumptions made should be provided.  To promote
confidence in risk-assumed cost measures, the estimation models and
data should be available to all parties involved in making budget
estimates and be subject to periodic external review.  As data become
available, OMB should undertake and report on evaluations of the
validity and reliability of the reported estimates. 


   AGENCY COMMENTS
---------------------------------------------------------- Chapter 0:7

OMB officials reviewed a draft of this report and agreed with GAO's
conclusion that budgeting for insurance programs should be based on
the government's long-term expected cost of the insurance
extended--the risk assumed by the government.  OMB also concurred
with the report's findings that the challenges involved in bringing
risk-assumed estimates into the budget are significant and that
additional effort to improve estimation methods is required.  OMB
officials noted that they would like to pursue such improvements but
are not doing so because they do not currently have the additional
expertise that would be required. 

OMB officials expressed concern about GAO's use of the terms "cash"
and "accrual" in this report to describe different approaches to
budgeting for insurance programs.  OMB officials suggested that the
current federal budget system is better characterized as
commitment-based or obligation-based budgeting and that the use of
risk-assumed cost estimates is consistent with this concept.  While
GAO agrees that this is a useful way of thinking about potential
changes in budgeting for insurance programs, it uses the term
"cash-based" because cash is the measurement basis for the amounts
shown in the budget for budget authority, obligations, outlays, and
receipts.  The estimates for these amounts generally are made in
terms of cash payments to be made or received.  The term
"accrual-based" is used in the report because the term "accrual" is
generally understood as a basis of measuring cost rather than cash
flows. 

GAO modified relevant sections of the report to clarify its
explanation of OMB's views on the budget treatment of deposit
insurance under an accrual-based approach.  GAO also dropped from
chapter 1 a brief discussion of early budget commissions'
recommendations regarding accrual accounting in the federal
government which was not necessary to convey its message.  Lastly,
OMB officials provided a number of technical comments, which were
incorporated into the report as appropriate. 


INTRODUCTION
============================================================ Chapter 1

During the last 50 years, many analysts and researchers have raised
concerns that cash-based budgeting does not provide complete
information on the cost of some federal programs.  Concerns that the
cash-based budget badly distorted information on credit programs led
to the inclusion of accrual-based costs in the budget for credit
programs as the result of the Federal Credit Reform Act of 1990. 
Similar concerns have been raised about other programs--most notably
insurance and federal employee pensions.  In 1992, the Bush
administration proposed to change the budget treatment of insurance
programs from a cash basis to an accrual basis.  Although the
proposal was not enacted, analysts continue to assess the merits of
accrual-based budgeting for such programs. 

Because of his concern that for some federal programs, the cash-based
budget does not provide a complete picture of the consequences of the
government's actions, the Chairman of the House Committee on the
Budget asked us to evaluate the use of accrual-based budgeting for
federal insurance programs.  He believes that making the government's
cost of these programs more visible will improve budget
decision-making. 


   BACKGROUND
---------------------------------------------------------- Chapter 1:1

The federal budget serves as the primary financial plan of the
government.  As such, difficult decisions concerning resource
allocation and fiscal policy are framed by the information provided
in the budget.  Historically, government outlays and receipts have
been reported on a cash or cash-equivalent basis.\1 Receipts are
recorded when received and expenditures are recorded when paid,
without regard to the period in which taxes and fees were assessed or
the costs incurred.  For most federal programs, cash-based reporting
provides adequate information on and control over the government's
spending commitments because the time between when a liability is
incurred and when it is paid is short.  Costs to the government are
known at the time the decision is made to provide budget authority,
and cash outlays generally capture the fiscal effects of the
government's spending.  However, for certain programs, such as
federal insurance in which the government's commitment can involve
cash flows to and from the government over many years, the actual
cost to the government may not be fully recognized with cash-based
reporting. 

The failure of the cash-based budget to provide timely signals to
policymakers on the rapidly deteriorating financial condition of the
nation's deposit insurance system and growing federal commitments for
deposit insurance during the 1980s has been widely cited as a vivid
illustration of the shortfalls of cash budgeting for federal
insurance programs.  Although GAO and some industry analysts raised
concerns about the rapidly rising deposit insurance costs that were
accruing to the government, corrective action was delayed and the
government's ultimate cost increased.  If the budget had recognized
the government's expected cost of deposit insurance, the government's
ultimate cost might have been lower if such recognition had prompted
earlier actions by policymakers to limit losses.  Instead, the budget
did not report these costs until institutions were closed and
depositors paid.  In addition, by not reflecting the government's
deposit insurance liabilities as they accrued, the cash-based budget
proved to be a poor gauge of the program's economic impact.  Delay in
recognizing these costs obscured the government's underlying fiscal
condition during and after the crisis.  Furthermore, not only was the
cash-based budget slow to recognize these costs, but it may have also
created incentives to delay closing insolvent institutions (to avoid
increasing the annual deficit), which increased the ultimate cost to
the government.\2 This experience with deposit insurance heightened
concerns that the cash-based budget was not providing adequate
information on the potential cost of other federal insurance
commitments. 

In a series of reports in the 1980s on managing the cost of
government, GAO advocated the use of some accrual cost measures in
the budget.\3 Specifically, we reported that due to the budget's
exclusive focus on cash transactions, the costs of some programs,
including retirement, insurance, and credit, may not be accurately
reflected in the budget.  However, given the limitations of
governmentwide accounting systems, we suggested that budget reporting
could be improved by recording annual accrued costs for selected
programs.  Since then, budget reporting has gradually been modified
using accrual measures to recognize the government's cost for certain
programs.  For example, in 1985, budgeting for military retirement
costs was moved to an accrual basis reflecting--at the program
level--the government's expected costs for retirement benefits as
they are earned.  These program level accrued amounts are offset so
that total budget outlays and the deficit are not affected by this
change.  Similarly, beginning in 1987, accruing retirement benefit
costs not covered by employee contributions are now charged to
employing agencies for civilian employees covered under the Federal
Employees Retirement System.  More recently, the Federal Credit
Reform Act of 1990 changed the method of controlling and accounting
for credit programs to an accrual basis. 


--------------------
\1 A long-standing exception to this is interest on public issues of
public debt, which is recorded as it accrues. 

\2 Budget Issues:  1991 Budget Estimates:  What Went Wrong
(GAO/OCG-92-1, January 15, 1992). 

\3 Managing the Cost of Government:  Building an Effective Financial
Management Structure (GAO/AFMD-85-35, February 1985) and Managing the
Cost of Government:  Proposals for Reforming Federal Budgeting
Practices (GAO/AFMD-90-1, October 1989). 


      CREDIT REFORM MARKED A
      SIGNIFICANT DEPARTURE FROM
      CASH-BASED BUDGETING
-------------------------------------------------------- Chapter 1:1.1

On November 5, 1990, the Federal Credit Reform Act was signed into
law, as Title 13B of the Omnibus Budget Reconciliation Act of 1990
(Public Law 101-508).  The act, which legislated changes GAO and
others advocated,\4 addressed many of the concerns raised by various
analysts by changing the budget reporting of the cost of credit
programs from a cash basis to an accrual basis.  Because the federal
government uses loans and loan guarantees to achieve numerous policy
objectives, the scope of this change was far-reaching.  In fiscal
year 1996, for example, the federal government entered into
commitments to make or guarantee loans totaling approximately $200
billion. 

Prior to credit reform, outlays for credit programs were reflected in
the budget only when cash was disbursed.  The full amount of a direct
loan was reported as an outlay, ignoring the fact that many would be
repaid.  In the case of loan guarantees, initially no outlays were
reported.  This ignored the fact that some of the guaranteed loans
would be defaulted upon and thus require future outlays. 
Consequently, the cash basis of reporting overstated the cost of
direct loans in the year they were made because it ignored repayments
and understated the cost of loan guarantees in the year they were
issued by ignoring defaults.  This deficient reporting skewed cost
comparisons between programs with similar purposes but different
funding approaches (i.e., direct loans, loan guarantees, or grants). 
Further, the relative cost of such programs in comparison to other
federal spending was also misrepresented.  By incorporating accrual
cost measures in the budget for loan and loan guarantee programs,
credit reform improved these cost comparisons. 

Credit reform addressed the shortfalls of cash-based reporting for
credit program costs by requiring the budget to include the estimated
cost to the federal government over the entire life of the loan or
loan guarantee, calculated on a net present value basis.\5

For purposes of the Credit Reform Act, the estimated cost of a direct
loan or loan guarantee is now the sum of all expected
costs--including interest rate subsidies and estimated default
losses--and all expected payments received by the government over the
life of the commitment, discounted by the interest rate on Treasury
securities of similar maturity to the loan or guarantee. 
Reestimation of the cost of loans disbursed or guaranteed in a given
year is required over the life of the commitment.  This more accurate
reporting of credit program costs allows for more efficient
allocation of budget resources and improved measurement of these
programs' economic impact. 


--------------------
\4 Budget Reform for the Federal Government (GAO/T-AFMD-88-13, June
7, 1988) and Budget Issues:  Budgetary Treatment of Federal Credit
Programs (GAO/AFMD-89-42, April 10, 1989). 

\5 Present value is the worth of a future stream of returns or costs
in terms of money paid today.  A dollar today is worth more than a
dollar at some date in the future because today's dollar could be
invested and earn interest in the interim. 


      CREDIT REFORM REQUIRES
      RECOGNITION OF ESTIMATED
      COSTS WHEN THE GOVERNMENT
      ENTERS INTO A COMMITMENT
-------------------------------------------------------- Chapter 1:1.2

Credit reform was significant not only because it changed the budget
reporting for credit programs from a cash basis to an accrual basis
but also because it prescribed a more prospective form of accrual
measurement than required by generally accepted accounting standards
used in financial statements prepared prior to credit reform.\6

Traditional accounting standards required the recognition of all
losses and expenses incurred during a reporting period, including
those that have occurred but have not yet been reported.  In other
words, a cost would be accrued when it was more likely than not that
a borrower had defaulted on his or her loan.  In contrast, credit
reform requires recognition of the expected costs of new loans and
guarantees (on a net present value basis) at the time the credit is
extended.  This "risk-assumed" approach recognizes the expected cost
to the government before the government commits itself to future
losses inherent in the credit issued. 

For example, prior to credit reform, if the government decided to
provide $3 million in direct loans, the cash budget would have shown
an outlay in the first year of $3 million.  Repayments by borrowers
would be recorded when received in future years, and, when some
borrowers defaulted, net payments received by the government would
simply be lower.  Under traditional accrual accounting no cost would
be shown in the first year since repayment is expected, but in
subsequent years when some borrowers defaulted, the unpaid principle
would be recognized as a cost.  Thus, in neither case was the
government's cost recognized correctly at the time the decision was
made to authorize the loans.  In contrast, using the risk-assumed
basis of credit reform, an estimate of the government's cost would be
recorded when the government made the commitment to provide the
loans. 

Recognizing that the shortcomings of the cash budget are not unique
to credit programs, the Congress in the Credit Reform Act directed
OMB and the Congressional Budget Office (CBO) to study the possible
application of accrual budget reporting for federal deposit insurance
programs.  In May 1991, CBO reported that the current cash-based
budgeting approach for deposit insurance could be improved either
through the use of accrual concepts or other reporting
alternatives.\7 CBO concluded the following: 

     Adopting a full credit reform approach to deposit insurance has
     one major advantage and one major disadvantage .  .  .  .  The
     advantage is that only the accrual recognition of costs will
     provide an early warning of financial disaster in the budget. 
     The disadvantage is that estimating the cost of deposit
     insurance--when cost is incurred--is very difficult. 

OMB also reported that accrual-based budget reporting for deposit
insurance could be an improvement over the current approach and
outlined specific financial and econometric models that could be used
to estimate deposit insurance costs as they arise.  OMB recommended
that these cost measures be further developed, tested, and validated
before deciding whether or how to bring accrual-based estimates into
the budget.\8


--------------------
\6 The Federal Accounting Standards Advisory Board (FASAB)
subsequently developed accounting standards for credit programs that
reflected and supported the prospective accrual measures called for
under credit reform. 

\7 Budgetary Treatment of Deposit Insurance:  A Framework for Reform,
Congressional Budget Office, May 1991. 

\8 Budgeting for Federal Deposit Insurance, Office of Management and
Budget, June 1991. 


      THE BUSH ADMINISTRATION
      PROPOSED EXTENDING CREDIT
      REFORM PRINCIPLES TO
      INSURANCE PROGRAMS
-------------------------------------------------------- Chapter 1:1.3

In the President's fiscal year 1993 budget, less than a year after
OMB and CBO reported on the budget treatment of deposit insurance,
the Bush administration proposed applying credit reform principles to
budgeting for deposit insurance and pension guarantees.  Under the
proposal, other insurance programs would be moved to an accrual basis
the following year.  The administration emphasized earlier concerns
that cash-based budgeting for insurance programs did not provide
clear and timely measurement of their cost to the government.  It
maintained that budget reporting for these programs on an accrual
basis would provide policymakers with the information and incentives
necessary to control their costs. 

The similarities between loan guarantees and federal insurance were
noted in the administration's proposal.  In both cases the government
commits to paying some or all of future costs under specified
conditions in exchange for a fee or premium.  As with the new
treatment of credit programs, the proposal called for the recognition
of the government's cost of new or expanded insurance coverage at the
time the insurance is extended.  The cost of the risk assumed by the
government would be estimated on a net present value basis and would
include all expected costs and collections related to the coverage
extended.  OMB showed estimates of the new accrual-based measures in
the budget for deposit insurance and pension guarantees.  These
measures were based on complex, newly developed estimation
methodologies using options pricing models.\9 Legislation to effect
the new budget reporting was introduced in the Congress. 

Despite continued concern about the cash basis of reporting for
insurance programs, both GAO and CBO objected to the administration's
proposal at the time.\10 GAO affirmed its long-standing support of
reporting accrual-based costs in the budget but concluded that the
proposal made at that time was flawed.  GAO and CBO questioned the
sufficiency of available data and estimation methodologies necessary
to make reasonably accurate accrual cost estimates.  Both agencies
expressed concern about the rush to implement a major conceptual and
technically challenging change in budget reporting without thorough
study.  CBO also reported that by changing the way shortfalls in
program funding would be financed, the proposal could have increased
taxpayer liability for these programs. 

Another major concern surrounding the initiative was the budget
treatment of savings stemming from deposit and pension insurance
program reforms that were also proposed.  On a cash basis, these
savings would not have been recognized for several years in the
budget, but, by recording their effects on an accrual basis, the
administration was able to show savings in fiscal years 1992 and 1993
to offset revenue lost from proposed tax reductions.  CBO concluded
that the savings achieved by the administration's program reforms
should not be available to pay for other policy initiatives.  As a
result, most observers viewed the accrual-based budgeting proposal as
an accounting gimmick rather than a way to improve budget reporting
for insurance programs.  The merits of accrual-based reporting for
these programs were overshadowed by these concerns.  No action was
taken by the Congress on the legislation. 

Since the Bush administration's proposal for changing the budget
treatment of insurance programs, OMB has continued work on developing
methodologies to estimate the risk-assumed cost of deposit insurance
and pension guarantees.  At the request of the Chairman of the House
Committee on the Budget and because of continued interest in this
area, we undertook this study to more thoroughly develop the issues
involved in changing the budget treatment of insurance programs. 


--------------------
\9 Options pricing models are mathematical models that employ
probabilistic functions to value contracts that give the owners the
right to buy or sell an asset (such as a stock) at a fixed price on
or before a given future date.  For an additional discussion of
options pricing see figure 5.1 in chapter 5. 

\10 See Accrual Budgeting (GAO/AFMD-92-49R, February 28, 1992) and An
Analysis of the President's Budgetary Proposals for Fiscal Year 1993,
Congressional Budget Office, March 1992. 


   OBJECTIVES, SCOPE, AND
   METHODOLOGY
---------------------------------------------------------- Chapter 1:2

The Chairman of the House Committee on the Budget asked us to review
the budget treatment of federal insurance programs to assess whether
the current cash-based budget provides complete information and
whether accrual concepts could be used to improve budgeting for these
programs.  Specifically, we were asked to (1) identify approaches for
using accrual concepts in budgeting for insurance programs, (2)
highlight trade-offs among different approaches, including the
current budget treatment, and (3) discuss potential implementation
issues, such as cost estimation. 

We limited the scope of our study to programs previously identified
by OMB and FASAB as federal insurance programs.  Programs included in
our study are shown in table 2.1.  OMB's list forms the basis of its
annual analysis of credit and insurance programs, which, in recent
years, has been part of the Analytical Perspectives volume of the
President's budget.  We added one program to the OMB list--the
Federal Employees' Group Life Insurance program.  This program was
included by FASAB as federal insurance in its recommended accounting
standards for federal liabilities.  Of the veterans life insurance
programs underwritten by the federal government, we include only
those which are still open to new participants. 

In undertaking this study, we acknowledge that there is not universal
agreement on which programs constitute federal insurance.  The
programs we included in our analysis share some, but not necessarily
all, the characteristics of private insurance.  Conversely, some
programs not on our list have some of the characteristics of programs
on our list.  The diversity of the programs undertaken by the federal
government could result in disagreement about what constitutes a
federal insurance program.  Valid arguments may be made for additions
to or deletions from the list of insurance programs to consider for
an accrual-based budgeting approach.  This is but one of many issues
policymakers face in incorporating accrual concepts in the budget. 

To accomplish our objectives, we focused our analysis on the
sufficiency of information provided for resource allocation and
fiscal policy with the recognition that budget reporting must be
understandable and facilitate budget control and accountability. 
This premise is grounded in the work of the 1967 President's
Commission on Budget Concepts, which stressed that resource
allocation and fiscal policy outweigh all other uses of the budget,
such as the cash and debt management activities of the Treasury and
analyses of the impact of federal activity on the financial markets. 
To assess the sufficiency of information in the budget for these
purposes, we reviewed the programs' current budget treatment,
consulted with budget experts, and analyzed historical data on
budgeted and actual insurance outlays. 

To develop approaches for using accrual concepts in the budget and to
identify trade-offs among approaches, we began by reviewing the Bush
administration's 1992 proposal to adopt accrual accounting for
federal insurance programs.  We surveyed existing research on the
budget treatment of insurance programs conducted by OMB, CBO, and
other budget analysts.  We examined various reports and documents
pertaining to the accrual-based approach for loan and loan guarantee
programs prescribed by the Federal Credit Reform Act of 1990.  We
studied the accounting standards for insurance activities promulgated
by the Financial Accounting Standards Board (FASB) for private sector
entities and FASAB for the federal government. 

To identify potential implementation issues, we convened panels of
federal insurance agency officials and staff to gather information on
the operation of the programs and the agencies' risk assessment
capabilities.  We also obtained their views on the potential benefits
and drawbacks to the use of accrual-based budgeting.  When we could
not convene a panel or when key agency personnel were unavailable, we
obtained written responses to our questions.  We also discussed
potential implementation issues with budget experts familiar with the
implementation of accrual-based budgeting for credit programs. 

To identify issues related to developing risk-assumed cost estimates,
we interviewed agency actuaries, economists, and other staff
responsible for risk assessment.  We also analyzed documentation
supplied by the agencies and prior GAO reports on individual
programs.  In addition, we retained the services of an independent
contractor to assist us in reviewing OMB's options pricing models for
deposit insurance and pension guarantees.  As part of the
contractor's review, it assessed the validity of using options
pricing concepts and techniques to estimate insurance liabilities,
the technical sophistication and data requirements of OMB's models,
and the reliability of OMB's model estimates for budget and policy
decision-making.  We did not test or validate (1) any of the other
estimation methodologies currently used by the agencies for risk
assessment or rate-setting or (2) any of the methodologies that could
potentially be used for these purposes. 

We performed our work in Washington, D.C., from September 1995
through November 1996 in accordance with generally accepted
government auditing standards.  We requested written comments on a
draft of this report from the Director of OMB or his designee.  The
Deputy Assistant Director, Budget Analysis and Systems, provided
comments, which are discussed in chapter 8 and are reprinted in
appendix V. 


OVERVIEW OF FEDERAL INSURANCE
PROGRAMS
============================================================ Chapter 2

Federal insurance programs are a diverse set of programs covering a
wide range of risks that the private sector has traditionally been
unable or unwilling to cover.  From a federal budgeting perspective,
these programs present significant challenges because the insured
events tend to be catastrophic or volatile in nature and may not
occur for years after the government's commitment is extended. 
Although several financial measures are available for insurance
programs, estimates of the risk assumed by the federal
government--the key information for budget decision-making--have been
limited.  Despite some common elements, these programs vary
significantly in several respects, including size, length of the
government's commitment, frequency of activation, and financing. 
These differences warrant consideration in determining the
appropriate budget treatment for these programs. 


   FEDERAL INSURANCE PROGRAMS
   COVER A WIDE VARIETY OF RISKS
---------------------------------------------------------- Chapter 2:1

The federal government insures individuals and firms against a wide
variety of risks ranging from natural disasters under the flood and
crop insurance programs to bank and employer bankruptcies under the
deposit and pension insurance programs.  Other federal insurance
programs provide life insurance for veterans and federal employees
and political risk insurance for overseas investment activities.  The
federal government also provides protection against war-related risks
and adverse reactions to vaccinations.  Further, in recent years,
proposals have called for extending federal insurance activities to
cover natural catastrophes, such as earthquakes and volcanic
eruptions.\1

Some federal insurance programs have a statutory intent to provide
subsidized coverage while others do not.  In some cases, the
government subsidizes insurance programs in order to achieve a public
policy objective.  For example, catastrophic coverage under the crop
insurance program is fully subsidized in an attempt to reduce
reliance on ad hoc disaster assistance.  The Service-Disabled
Veterans Insurance Program provides life insurance coverage to
veterans with service-connected disabilities based on rates for
healthy individuals or free to totally disabled veterans.  In other
cases, as noted later in this chapter, the federal government may
intend to provide unsubsidized insurance.  However, regardless of
statutory intent, whenever federal insurance is underpriced relative
to its long-run cost, those who are insured receive a subsidy because
premiums will not cover program costs.  Table 2.1 provides an
overview of the programs included in our study.  More detailed
program summaries are provided in the appendixes to this report. 



                               Table 2.1
                
                 Overview of Federal Insurance Programs
                                Reviewed

                                                    Statutory intent
                                                    for government
Program                         Description         subsidy
------------------------------  ------------------  ------------------
Aviation War-Risk Insurance     Insures against     No; expectation of
                                losses resulting    legislation was
                                from war,           that it would
                                terrorism, and      probably be self-
                                other hostile acts  financing from
                                when commercial     premiums for
                                insurance is        assumption of
                                unavailable on      anticipated risks.
                                reasonable terms
                                and conditions and
                                continued air
                                service is in the
                                interest of U.S.
                                policy.

Bank Deposit Insurance          Insures deposits    Intent unclear;
                                at commercial       deposits backed by
                                banks and some      the full faith and
                                savings banks       credit of the U.S.
                                against losses in   government.
                                the event of
                                insolvency.

Federal Crop Insurance          Insures against     Yes.
                                crop damage from
                                unavoidable risks
                                associated with
                                adverse weather,
                                plant diseases,
                                and insect
                                infestations.

Federal Employees' Group Life   Provides life       No.
Insurance                       insurance to
                                federal employees,
                                annuitants, and
                                their families for
                                accidental death
                                and dismemberment.

Maritime War-Risk Insurance     Insures losses      No; expectation of
                                resulting from      legislation was
                                war, terrorism,     that it would
                                and other hostile   probably be self-
                                acts when           financing from
                                commercial          premiums for
                                insurance is        assumption of
                                unavailable on      anticipated risks.
                                reasonable terms
                                and conditions and
                                continued service
                                is in the interest
                                of U.S. policy.

National Flood Insurance        Insures buildings   Yes; implicit
                                and contents        subsidy by
                                against losses due  statutory design.
                                to flooding in
                                communities
                                nationwide that
                                enact and enforce
                                appropriate flood
                                plain management
                                measures.

National Credit Union Share     Insures member      Intent unclear;
Insurance                       shares (deposits)   deposits backed by
                                at credit unions    the full faith and
                                against losses in   credit of the U.S.
                                the event of        government.
                                insolvency.

OPIC's Political Risk           Insures the         No; statutory
Insurance                       investments of      intention for
                                U.S. companies in   self-financing but
                                developing          guaranteed by the
                                countries against   full faith and
                                several political   credit of the U.S.
                                risks, including    government.
                                expropriation,
                                currency
                                inconvertibility,
                                and political
                                violence.

PBGC's Pension Insurance        Insures retirement  No; statutory
                                benefits of         intent for self-
                                workers and         financing from
                                beneficiaries       premiums paid by
                                covered by private  employers on
                                sector defined      behalf of their
                                benefit pension     employees.
                                plans.

Savings Association Deposit     Insures deposits    Intent unclear;
Insurance                       at savings and      deposits backed by
                                loans and savings   the full faith and
                                banks against       credit of the U.S.
                                losses in the       government.
                                event of
                                insolvency.

Service-Disabled Veterans       Provides life       Yes.
Insurance                       insurance to
                                veterans with
                                service-connected
                                disabilities.

National Vaccine Injury         Provides            No.
Compensation                    compensation for
                                vaccine-related
                                injury and death.

Veterans Mortgage Life          Provides life       Yes.
Insurance                       insurance to
                                disabled veterans
                                who have received
                                grants for
                                specially-adapted
                                housing.
----------------------------------------------------------------------

--------------------
\1 Natural Disaster Insurance:  Federal Government's Interests
Insufficiently Protected Given Its Potential Financial Exposure
(GAO/T-GGD-96-41, December 5, 1995); Federal Disaster Insurance: 
Goals Are Good, but Insurance Programs Would Expose the Federal
Government to Large Potential Losses (GAO/T-GGD-94-153, May 26,
1994); and Bipartisan Task Force on Funding Disaster Relief, Federal
Disaster Assistance, S.  Doc.  No.  4, 104th Congress, 1st Sess. 
(1995). 


   CHARACTERISTICS OF FEDERAL
   INSURANCE PROGRAMS COMPLICATE
   BUDGET TREATMENT
---------------------------------------------------------- Chapter 2:2

The budget treatment of federal insurance programs is complicated by
the characteristics of the risks covered.  In general, these programs
assume risks that the private sector has historically been unable or
unwilling to undertake.  Ideally, individual risks should be
independent and of sufficient number to reasonably project losses and
adequately pool risk\2 to be insurable.  In addition, the occurrence
of losses should be accidental or unintentional in nature and capable
of being measured.\3 Many of the risks undertaken by the federal
government lack these key conditions for ideal insurability.  Without
these insurable conditions, establishing an actuarially sound rate
structure is difficult and the likelihood of adverse selection\4 and
moral hazard increases.\5 From a federal budget perspective, the lack
of these insurable conditions presents significant challenges. 

The risks insured by the government are often hard to predict and
catastrophic in size.  In general, the lack of an actuarial base,\6
an ever-changing environment, and low participation rates make it
difficult to assess risk assumed and set premiums commensurate with
the risk insured.  For example, officials at the Overseas Private
Investment Corporation (OPIC) cited the lack of an actuarial base as
a factor in the limited availability of private sector political risk
insurance.\7 Further, some risks assumed by the federal government
are not independent in that losses may strike a large number of
insureds at the same time.  For example, weather-related events may
reduce crop yields over large areas of the nation in the same year. 
Similarly, changes in macroeconomic conditions may have widespread
effects on banks and pension plans covered under the deposit and
pension insurance programs. 

Achieving adequate participation to spread risks may also be
problematic.  For example, our previous work found that the majority
of federal crop insurance policies are in the contiguous areas of the
Midwest and the Plains States.\8 Similarly, those living in
concentrated areas with the greatest risk of flooding are most likely
to buy flood insurance while those with lower risk are not.  Finally,
according to agency officials, the war-risk and political risk
insurance programs provide only a limited number of policies covering
diverse events with strong individual and case-specific identities. 

The catastrophic nature of these risks and the impediments to
broad-based participation reduce the ability of an insurer to pool
risk--an important way insurers reduce the costs of bearing risk. 
When insured events affect a large number of the insured population
at the same time, the likelihood that the insurer would have to make
large claim payments in a relatively short period of time increases. 
When there are only a few insured, the insurer is unable to pool risk
and thus may be subject to virtually the same uncertainties of random
experience as the insured. 

OMB has cited the government's size and sovereign power as providing
it with the unique ability to offer insurance when the private market
is unable or unwilling to do so.  Some analysts contend that the size
of the government makes it better able to absorb large losses if
insurance reserves are not sufficient.  Over time, by providing
ongoing insurance, the government may be able to recoup some of these
losses with future premium collections, thus in effect pooling risks
over time.  In addition, the government can attempt to spread the
cost of these risks by providing insurance nationwide and/or
mandating participation.  Further, some analysts cited the
government's unique status as advantageous in monitoring and
mitigating these types of risks.  For example, for a community to
participate in the flood insurance program, it must enact and enforce
minimum flood plain management standards.  Similarly, federally
insured banks and thrift institutions must adhere to numerous
regulations and periodic examinations. 

Whatever the merits of the federal government as an insurer, the same
characteristics that inhibit private insurance firms from covering
these risks also complicate budgeting for them at the federal level. 
In some cases, the volatile and/or catastrophic nature of the insured
risks make pooling risk and estimating claims on an annual basis
difficult, if not impossible.  For some programs, such as life and
pension insurance, claims may not be expected to occur for years or
even decades after the government's commitment is made.  Thus, a key
budget consideration is how and when the government's costs for these
programs should be recognized in the budget. 


--------------------
\2 Pooling risk refers to the spreading of risk among a large number
of insureds in order to reduce the cost of bearing the risk. 

\3 Additional factors, such as the ability to diversify risk, are
likely to affect the private sector's willingness to provide certain
types of coverage. 

\4 Adverse selection is the tendency for those with the highest
probability of loss to purchase insurance and those with the least
risk of loss to opt not to purchase insurance. 

\5 Moral hazard is the incentive for those insured to undertake
greater risk than if they were uninsured because the negative
consequences of such actions are passed through to the insurer.  For
example, in the 1980s when government regulators allowed thrifts to
remain open with low levels of capital, the temptation of moral
hazard was increased.  Thrifts with nearly depleted capital had
little to lose by making very risky loans in the hope of large
profits. 

\6 An actuarial base is an historical pattern of insured events under
similar conditions that is of sufficient number to reasonably project
losses and pool risks. 

\7 For a description of political risk insurance, see appendix IV. 

\8 Crop Insurance:  Additional Actions Could Further Improve
Program's Financial Condition (GAO/RCED-95-269, September 28, 1995). 


   KEY INFORMATION FOR BUDGET
   DECISION-MAKING--
   THE RISK ASSUMED BY THE
   GOVERNMENT--IS NOT READILY
   AVAILABLE
---------------------------------------------------------- Chapter 2:3

As a general principle, decision-making is best informed if the
government recognizes the costs of its commitments at the time it
makes them.  However, despite numerous financial measures, in most
cases, the expected cost of the government's insurance commitments is
not readily available.  Table 2.2 provides several financial measures
for the programs in our study including face value, net outlays,
liability for claims, and net position.  As discussed in the
following sections, each of these measures provides useful
information but, in most cases, does not adequately capture and
isolate the cost of the risk assumed by the federal government at the
time the insurance is extended.  However, to the extent practicable,
the government's ultimate cost is key information that ought to be
considered in making budget decisions. 



                                    Table 2.2
                     
                            Federal Insurance Programs

                              (Dollars in millions)

                                                        Claims
                                                   liability\b    Net position\b
              Face value\a     Net outlays\a     September 30,     September 30,
Program   Fiscal year 1995  Fiscal year 1995              1995              1995
--------  ----------------  ----------------  ----------------  ----------------
Aviation             2,000                -2               0\a              60\a
 War-
 Risk
 Insuran
 ce
Bank             1,919,000            -6,916               493            25,454
 Deposit
 Insuran
 ce
National           266,000              -297               122             3,250
 Credit
 Union
 Share
 Insuran
 ce
Federal             26,000               387             1,237               915
 Crop
 Insuran
 ce
Federal            353,000              -916            20,090            -3,472
 Employe
 es'
 Group
 Life
 Insuran
 ce
Maritime             2,000                -2               0\a              24\a
 War-
 Risk
 Insuran
 ce
National           325,000               459               162            -1,027
 Flood
 Insuran
 ce
OPIC's            21,300\c              -208                79           2,462\d
 Politic
 al Risk
 Insuran
 ce
PBGC's             853,000              -430            10,398              -123
 Pension
 Insuran
 ce
Savings            709,000            -1,101               111             3,358
 Associa
 tion
 Deposit
 Insuran
 ce
Service-             1,500                62             516\a            -463\a
 Disabled
 Veteran
 s
 Insuran
 ce
National               700                51            n.a.\e             945\a
 Vaccine
 Injury
 Compens
 ation
Veterans               200            n.a.\e            n.a.\e            n.a.\e
 Mortgage
 Life
 Insuran
 ce
--------------------------------------------------------------------------------
\a Budget of the United States Government, Fiscal Year 1997 and OMB. 
GAO did not independently verify. 

\b Agency audited financial statements, unless otherwise noted. 

\c Under most outstanding insurance contracts, investors and lenders
may obtain three different types of insurance coverage, but aggregate
claim payments may not exceed the single highest coverage amount.  In
addition, face value includes a provision for standby coverage for
which OPIC is currently not at risk.  OPIC calculated its "Current
Exposure to Claims" for 1995 as $6.6 billion. 

\d Capital and retained earnings. 

\e Not available. 

Face value represents the total amount of insurance outstanding.  For
example, the face value of deposit insurance is the total insured
deposits held by financial institutions.  As such, it provides a
measure of the maximum exposure undertaken by the federal government. 
As shown in figure 2.1 and table 2.3, the face value of federal
insurance (in constant dollars) grew substantially between 1975 and
1990.  The majority of this increase is attributable to the two
largest insurance programs, pension and deposit insurance.  For
fiscal year 1995, the estimated face value of major federal insurance
programs was approximately $5 trillion--more than half of which was
deposit insurance.  Figure 2.1 shows the trend in the face value of
major federal insurance programs. 

   Figure 2.1:  Face Value of
   Major Federal Insurance
   Programs by Type

   (See figure in printed
   edition.)

While face value provides one measure of program size, it overstates
the potential cost to the government.  The probable cost to the
government is most likely some percentage of the total face value. 
However, a single fixed percentage cannot be used as a proxy for
exposure since the government's risk varies based on a variety of
factors, such as the nature of insured risk and the extent to which
premium collections offset costs.  Thus, a self-supporting insurance
program with a relatively high face value may have a lower potential
cost to the government than a subsidized insurance program with lower
face value. 



                               Table 2.3
                
                 Face Value of Major Federal Insurance
                                Programs

                         (Dollars in billions)

                                 Constant fiscal year 1995 dollars
                              ----------------------------------------
Program                        1970   1975   1980   1985   1990   1995
----------------------------  -----  -----  -----  -----  -----  -----
Federal Deposit Insurance     1,707  2,256  2,637  3,212  3,250  2,894
                                 .2     .7     .0     .6     .5     .0
Pension Benefit Guaranty         \b   n.a.  734.6  820.4  1,008  853.0
 Corporation\a                                               .1
National Flood Insurance       n.a.   36.2  160.3  183.5  234.9  325.0
Federal Crop Insurance          2.9    3.2    4.9    9.1   14.8   26.0
Aviation War-Risk Insurance   190.3  125.9  335.1  251.0  547.5  2.0\c
Maritime War-Risk Insurance    60.9   66.0   39.3   n.a.   12.7  2.0\c
Veterans Life Insurance\d     319.0  273.1  191.1  212.6  247.6  490.1
Federal Employees' Group      141.2  157.5  136.6  273.0  318.7  353.0
 Life Insurance
Overseas Private Investment   28.0\   15.3    9.8   15.1   11.4   21.3
 Corporation                      e
National Vaccine Injury          \b     \b     \b     \b   n.a.    0.7
 Compensation\f
Nuclear Risk Insurance        358.4  154.4  162.1   96.7     \b     \b
======================================================================
Total insurance in force      2,807  3,088  4,410  5,074  5,646  4,967
                                 .9     .3     .8     .0     .2     .1
----------------------------------------------------------------------
Source:  OMB data adjusted for inflation.  GAO did not independently
verify. 

\a Established in 1974. 

\b Not in existence. 

\c Methodology for calculating face value changed in 1995 to more
realistically reflect program operation. 

\d Includes all veterans' life insurance programs.  Only the
Service-Disabled Veterans Life Insurance program and the Veterans
Mortgage Life Insurance program are included in our study. 

\e Includes insurance issued by the Agency for International
Development (AID).  The Foreign Assistance Act of 1969 established
OPIC and transferred AID's insurance and credit programs to OPIC. 

\f Program established in 1986. 

Other financial measures may also be of limited help in assessing the
cost of the risk assumed by the government at the time the insurance
commitment is extended.  For example, the outlays recorded in the
President's budget provide a measure of an insurance program's
estimated and actual annual cash flows but, in most cases, does not
capture the government's cost of insurance commitments at the time
they are extended.  In addition, cash outlays may be subject to
significant volatility due to the irregular and catastrophic nature
of some insured risks, such as natural disasters.  Chapter 3
discusses in detail the shortcomings of the current cash-based budget
reporting for federal insurance programs. 

Further, while the claims liability and net position reported in the
financial statements for federal insurance programs provide useful
measures of the programs' financial condition based on insured events
that have occurred, these measures do not, in most cases, capture the
expected cost of claims inherent in the government's commitment.  In
general, the financial statement liability is an estimate of the
amount needed to settle unpaid and expected claims related to insured
events that have occurred on or before the reporting date.  Net
position is the difference between an entity's assets and
liabilities.  The Federal Accounting Standards Advisory Board
(FASAB)\9 recently developed standards calling for the supplemental
disclosure of estimates of the risk assumed by the federal government
for its insurance programs.  This action, which is discussed in
greater detail in chapter 4, will help improve information on these
costs. 


--------------------
\9 FASAB was established in October 1990 by the Secretary of the
Treasury, the Director of the Office of Management and Budget, and
the Comptroller General.  The nine-member Board was created to
consider and recommend accounting principles for the federal
government. 


   SNAPSHOT OF CURRENT BUDGET
   TREATMENT
---------------------------------------------------------- Chapter 2:4

All the federal insurance programs reviewed record collections and
payments in net outlays on a cash basis and thus influence the
deficit in the year cash flows occur, regardless of when the
commitments are made.  With one exception,\10 the premiums and fees
paid by participants are held in revolving funds--trust or public
enterprise--and, in most cases, administrative expenses are also paid
out of these funds.  To the extent that the budget authority in these
funds exceeds current cash outlay needs and remains available for
future claims, most insurance programs have some level of reserves. 
Six of the 13 programs have permanent borrowing authority to cover
the cost of claims, and 4 have received general fund appropriations
within the last 10 years to pay claims in excess of available
resources. 

Budgetary characteristics, such as the classification of a program's
spending under the Budget Enforcement Act of 1990 (BEA),\11 will
affect the extent to which an accrual-based approach would change the
information and incentives provided to policymakers.  An examination
of the programs included in our study shows that the majority are
classified as mandatory\12 spending under BEA.  Claim payments for
only 3 of the 13 programs--Aviation War-Risk Insurance, Maritime
War-Risk Insurance, and OPIC's political risk insurance--are
classified as discretionary spending.\13 Table 2.4 summarizes key
budget information for the insurance programs reviewed. 

The programs also differ in the extent to which costs are currently
recognized in budget authority and obligated based on accrual
concepts.\14 For example, two programs--the Federal Crop Insurance
Program and OPIC's political risk insurance program--currently
obligate budget reserves based on accrual concepts.  According to
OMB, the crop insurance program obligates funds based on an estimate
of claims incurred or expected to be incurred for outstanding
policies at the end of the fiscal year.  OPIC currently obligates
budget reserves for its insurance program based on an estimate of the
losses inherent in insurance outstanding. 



                                                                      Table 2.4
                                                       
                                                         Budget Information for Major Federal
                                                                  Insurance Programs

                                   BEA classification
                              ----------------------------
                                                                                                                     General
                              Claim          Administrativ                                                           appropriations    Borrowing
Program                       payments       e costs        Fund type       Sources of financing                     in last 10 years  authority
----------------------------  -------------  -------------  --------------  ---------------------------------------  ----------------  --------------
Bank                          Mandatory      Mandatory      Public          Premiums, recovery of assets acquired    No                $30 billion
Deposit Insurance                                           enterprise      in receivership, deposit assumption
                                                                            transactions, and interest earnings

Savings Association           Mandatory      Mandatory      Public          Premiums, recovery of assets acquired    Yes               $30 billion
Deposit Insurance                                           enterprise      in receivership, deposit assumption
                                                                            transactions, and interest earnings

National Credit Union         Mandatory      Mandatory      Public          Premiums, interest earnings, and 1-      No                $100 million
Share Insurance                                             enterprise      percent deposit from insured credit
                                                                            unions

PBGC's                        Mandatory      Mandatory      Public          Premiums, assets from terminated plans,  No                $100 million
Pension Insurance                                           enterprise      and investment income

National Flood                Mandatory      Discretionary  Public          Premiums, collection of program          No                $1 billion\a
Insurance                                                   enterprise      expenses, and interest earnings

Federal Crop                  Mandatory      Discretionary  Public          Premiums and appropriations\c            Yes               No
Insurance                                    \b             enterprise

Aviation                      Discretionary  Discretionary  Public          Premiums, interest earnings, and one-    No                No
War-Risk Insurance                                          enterprise      time registration fees for nonpremium
                                                                            insurance

Maritime                      Discretionary  Discretionary  Public          Premiums, interest earnings, binder      No                No
War-Risk Insurance                                          enterprise      fees, and claim reimbursements

Service-Disabled              Mandatory      Mandatory      Public          Premiums, interest on policy loans,      Yes               No
Veterans Insurance                                          enterprise      policy loan repayments, and
                                                                            appropriations

Veterans Mortgage             Mandatory      Mandatory      General         Premiums, interest on policy loans,      Yes               No
Life Insurance                                                              policy loan repayments, and
                                                                            appropriations

Federal Employees'            Mandatory      Mandatory      Trust           Premiums and interest earnings           No                No
Group Life Insurance

OPIC's Political              Discretionary  Discretionary  Public          Premiums, insurance claim recoveries,    No                $100 million
Risk Insurance                                              enterprise      and interest earnings

National Vaccine Injury       Mandatory      Mandatory,     Trust           Excise tax on manufacturers and          No                No
Compensation                                 Discretionary                  interest earnings
-----------------------------------------------------------------------------------------------------------------------------------------------------

\a For fiscal year 1997 only, the program is authorized to borrow
$1.5 billion. 

\b Before fiscal year 1994, administrative and operating expenses
were classified as discretionary spending.  Under the 1994 crop
insurance reforms, these expenses were classified as mandatory for
fiscal year 1995 and fiscal year 1996 and discretionary spending for
fiscal year 1997.  However, the most recent Freedom to Farm
legislation classifies these expenses as mandatory for fiscal year
1997 and then splits them between mandatory and discretionary for the
years thereafter. 

\c The Federal Crop Insurance Corporation is authorized under the
Federal Crop Insurance Act, as amended, to use funds from the
issuance of capital stock, which provides working capital for the
corporation. 


--------------------
\10 Veterans Mortgage Life Insurance is included in the Veterans
Insurance and Indemnities account, which is a general fund account. 

\11 Title XIII of Public Law 101-508. 

\12 Under BEA, mandatory spending (also known as direct spending) is
subject to pay-as-you-go (PAYGO) provisions.  PAYGO provisions do not
set limits on mandatory spending but rather control the enactment of
new authorizing legislation for mandatory spending.  Under PAYGO
provisions, legislation enacted during a session of the Congress that
increase mandatory spending or decrease revenues must be at least
deficit neutral in the aggregate.  Deposit insurance spending was
specifically exempted from PAYGO restrictions. 

\13 The aviation and maritime war-risk programs have permanent
authority to spend offsetting collections. 

\14 Under some accrual-based budgeting approaches, expected costs
would be recorded in budget authority and obligated when the
insurance is extended.  (See chapter 6.)


A BUDGET PERSPECTIVE ON FEDERAL
INSURANCE
============================================================ Chapter 3

Budget reporting influences decision-making because it determines how
critical choices are framed and how the deficit is measured.  The
method of budget reporting reflects choices about the uses and
functions of the federal budget.  Ideally, budget reporting should
fully inform resource allocation and fiscal policy choices. 
Unfortunately, the current budget's focus on annual cash flows
provides potentially incomplete and misleading information on the
cost of federal insurance programs.  As a result, the information and
incentives for sound resource allocation decisions and information on
the timing and magnitude of the economic impact of these programs may
be distorted.  However, the impact of these shortcomings on budget
decision-making varies significantly across the federal insurance
programs we reviewed. 


   BUDGET REPORTING REFLECTS
   CHOICES ABOUT THE USES AND
   FUNCTIONS OF THE BUDGET
---------------------------------------------------------- Chapter 3:1

In practice, the federal budget serves multiple functions.  The
budget is used to plan and control resources, assess and guide fiscal
policy, measure the government's borrowing needs, and communicate the
government's policies and priorities.  The budget is both an internal
management tool of the government and a public policy statement.  The
many uses of the budget lead to multiple and often conflicting
objectives for budget reporting.  For example, the budget should be
understandable to policymakers and the public yet comprehensive
enough to fully inform resource allocation decisions.  Since no one
method of budgetary reporting can fully satisfy all uses, the choice
ultimately reflects a prioritization of the budget's various uses. 

The method of budget reporting influences decision-making because the
way budget transactions are recorded determines how critical choices
are framed and how the deficit is measured.  For example, suppose the
federal government extends insurance for which it collects $1 million
in premiums and expects total losses of $3 million to be incurred in
future years.  If the primary objective of the budget is to track
annual cash flows, then it is appropriate to record the $1 million
cash inflow and to offset the aggregate deficit accordingly, as is
currently the case.  However, if the objective is to provide
information on the government's cost when program decisions are made
then it is appropriate to recognize a net cost of the present value
of $2 million in the year the insurance is extended.  Clearly, the
two methods of reporting provide policymakers with very different
information and so may affect budget choices differently.  While both
methods provide useful information and can be tracked simultaneously,
only one can be the primary basis upon which budget decisions are
made.\1


--------------------
\1 For additional discussion, see Budgetary Treatment of Deposit
Insurance:  A Framework for Reform, May 1991, Congress of the United
States, Congressional Budget Office. 


      BUDGET REPORTING SHOULD
      FULLY INFORM RESOURCE
      ALLOCATION AND FISCAL POLICY
      DECISIONS
-------------------------------------------------------- Chapter 3:1.1

An essential step in assessing the adequacy of a program's budget
treatment is determining the information necessary for sound
decision-making.  Although the federal budget has multiple functions,
it is generally recognized that the allocation of resources and
measure of fiscal policy are primary.  In 1967, the President's
Commission on Budget Concepts first identified resource allocation
and fiscal policy as the primary purposes of the budget.\2

In doing so, the Commission acknowledged that no one method of budget
reporting can adequately serve all possible purposes of the budget or
all users' needs but concluded that these other uses were subordinate
to the needs of resource allocation and fiscal policy.  The
Commission reported that

     Of the various purposes for which the President's budget is
     prepared, two closely related purposes outweigh the rest.  .  . 
     .  In short, the budget must serve simultaneously as an aid in
     decisions about the efficient allocation of resources among
     competing claims and economic stabilization and growth.\3

Our assessment of the budget treatment of federal insurance programs
focuses on the adequacy of budget information for resource allocation
and fiscal policy.  However, to support these purposes, budget
reporting must be understandable and provide for budget control and
accountability.\4 As a result, implementation issues, such as
estimation uncertainties and reporting complexities, may offset or
even negate the potential benefits of some changes that would seem to
support resource allocation and fiscal policy decisions.  That is,
decisions on budget treatment must balance the ideal of better
information with the realities of implementation. 

Information on the cost of the government's commitments is vital for
sound resource allocation decisions.  In an environment of limited
resources, decisions are based on the relative budgetary cost\5 of
each potential use.  To permit fully informed choices and provide for
budget control, the cost for each alternative use of federal
resources must be clear and directly related to the commitments
undertaken by the government.  Full cost information in the budget
not only allows policymakers to make relative cost comparisons but
can also warn of estimated increases in costs when they are still
controllable.  To do this, budget reporting ideally would provide
reliable information on the cost of commitments made in a given year. 
However, in practice, there is often no clear bright line at which
this commitment point is made for any particular program. 

For federal insurance programs, key information relevant to
policymakers is the balance between collections and costs over time
flowing from a commitment.  Amounts not covered by program
collections represent the government's subsidy cost for the program. 
Because of the wide variety of risks covered by different federal
insurance programs, the application of the risk-assumed concept is
likely to differ depending on the nature of the program.  For
example, the extent to which a model can capture the full long-term
expected cost of the government's deposit insurance commitments,
including rare catastrophic events such as the savings and loan
crisis, is open to debate. 

Fiscal policy decisions require information on the timing and
magnitude of the economic impact of the government's actions. 
Economic impact is generally considered to be the impact on aggregate
demand and the allocation of resources between private and public
markets.  In general, the budget deficit (or surplus) is considered
to be an appropriate measure of the macroeconomic impact of aggregate
federal fiscal policy on the economy and for most programs cash-based
reporting adequately captures the fiscal impact of budget decisions. 
However, for insurance programs cash-based reporting may misstate the
economic impact of the government subsidy by recording cost when cash
flows occur rather than when the insurance commitment is made. 
Although discerning the economic impact of insurance programs can be
difficult, private economic behavior generally is affected when the
government commits to providing insurance coverage and thus lowers
the risk to the insureds.  Therefore, to fully inform
decision-making, the budget recognition of an activity's expected
costs ideally should coincide with the timing and magnitude of its
economic effects.  Similarly, financial transactions that have no
impact on the cost to the government--such as temporary working
capital needs--should not be recognized in the budget. 


--------------------
\2 Report of the President's Commission on Budget Concepts,
President's Commission on Budget Concepts, U.S.  Government Printing
Office, October 1967. 

\3 Ibid., p.12. 

\4 For additional information on the objectives of the budget
process, see Budget Process:  History and Future Directions
(GAO/T-AIMD-95-214, July 13, 1995). 

\5 Relative budgetary cost refers to the cost recorded in the budget
for one federal program in relation to another.  To the extent that
the method of budget reporting does not measure program costs on a
comparable basis, relative costs and thus resource allocation
decisions will be distorted. 


   CASH-BASED BUDGETING GENERALLY
   PROVIDES INCOMPLETE INFORMATION
   ON FEDERAL INSURANCE PROGRAMS
---------------------------------------------------------- Chapter 3:2

In general, the information provided by cash-based budgeting for
federal insurance programs may be incomplete or misleading for both
resource allocation and fiscal policy decisions.  In most cases, the
cash-based budget does not adequately reflect on a timely basis
either the government's cost or the economic impact of these programs
because costs are recognized when claims are paid rather than when
the commitment is made and when economic behavior is generally
changed.  Thus, the budget may provide neither complete cost
information for budget decision-making nor the incentives necessary
to control costs or ensure that adequate resources are available for
future claims at the time the decision to extend the insurance is
made. 

In general, cash-based budgeting for insurance programs presents
several problems.  In most instances, it focuses on single period
cash flows that may distort the program's cost to the government and
thus may

  -- distort the information and incentives for resource allocation
     decisions,

  -- not accurately reflect the program's economic impact, and

  -- cause deficit fluctuations unrelated to long-term fiscal
     balance. 

However, the magnitude of this problem and the implications for
budget decision-making vary significantly across the insurance
programs.  This is due primarily to differences in the size and
length of the government's commitment, the nature of the insured
risk, and the extent to which costs are currently recognized in the
budget at the time decisions are made. 


      SINGLE PERIOD CASH FLOWS
      DISTORT THE GOVERNMENT'S
      COST FOR FEDERAL INSURANCE
      PROGRAMS
-------------------------------------------------------- Chapter 3:2.1

With limited exceptions, current budget reporting focuses on annual
cash flows.  Collections for insurance programs are recorded in the
budget when received and costs are recorded in outlays and the
deficit when claims are paid.  Yet the focus on annual cash flows may
not adequately reflect the government's cost for federal insurance
programs because the time between the extension of the insurance, the
receipt of program collections, the occurrence of an insured event,
and the payment of claims may extend over several budget periods.  As
a result, the government's cost may be understated in years that a
program's current collections exceed current payments and overstated
in years that current claim payments exceed current collections. 
These distortions occur even if the collections and payments for an
insurance commitment balance over time. 

The timing differences between an insurance activity's collections
and payments on a cash basis are complicated by combining, in a
single account, transactions that represent a cost to the government
and transactions that merely represent cash flows that net out over
time.  A key feature of credit reform was the separation of the
government's cost, called the subsidy cost, from unsubsidized program
costs.  Similarly, federal insurance programs that do not set
premiums high enough to cover expected future claims represent a cost
to the government.  Claim payments to the extent covered by
collections and temporary transactions,\6 such as the acquisition and
sale of assets obtained in settlements, are examples of cash flows
that over time, do not impose a cost to the government.  However,
since the current budget treatment focuses on annual cash flows
rather than a program's long-term financial balance, the cost to the
government--the key information that should be used in budget
decision-making--may be obscured.  The cost of current decisions is
further obscured because single period cash flows often reflect a mix
of old and new insurance business. 

As shown in table 3.1, the timing differences between cash flows for
insurance programs occur for several reasons that vary across the
programs.  The length of the government's commitment (policy
duration) or the time between the occurrence of an insured event and
the payment of claims (the claim settlement period) may extend over
several years.  In addition, erratic cash flows may result from
temporary (working capital) transactions or from the nature and
timing of insured events.  The different reasons for the time lags
between collections and payments among the various insurance programs
are important because they influence both the extent to which
cash-based budgeting is a deficient measure of program costs and the
effectiveness of alternative accrual cost measures in overcoming
these deficiencies. 



                                    Table 3.1
                     
                       Reasons for Mismatch Between Program
                             Collections and Payments

                      Policy                          Claim         Offsetting
                     duration      Loss pattern     settlement     transactions
                  --------------  --------------  --------------  --------------
                                                  Time between    Cash flows
                                                  insured event   resulting from
                                                  and claims      offsetting
                  Government's    Insured events  payment         transactions
                  commitment      tend to be      extends over    such as
                  extends over    sporadic or     several budget  working
Program           several years   catastrophic    periods         capital needs
----------------  --------------  --------------  --------------  --------------
Aviation War-                     X
Risk Insurance

Deposit           X               X               X\a             X
Insurance

Federal Crop                      X
Insurance

Federal           X
Employees'
Group Life
Insurance

Maritime War-                     X
Risk Insurance

National Flood                    X
Insurance

OPIC's Political  X               X               X
Risk Insurance

PBGC's Pension    X               X               X
Insurance

Service-          X
Disabled
Veterans
Life Insurance

National Vaccine                  \b              X
Injury
Compensation

Veterans          X
Mortgage Life
Insurance
--------------------------------------------------------------------------------
\a Based on the experience in the late 1980s in which financial
institutions were allowed to remain open for months or years after
becoming insolvent.  Future experience may be different.  The Federal
Deposit Insurance Corporation Improvement Act of 1991 requires the
prompt closure of severely undercapitalized financial institutions. 

\b Program information insufficient to determine claim pattern. 

A mismatch between collections and payments may occur when the
government's commitment extends over multiple years or budget
periods.  As table 3.1 shows, several federal insurance
programs--those offering multiyear fixed term, renewable term, or
noncancelable coverage--commit the government for extended periods. 
For example, OPIC provides multiyear political risk coverage for up
to 20 years.  In addition, some budget and financial experts view
PBGC's pension guarantee as a long-term, renewable, or noncancelable
commitment.  In all these cases, the extension of the insurance and
the collection of premiums may occur years, even decades, before the
insured event occurs and claim payments come due.  As a result, there
can be years in which an insurance program's current cash collections
are estimated to exceed current cash payments, and the program
appears to be profitable regardless of its expected long-term cost to
the government. 

Time lags between the occurrence of an insured event and the payment
of claims may also result in a mismatch between collections and
payments.  While some insurance programs pay claims within one or two
budget cycles, several do not.  For example, during the savings and
loan crisis, a number of factors, such as inadequate regulatory
oversight and the insurance fund's lack of cash, delayed action to
close failed institutions and pay depositors.  A different set of
factors create a delay in the pension guarantee program.  Benefit
payments of terminated plans assumed by the PBGC may not be made for
years, even decades, because plan participants generally are not
eligible to receive pension benefits until they reach age 65.  Once
eligible, these benefits are paid over a period of years or even
decades.  Payment of claim awards under the Vaccine Injury
Compensation Program (VICP) may not be made for several years after
the injury occurs or not at all.  The total time lag is the sum of
(1) the time between the occurrence of the adverse event and the
filing of a petition for payment, (2) the time taken to reach a
judicial decision with respect to the petition, and (3) the time
between the decision to grant an award and payment.  As of May 1994,
the average time between vaccination and payment for VICP cases
arising from 1989 vaccinations was 1,053 days. 

In some cases, temporary transactions that occur over time may impede
the proper matching of insurance collections and payments on a cash
basis.  During the savings and loan crisis, large temporary cash
flows (working capital) resulting from the acquisition and sale of
assets from failed institutions distorted the government's cost for
deposit insurance in the cash-based budget.  In years that assets
were acquired, the full amount of cash required was recorded as an
outlay; later, when the assets were sold, the proceeds were recorded
as income.  As a result, the cash-based budget overstated the cost of
the deposit insurance in some years and understated it others. 

The catastrophic or uneven occurrence of some insured events also
increases the difficulty in achieving the proper matching of
insurance collections and payments on an annual cash flow basis.  The
focus on annual cash flows generally is not compatible with budgeting
for these types of events because it is difficult to estimate the
occurrence of the insured events and pool risk on an annual basis. 
This is true even when it is possible to estimate the long-term
expected cost of the program.  For example, while it is possible to
estimate with a fair degree of accuracy the probability that floods
will occur over a considerable number of years, predicting the timing
and magnitude of any particular flood by more than a few days is
considered impossible.  Thus, even if long-term flood losses are
correctly estimated, losses in a considerable number of years may
deviate significantly from the long-term average.\7 This means that
in some years cash flows shown in the budget may neither adequately
reflect the program's cost to the government nor recognize the need
to establish reserves over time to cover costs in high-loss years. 


--------------------
\6 For purposes of this report, temporary transactions are defined as
transactions that result in offsetting cash flows that net over time
and, consequently, do not impose a cost on the government. 

\7 Insurance and Other Programs for Financial Assistance to Flood
Victims:  A Report from the Secretary of the Department of Housing
and Urban Development to the President, as Required by the Southeast
Hurricane Disaster Relief Act of 1965, Senate Committee on Banking
and Currency, 89th Congress, 2nd Session, September 1966. 


      FAILURE TO REFLECT THE
      GOVERNMENT'S COST DISTORTS
      THE BUDGET INFORMATION AND
      INCENTIVES NECESSARY FOR
      SOUND RESOURCE ALLOCATION
      DECISIONS
-------------------------------------------------------- Chapter 3:2.2

When insurance costs measured on an annual net cash flow basis do not
capture the cost of the government's insurance commitments,
policymakers may be basing decisions on incomplete or misleading
information.  The failure to isolate and recognize the government's
cost--the key information that should be used for resource
allocation--at the time decisions are made can have significant
implications.  Generally speaking, cash-based budgeting for federal
insurance programs may provide neither the information nor incentives
necessary to signal emerging problems, make adequate cost
comparisons, control costs, or ensure the availability of resources
to pay future claims. 


         CASH-BASED BUDGETING
         NEITHER PROVIDES COMPLETE
         COST INFORMATION WHEN
         DECISIONS ARE MADE NOR
         SIGNALS EMERGING PROBLEMS
------------------------------------------------------ Chapter 3:2.2.1

In most cases, the cash-based budget does not prompt decisionmakers
to consider an insurance program's actuarial soundness.\8 When costs
are not recognized and fully funded in the budget, policymakers may
not receive adequate information on a program's relative budgetary
cost or incentives to address emerging problems.  As a result, the
government's subsidy costs may be obscured until claim payments come
due. 

For example, the National Flood Insurance Program (NFIP) provides
subsidized coverage without triggering recognition of potential
subsidy costs to the government.  Under current policy, the Congress
has authorized the Federal Insurance Administration (FIA) to
subsidize a significant portion (approximately 38 percent) of the
total policies in force without providing annual appropriations to
cover these subsidies.  Although FIA has been self-supporting since
the mid-1980s--either paying claims from premiums or borrowing and
repaying funds to the Treasury--the program has not been able to
establish sufficient reserves to cover catastrophic losses\9

and, therefore, cannot be considered actuarially sound.  Similarly,
the two veteran's life insurance programs included in our
study--Service-Disabled Veterans Insurance (SDVI) and Veterans
Mortgage Life Insurance (VMLI)--also provide subsidized coverage
without accruing the annual cost of the subsidy in the budget in the
year the coverage is extended.\10

The implications of the failure of cash-based budgeting to recognize
potential costs and signal policymakers of emerging problems was most
apparent during the 1980s and early 1990s as the condition of the two
largest federal insurance programs--deposit insurance and pension
insurance--deteriorated while the budget continued to present a
favorable scenario.  For decades, the deposit insurance program
appeared to provide an efficient and self-financing form of
protection.  During this period, the program had positive cash flows
and reduced the federal budget deficit.  Yet, in the 1980s and early
1990s, over 1,600 banks and nearly 1,300 thrifts failed, resulting in
direct costs to taxpayers of $125 billion.  Although GAO and others
raised concerns about these rapidly growing costs, corrective actions
were delayed.  The cash-based budget was slow to recognize the
government's mounting cost of resolving insolvent institutions
because cash outlays were not required until actions were taken to
close them and protect depositors.  These costs had already been
incurred by the time they were disclosed in the budget.  Furthermore,
the cash-based budget may have also created an incentive to delay
closing insolvent institutions (to avoid increasing the annual
deficit), which increased the government's ultimate cost of resolving
the crisis.  Since the crisis, the condition of the deposit insurance
funds has improved dramatically.  Once again, the deposit insurance
programs appear healthy and are generating budgetary cash income\11
--approximately $8.4 billion for fiscal year 1996--that offset the
aggregate deficit. 

In a similar pattern, the cash-based budget did not signal the
deteriorating financial condition of PBGC.  As shown in figure 3.1,
the cash-based budget consistently has reported collections exceeding
payments (negative outlays), while the program's financial
statements, which take into account the present value of insured
benefits the government has incurred, reported an accumulated
deficit.\12 For example, in 1981 when the account containing PBGC's
cash flows was put on-budget, the cash-based budget showed cash
income of $29 million while the financial statements showed an
accumulated deficit of about $190 million.  Over a decade later, in
fiscal year 1992, the cash-based budget continued to provide an
optimistic picture, showing cash income of $654 million, while the
financial statements reported a larger accumulated deficit of about
$2.4 billion. 

   Figure 3.1:  Budgetary Cash
   Flows Versus Accumulated
   Deficit

   (See figure in printed
   edition.)

The cash-based budget has continued to be a poor gauge of PBGC's
financial condition in recent years.  In fiscal year 1996, PBGC
reported a surplus for the first time in its history of $993 million. 
This surplus contrasts sharply with the $2.6 billion accumulated
deficit reported in fiscal year 1993.  The cash-based budget,
however, did not reflect this turnaround.  In fact, cash income
reported in the budget during this period was, on average, lower than
in the previous 4 years when PBGC's financial condition was
deteriorating.  Further, despite the improvement in PBGC's financial
condition, OMB's more prospective estimate of the program's future
cost, included in the Analytical Perspectives of the President's
Fiscal Year 1997 Budget, ranges from $30 to $60 billion.\13 Most
important, if the program's condition were to worsen in the future,
the cash-based budget may not provide timely warning of the program's
deteriorating condition. 

The Federal Employees' Group Life Insurance program also demonstrates
the disparity in the signals provided to policymakers from cash basis
versus accrual basis data.  For example, in fiscal year 1993, when
the program's financial condition shifted on an accrual basis from
having a surplus to a deficit, the cash-based budget failed to signal
this change in the program's financial condition.  While the
program's accrual-based net position shifted from a surplus of about
$1.6 billion to a deficit of $5.8 billion,\14 the cash-based budget
showed cash income of just over $1 billion.  More recently, the
fiscal year 1998 budget year estimates show cash income of $1.2
billion, while the program's balance sheet provided in the budget
appendix reveals an increase in the actuarial liability\15 of
approximately $1 billion and a deficit of about $3.1 billion. 

In addition, the Aviation War-Risk Program appears financially sound
on a cash basis while exposing the government to potentially large
unfunded claims when insurance is in force.\16 Despite a current fund
balance of approximately $67 million, the program's resources may not
be sufficient to cover potential insurance claims.  One major
loss--such as a Boeing 747-400, which can cost over $100
million--could liquidate all the available funds and leave a
substantial portion of the claim unfunded.  If a loss exceeded the
available funds, the Federal Aviation Administration (FAA) would have
to seek supplemental funding to cover the claim.\17


--------------------
\8 In order to be actuarially sound, a program's funding would need
to be sufficient to cover expected future payments for claims and
administrative expenses. 

\9 The Federal Insurance Administration estimates that a catastrophic
loss year resulting in $3 billion to $4 billion in claim losses has a
1 in 1,000 chance of occurring each year. 

\10 The SDVI and VMLI programs reported accrual-based deficits of
$457 million and $93 million, respectively, as of September 30, 1996. 

\11 Budgetary cash income refers to the cash flows shown as negative
net outlays in the budget. 

\12 The liability includes an estimate for future pension benefits
that PBGC is or will be obligated to pay with respect to trusteed
plans and terminated plans pending trusteeship.  In addition, it
includes an estimate of the liabilities attributable to plans that
are likely to terminate in a future year based on conditions that
exist at the end of the fiscal year. 

\13 Unlike the financial statement liability, OMB's estimate includes
costs resulting from expected future terminations of underfunded
plans sponsored by currently healthy firms. 

\14 According to agency officials, an unfunded liability of about $5
billion resulted from an unexpected increase in the number of
enrollees in option B--the most costly option--during an open season
for both basic and optional coverage in fiscal year 1993.  The
accumulated deficit is based on statutory reporting requirements,
which, according to agency officials, might overstate the program's
liability. 

\15 The actuarial liability represents the excess of the present
value of estimated benefits to be paid less the present value of
estimated premiums. 

\16 The Aviation Program is only activated when commercial insurance
is unavailable on reasonable terms and conditions and continued
service is in the interest of U.S.  policy.  This limited and
sporadic operation may reduce the feasibility of accrual-based
budgeting for this program.  See appendix IV for a more detailed
description of the program. 

\17 For additional information on the Aviation War-Risk Program, see
Aviation Insurance:  Federal Insurance Program Needs Improvements to
Ensure Success (GAO/RCED-94-151, July 15, 1994). 


         CASH-BASED BUDGETING
         GENERALLY PROVIDES
         NEITHER THE INFORMATION
         NOR THE INCENTIVES TO
         CONTROL COSTS
------------------------------------------------------ Chapter 3:2.2.2

Because the cash-based budget delays recognition of emerging
problems, it may provide little or no incentive to address potential
funding shortfalls before claim payments come due.  Policymakers may
not be alerted to the need to address programmatic design issues
because, in most cases, the budget does not encourage them to
consider the future costs of federal insurance commitments.  Thus,
reforms aimed at reducing costs may be delayed.  In most cases, by
the time costs are recorded in the budget, policymakers do not have
time to ensure that adequate resources are accumulated or to take
actions to control costs.  Delayed recognition of these costs can
reduce the number of viable options available to policymakers,
ultimately increasing the cost to the government. 

Further, in some cases, the cash-based budget not only fails to
provide incentives to control costs, it may also create a
disincentive for cost control.  Deposit insurance is a key example. 
Many analysts believe that the cash-based budget treatment of deposit
insurance exacerbated the saving and loan crisis by creating a
disincentive to close failed institutions.  Since costs were not
recognized in the budget until cash payments were made, leaving
insolvent institutions open avoided recording outlays in the budget
and raising the annual deficit but ultimately increased the total
cost to the government. 

In the past, the cash-based budget treatment and budget scoring rules
also have been cited as creating disincentives for implementing
pension insurance reforms.  For example, CBO reported that the Bush
administration's 1992 program reform proposals would have reduced
PBGC's funding shortfall and enhanced the financial stability of the
program.\18 However, these reforms--specifically the one raising the
minimum contributions required of sponsors of insured pension
plans--would have reduced income tax revenues (because contributions
are tax deductible) and added to the federal deficit in the near
term.  Thus, under the pay-as-you-go (PAYGO) provisions of BEA,\19
these reforms would have required reductions in other spending or
increases in other revenues. 


--------------------
\18 CBO Testimony, Congressional Budget Office, August 11, 1992. 

\19 Under PAYGO provisions of BEA, legislation enacted during a
session of the Congress that increases mandatory spending or
decreases revenues must be at least deficit neutral in the aggregate. 


         FAILURE TO RECOGNIZE COST
         AND SIGNAL EMERGING
         PROBLEMS MAY DISTORT
         RESOURCE ALLOCATION AND
         CONSTRAIN FISCAL POLICY
------------------------------------------------------ Chapter 3:2.2.3

To the extent that the cash-based budget fails to capture the cost
implicit in the government's commitment and signal emerging problems,
the relative budgetary costs of an insurance program will be
distorted.  In some cases, this may simply result in the delayed
recognition of intended choices, but, in other cases, it may lead to
unintended resource allocation and fiscal policy.  For example, by
the time claims come due the government may be faced with little
choice but to increase the deficit, raise taxes, or cut other
spending in order to honor these commitments.  The lack of cost
recognition may delay programmatic changes aimed at reducing costs at
a point when they are manageable.  In summary, the failure to
recognize these costs when decisions are made may not only distort
current budget choices among competing uses, but may also reduce
options for cost control and future budget flexibility when bills
come due. 


      CASH-BASED BUDGETING MAY NOT
      REFLECT THE ECONOMIC IMPACT
      OF FEDERAL INSURANCE
      PROGRAMS
-------------------------------------------------------- Chapter 3:2.3

In addition to not providing sufficient information and incentives
for resource allocation, the cash-based budget also may not be a very
accurate gauge of the economic impact of federal insurance programs. 
Although discerning the economic impact of federal insurance programs
can be difficult, private economic behavior generally is affected
when the government commits to providing insurance coverage.  It is
at this point that insured individuals or organizations alter their
behavior as a result of insurance.  However, as noted above, the
current cash-based budget records costs not at that point but rather
when payments are made to claimants.  Federal payments for insurance
claims may have little or no macroeconomic effect because these
payments generally do not increase the wealth or incomes of the
insured.  They are merely intended to restore the insured to the
approximate financial position he or she would have been in absent
the occurrence of the insured event. 

For example, most analysts agree that the cash-based budget provided
misleading information on the timing and magnitude of the economic
effects of deposit insurance.  A 1992 CBO study of the economic
effects of the savings and loan crisis concluded that the economic
impact of deposit insurance is more directly related to the accrual
of new federal obligations for deposit insurance than to cash
payments made under the program.\20 While federal costs, on an
accrual basis, mounted steadily during the 1980s as hundreds of
thrift institutions became insolvent, the budget did not record any
costs until institutions were closed and depositors paid.  Although
unrecognized in the budget, these accruing liabilities had economic
effects at the time similar to conventional expansionary policy in
that aggregate demand was maintained at a higher level than it would
have been if the depositors had sustained losses.\21 Conversely, the
budget outlays made to restore saving and loan depositors' accounts
had little effect on overall demand because the wealth or income of
depositors was not increased.  Further, many analysts have concluded
that unlike most spending on other federal programs, the federal
borrowing to fund the payments for deposit insurance did not
significantly increase interest rates because it did not lead to any
increase in the demand for goods or services.  Instead, interest
rates tended to increase as the government's deposit insurance
liabilities accrued. 


--------------------
\20 The Economic Effects of the Savings and Loan Crisis,
Congressional Budget Office, January 1992. 

\21 In general, conventional expansionary fiscal policy raises the
income of some groups, leading to increased consumption and aggregate
demand.  Since deposit insurance protects the wealth of depositors in
the event of a bank or thrift failure, it increases the consumption
of insured depositors and raises overall demand. 


      CASH-BASED BUDGETING FOR
      INSURANCE MAY PRODUCE
      FLUCTUATIONS IN THE FEDERAL
      DEFICIT UNRELATED TO
      LONG-TERM FISCAL BALANCE
-------------------------------------------------------- Chapter 3:2.4

Uneven cash flow patterns of some federal insurance programs can
result in fluctuations in the federal deficit unrelated to the
budget's long-term fiscal balance.  As noted earlier, uneven cash
flows may result from both the erratic nature of some insured risks
or temporary (working capital) transactions.  For example, natural
disasters, such as severe floods and droughts, may create spikes in
spending patterns that are not indicative of long-term fiscal
balance.  In addition, the working capital used to resolve failed
institutions under the deposit insurance program resulted in large
temporary cash flows that distorted the aggregate deficit as a
measure of the government's long-term fiscal imbalance. 

Insurance programs with long-term commitments, such as PBGC and life
insurance programs, also may distort the budget's long-term fiscal
balance by looking like revenue generators and reducing the aggregate
deficit in years that collections exceed payments without recognizing
the programs' expected costs.  On a cash basis, premium income can
divert attention from such programs' financial condition.  For
example, although the PBGC reforms that were enacted in 1994 as part
of the General Agreement on Tariffs and Trade (GATT) legislation will
likely improve the financial condition of the program, they were
adopted at least in part because on a cash basis they raised
revenues.  The increase in revenue, primarily resulting from the
phase-out of the cap on premiums charged underfunded plans, was
necessary under PAYGO budget rules to offset revenue lost from
changes in various tariffs affected by the trade agreement.  This
budget accounting, however, does not recognize that these premiums
will be needed to pay PBGC's costs in the future. 


   THE IMPLICATIONS OF CASH-BASED
   BUDGETING FOR DECISION-MAKING
   VARY ACROSS PROGRAMS
---------------------------------------------------------- Chapter 3:3

While annual cash flows for federal insurance programs generally do
not provide complete information for resource allocation and fiscal
policy, the magnitude of the problem and the implications for budget
decision-making vary across the insurance programs reviewed. 
Specifically, the size and length of the government's commitment, the
nature of the insured risk, and the extent to which these costs are
currently captured in cash-based budget estimates influence the
degree to which cash-based budgeting is incomplete or misleading for
a particular federal insurance program. 

The size of a program relative to total federal spending and the
potential magnitude of unrecognized costs are key factors in judging
the severity of the shortcomings of cash-based budgeting.  For
example, the implications of the shortcomings of the current budget
treatment appear greatest for the largest insurance programs, pension
and deposit insurance.  The large size of these programs means that
incomplete or misleading information about their cost could distort
resource allocation and fiscal policy significantly, making the
limitations of cash-based budgeting more pronounced than for other
federal insurance programs. 

The limitations of cash-based budgeting are most apparent when the
government's commitment extends over a long period of time--e.g.,
life or pension insurance-- and/or the insured events are infrequent
or catastrophic in nature, such as severe flooding or depository
losses.  As discussed earlier, the cash-based budget may not provide
timely recognition of the government's costs for these commitments
because of the time lags between the extension of the insurance and
the payment of claims as well as the difficulty in estimating and
pooling risk on an annual basis.  As a result, the cash-based budget
may provide misleading or incomplete cost information for extended
periods, thus not signaling policymakers of emerging problems when
costs are controllable.  In these cases, both the direction--positive
or negative--and the magnitude of the government's costs may be
distorted on an annual cash flow basis. 

Conversely, the deficiencies of cash-based budgeting may not be as
problematic when the length of the government's commitment is short
and claims occur relatively frequently, such as the occurrence of
normal losses under crop and flood insurance programs.  In these
cases, because the length of time between the government's commitment
and the occurrence and payment of claims is relatively short, the
accumulation of unrecognized losses over an extended period of time
is less likely. 

In addition, the implications for budget decision-making may be less
severe if relatively frequent claim payments prompt policymakers to
consider the financial condition and funding needs of the program. 
For example, some insurance programs, such as flood and crop
insurance, use the average or normal annual loss to make annual
budget estimates.  Even so, this approach does not isolate and may
not completely capture the programs' full costs, including the need
to establish reserves for catastrophic losses.  While these estimates
provide policymakers some signals about potential costs at the time
decisions are made, the programs' relative costs may still be
understated.  For example, in the case of flood insurance, premiums
based on the historical average loss year may not be sufficient to
establish reserves to cover catastrophic losses because the loss
experience period used does not include a catastrophic loss year.  As
a result, the program's cost is understated and the government's cost
may not be recognized until the bills come due. 


ACCRUAL-BASED BUDGETING HAS THE
POTENTIAL TO IMPROVE BUDGET
INFORMATION AND INCENTIVES FOR
MOST FEDERAL INSURANCE PROGRAMS
============================================================ Chapter 4

Several characteristics of federal insurance programs support the use
of accrual-based budgeting.  Two general approaches for measuring
accrual-based costs of insurance programs are (1) the risk-assumed
concept, which recognizes the cost of claims inherent in the
government's commitment at the time of commitment,\1 and (2)
traditional financial reporting standards for claims liabilities,
which generally recognize the cost of claims inherent in events that
have already occurred.  The risk-assumed basis would be more useful
for budgeting because it looks further ahead at the time the
commitment is made rather than waiting for claims-producing events. 
Thus, the information and incentives for resource allocation and
fiscal policy could be improved--assuming it is possible to make
reasonable cost estimates.  While moving to accrual-based budgeting
based on the risk-assumed concept would offer several benefits, the
magnitude of the change in information and incentives provided to
policymakers varies across insurance programs and depends on the
design of the accrual-based budgeting approach used. 


--------------------
\1 As discussed in chapter 7, the extent to which administrative and
other operating expenses should be included in the calculation of the
risk-assumed accrual cost needs to be determined. 


   CHARACTERISTICS OF FEDERAL
   INSURANCE PROGRAMS SUPPORT USE
   OF ACCRUAL CONCEPTS
---------------------------------------------------------- Chapter 4:1

As discussed in the previous chapters, several characteristics of
federal insurance programs complicate their budget reporting.  In
some respects, the difficulties in budgeting for insurance programs
are similar to those for loan guarantees.  Both insurance and
guarantees commit the government to pay future losses inherent in the
coverage provided.  Both represent contingent liabilities\2 that
generally are not adequately reflected on a cash basis because the
government's full cost is not recognized when the commitment is made. 
While credit reform dealt with this problem for loan guarantees and
improved the budget recognition of their cost, the cost of most
federal insurance programs is not fully recognized in the budget at
the time the insurance commitment is extended. 

The analogy to credit programs suggests that some form of
accrual-based budgeting could improve the budget treatment of federal
insurance programs.  Specifically, two features of federal insurance
programs support the use of accrual-based budgeting for these
programs:  (1) the promise to cover future losses that may occur
beyond the current budget period and (2) the difficulty in estimating
and pooling some risks on an annual basis.  Accrual-based budgeting
would allow for the recognition of future costs at the time
commitments are made.  However, insurance is dissimilar to loan
guarantees in some ways that present additional challenges for cost
estimation and budget control.  These issues need to be dealt with
carefully before accrual-based budgeting can be applied to federal
insurance programs.  For example, the benefits of accrual-based
budgeting depend heavily on whether reasonable cost estimates are
currently available or can be developed.  In some cases, estimating
the risk assumed by insurance programs may be a greater challenge
than for some credit programs. 

Accrual-based reporting recognizes transactions or events when they
occur regardless of when cash flows take place.  An important feature
of accrual-based reporting is the matching of expenses and revenues
whenever it is reasonable and practicable to do so.  In the case of
insurance, accrual concepts would recognize the cost for future claim
payments and the establishment of reserves to pay those costs.  Thus,
the use of accrual concepts in the budget has the potential to
overcome the time lag between the extension of an insurance
commitment and the payment of claims that currently distorts the
government's cost for these programs on an annual cash flow basis. 
To the extent practicable, the government's ultimate cost is the key
information for budget decision-making. 

The Federal Accounting Standards Advisory Board (FASAB) has done
significant work to develop financial reporting standards to meet the
needs of the various users of federal financial statements.\3 The
accounting principles developed by FASAB provide a sound foundation
for federal financial statements that are useful and relevant to
needs of the federal environment.  FASAB's work also provides a
useful framework for understanding the use of accrual cost measures
for budgeting for federal insurance programs.  As such, efforts to
apply accrual-based budgeting for federal insurance should build on
and further adapt this work for budget purposes. 


--------------------
\2 Contingent liabilities are obligations that are dependent upon the
occurrence or nonoccurrence of one or more future events. 

\3 For a detailed discussion of uses, user needs, and objectives of
federal financial reporting, see Objectives of Federal Financial
Reporting:  Statement of Recommended Accounting and Reporting
Concepts, Federal Accounting Standards Advisory Board, July 1993. 


   THE RISK-ASSUMED CONCEPT IS
   MOST APPROPRIATE FOR BUDGETING
   FOR FEDERAL INSURANCE PROGRAMS
---------------------------------------------------------- Chapter 4:2

The focus and purpose of federal budget reporting argues for the use
of forward-looking cost measures for federal insurance programs if
reasonably reliable ones can be developed.  In order to support
current and future resource allocation decisions and formulate fiscal
policy, the federal budget needs to be a prospective document that
enables and encourages users to weigh the future consequences of
current decisions.  To do this, the budget should provide the
information and incentives necessary to assess the future
implications of various choices.  For federal insurance programs, the
information needed for budgeting is roughly analogous to the
insurance rate-setting process because the relevant question in
assessing the government's ultimate cost is whether premiums over the
long term will be sufficient to cover losses and, if not, what
subsidy the government is agreeing to provide.  That is, when the
federal government decides to undertake the role of the insurer,
policymakers need information on the cost of the risk inherent in the
government's commitment. 

The risk-assumed cost measure\4 would provide the prospective
information necessary for budget decisions about insurance programs. 
For insurance programs, risk assumed generally refers to the portion
of the full risk premium based on the expected cost of losses
inherent in the government's commitment that is not charged to the
insured.\5 As a result, the government's subsidy cost--the difference
between the full-risk premium and actual premiums charged--may be
more visible in the budget process.  Thus, the use of risk-assumed
estimates in the budget would provide the information necessary for
assessing the cost of establishing reserves and the ability of an
insurance program to pay future losses.  This approach is similar to
that used under credit reform to measure the cost of direct loans and
loan guarantees.  However, because of the wide variety of risks
covered by federal insurance programs, the risk-assumed concept may
be interpreted differently depending on the nature of the program. 
For example, the time horizon used to estimate the risk assumed by
the government under deposit insurance may be shorter than that used
to estimate the risk assumed in providing life insurance coverage to
federal employees. 

The risk-assumed concept expands upon the standards used for
financial statement reporting.  Except in the case of life insurance,
the risk-assumed concept takes a longer look forward than standards
used to recognize insurance liabilities in financial statements. 
Under standards developed by FASAB, the financial statements for all
federal insurance programs must recognize a financial statement
liability based on insured events that have been identified by the
end of the accounting period.\6 The standard requires recognition of
the expected unpaid net claims inherent in insured events that have
already occurred, including (1) reported claims, (2) claims incurred
but not yet reported,\7 and (3) any changes in contingent liabilities
that meet the criteria for recognition.\8 Life insurance programs are
required to recognize a liability for future policy benefits\9 in
addition to the liability for unpaid claims.  This means that except
for life insurance, no liability for an insurance cost is recognized
in the financial statements until it is probable that a cost has
actually been incurred and the amount of the cost can be reasonably
estimated.  These liability reporting requirements closely parallel
the liability reporting requirements for private sector insurance
companies\10 and are based on the principles that are essential to
support the purposes of financial statement reporting.\11

In developing these standards, FASAB also recognized the importance
of the risk-assumed measure for federal insurance programs.  Because
the risk-assumed measure provides important information beyond that
included in the financial statement liability, FASAB recommended and
the final standards require that this information be disclosed as
supplemental information beginning with financial statements for
fiscal year 1997.\12 However, concerns about the measurability and
the exact nature of some risks assumed by the government dissuaded
FASAB from recommending the use of risk-assumed estimates as the
basis for liability recognition in the financial statements. 
Disclosure of risk-assumed estimates provides users with a broader
and prospective cost measure that FASAB believes is relevant in
assessing whether future budget resources will be sufficient to
sustain public services and meet obligations. 

The accrual-based cost measures appropriate for the budget differ
from those appropriate for financial statements largely because of
differences in the primary purposes of the information, the nature of
the federal budget environment, and differences in the acceptable
level of uncertainty for financial statements and budget projections. 
In the past, CBO\13 and OMB\14 have expressed concerns about the
limitations of traditional financial reporting standards for
assessing future budgetary costs of insurance programs.  Generally
speaking, traditional financial statement reporting is of limited use
for budget purposes because, in most cases, it does not recognize the
potential costs of claims that have not yet been incurred\15 or the
present value of future premiums\16 that offsets future budgetary
costs.  Federal accounting standards requiring supplemental
disclosure of an estimate of the risk assumed should improve the
recognition of these potential costs in federal financial statements. 

The benefits achieved by budgeting using financial statement
liability recognition standards would vary across the insurance
programs.  The benefits achieved depend primarily on the length of
time between the occurrence of insured events and payment of claims. 
For some programs, the traditional liability recognition standards
may yield information not very different from what is currently
reported on a cash basis in the budget.  However, for programs with
long time lags between the occurrence of the insured event and the
payment of claims, such as pension and deposit insurance,\17

budgeting based on financial statement liability standards might
provide earlier budget recognition of the costs incurred than does
cash-based budgeting.  In these cases, the earlier recognition could
reduce the incentive to delay the payment of claims and would allow
for some earlier action to reduce future costs.  In most cases,
however, this approach would not be as forward-looking as the
risk-assumed concept and, therefore, would not provide recognition of
the risks inherent in the government's commitment at the time that
the insurance is extended.  It is at that time that decisions can be
made to change the extent of the risk being assumed by the
government.  Since the financial statement liability standards
generally report costs that have been incurred as the result of past
decisions, using that standard for estimating the government's cost
in the upcoming budget year may not provide signals of the
government's risk exposure early enough so as to maximize options
available for limiting program costs.  This is true because the range
of options for changing the program to reduce the government's costs
may be more limited after the cost has been incurred than it would
have been when the insurance was extended.  Nevertheless, until
risk-assumed estimates are fully developed, insurance programs'
financial statements, which are included in the budget appendix,
provide policymakers with valuable information on insured events
(losses) that are probable and measurable as of a given date and
should be considered in budget discussions. 

Table 4.1 compares the potential benefits of accrual-based budgeting
using these two cost recognition standards.  The potential benefits
of accrual-based budgeting based on the risk-assumed concept included
in table 4.1 are discussed in the following section. 



                                                                     Table 4.1
                                                      
                                                      Usefulness of Cost Recognition Approach
                                                           for Improving Budget Treatment

                                  Life insurance\a                     PBGC                   Deposit insurance              Other insurance
                            -----------------------------  ----------------------------  ----------------------------  ----------------------------
                            Financial                      Financial                     Financial                     Financial
                            statement                      statement                     statement                     statement
Benefit of change in        liability                      liability                     liability                     liability
budget treatment            recognition     Risk assumed   recognition    Risk assumed   recognition    Risk assumed   recognition    Risk assumed
--------------------------  --------------  -------------  -------------  -------------  -------------  -------------  -------------  -------------
Recognizes the risk         X               X                             X                             X                             X
assumed by the government
at the time the commitment
is made

Improves the information    X               X              \b             X              \b             X              \b             X
and incentives for
managing insurance costs

Provides comparable cost    X               X                             X                             X                             X
information at the time
decisions are made

Establishes "reserve" for   X               X                             X                             X                             X
sporadic/catastrophic
events

Reflects the timing and     X               X                             X                             X                             X
magnitude of the program's
economic impact
---------------------------------------------------------------------------------------------------------------------------------------------------
Note:  Assumes that reasonably reliable risk-assumed cost measures
can be developed. 

\a The financial statement liability for life insurance is measured
on a risk-assumed basis. 

\b Method may improve information provided in the budget but not to
the extent of risk-assumed information. 


--------------------
\4 As noted earlier, there are two general ways that measure the cost
of future claim payments for budgeting purposes:  a measure based on
the risk inherent in the insurance and a measure based on the
occurrence of an insured event.  The latter measure is used to record
claims liability in financial statements for most federal insurance
programs. 

\5 As will be discussed in chapter 5, the concept of risk
assumed--losses inherent in the government commitment--is consistent
for all federal insurance programs, but how risk assumed is
calculated, such as the time period considered, may vary across
insurance programs.  In some cases, estimating the full-risk premium
might prove to be prohibitively difficult and modifications to the
risk-assumed concept may be necessary. 

\6 Statement of Federal Financial Accounting Standards No.  5,
Accounting for Liabilities of the Federal Government. 

\7 Claims relating to insured events that have occurred but have not
yet been reported to the insurer as of the date of the financial
statements. 

\8 Under FASAB standards, contingent losses are only reported as a
liability and charged to expenses if a past transaction or event has
occurred and the loss is both probable (events are likely to occur)
and measurable. 

\9 The liability for future policy benefits represents the expected
present value of future outflows paid to, or on behalf of, existing
policyholders, less the expected present value of future net premiums
to be collected from those policyholders. 

\10 Standards for private sector entities are promulgated by the
Financial Accounting Standards Board (FASB).  Applicable standards
include Statement of Finanical Accounting Standards (SFAS) No.  5: 
Accounting for Contingencies, SFAS No.  60:  Accounting and Reporting
by Insurance Enterprises, and SFAS No.  97:  Accounting and Reporting
by Insurance Enterprises for Certain Long-Duration Contracts and for
Realized Gains and Losses From the Sale of Investments.  FASB
considered and rejected catastrophic reserve accounting for property
and casualty insurance.  A key consideration in this decision was the
ability of private sector insurance companies to recoup losses in
premiums charged to policyholders.  FASB viewed the long-run nature
of pricing premiums as separate from and not a determinant of when a
liability should be recorded.  This reflects the different
requirements and needs of traditional accrual liability recognition
and the needs of federal budgeting.  As noted earlier, accrual-based
budgeting for insurance programs is more closely related to the
premium rate-setting (internal management) process of private sector
companies than to their external (liability) reporting process. 

\11 Examples of these principles include reliability, relevance,
consistency, comparability, and materiality. 

\12 Risk-assumed estimates for all insurance and guarantee programs
will be reported as required supplementary stewardship information. 
For insurance programs administered by government corporations, which
follow FASB (private sector) accounting standards, risk-assumed
estimates will be reported only when financial information on the
government corporation is consolidated into general purpose financial
reports of a larger federal reporting entity. 

\13 Budgetary Treatment of Deposit Insurance:  A Framework for
Reform, Congressional Budget Office, May 1991. 

\14 Budgeting for Federal Deposit Insurance, Office of Management and
Budget, June 1991. 

\15 In 1991, the Federal Deposit Insurance Corporation adopted a
somewhat more prospective view of what constitutes an accountable
event for the purpose of recognizing estimated future deposit
insurance losses.  It now includes an estimated loss from
institutions that are solvent at year-end, but which have adverse
financial trends and will probably become insolvent in the future. 

\16 Under traditional financial accounting standards, revenue
generally cannot be recognized until it is earned. 

\17 This is based on the experience in the late 1980s in which
financial institutions were allowed to remain open for months or
years after becoming insolvent.  Future experience may be different. 
The Federal Deposit Insurance Corporation Improvement Act of 1991
requires the prompt closure of severely under-capitalized financial
institutions. 


   ACCRUAL-BASED BUDGETING HAS THE
   POTENTIAL TO IMPROVE RESOURCE
   ALLOCATION AND FISCAL POLICY
   DECISIONS
---------------------------------------------------------- Chapter 4:3

Accrual-based budgeting for federal insurance programs based on the
risk-assumed concept\18 has the potential to improve the information
and incentives for resource allocation and fiscal policy by
overcoming many of the deficiencies of cash-based budgeting. 
Specifically, the potential benefits of accrual-based budgeting for
federal insurance programs include

  -- providing more accurate and timely recognition of the
     government's cost of insurance commitments,

  -- improving the information and incentives for managing insurance
     costs,

  -- making cost information for insurance programs more readily
     comparable to other federal programs,

  -- providing a mechanism to establish reserves for high or
     catastrophic loss years, and

  -- reflecting more accurately the economic impact of insurance
     programs. 

However, the extent to which a shift to accrual-based budgeting will
change the information and incentives varies across insurance
programs.  This is due primarily to differences in the size and
length of the government's commitment, the nature of the insured
risk, and the extent to which costs are currently recognized in
budget authority and obligations at the time the budget decisions are
made.  In addition, the approach used to incorporate risk-assumed
estimates into the budget will affect the degree to which each
achieves these benefits.  Three general approaches to using
accrual-based estimates in the budget will be discussed in chapter 6. 


--------------------
\18 In the remainder of this report, all references to accrual-based
budgeting assume the use of the risk-assumed measurement basis. 


      ACCRUAL-BASED BUDGETING
      WOULD PROVIDE MORE TIMELY
      RECOGNITION OF THE
      GOVERNMENT'S COST FOR
      INSURANCE COMMITMENTS
-------------------------------------------------------- Chapter 4:3.1

Accrual-based budgeting for federal insurance programs has the
potential to reduce the cost distortions that occur in the cash-based
budget by improving the match between estimated revenues and claims
of insurance commitments.  By doing so, accrual-based budgeting would
recognize any imbalance or net cost to the government--the key
information that should be considered in budget decision-making--in
the year the insurance is extended. 

The prospective recognition of insurance costs is the key advantage
of accrual-based budgeting for federal insurance programs.  Unlike
the current cash-based budget, accrual-based budgeting would
recognize and report the government's costs for insurance commitments
at the time decisions are made and costs are controllable.  As a
result, the adoption of accrual-based budgeting for federal insurance
programs would shift the focus of the budget from retrospective
reporting to prospective cost estimation. 


      ACCRUAL-BASED BUDGETING MAY
      IMPROVE THE INFORMATION AND
      INCENTIVES FOR MANAGING
      INSURANCE COSTS
-------------------------------------------------------- Chapter 4:3.2

The prospective focus of accrual-based budgeting has the potential to
improve both the opportunities and incentives for controlling
insurance costs by providing more timely warning of emerging
problems.  Under accrual budgeting, the subsidy costs--the difference
between expected losses and expected income--would be included in the
budget and serve as a gauge of the government's risk exposure.  Thus,
policymakers would be encouraged to examine the underlying benefits
and structure of insurance programs before large losses accumulate. 
Since policymakers are prompted to take action to reduce costs when
costs are still controllable, the potential for unintended subsidies
may be reduced.  For example, according to OMB, the subsidy conveyed
by deposit insurance rises with increased exposure, such as an
increase in the number of weak institutions, and falls as policies
are put in place that effectively limit risk-taking with insured
funds.  Thus, if properly recognized in the budget, the growing
subsidy cost for deposit insurance would have signaled policymakers
in the 1980s that thrifts and banks were undertaking greater risks
and depending more heavily on deposit insurance guarantees. 

In cases where the Congress intends to provide a subsidy in order to
achieve some public policy objective--as is the case for some
veterans life insurance programs and the flood insurance
program--accrual-based budgeting would prompt recognition of the
subsidy cost at the time the coverage is extended.  Thus, the cost
recognition in the budget would be more clearly linked to the
decision to provide subsidized coverage rather than merely reflecting
the unfunded bills when they come due. 

The earlier reporting of costs on an accrual basis not only changes
the information available to policymakers but also changes budget
incentives if actually incorporated into outlays and/or budget
authority.  Unlike cash-based budgeting that delays cost recognition
and does not encourage early action to control cost, the earlier cost
recognition provided by an accrual basis shifts the budget incentives
in favor of reforms aimed at controlling costs.  For example, under
some accrual-based budgeting approaches, policymakers would be faced
with a choice of providing additional government funding to cover
shortfalls, raising premiums, or otherwise reducing program benefits
to reduce future costs.  However, the extent to which budget
incentives are changed depends on the nature of the particular
insurance program, the accrual-based budgeting approach used, and the
extent to which budget recognition leads to choices between
additional funding and programmatic changes.  These issues are
discussed in more detail in chapters 6 and 7. 


      ACCRUAL-BASED BUDGETING MAY
      IMPROVE RELATIVE COST
      INFORMATION
-------------------------------------------------------- Chapter 4:3.3

The use of accrual-based budgeting for federal insurance programs
also has the potential to improve the information available to make
relative cost comparisons.  As discussed in chapter 3, the cash-based
budget may misstate the government's cost for insurance commitments. 
On a cash basis, some insurance programs may appear profitable while
subjecting the government to long-term costs.  As a result, cost
comparisons with programs whose costs are fully reflected on a cash
basis will be distorted.  Accrual-based budgeting allows for better
relative cost comparisons by recognizing the government's expected
costs for insurance commitments at the time decisions are made.  For
example, for fiscal year 1993, an accrual-based budget would have
shown that PBGC had a potential future cost to the government rather
than being an income generator as reflected in the cash-based budget. 
As a result, pension insurance would have competed for budget
resources with other federal programs based on the government's
expected cost rather than appearing to be a source of income in
budget terms. 


      ACCRUAL-BASED BUDGETING FOR
      INSURANCE PROGRAMS WOULD
      PROVIDE A MECHANISM FOR
      ESTABLISHING PROGRAM
      RESERVES
-------------------------------------------------------- Chapter 4:3.4

In addition to improving cost recognition and resource allocation,
accrual-based budgeting for insurance programs would provide
policymakers with a mechanism for establishing program reserves for
expected insurance losses.  One outcome of budgeting based upon the
full risk assumed by the government would be that in some years
premiums collected and subsidies provided by the government would
exceed cash payments for insured losses.  This would occur because
losses for some programs are highly variable from year to year and
for other programs may not occur for many years.  As a result, when
premiums and the government subsidy exceed claim payments, funds
could be held in reserve for expected future claims. 

The establishment of reserves may be particularly important given
that many of the risks insured by the federal government are
catastrophic in size and/or erratic in occurrence.  The uneven
occurrence of these risks makes estimating funding needs on an annual
basis difficult because actual losses in any particular year may
vary, in some cases significantly, from the estimated annual cost
based on the long-term expected risk.  For example, a widespread
drought can result in claim payments in a single year to a large
proportion of farmers insured under the crop insurance program. 
Charging premiums sufficient to cover a catastrophic loss in any one
year would be prohibitively expensive.  As a result, in order for the
program to be financially sound, amounts sufficient to cover high or
catastrophic losses need to be accumulated over a number of years. 
Other federal insurance programs, such as life and pension insurance,
commit the government to making payments many years in the future. 
As a result, premiums collected over the duration of the policy must
be held in reserve to pay the promised benefits at some future date. 
If, over time, sufficient reserves are accumulated to pay expected
costs, the program would be fully funded.\19 However, a program could
require additional funds--or borrowing authority--if significant
losses occur before sufficient reserves are accumulated even if
annual funding is based on the long-term expected cost. 

While accrual-based budgeting for insurance programs would recognize
any government subsidy at the time insurance is extended and hold
such amounts in reserve,\20 the government's financing needs would
not change.  As is the current practice with insurance and many other
funds, when a program's collections exceed its cash needs for
payments, reserves are held in Treasury securities (i.e., lent to the
government), which, from a governmentwide perspective, satisfies some
of the government's borrowing needs.  Under accrual-based budgeting,
federal borrowing (or a reduction in other spending) would still be
necessary when the insurance fund redeemed its Treasury securities to
make cash payments if insurance claims exceeded premiums collected
from the public in a given year.  However, under accrual-based
budgeting, the government's cost for the program would have already
been recognized in the budget when the commitment was extended. 


--------------------
\19 The government's cost would be funded from the perspective of the
program but not of the government as a whole since under current
practice reserves are held in Treasury securities (i.e., borrowed by
the Treasury to finance other government spending). 

\20 The degree to which the government's cost is recognized in budget
authority, outlays, and the deficit depends on the approach used to
incorporate accrual-based measures in the budget.  The advantages and
disadvantages of different approaches are discussed in chapter 6. 


      ACCRUAL-BASED BUDGETING MAY
      IMPROVE THE INFORMATION ON
      THE FISCAL IMPACT OF
      INSURANCE PROGRAMS
-------------------------------------------------------- Chapter 4:3.5

In addition to improving the information and incentives for resource
allocation, accrual-based budgeting would better reflect the fiscal
impact of federal insurance programs.  Although accrual-based
reporting would lessen the extent to which the budget reflects the
government's borrowing needs, it would generally reflect the effects
of an insurance program on the economy closer to the time when they
occur by incorporating a prospective estimate of the program's
accruing cost.  Discerning the economic impact of insurance programs
can be difficult, but private economic behavior generally is affected
when the government commits to providing insurance coverage and thus
lowers the risk to the insureds.  Therefore, accrual-based budgeting,
which, by recognizing the government's costs at the time the
insurance is extended, would better reflect the timing and magnitude
of the economic impact of these programs than the current cash-based
reporting of outlays in the budget.  Further, approaches to
accrual-based budgeting that recognize accrued cost in net outlays
would remove the uneven cash flow patterns of insurance programs from
the budget deficit.  By removing temporary working capital needs of
deposit insurance programs and large sporadic payments for disaster
insurance claims, accrued cost measures would provide a truer measure
of the government's underlying fiscal condition. 


   BENEFITS OF ACCRUAL-BASED
   BUDGETING FOR INDIVIDUAL
   FEDERAL INSURANCE PROGRAMS WILL
   DEPEND ON SEVERAL FACTORS
---------------------------------------------------------- Chapter 4:4

Although the use of accrual-based budgeting for federal insurance
programs has the potential to overcome a number of the shortcomings
of cash-based budgeting for these programs, a number of factors
influence the extent to which the information and incentives for a
particular insurance program would be changed.  These factors include
individual program characteristics, a program's BEA spending
classification, the extent to which costs are already recognized in
cash-based estimates, and the approach used to incorporate accrual
measures in the budget.  Further, the effective implementation of an
accrual-based budgeting approach depends on the ability to generate
reliable risk-assumed estimates. 

The characteristics of individual insurance programs will influence
the potential benefits achieved under accrual-based budgeting.  As
noted in chapter 3, the larger the government's commitment relative
to total federal spending, the greater the potential for budget and
fiscal policy distortions and the greater the need to capture the
government's cost at the time the commitment is made.  Thus, the
larger size of the deposit and pension insurance programs make the
benefits of accrual-based budgeting more pronounced than for other
smaller programs. 

In general, the effects of shifting to an accrual-based approach
would be beneficial for long-duration insurance programs with large
subsidies.  In these cases, the shift to accrual-based budgeting may
affect the magnitude of the reported program cost in the budget, or
whether the program is reported as having a cost rather than cash
income.  However, it does not appear that the benefits of
accrual-based budgeting would be as great for programs that offer
short-duration insurance coverage and experience relatively frequent
claims, such as crop or flood insurance.  For these programs, the
benefits of accrual-based budgeting primarily would be in recognizing
the cost of less frequent catastrophic losses and eliminating the
effect of programs' uneven cash flows on the budget deficit. 

As discussed later in the report, whether the program is classified
as mandatory or discretionary under BEA will also influence the
degree to which increased cost recognition is likely to influence
budget incentives.\21

For mandatory programs, accrual-based budgeting's effect on decisions
would be most apparent when legislated program changes--such as an
increase in benefits--are considered.  For discretionary programs,
accrual-based budgeting may have a more significant influence on
budget incentives, as their full cost becomes apparent and must be
provided for each year. 

The extent to which costs are currently recognized in budget
authority and obligations also influences the degree to which budget
information will change due to a shift to accrual-based budgeting. 
For example, according to OMB, the crop insurance program currently
estimates annual funding needs based on the normal loss year at the
time decisions are made.  In addition, OPIC currently receives budget
authority for and obligates loss provisions in the year the
provisions are recognized.  In both cases, program officials and
analysts believe that the current budget treatment adequately
reflects the program's expected costs at the time budget decisions
are made. 

If accrual-based budgeting were to be undertaken, the approach used
to incorporate accruals into the budget will also have an impact on
the extent to which budget information and incentives are changed by
a shift from cash-based budgeting to accrual-based budgeting.  As
will be discussed in chapter 6, different approaches to accrual-based
budgeting incorporate these costs into the primary budget
data--budget authority, net outlays, and the deficit--to varying
degrees.  Finally, the feasibility of accrual-based budgeting will
depend on whether reasonable unbiased estimates of the risk assumed
by the government for the various programs are available or can be
developed.  Estimation challenges and other implementation issues
that will have to be addressed in order to achieve the potential
benefits of accrual-based budgeting will be discussed in the chapters
that follow. 


--------------------
\21 Under BEA, budgetary resources are classified as either
discretionary or mandatory.  Discretionary refers to program spending
that is controllable through annual appropriation acts.  Mandatory
refers to program spending that is relatively uncontrollable without
changing existing substantive law. 


ESTIMATION LIMITATIONS AT CENTER
OF ACCRUAL BUDGETING DEBATE
============================================================ Chapter 5

A crucial component in the effective implementation of accrual-based
budgeting for federal insurance programs is the ability to generate
reasonable, unbiased estimates of the risk assumed by the federal
government.  Although in most cases the risk-assumed concept is
relatively straightforward, generating estimates of these costs is
considerably more complex.  The development and acceptance of
methodologies to estimate the risk assumed by the government varies
significantly across the federal insurance programs we reviewed.  The
following sections discuss some limitations of existing risk
assessment approaches that might be used to generate risk-assumed
cost estimates under an accrual-based budgeting approach. 


   CALCULATION OF RISK ASSUMED BY
   THE GOVERNMENT IS COMPLEX
---------------------------------------------------------- Chapter 5:1

The risk assumed by the government is most easily thought of as the
difference between the actual premiums paid by the insured and the
premiums necessary to fully cover losses inherent in the coverage
provided.  This difference between the full risk premium and the
actual premium charged--the "missing premium"-- represents the
government's subsidy cost for the insurance program.  In general,
decision-making is best informed if this subsidy cost is known at the
time a commitment is made.  This would suggest that to the extent
practicable, the budget ought to reflect this subsidy cost.  Under an
accrual-based budgeting approach, it would be recognized\1 at the
time the government extends insurance coverage.  The ability to
assess the risk covered by the various insurance programs is central
to being able to determine the subsidy cost to the government.  This
task is made difficult by the nature of the risks insured by the
government and the methodological and data limitations discussed
below. 

For insurance, the accuracy of estimated future claims is determined
by the extent to which the probability of all potential outcomes can
be determined.  Unfortunately, these probabilities are not known with
certainty for most activities more complex than the toss of a fair
coin.  However, for activities in which data on actual outcomes
exist, the underlying probabilities can be estimated based on the law
of large numbers.\2

When these conditions are understood and the probabilities of future
events can be inferred, estimates are said to be made under the
condition of risk.  In other words, since the possibility of each
outcome can be estimated, the risk undertaken by the insurer can be
measured.  However, when underlying conditions are not fully
understood estimates are made under uncertainty.  For most federal
insurance programs, this latter case holds due to the nature of the
risks insured, program modifications, and other changes in conditions
that affect potential losses.  Thus, estimating the government's
subsidy cost, including the time period considered, may vary by
program out of necessity. 

Complete data on the occurrence of insured events over a sufficiently
long period and under similar conditions are generally not available
for many federal insurance programs.  Frequent program modifications
as well as fundamental changes in the activities insured reduce the
predictive value of historical data and further complicate risk
estimation.  For example, the crop insurance program has been
modified a number of times by the Congress in the last 15 years
affecting key conditions, such as participation rates.  Similarly,
advances in technology and new competitive pressures have
significantly transformed the banking and thrift industries. 

These factors, which limit the ability to predict losses and the
potential for catastrophic losses, have been cited as preventing the
development of commercial insurance markets for the risks covered by
federal insurance programs.  As a result, private sector comparisons
are generally unavailable to aid in the risk estimation process for
these programs.  For example, although several private sector
companies offer aviation war-risk insurance, the coverage is
generally limited to random acts of terrorism and often excludes
areas of military conflict.  Federal war-risk insurance is only made
available when commercial insurance cannot be obtained or is
available only on unreasonable terms and conditions and it is in the
national interest to provide air service to a particularly risky
area.  The risk inherent in these two situations is not comparable. 

Some have suggested the use of simple historical loss averages as an
alternative to the complex estimation methodologies.  However, the
same conditions discussed above that make risk estimation difficult
may reduce the usefulness of this alternative.  Losses incurred under
most of the federal insurance programs over a 10- or 20-year period
may not adequately capture the risk inherent in the insurance because
such relatively short experience periods do not encompass the full
range of possible outcomes, including infrequent catastrophic events. 
Historical averages also may not be reflective of future losses if
there have been program changes or changes in underlying conditions
that may affect outcomes. 


--------------------
\1 The question of whether the subsidy is included as supplemental
information, budget authority only, or in both budget authority and
outlays is discussed in the next chapter. 

\2 The law of large numbers holds that as the number of independent
observations increases, the proportion of times a certain outcome
occurs tends to approach the underlying probability assuming that the
same conditions exist.  If statistical evidence is available, changes
in underlying conditions can be taken into account. 


   ABILITY TO ESTIMATE THE RISK
   ASSUMED BY THE GOVERNMENT
   VARIES ACROSS PROGRAMS
---------------------------------------------------------- Chapter 5:2

The extent to which risk assessment methodologies are currently
developed and accepted varies significantly across federal insurance
programs.  Some federal insurance programs, such as the life
insurance programs, cover risks that are commonly insured by the
private sector and are based on widely accepted actuarial science. 
However, as discussed in earlier chapters, most federal insurance
programs cover catastrophic or case-specific risks that the private
sector has been unwilling or unable to cover.  Risk assessment for
these programs is considerably more challenging.  For some insurance
programs, such as deposit insurance, several quantitative risk
assessment techniques have been developed but there is no strong
consensus supporting any particular technique.  For other federal
insurance programs, such as the war-risk insurance programs and
OPIC's political risk insurance, risk assessment currently relies
heavily on expert judgment rather than highly quantitative or
standardized risk assessment methods. 

Given these estimation challenges and the shortcomings of cash-based
budgeting, consideration of the adequacy of risk-assumed estimates
for budget purposes is likely to be most beneficial when the focus of
the discussion is on whether these estimates would provide
policymakers with more timely information and signals about the
underlying insurance programs.  For these purposes, what is important
is that the estimates are based on the best information available at
the time the insurance commitment is extended.  In this sense, it may
be most important that the budget information and incentives provided
to policymakers be "more approximately right rather than precisely
wrong."\3

The remainder of this chapter discusses risk assessment for the
various types of federal insurance programs we reviewed: 

  -- life insurance;

  -- disaster insurance (flood and crop insurance);

  -- deposit insurance;

  -- pension insurance; and

  -- other insurance (war-risk, political risk, and vaccine injury
     insurance). 

The sections that follow are ordered approximately according to the
current level of the development and acceptance of methodologies that
could be used to estimate the risk assumed by the government.  The
first programs discussed--life insurance--have a methodology that is
well established in actuarial science.  Next, we discuss disaster
insurance programs for which methodologies have been developed and
used to set risk-related premiums.  These methodologies may provide a
useful foundation for estimating the risk-assumed costs for these
programs.  For the large programs--deposit and pension
insurance--competing methodologies exist or are under development
that potentially could be used to estimate risk-assumed costs;
however, little consensus exists on any one model.  The remaining
programs--overseas private investment, vaccine injury, and war-risk
insurance--present significant estimation challenges and rely heavily
on expert judgment. 


--------------------
\3 The Congressional Budget Office used this phrase to characterize
the difference between accrual-based cost estimates and cash-based
reporting in advocating accrual-based budgeting for credit programs. 
See Credit Reform:  Comparable Budget Costs for Cash and Credit,
Congressional Budget Office, December 1989. 


   RISK ASSESSMENT FOR LIFE
   INSURANCE HAS ITS FOUNDATION IN
   ACTUARIAL SCIENCE
---------------------------------------------------------- Chapter 5:3

The methodology for measuring the risk assumed by the government
under life insurance programs for government employees and
service-disabled veterans is well established in actuarial science. 
The certainty of death and the compilation of extensive data on
mortality have made it possible to estimate future life insurance
claims with a high level of accuracy.  The Department of Veterans
Affairs (VA) and the Office of Personnel Management (OPM) currently
use actuarial approaches that are the standard practice of the life
insurance industry.  Although modifications are made to reflect the
unique characteristics of the insured groups, the basic assumptions
used are comparable to those used by commercial life insurance
companies. 

By applying the laws of probability to mortality statistics,
actuarial science provides a methodology to estimate future rates of
death.  A basic principle of actuarial science states that by
studying the rate of death within any large group of people and
gathering information on all factors that may affect that rate, it is
valid to anticipate that any future group of persons with
approximately the same factors will experience the same rate of
death.  Mortality tables are constructed to reflect probabilities of
death at each age.  The accuracy with which the estimated future
claims approximate actual experience depends upon two key factors: 
(1) the accuracy and appropriateness of the underlying mortality
statistics and (2) the number of observations the estimate is based
on and the number of individuals insured. 

Most mortality tables in use today are based upon the experience of
commercially insured individuals.  Because the mortality experience
of federal employees appears to be different from the experience used
to construct these tables, OPM constructs its own mortality tables
based on program experience to more accurately capture the insurance
risk.  VA also conducts periodic studies of mortality and disability
to ensure that its assumptions are sufficiently conservative.  The
information on mortality, together with data on policy benefits and
interest rate assumptions, makes it possible to calculate the present
value of future insurance claims.  The extent to which this amount
differs from premium and investment income would constitute the risk
assumed by the government or the government subsidy.  However, as is
the case with most long-term forecasts, estimates of a life insurance
program's income are sensitive to interest rates.  For example,
interest earnings on funds collected from policyholders are a
significant source of revenue in the Federal Employees' Group Life
Insurance program.  OPM officials cited the difficulty in forecasting
fluctuations in interest rates over the long term as a weak point in
the estimation process. 


   DISASTER INSURANCE PROGRAMS
   HAVE ESTABLISHED RATE-SETTING
   METHODOLOGIES
---------------------------------------------------------- Chapter 5:4

The two disaster insurance programs we reviewed--the National Flood
Insurance program and the Federal Crop Insurance program--currently
have established methodologies for setting risk-related premium
rates.  These methodologies and the corresponding agency risk
assessment experience should provide a useful foundation for
estimating the cost of the risk assumed by these programs if an
accrual-based budgeting approach is adopted.  However, some
modifications and refinements to the methodologies and other
implementation challenges should be expected.  Further, as is the
case with all modeling efforts, professional judgment and underlying
assumptions are necessary components of these methodologies. 


      FLOOD INSURANCE LOSSES ARE
      ERRATIC BUT MEASURABLE OVER
      THE LONG TERM WITH A FAIR
      DEGREE OF ACCURACY
-------------------------------------------------------- Chapter 5:4.1

The Federal Insurance Administration (FIA) has an established
rate-setting model, which, according to FIA, could be used to assess
the risk assumed for policies issued by the National Flood Insurance
Program (NFIP).\4 FIA officials told us that this model is based on
generally accepted actuarial principles and has been used by the
agency for years to set premium rates for unsubsidized insurance
policies.\5 They told us that the model and its output, however, do
not undergo regular external reviews.  In addition to the
rate-setting methodology, the FIA has worked on developing
catastrophic loss estimates that may prove useful in assessing the
appropriateness of reserve levels under an accrual-based budgeting
approach.  As an alternative to the detailed rate-setting model, some
budget experts suggested that historical averages--which do not
include catastrophic losses unless they have occurred during the
chosen experience period--may provide a sufficient basis for
measuring the program's accrual-based costs. 

Flood hazards have several characteristics that are important in
considering risk assessment and budgeting for the NFIP.  Although the
timing and magnitude of floods is considered unpredictable by more
than a few days or hours, the probability that they will occur is
measurable with a fair degree of accuracy.  Flood losses are very
predictable in that they occur in well-defined areas and are
inevitable in these areas over the long run.  However, as noted in
chapter 3, the erratic nature of floods can have serious implications
for risk assessment and budgeting.  A Department of Housing and Urban
Development (HUD) study points out that while most property and
casualty insurance is based on realized losses over a period of time,
this approach is not applicable to flood insurance because of the
highly skewed nature of flooding losses.\6 Rather than following the
normal bell-shaped statistical curve, there are many small to
moderate floods, some larger floods, and a few extremely large ones. 
As a result, an average of even a considerable number of years may
differ significantly from the true long-term average. 

The NFIP illustrates this point.  The NFIP is not actuarially sound
even though it has achieved a goal of collecting premium income
sufficient to at least cover expenses and expected losses for an
average historical loss year.  This is because the historical
experience period, beginning in 1978, does not include any loss years
that can be considered to be of a catastrophic level for the
program.\7 As a result, the average historical loss year involves
fewer claim losses than the expected per annum claim losses in future
years.  Thus, the premium income currently collected by the program
may not be sufficient to build reserves for potential catastrophic
losses. 

Despite this limitation, some budget experts indicated that they
thought it acceptable to use historical averages as the basis for
measuring the program's accrual cost in the budget while providing
some additional funding mechanism, such as the program's borrowing
authority, to cover catastrophic losses.  Funding this amount would
allow for the accumulation of reserves during years where losses are
less than the historical average.\8 According to the FIA, this level
of funding in conjunction with the program's borrowing authority of
$1 billion should be sufficient to cover costs approximately 85
percent to 90 percent of the time.  Further, the average historical
loss year is not a static measure and could be expected to move
toward the long-term average as the experience period increases over
time.  Nevertheless, if the objective of adopting accrual-based
budgeting is to recognize the currently unrecognized government
subsidy and/or to accumulate reserves to cover future losses,
including catastrophic losses, then the program's long-term expected
cost is the most appropriate measure to use as the basis for
measuring the government's cost in the budget. 

FIA officials told us that they were reasonably confident that the
actuarial rate-setting method currently used to establish premium
rates for unsubsidized polices could be used to generate reasonable
estimates of the expected long-term risk for all policies.  The
difference between the program's expected long-term risk and the
actual premium rates would then provide an estimate of the risk
assumed by the government.  The FIA estimated this difference or
"missing premium" at approximately $520 million per year.\9

FIA's method for establishing rates for unsubsidized policies follows
a hydrological method based on studies performed by the U.S.  Army
Corps of Engineers and private engineering companies.  These rates
are based on available hydrological data, flood insurance claims and
simulations, as well as engineering and actuarial judgment.\10

According to FIA officials, the key components of the method are (1)
probability estimates of the frequency with which floods of different
severity will occur and (2) estimates of associated structural
property damage incurred due to different types of floods.  Program
expense items, such as administrative costs, are also accounted for
in the actuarial rates.  These rates are based on actual risk
exposures and generally vary according to risk-related features, such
as the flood zone, the elevation of the structure, and the amount of
insurance purchased. 

As is often the case in modeling, professional judgments and
assumptions are necessary to overcome data limitations.  For example,
the flood histories used to develop the original estimates of the
probability of floods of different severity were generally not very
long.  Consequently, modifications had to be made to prevent
statistical bias.  In our discussions, FIA officials described the
measurement of flood frequency as "good as the state of the art" but
noted that not every area has been studied in depth due to resource
constraints.  Agency officials said that in these cases, the
frequency estimates are based on various histories and statistical
analysis which ad hoc studies have shown to yield reliable results. 
In addition, the original estimates of the structural damage caused
by floods of various depths were based on engineering studies and
available flood claims.  According to agency officials, these
estimates are regularly updated with claims data, and credibility
analysis\11 is used to check validity.  Appendix IV provides a more
detailed description of the model and its key data elements. 

According to FIA, additional assumptions and judgments would be
necessary if the model were to be used for the entire program because
there is currently a lack of information on pre-flood insurance rate
map (FIRM) properties.\12 FIA said that a current study on the impact
of charging actuarial rates for pre-FIRM properties will be gathering
additional information on these properties.  Agency officials also
noted that a trade-off exists between the benefits of more precise
estimates and the cost of making these improvements.  While factors
such as the frequency of remapping and rezoning may affect the
quality of the model's estimates, the costs of these studies should
also be considered.  For example, agency officials stated that even
if sampling were done to get more precise estimates of the
evaluations for pre-FIRM properties, the cost of this type of
refinement could outweigh the potential benefits of improved
estimates. 

FIA has also done some work on developing catastrophic loss estimates
that may be useful in assessing the program's appropriate reserve
levels under an accrual-based budgeting approach.\13 According to
FIA, the method used to estimate catastrophic funding levels uses
additional statistical analysis and simulations in conjunction with
the hydrologic method and is more complex than the actuarial
rate-setting method alone.  An FIA official estimated that
catastrophic reserve levels of $4.5 billion to $5 billion should be
sufficient 99.9 percent of the time.  In recent years, FIA has been
unable to establish reserves and has had to borrow funds from and
repay funds to the Treasury to cover excess losses.  These estimates
of catastrophic reserve requirements were described as orders of
magnitude, suitable for program planning purposes, rather than
precise estimates.  The FIA official explained that the closer the
actual reserve funding level gets to the estimated amount, the more
important data limitations and underlying assumptions become. 

In summary, FIA has an established method for setting risk-related
premiums for its unsubsidized policies.  According to FIA, this
methodology could be extended to generate risk-assumed estimates for
the entire program.  In addition, FIA's work on catastrophic reserve
requirements may prove useful in assessing the appropriateness of
reserve levels under an accrual-based budgeting approach.  However,
while this work should provide a useful foundation for developing
risk-assumed estimates, FIA indicated that some modifications and
refinements would be desirable before these estimates are used for
accrual-based budgeting purposes and fully funding a reserve level
target.  One FIA official noted that this may involve considerable
effort. 


--------------------
\4 For additional information on the Flood Insurance Program and its
rate-setting methods, see Flood Insurance:  Financial Resources May
Not Be Sufficient to Meet Future Expected Losses (GAO/RCED-94-80,
March 21, 1994) and Flood Insurance:  Information on Various Aspects
of the National Flood Insurance Program (GAO/T-RCED-93-70, September
14, 1993). 

\5 NFIP flood insurance premiums are either based on actuarial
principles or are subsidized depending on when the insured structure
was built.  Subsidized rates are available for structures built
before rate maps were prepared for the areas in which they are
located.  Premiums for properties constructed after the rate maps
were prepared must be set in relation to risk.  See appendix IV for a
more detailed description of the program. 

\6 Insurance and Other Programs for Financial Assistance to Flood
Victims:  A Report From the Secretary of the Department of Housing
and Urban Development to the President, as required by the Southeast
Hurricane Disaster Relief Act of 1965, Senate Committee on Banking
and Currency, 89th Congress, 2nd Session, September 1966. 

\7 According to FIA, the probable maximum loss resulting in $4.5
billion to $5 billion in claim losses has a 1 in 1,000 chance of
occurring.  For comparison purposes, Hurricane Hugo resulted in
claims of $0.4 billion. 

\8 FIA officials told us that in any particular year, there is about
a 40-percent to 45-percent chance that flooding losses will be less
than the historical average and about a 50-percent to 55-percent
chance that flooding losses will exceed the historical average. 

\9 According to FIA, $520 million represents the amount of annual
general fund appropriations necessary to build reserves to cover
catastrophic losses.  FIA told us that if the shortfall was covered
by an increase in premiums, then the annual amount needed would be
larger due to corresponding increases in commission payments and
other expenses. 

\10 We have not independently reviewed the studies on which FIA's
data for actuarial rate-setting are based. 

\11 Credibility analysis is a statistical technique used to determine
the degree to which the accuracy of the experience can be relied on. 

\12 Pre-FIRM properties are structures that were constructed before
the initial mapping studies for the FIRMs were completed.  The rates
for these properties are currently subsidized. 

\13 As discussed in chapter 7, reserve levels are an important
consideration because, in order to cover future losses, reserves need
to be based on the long-term expected risk of an insurance program
rather than policies issued in a given year. 


      CROP INSURANCE PROGRAM HAS
      AN ESTABLISHED RATE-SETTING
      METHODOLOGY
-------------------------------------------------------- Chapter 5:4.2

The Federal Crop Insurance Corporation (FCIC) has an established
premium rate-setting methodology and considerable experience
assessing crop risk.  According to FCIC, the basic rate-setting
methodology used by the agency--the loss cost ratio method\14 --is
commonly used in the insurance industry.  Within agriculture, it is
used by the rating bureaus that support the crop-hail insurance
industry.  Although this methodology could be used as the basis for
measuring the program's cost in an accrual-based budget, several
implementation issues stemming largely from the diversity of crop
risks and timing differences between the budget cycle, the
rate-setting process, and the exposure period would have to be
overcome.  In addition, some limitations in the methodology's
underlying data and assumptions have been identified.  According to
the FCIC, internal reviews of the model and its key assumptions are
ongoing and an external review of the model by an actuarial firm is
underway.  This external review should provide additional insights to
help evaluate the model's ability to generate risk-assumed estimates
for accrual-based budgeting and identify areas for continued
improvement.  The last comprehensive review of the methodology was
completed in 1983 by the same firm. 

The nature of crop losses makes risk assessment challenging.  In
prior reports, we have identified inherent problems in the ability to
pool and assess the risk of crop losses.\15 Crop losses are not
normally independent--some perils are likely to strike a large number
of insured farmers in the same crop year.  Further, it is difficult
to align premium rates directly with risk because the risk associated
with growing a particular crop varies by county, farm, and farmer. 
For example, the risk associated with a particular farmer is
influenced by a variety of factors, including farm management
practices, soil type, and the productivity of individual tracts of
land.  Monitoring these individual risks may be neither feasible nor
cost-effective.  In addition, crop risks are volatile.  The
performance of any specific crop or any area of the country is
subject to wide variations depending on the state of nature.  In
general, the performance of any particular crop in a county is
characterized by relatively infrequent catastrophes of moderate to
extreme severity and a number of annual spot losses resulting from
noncatastrophic events, such as hail. 

As a result, the program's rate-setting methodology is complex.  To
align crop insurance premium rates with the associated risk, FCIC
establishes rates that vary by crop, location (county), farm, and
farmer.  Because of all the combinations involved, literally hundreds
of thousands of premium rates are in place.  These rates are adjusted
annually using a multistep process involving considerable computer
analysis and professional judgment.  FCIC begins the process each
year by looking at crop insurance experience over the past 20 years
(if available) for each county and state.  On the basis of county and
state historical experience, FCIC sets basic rates for each crop in
each county at the 65-percent coverage level for average production. 
These basic rates are adjusted for specific risk classifications
including each farming type (such as whether the insured acreage is
irrigated or dry land) and for each crop type (such as winter wheat
or spring wheat).  Using these basic rates, FCIC makes several
adjustments to establish rates for other coverage levels and for
farmers whose production levels differ from the county's average. 
FCIC's rate-setting methodology is described in more detail in
appendix IV. 

FCIC agreed that this methodology could be used to estimate the
program's expected risk but noted that data for detailed rate-setting
are not available at the time the budget year submission is prepared. 
Because the risk of crop damage is closely aligned with specific
characteristics of the crop, county, and individual farming
practices, detailed information on the composition of coverage
extended is necessary to provide projections of the full risk assumed
for policies issued in any given year.  However, detailed information
on the composition and volume of policies and updated premium rates
are not available at the time the budget year submissions are made. 
FCIC said that reasonable projections of insurance coverage can be
made on a national scale at a higher level of aggregation, such as by
crop type, but considerable uncertainty surrounds more detailed
projections of specific policy coverage, such as crop-county
combinations.  Since estimates based on the actual composition of
policies provide a more appropriate measure of the risk assumed,
adjustments based on more detailed information may have to be made
when the information becomes available. 

According to FCIC's senior actuary, FCIC currently bases its budget
estimate on current year sales data adjusted for several factors,
such as projections of commodity prices, planted acres, and
anticipated changes in premium rates.  He agreed that this more
aggregated approach would provide a sufficient basis for an
accrual-based budget year estimate, even though the data for detailed
rate-setting, including crop/county level data, are not available
when the budget submission is prepared.  He said that the more
detailed rate-setting methodology could subsequently be used to
re-estimate the government's risk for actual policies issued during
the year by "repricing" these polices using the full risk premium
rates when sufficient information becomes available.  The difference
between the full risk premium for policies issued and actual premium
collections would constitute the government's subsidy costs for
policies issued in a given year.  He agreed that the full risk
premiums, which take into account specific risk and generally include
amounts for catastrophic losses, are probably the appropriate basis
for establishing reserves under an accrual-based budgeting approach. 
Even with this, there are decisions to be made about the confidence
placed in these estimates, which--like other risk assessment
models--depends on the acceptance of the methodology's underlying
assumptions and data limitations. 

A key assumption of the FCIC's rate-setting methodology is that the
20-year base period is sufficient to assess the program's future
costs.  The FCIC's senior actuary acknowledged that the appropriate
period to use is debatable and that any finite number of years is
inadequate to observe all possible states of nature or to assess the
probability of each state of nature.  He explained that the 20-year
rolling average has been used primarily due to concerns about the
availability, quality, and applicability of data from the early years
of the program. 

Our previous work found that premium rates depend heavily on the
number of years included in the experience period and the weight
assigned to each year.  For example, in 1983 USDA's consultant
suggested changing from the current methodology of giving equal
weight to each of the 20 years' experience to giving greater weight
to more recent years' experience.  We found that the consultant's
approach had a significant impact on the premium rates for three
crops of the six major crops we reviewed.\16

The USDA consultant is evaluating whether the trend in losses in
recent years requires a change in the methodology. 

Another key assumption is that the sample of previous buyers of crop
insurance is adequately representative of future buyers.  Recent
changes in the program have resulted in changes in the
characteristics of the buyer population.  As such, there is
considerable uncertainty surrounding the future composition of the
insurance portfolio.  The model is also based on a number of
secondary assumptions.  These include the choice of parameters used
to (1) allocate basic rates to the various components of risk, such
as crop type and planting practice, and (2) adjust the basic rates
for different coverage and production levels. 

Our previous work raised concerns about the adjustments made to the
basic rate level to arrive at rates for coverage and production
levels that differ from those of the basic rate.  These adjustments
are important because the majority of all crop insurance is purchased
at rates for coverage and production levels that differ from those
covered under the basic rates.  However, our previous work found that
these adjustments did not result in rates that are aligned with risk. 
For example, to set the rates for the 75-percent and 50-percent
coverage levels, FCIC applies preestablished mathematical factors to
the basic rate.\17 According to our analysis of six major crops, the
rates for this insurance were too high at the 75-percent coverage
level and too low at the 50-percent coverage level in relationship to
the basic rates.  FCIC, using a mathematical model that sets rates
according to preestablished relationships between production levels,
also adjusts the basic rates for production for farmers whose
historical production levels are above or below the county's average. 
However, as with the varying rates for coverage levels, we found that
these adjustments did not result in rates that accurately reflect the
risk involved at each production level.\18 In the past, agency
officials cited a lack of time and resources as a barrier to revising
the formulas applied to the basic rates to calculate these other
rates.  Currently, however, FCIC and its consultant are reviewing
these factors and FCIC anticipates making adjustments in the future. 
Further, the FCIC noted that these differences may be offsetting in
the aggregate and thus may not be as important for budget purposes as
for setting individual farmers' premium rates. 

Additional modeling techniques to aid in assessing the risk of crop
losses may become available in the future.  For example, FCIC said it
is doing some work with multistage econometric models and OMB
suggested that options pricing\19 could potentially be used to
estimate the risk assumed by the government.  However, these methods
are only in the conceptual or early stages of development.  The
FCIC's senior actuary told us that the alternative models the FCIC is
working on may provide useful supplemental information but are not
reliable or useful enough for budget purposes.  In addition, the use
of these methods would require a significant upgrade in staff skills. 

In summary, the FCIC has an established rate-setting method that
could serve as the basis for estimating the risk assumed by the
government.  However, risk assessment for crop insurance is
complicated by the variation of risk associated with different
combinations of crops, counties, and farming conditions and
practices.  Detailed information to estimate the risk assumed based
on specific characteristics of the insureds is not known at the time
of the budget year submission.  Therefore, a higher level of
aggregation--perhaps similar to what is used for current budget
estimates--could be used for the budget year estimates.  Reestimates
for the risk assumed based on actual polices issued in a particular
year could then be achieved by repricing the polices based on the
full risk premium when necessary information becomes available.  Many
of the methodology's assumptions and underlying data limitations are
currently under review.  This review should provide additional
insights into the reasonableness of the methodology and its use for
accrual-based budgeting. 


--------------------
\14 The loss cost ratio is calculated by dividing total claim
payments by the total insurance in force. 

\15 See Crop Insurance:  Federal Program Faces Insurability and
Design Problems (GAO/RCED-93-98, May 24, 1993) and Crop Insurance: 
Additional Actions Could Further Improve Program's Financial
Condition (GAO/RCED-95-269, September 28, 1995). 

\16 Crop Insurance:  Additional Actions Could Further Improve
Program's Financial Condition (GAO/RCED-95-269, September 28, 1995). 

\17 FCIC multiplies the basic rate at the 65-percent level by 154
percent to arrive at the rate for 75-percent coverage and by 72
percent to arrive at the rate for 50-percent coverage. 

\18 According to the FCIC's senior actuary, recent analysis by FCIC's
actuarial consultant show there is considerable variation in these
relationships by crop and area of the county.  There also is debate
about appropriate methodology--experience-based, as done by the GAO,
or yield-based, as done by the consultant. 

\19 For a discussion of options pricing, see figure 5.1. 


   LACK OF CONSENSUS ON RISK
   ASSESSMENT METHODOLOGY FOR
   DEPOSIT INSURANCE
---------------------------------------------------------- Chapter 5:5

The historic number of thrift and bank failures in the late 1980s and
early 1990s and the costs associated with resolving these
institutions motivated the development of methodologies to estimate
future failures and their expected costs to the government's deposit
insurance funds.  In a prior report, we reviewed methodologies used
by various federal agencies and private forecasters.\20 We found that
different estimation approaches produced widely disparate results due
in part to heavy reliance on professional judgment in specifying
critical assumptions, such as estimates of market value, and the
historical period used to project expected future losses.  An
analysis by staff of the Office of the Comptroller of the Currency
(OCC) concluded that different estimation methodologies have
strengths and weaknesses but no one approach appears to be
superior.\21 Appendix II contains a description of six loss
estimation methodologies. 

The health of the bank, thrift, and credit union industries is
subject to many variables that are extremely difficult to predict. 
These include variables related to local and national economic
conditions, behavior of regulators and management, and structural
changes in the industries.  Thus, attempting to predict the future
prospects of financial institutions and estimate future losses to the
insurance funds is an intrinsically uncertain proposition.  In
preparing loss estimates, there is no empirical formula for
forecasters to follow that would enable them to know with certainty
what approach or assumptions can most accurately reflect both present
and future conditions and events that can play a significant role in
the solvency of a financial institution.  Only by making many
assumptions can the available methodologies generate estimates of the
impact of changes in the economy, industries, or regulatory behavior
on the government's cost of providing deposit insurance.  In
addition, small changes in these key assumptions can produce large
changes in cost estimates. 

Determining the value of an institution's assets is one of the most
challenging steps in estimating the government's cost of deposit
insurance losses from failed institutions.  Economic insolvency of a
financial institution occurs and costs accrue to the insurance funds
when the value of the institution's liabilities exceeds the market
value of its assets.  While the value of an institution's
liabilities--primarily deposits--is generally known, the value of its
assets--primarily loans--is much more uncertain.  Most loss
estimation methodologies rely on unaudited financial data that
financial institutions are required to report--in call reports--to
regulatory agencies.  Experience has shown that these data do not
always provide an accurate picture of the value of an institution's
assets.  For example, in 1991 we reported that asset valuations
prepared by FDIC for 39 failed banks revealed $7.3 billion in
additional deterioration in asset values (losses) compared to the
last quarterly call reports filed by the institutions.\22 As a
result, most estimation approaches adjust call report data in an
attempt to approximate the market value of an institution's assets. 

Estimating the market value of an institution's assets allows for
earlier recognition of the government's deposit insurance losses than
does reliance on historical book value measures reported in call
reports.  However, the use of market-value accounting is still
controversial.  Market values are not readily available for all
categories of bank and thrift assets and liabilities.  Analysts are
divided over whether market-value accounting is precise enough for
financial statements or whether it provides better estimates than
book-value accounting.  In addition, some models, such as OMB's,
include a closure rule as a policy variable defined in terms of the
asset-to-liability ratio of an institution.  This variable can (1) be
set based on observed behavior of regulators in a given period or (2)
reflect prompt closure as mandated by the Federal Deposit Insurance
Corporation Improvement Act of 1991.  Delay in closing an institution
after it has become insolvent has been shown to increase resolution
costs. 

Because of the nature of deposit insurance and the significant
challenges associated with estimating the program's full long-term
risk described previously, some departure from the pure risk-assumed
cost concept may be appropriate in calculating risk-assumed
estimates.  For example, experts we consulted with held differing
views on the degree to which OMB's model accounts for the full range
of possible future outcomes, such as the catastrophic losses
associated with the savings and loan crisis.  However, the model's
market-value-based accrual cost estimates would have provided
policymakers with earlier recognition of deposit insurance costs than
cash-based reporting. 

The following sections discuss some of the limitations of the various
types of models that are currently used by different forecasters to
project losses to the deposit insurance funds.  An assessment of the
applicability and accuracy of any model for estimating the
government's cost for deposit insurance can only be made in the
context of alternative models so that the benefits and limitations of
different approaches can be compared.  The first section highlights
some of the limitations of OMB's options pricing model, which the
Bush administration proposed using for accrual-based budget
reporting.  The second section discusses some of the weaknesses of
other loss estimation methodologies.  A description of each of the
methodologies is included in appendix II. 


--------------------
\20 Bank Insurance Fund:  Review of Loss Estimation Methodologies
(GAO/AIMD-94-48, December 9, 1993). 

\21 A Comparison of Different Approaches to Projecting Bank
Resolutions, Draft Working Paper, Thomas J.  Lutton, Robert DeYoung,
and David Becher, Office of the Comptroller of the Currency, November
1993. 

\22 Failed Banks:  Accounting and Auditing Reforms Urgently Needed
(GAO/AFMD-91-43, April 22, 1991). 


      OMB'S OPTIONS PRICING MODEL
-------------------------------------------------------- Chapter 5:5.1

Although any of the methodologies described in appendix II could be
adapted to estimate the government's annual costs of deposit
insurance, the focus has been on OMB's options pricing model because
it provides a direct computation of accruing deposit insurance costs. 
Under OMB's estimation approach, deposit insurance is treated as
giving the owners of a bank or thrift institution the option to
transfer its liabilities to the government if the value of its assets
falls below that of its liabilities.  A brief overview of options
pricing theory is provided in figure 5.1.  OMB's deposit insurance
model has two distinct components.  The first part attempts to
estimate the financial condition, or market value, of every
institution with liabilities over $100 million and then simulate
their financial condition for future years.  The second part uses
options pricing techniques to calculate the expected costs of deposit
insurance. 

   Figure 5.1:  Options Pricing
   Theory and Federal Insurance

   (See figure in printed
   edition.)

OMB's model is conceptually sound and OMB's efforts have made a
significant contribution in extending the use of options pricing
theory to estimate government insurance costs.  However, a number of
modifications to the model should be considered in order to improve
its ability to estimate the government's deposit insurance cost. 
These refinements are geared toward addressing some of the concerns
raised by experts about the model, including (1) the approach used to
value financial institutions' assets, (2) the treatment of interest
rate risk, and (3) the sensitivity of estimates to the specification
of model assumptions and parameter values. 

A key assumption of options pricing theory is that asset values are
observable, measurable, and vary randomly over time.  For assets for
which there are efficient, well-developed markets, such as stocks,
bonds, agricultural commodities, and foreign currencies, this
assumption is not problematic.  However, the value of a financial
institution's assets is not readily observable or measurable. 
Although stocks of the very largest banks are traded actively,
adequate market value data are not available for the many smaller and
non-publicly-traded institutions.  As such, the unavailability of
market value data on bank and thrift assets is a limitation of OMB's
options-based approach to estimating deposit insurance costs.  In
order to calculate the government's cost for insuring all
institutions, OMB uses call report data to estimate the market value
and volatility in the rate of return of bank and thrift assets. 

OMB's use of estimated asset values and its method for calculating
these estimates have been criticized.  Some financial economists we
spoke with questioned the practical application of options theory in
the absence of observable and measurable market data that are
generally available in more common uses of options theory.  The use
of an estimate of asset values is problematic because it may
introduce measurement errors if the input data are not unbiased and
efficient estimates of market value.  The lack of an explicitly
stated and observable exercise price of the option makes it difficult
to determine when the option would be exercised.  In the OMB model,
the exercise price--economic insolvency--is expressed as a ratio of
an institution's estimated assets and liabilities.  The valuation of
assets and liabilities is often difficult and depends on the
measurement basis used, thus identifying the timing of when a firm's
liabilities exceed its assets can be problematic. 

Other financial economists argue that the lack of observable and
measurable market value data for many financial institutions makes
the use of call report data to infer market values a reasonable
approach.  OMB bases its estimate of market value on an institution's
current cash flow\23 from call report data and a set of
econometrically determined variables.  This approach, however, has
been criticized because cash flows are very sensitive to business
cycles, which may result in overestimating or underestimating the
market value of an institution.  Furthermore, in estimating the cash
flows for individual banks, OMB divides all banks into four groups
and estimates parameters, such as cash flow volatility and loan
chargeoffs, for each group.  These group parameters are applied to
individual bank earnings in order to project future cash flows.  This
introduces correlation among banks in each of the groups when none in
fact may exist.  The limitations of OMB's asset valuation process
were evident from its initial estimates of the parameters, which
implied a negative net worth for all banks.  This occurred because
the estimation period included a banking recession.  To correct for
this, OMB adjusted its estimates of market values using stock market
data on the largest publicly traded bank holding companies. 

The OMB deposit insurance model has also been criticized because it
does not explicitly take into account interest rate risk.  The
profitability of banks, thrifts, and credit unions is heavily
dependent on both short- and long-term interest rates.  The OMB model
implicitly incorporates interest rate risk and other risks that
affect an institution's profitability through assumptions made about
asset value and volatility of asset earnings.  However, the financial
economists we consulted with suggested that because of the importance
of interest rates to the financial health of a depository
institution, explicit modeling of interest rates would be desirable. 
Some recent research in the options pricing area incorporates
interest rate risk in an options pricing framework to estimate the
government's deposit insurance liability.\24

Another concern raised by experts is the sensitivity of the deposit
insurance cost estimates generated by OMB's model to changes in key
assumptions and parameters.  Although the sensitivity of a model's
output to changes in parameter values is not necessarily a negative
attribute of a model, it heightens the need for unbiased assumptions
and parameter estimates.  For example, insurance cost estimates
generated by OMB's model are particularly sensitive to assumptions
about the future value of financial institutions' assets.  Two key
assumptions affecting estimates of future asset values are asset
volatility\25 and mean-reversion of earnings.\26 Analysis of OMB's
model using alternative values for these assumptions produced
significant changes in the estimated cost of deposit insurance. 
Using an Office of Thrift Supervision (OTS) estimate of
mean-reversion reduced the government's estimated 5-year accrued
costs by 49 percent compared to the estimated cost using OMB's
assumptions.  Using OTS estimated standard deviation of thrift assets
increased the estimated 5-year cost by 56 percent.  The differences
in assumptions made by various financial institution experts and the
resulting impact on the cost estimates demonstrates that additional
research on the appropriate assumptions and parameter values is
desirable. 

Cost estimates generated by OMB's deposit insurance model are also
highly sensitive to initial period financial data on depository
institutions.  OMB's model uses only the most recent four quarters of
data on an institution's earnings to estimate its market value.  The
model projects the existing financial condition of institutions into
the future and does not account for wide swings in the general
financial health of the industry.  As a result, input data from
relatively good economic times will tend to underestimate future
costs, while input data from an economic downturn will overestimate
future costs.  For example, using financial data from 1992, a
recessionary year, the OMB model estimated that in 1994 the
government would assume liabilities of $51 billion from failed banks
with the cost to the government being a percentage of this amount. 
Actual liabilities from failed banks in 1994 were approximately $1
billion.  Forecasting turning points in the economy is difficult for
all forecasters--not only OMB's options model--but is one of the
major hurdles to generating risk-assumed estimates for deposit
insurance. 

Financial economists at bank regulatory agencies were divided in
their views on the use of OMB's model for accrual-based budgeting. 
An official at one agency stated that he did not believe that the
model estimates are valid and reliable enough for budget and policy
decisions.  Some banking agency officials expressed concern that the
complexity of OMB's model made it difficult to replicate and analyze
the reliability of the cost estimates.  On the other hand, an
official at another banking agency stated that the concept of
accrual-based budgeting makes sense given that once the government
extends the insurance it has already accrued a cost.  He stated
further that some estimation uncertainty may be acceptable in the
reported accrual-based cost for deposit insurance in the budget as
long as there is a general understanding of the limitations involved. 
In contrast, he suggested that the same level of uncertainty would
not be appropriate for setting insurance premium rates for individual
banks. 


--------------------
\23 Cash generated and used in operations. 

\24 See, for example, Jin-Chuan Duan, Arthur F.  Moreau, and C.  W. 
Sealey, "Deposit Insurance and Bank Interest Rate Risk:  Pricing and
Regulatory Implications," Journal of Banking and Finance, vol.  19,
no.  6, September 1995, pp.  1091-1108. 

\25 Asset volatility is the fluctuation over time of the value of an
institution's assets--primarily loans.  Assumptions about future
asset volatility are generally based on observed historical
volatility. 

\26 Mean-reversion of earnings is the tendency of very high or low
earnings to revert toward the industry's long-term average rate of
return. 


      OTHER LOSS ESTIMATION
      METHODOLOGIES FOR DEPOSIT
      INSURANCE
-------------------------------------------------------- Chapter 5:5.2

Although the OMB options pricing model is the only methodology that
directly provides an estimate of the government's accrued cost of
deposit insurance, alternative models exist that provide forecasts of
future bank insolvencies.  These forecasts could be used to estimate
the government's accruing costs.  Appendix II provides a brief
summary of the actuarial, transition matrix, asset markdown,
proportional hazards, and pro forma projection models currently being
used by various researchers to estimate bank and thrift institution
losses. 

All of the estimation models are limited by their high degree of
reliance on professional judgment in setting assumptions and in their
use of unaudited call report data.  For example, actuarial models
generate loss estimates based solely on historical incidence of
resolution.  Accordingly, the model estimates are highly sensitive to
the choice of the historical period used to set these probabilities. 
For example, at the same time that the financial condition of the
bank and thrift industries was improving in recent years, expected
future loss estimates based on resolutions during the 1987 through
1992 period tended to be very high because they reflected the dismal
performance of the industry during this period. 

Actuarial approaches are also limited in that the effects of only two
or three variables can easily be incorporated into the model. 
Estimates are thus highly sensitive to analysts' choice of the
variables used and the grouping of institutions by these variables. 
Transition matrix models, a variation of actuarial models, implicitly
incorporate more information into loss estimates by using regulatory
ratings of financial institutions to estimate the probability of
resolution for different categories of institutions.  Regulators
assign financial institutions a rating to reflect their financial and
operating condition, determined through on-site examinations and
examiners' assessment of risk.  However, in addition to the
limitations described above for all actuarial-based approaches,
transition matrix models assume that regulatory ratings are the sole
determinant of an institution's failure. 

Asset mark-down approaches to estimating the cost of bank and thrift
insolvencies are based on the premise that the market value of an
institution's assets and equity can be used to identify potentially
insolvent institutions.  These types of approaches are limited in
that they are very data intensive, relying heavily on call report
data and other data not readily available for all institutions.  For
example, some asset mark-down approaches attempt to discount the cash
flows of several categories of an institution's assets and
liabilities over their expected lives.  In addition to detailed call
report data, this process requires information, such as the maturity
or duration of different types of loans, differences between loan
contract and market interest rates, and expected prepayments.  Other
less rigorous approaches simply use analysts' judgments to adjust
reported asset values and earnings growth.  Even if these assumptions
appear reasonable, it is difficult to reproduce or verify the
adjustments. 

Proportional hazard models, another type of econometric approach,
attempt to predict the failure of an institution based on financial
characteristics, regulatory ratings, and economic indicators.  As
with other methodologies, the analyst's ability to identify and
measure the variables is fundamental.  Central to proportional hazard
models are historical data on failed institutions and the timing of
regulatory action to close the institutions.  Measuring time to
failure can be problematic due to the history of delay in closing
many insolvent institutions in the late 1980s.  Use of this type of
an approach has generally been limited to short-term forecasts
although some recent research has attempted to forecast failures over
a 5-year period. 

Forecasts based on simple projections of current income and capital
levels--pro forma projections--are also highly sensitive to
assumptions and limitations of call report data.  Such approaches
assume that an institution's earnings are its only source of funds
and therefore are highly sensitive to reported income and capital. 

The existence and diversity of alternative loss estimation
methodologies for deposit insurance provide a rich body of experience
to draw upon in estimating costs under an accrual-based budgeting
approach.  However, such significantly different designs and the
widely disparate cost estimates highlight the difficulty and
uncertainty inherent in estimating deposit insurance costs.  The
current use of different approaches by federal agencies will also
complicate efforts to reach consensus on the appropriate method to
use to accrue costs in the budget.  The uncertainties and limitations
of the various estimation methodologies also underscore the need to
have well-capitalized insurance funds to absorb losses from failed
institutions that are intrinsically difficult to estimate over a
long-term period. 


   RISK-ASSUMED ESTIMATION
   METHODOLOGIES FOR PENSION
   INSURANCE NOT FULLY DEVELOPED
   AND TESTED
---------------------------------------------------------- Chapter 5:6

Methodologies that could be used to estimate the full risk assumed by
the government in insuring private pension plans are not fully
developed and validated.  However, considerable research and
development has been invested in two potential approaches--an options
pricing approach and a simulation approach.  OMB has built upon the
work of several academic researchers and applied options pricing
theory to estimate the government's liability for pension insurance. 
The Pension Benefit Guaranty Corporation has extended the research of
Federal Reserve Bank of New York economists to simulate funding
necessary for future plan terminations.  Appendix III provides a
brief overview of these two estimation approaches. 

Estimating the cost to the government for the risk assumed in the
pension insurance programs is a difficult exercise primarily because
it entails forecasting the failure of firms with underfunded pension
plans.  Firm failure is dependent on a number of economic,
industry-specific, and behavioral factors, which are highly uncertain
and interrelated.  In addition to forecasting firm bankruptcy,
estimating the cost of pension insurance is complicated by the need
to forecast the financial condition of pension plans.  The health of
a pension plan is greatly affected by the value of its assets, which
depend upon uncertain market conditions and interest rates.  In
addition, the financial condition of the firm also affects the
liabilities of the plan through factors such as employment and
benefit levels as well as statutorily defined minimum funding
requirements.  Under current law, PBGC is allowed to charge plan
sponsors a variable premium based only on its level of unfunded
vested benefits. 


      OMB'S OPTIONS PRICING
      APPROACH
-------------------------------------------------------- Chapter 5:6.1

In recent years, OMB has invested significant effort in using options
pricing theory\27 to estimate the government's cost of federal
pension guarantees.  The Bush administration's 1992 accrual budgeting
initiative proposed using options pricing methodologies for
estimating the accrual costs for both deposit insurance and pension
guarantees.  In OMB's pension model, the government's guarantee is
treated as giving the owners of a firm the option to transfer the
pension plan liabilities to PBGC when the firm becomes insolvent. 
This is similar to the concept used by OMB for estimating the cost of
deposit insurance.  However, since the cost to the government for the
pension guarantee is contingent on the financial conditions of both
the pension plan and the plan's sponsoring firm, OMB's pension model
specifies a probabilistic process for deriving the future value of
both the pension plan's and the sponsoring firm's assets and
liabilities. 

OMB's options pricing model for estimating the government's cost of
pension guarantees requires certain modifications and assumptions
that go beyond common applications of options theory.  In most
applications of options pricing, the time to expiration of the option
is typically short.  As such, assuming that the value of assets,
liabilities, or interest rates will change at a constant rate over
time is not problematic.  However, for pension insurance, the
duration of the option is long, making such standard assumptions
unrealistic.  For example, the OMB model assumes that the value of a
firm's assets will vary in the future but always at the same rate. 
This assumption is important in determining the future value of firm
and pension assets and, ultimately, the value of the option and the
government's cost.  Experts with whom we consulted pointed out that
common applications of options pricing for long-lived options
typically use probabilistic functions, which allow for large changes
in asset values.  Thus, the OMB model potentially could be improved
by specifying a probabilistic process that would allow greater
volatility in future asset values.  PBGC officials also noted that
OMB's model does not take into account Internal Revenue Code rules
that specify minimum and maximum pension plan funding that may dampen
actual volatility in the growth of pension plan assets. 

Modifications to the OMB model with regard to its treatment of
interest rates could strengthen its estimation capability.  The
values of pension liability are dependent on prevailing and future
interest rates because these liabilities are due in the future.  The
assumption that future interest rates are determined today, as
assumed in OMB's model, will cause inaccuracies--especially in a
long-term estimate.  The experts that we consulted also recommended
that since interest rate risk has significant implications for
pension liabilities, a separate recognition of interest rate risk
within the OMB model should be considered.  Modeling variations in
future interest rates using an appropriate probabilistic process
would introduce variation in a firm's future pension fund liabilities
and potentially improve estimates of the government's liability. 

Sensitivity analysis performed on the OMB model demonstrated that
assumptions about how much the value of firm and pension plan assets
will vary over time have significant impact on the model estimates. 
For example, PBGC's net liability decreased 80 percent when the
volatility of a firm's assets assumed by OMB was cut in half.  When
volatility estimates derived by other researchers were plugged into
the model, the government's estimated net liability dropped by 92
percent.  Such significant differences in the model estimates
indicate that additional research on key model parameters and
assumptions should be undertaken.  Other parameter assumptions, such
as the level of net worth at bankruptcy, have also been questioned
and should be grounded in empirical research. 

In addition to the modifications described previously, additional
research and development would be necessary before estimates from
OMB's model could be used for accrual-based budgeting.  In its
current form, the OMB model only generates an estimate of the
government's cost of pension insurance over an indefinite period.  In
order to use this cost estimate in an annual budget context, a
methodology to amortize the cost on a yearly basis is necessary. 

In discussing OMB's model with PBGC officials, concerns about the
complexity of the model were raised.  Even though OMB's research has
been published, and officials have been open in providing researchers
access to the model, PBGC's chief economist characterized the OMB
model as a black box.  He noted that the model is too complex for
most analysts and economists to fully understand and does not provide
an intuitive understanding of the factors influencing the
government's cost.  In part, these concerns led PBGC to pursue a
simulation-based approach to model the financial condition of its
insurance program under a range of economic scenarios.  PBGC
officials asserted that less restrictive computer simulation models
are increasingly taking the place of options pricing approaches in
financial markets as financial instruments have become more complex
and computing power less expensive. 


--------------------
\27 See figure 5.1. 


      PBGC'S SIMULATION APPROACH
-------------------------------------------------------- Chapter 5:6.2

Over the last several years, PBGC has been developing a computer
simulation model, called the Pension Insurance Modeling System
(PIMS), to improve its capacity to estimate future claims and
evaluate the impact of proposed legislative or regulatory changes on
its financial condition.  PIMS allows PBGC to model a large number of
firm and pension plan attributes, including interest rates, asset
returns, and bankruptcy rates, over a wide set of possible economic
scenarios.  PBGC believes that its simulation approach is a better
tool for policy analysis than OMB's options pricing model, but agency
officials we spoke to were divided over its use for accrual
budgeting.  OMB has indicated that it would like insurance program
agencies to have responsibility for developing accrual-based budget
estimates and views PBGC's PIMS research efforts as a step in that
direction. 

Opinions about the usefulness of PIMS for accrual-based budgeting
purposes differ.  PBGC's chief economist expressed concern about
using PIMS or any model for accrual-based budgeting purposes.  He
said that the future expected cost of PBGC's pension insurance is
very sensitive to changes in key assumptions.  For example, he said
using interest rate experience from the period 1926 to 1991 produces
a considerable change in the expected cost compared with using the
experience of the 1970 to 1991 period.  Although the information
provided by different simulations is very useful for policy analysis,
he does not think it is stable enough for budgeting or accounting. 
The chief economist suggested that some of his concern could be
alleviated if assumptions were set by a neutral body to minimize the
potential for manipulation of the cost estimates. 

PBGC's chief actuary stated that actuaries look for the best estimate
and not the "right" number.  He pointed out that all budget estimates
are imperfect and PIMS has real value as an estimating tool. 
Estimating the exposure undertaken by the government is
self-correcting--gains and losses over time offset each other. 
However, if over a number of years gains or losses start adding up
and exceed a certain level, then the methodology would have to be
reassessed.  He stated that this approach has been used by insurance
companies to estimate risk-based reserves. 


   OTHER INSURANCE PROGRAMS ALSO
   PRESENT ESTIMATION CHALLENGES
---------------------------------------------------------- Chapter 5:7

The other insurance programs we reviewed--the war-risk insurance
programs, the Overseas Private Investment Corporation's (OPIC)
political risk insurance, and the Vaccine Injury Compensation Program
(VICP)--will likely present significant estimation challenges under
an accrual-based budgeting approach.  The unique role of these
programs, the subjective or volatile nature of the insured risks, or
a lack of relevant historical data complicate risk assessment. 
According to agency officials, none of these programs currently rely
on heavily quantitative or systematic risk assessment tools. 


      OPIC'S RISK ASSESSMENT
      PROCESS RELIES HEAVILY ON
      EXPERT JUDGMENT
-------------------------------------------------------- Chapter 5:7.1

OPIC relies heavily on expert judgment to assess the risk it
undertakes in insuring investments of U.S.  companies abroad against
expropriation, currency inconvertibility, and political violence. 
Although the use of more quantitative methods, such as econometric
modeling or options pricing, has been suggested by some budget
experts, no specific comprehensive models have been developed. 
Further, OPIC officials and some analysts expressed skepticism about
the usefulness of this type of modeling for OPIC's insurance
activities. 

The complexity and subjectivity of political risks along with a lack
of relevant data make risk assessment difficult.  Political risks
tend to be country-, industry-, company-, and project-specific. 
Political risks are subject to many variables that are inherently
difficult to predict, such as the political stability of governments,
long-term macroeconomic conditions, changes in future foreign
relations, or the acceptability of a given project or industry to the
host country.  Thus, a risk for one industry may not be relevant for
another industry in the same country.  Further, because there is a
lack of empirical evidence, assessment of the potential implications
of various events and conditions is based on primarily subjective
evaluation.  An OPIC official stressed that there can be considerable
uncertainty surrounding the business environment and other factors
that influence the risk associated with a particular project.  For
example, agency officials stressed that many of the countries covered
by OPIC do not have an extensive history of private sector
development and economic reform programs that would be necessary to
develop a useful model.  One official noted that in some areas, such
as Eastern Europe and Russia, there is no historical experience to
draw on.  Agency officials said that for these reasons, there are no
effective quantitative models or actuarial tables for OPIC's
political risk insurance. 

Currently, the risk assessment methods used by OPIC to set premium
rates and establish insurance reserve levels rely heavily on expert
judgment.  However, according to OPIC officials, while the risk
assessment process is not highly quantitative, efforts are made to
establish premium rates based on the risk assumed for a particular
project.  In determining the risk associated with a given project,
OPIC considers the project-specific risk, such as the structure of
the project and the experience of the project's sponsors, and
country-based risk, such as projections of the country's general
economic condition, including balance of payments and foreign
exchange reserve levels.  According to OPIC officials, each
investment is negotiated and underwritten individually.  They
stressed that this process is important in controlling OPIC's risk
exposure precisely because predicting political risk over long
periods\28 is so difficult. 

OPIC establishes general reserves based on losses inherent in its
entire portfolio.  For budget purposes, budget authority is obligated
for these reserves when identified.  Reserve levels are developed by
OPIC's management in consultation with the agency's independent
auditors and are based on historical experience and an assessment of
other factors, including changes in the composition and volume of the
insurance outstanding and worldwide economic and political
conditions.  However, according to agency officials, there is little
historical data or 20-year trend information that can be used to
develop actuarial tables to accurately predict risk.  They noted that
the program's entire historical experience is considered because
claims have been sporadic over the life of the program and no
discernable patterns exist.  Further, they emphasized that although
historical data provides a starting point, adjustments are made to
account for OPIC's new business and other factors that affect the
level of risk undertaken.  As a result, OPIC officials stressed that
management's judgment is a key factor in determining the appropriate
reserve levels. 

OPIC officials expressed serious concerns about the feasibility and
usefulness of generating risk-assumed estimates for budget outlays on
either a project-specific or annual cohort basis.  They pointed out
that since only a few (about 150) policies for an even fewer number
of projects are issued each year, adequately pooling risk in any year
is extremely difficult.  According to agency officials, a primary
concern in minimizing overall risk is maintaining an appropriate
balance across clients, business sectors, and countries.  They
stressed that in their opinion, the focus of management's efforts and
decision-making should be on "good portfolio management," such as
using contract provisions and client diversification to mitigate the
aggregate risk undertaken by the program.  Agency officials did not
believe that a focus on annual cohorts--rather than on OPIC's entire
portfolio--was conducive to this broad management focus. 

Agency officials also noted that a number of factors make determining
the net cost to the government at the time insurance is extended
difficult.  For example, they explained that the amount of recoveries
associated with specific projects is very uncertain but not including
these amounts would overstate the government's potential cost.  They
also said that it would be very difficult to determine how to account
for and allocate the benefits of contract provisions that limit total
covered losses for multiple projects by the same company.  Overall,
OPIC officials said that they strongly opposed the use of
cohort-based budget estimates and were skeptical of whether a
comprehensive risk assessment model for their insurance activities
could be developed.  They maintain that their current practice of
obligating reserves based on losses inherent in their portfolio when
identified is a reasonable approach. 


--------------------
\28 According to OPIC officials, the majority of contracts are
written for 20 years. 


      UNIQUE ROLE OF THE WAR-RISK
      PROGRAMS COMPLICATES RISK
      ASSESSMENT
-------------------------------------------------------- Chapter 5:7.2

The unique role of the maritime and aviation war-risk insurance
programs complicates risk assessment.  The war-risk insurance
programs provide insurance to commercial airlines and ship owners
during extraordinary circumstances, such as war and other
hostilities, in order to support the foreign policy interests of the
United States.  Both programs provide coverage only when commercial
insurance is not available or is available only on unreasonable terms
and conditions.  This unique role complicates risk assessment because
by design (1) the insured risks tend to be case-specific and highly
variable, (2) historical program data are limited, and (3) commercial
sector war-risk insurance data are unlikely to be directly applicable
to the risk assumed by these federal programs.  Currently, risk
assessment for both programs relies heavily on expert judgment. 
Neither program uses quantitative modeling or standard risk
assessment procedures. 

Officials from both agencies told us that because of the programs'
infrequent activation and extremely rare losses, there is a lack of
historical program data for risk assessment.  For example, according
to Federal Aviation Administration (FAA) officials, aviation war-risk
insurance has only been issued during a few brief periods since 1975. 
Maritime Administration (MARAD) officials also stated that their
war-risk insurance is activated very infrequently and remains active
for short durations, usually less than a year.  Claims under the
programs are also extremely rare.  In addition, agency officials told
us that historical information from commercial war-risk insurance may
not be useful in assessing the risk undertaken by their war-risk
insurance programs because commercial information often is not
readily available or applicable.  For example, officials at both
agencies told us that premium information is generally not released
by commercial sector war-risk insurers. 

Because of the above limitations, risk assessment for the federal
war-risk programs currently relies heavily on expert judgment. 
Premiums for both programs are set in consideration of the risk
involved and U.S.  policy interests and to encourage the
participation of commercial insurers.  In general, risk assessment
involves the subjective evaluation of the numerous factors associated
with a particular flight or voyage.  For example, according to FAA
officials, they consider factors such as (1) the hull value, (2) the
potential liability for passengers, crew, cargo, and losses on the
ground, and (3) the apparent danger associated with flights into the
area(s) excluded by commercial insurers.  They told us that in
assessing the risks associated with a particular area, they consider
available information on potential dangers, such as intelligence
information on terrorist groups and the types of weapons involved in
the conflict.  MARAD officials also described their risk assessment
process as ad hoc and subjective.  They said that a number of factors
are considered in assessing risk, such as (1) the destination of the
vessels, (2) the extent of the military threat, (3) the current
commercial rates, and (4) the value of the vessels.  According to
agency officials, an outside consultant, the American War-Risk
Agency, has provided advice on risk assessment. 

Overall, officials from both war-risk programs expressed concerns
that accrual-based budgeting may not be feasible for their programs. 
Officials at both agencies described the infrequent and limited
issuance of insurance and the resulting lack of historical experience
as key obstacles to developing risk-assumed estimates and using
accrual-based budgeting for these programs.  The emergency--or
stand-by--nature of the programs makes it difficult to know in
advance when they will be activated and limits the time available for
risk assessment.  FAA officials stated that in their opinion it was
not feasible to generate reliable risk-assumed estimates for the
budget.  MARAD officials provided a similar assessment for their
war-risk program, stating that given the nature of the program,
reliable estimates of the risk assumed could not be developed. 


      VACCINE INJURY COMPENSATION
      PROGRAM'S LIMITED HISTORICAL
      EXPERIENCE MAY IMPEDE RISK
      ASSESSMENT
-------------------------------------------------------- Chapter 5:7.3

According to Health Resources and Services Administration (HRSA)
officials within the Department of Health and Human Services (HHS),
systematic risk assessment is not currently undertaken for VICP.  The
program's limited historical experience was cited as a key factor in
the uncertainty surrounding its future costs.  HRSA officials
stressed that in their opinion, there is not sufficient historical
evidence on the cost of claims to produce meaningful estimates of the
program's future costs because the program has only been in operation
since 1989.  A 1994 Treasury report also concluded that VICP had not
been in existence long enough to project future outlays with
confidence.\29

The lack of scientific evidence linking adverse events to vaccines
and the dynamic or subjective nature of some variables, such as the
amount of settlement awards, have also been cited as factors
complicating risk assessment.  According to the Treasury study, "the
scientific literature indicates that most injuries and deaths of the
type compensable under VICP cannot be said with certainty to be
caused by vaccinations."\30

In addition, HRSA officials expressed concern that the dynamic or
subjective nature of some variables make it difficult, if not
impossible, to generate reasonable projections of the program's
future claims.  For example, the introduction of new vaccines and the
increasing use of combined antigens in a single vaccination make it
more difficult to determine risk.  Also, HRSA officials described
settlement amounts awarded by the courts as case-specific and
subjective. 

Overall, HRSA officials expressed serious reservations about the
feasibility of producing reasonable projections of the program's
future costs and the use of accrual-based budgeting for VICP.  The
Treasury report concurred that until the program matures, program
outlays cannot be estimated with confidence, but noted that "as the
program matures sufficient program data will become available to
permit more sophisticated methods of estimating future outlays to be
used."\31 For example, a Treasury analyst noted that it may not be
necessary to establish causation between the vaccine and the adverse
event in order to establish an estimate of the program's future
outlays.  As more cases are settled, it may be possible to establish
a pattern between adverse events and award amounts based on
historical data.  However, changes in variables over time, such as
injury coverage and the introduction of new vaccines, will have an
impact on the usefulness of cost estimates based on historical data. 


--------------------
\29 National Vaccine Injury Compensation Program:  Financing the
Post-1988 Program and Other Issues, Department of the Treasury,
August 1994. 

\30 Ibid., v. 

\31 Ibid., v. 


   ESTIMATION CHALLENGES ARE THE
   CRITICAL FACTOR IN USE OF
   RISK-ASSUMED ESTIMATES
---------------------------------------------------------- Chapter 5:8

The ability to generate reasonable, unbiased estimates of the risk
assumed by the federal government is of primary importance in the
effective implementation of accrual-based budgeting for federal
insurance programs.  However, as the discussion in the preceding
sections shows, the current development and acceptance of risk
assessment methodologies varies significantly across these programs. 
This variation reflects the diversity and nature of the risks insured
by the federal government.  For some programs, such as the Service
Disabled Veterans Life Insurance program, estimates are sufficiently
established so that the government's cost--"the missing premium"--can
be reasonably estimated.  For other programs, such as deposit
insurance, alternative models with different theoretical and
practical approaches result in an array of estimates of the
government's costs.  Other insurance programs, such as the war-risk
insurance and vaccine compensation programs, have no or limited
systematic risk assessment experience.  To date, no formal or
quantitative risk assessment methodologies have been established for
these programs.  All risk assessment approaches, regardless of their
technical sophistication, are based upon many judgments and the
quality and quantity of available data.  As such, assumptions and the
process used to arrive at estimates need to be well documented.  In
this regard, the use of econometrics or other quantitative methods
can facilitate the replication of estimates by other analysts and
auditors. 

The estimation challenges highlighted in this chapter are at the
center of the accrual-based budgeting debate.  Within the budget
community, there are a variety of views on the acceptable level of
uncertainty and complexity to introduce into the federal budget.  As
discussed in chapter 7, consideration of these issues is likely to be
most beneficial when the focus of the discussion is on whether or not
the inclusion of risk-assumed estimates would provide policymakers
with more accurate information and signals about the underlying
insurance programs, rather than on whether an estimate is the "right"
number.  It may be most important that the budget information and
incentives provided to policymakers be "approximately right rather
than precisely wrong."


APPROACHES FOR INCORPORATING
ACCRUAL-BASED ESTIMATES INTO THE
BUDGET FOR INSURANCE PROGRAMS
============================================================ Chapter 6

The way in which accrual-based cost information for federal insurance
programs is incorporated into the budget will influence the extent to
which budget information and incentives are changed and the
limitations of cash-based budgeting are overcome.  A key issue
surrounding the use of accruals in the budget is the extent to which
earlier cost recognition is linked to increased cost control.  It is
clear that risk-assumed estimates are not yet sufficiently developed
to be incorporated directly into the primary budget data--budget
authority, outlays, and the deficit.  Supplemental reporting of these
estimates as they develop would provide policymakers additional
information and serve as the basis for future evaluation of whether
to incorporate the estimates into the primary budget data.  If, over
time, reasonable unbiased estimates are developed and a decision is
made to use them in the primary budget data, approaches to doing so
need to be considered.  In this chapter, we examine three general
ways of using accrual-based estimates in the budget and their
respective advantages and disadvantages.\1 In chapter 7, we discuss
other issues related to the implementation of accrual-based budgeting
for insurance programs. 


--------------------
\1 CBO and OMB discuss several options for incorporating accrual cost
measures into the budget for deposit insurance in their respective
studies, Budgetary Treatment of Deposit Insurance:  A Framework for
Reform, Congressional Budget Office, May 1991, and Budgeting for
Federal Deposit Insurance, Office of Management and Budget, June
1991. 


   THREE GENERAL APPROACHES OFFER
   PROGRESSIVE INTEGRATION OF
   ACCRUAL-BASED INFORMATION INTO
   THE BUDGET
---------------------------------------------------------- Chapter 6:1

Three general approaches to using accrual-based estimates in the
budget demonstrate how these measures might be progressively
integrated into the primary budget data-- budget authority, net
outlays, and the budget deficit.  Each approach would have a
different effect on the aggregate budget totals. 

Supplemental Approach:  Under this approach, accrual-based cost
measures would be included as supplemental information in the budget
documents.  The current basis of reporting budget authority, net
outlays, and the budget deficit would not be changed. 

Aggregate Budget Authority Approach:  Under this approach,
accrual-based cost measures would be included in budget authority for
the insurance program account and in the aggregate budget totals. 
Net outlays--and hence the budget deficit--would continue to be
reported on a cash basis. 

Aggregate Outlay Approach:  Under this approach, accrual-based cost
measures would be incorporated into both budget authority and net
outlays for the insurance program account and therefore in the
aggregate budget totals.  Thus, unlike the other approaches, the
budget deficit would include the accrual-based cost for federal
insurance programs.  This approach is the most comprehensive and
would be similar to the approach used for credit programs under
credit reform.\2


--------------------
\2 Appendix I provides an overview of the treatment of direct loans
and loan guarantees under credit reform. 


   SUPPLEMENTAL APPROACH WOULD
   RETAIN CURRENT BASIS OF BUDGET
   REPORTING FOR FEDERAL INSURANCE
   PROGRAMS
---------------------------------------------------------- Chapter 6:2

Under the supplemental approach, estimates of the risk assumed by the
government would be included in the budget documents as additional
information.  Given the existing state of the art in making
accrual-based estimates, this approach is the most feasible at this
time.  The current basis of reporting budget authority, net outlays,
and the deficit would not be changed.  Including accrual-based
information in the budget to supplement the traditional cash-based
budget reporting for federal insurance programs would increase the
information available to decisionmakers by helping to highlight the
potential costs of these programs.  This approach would also allow
time to test and improve estimation methodologies and increase the
comfort level of users before considering whether to move to a more
comprehensive approach.  In a similar fashion, information on federal
credit programs and estimates of the government's subsidy costs were
reported for years in the Special Analyses volume of the President's
budget prior to the enactment of credit reform. 

However, this approach might not have a significant impact on the
budget decision- making process because the accrual-based cost
information would not directly affect the budget totals and the
budget allocations to congressional committees.  Furthermore, there
may be little incentive to improve cost estimates and/or risk
assessment methodologies for the various insurance programs since
this information would not be the basis for budget decisions.  Figure
6.1 provides a summary of the key advantages and disadvantages of the
supplemental information approach. 

   Figure 6.1:  Advantages and
   Disadvantages of the
   Supplemental Approach

   (See figure in printed
   edition.)

Accrual-based costs for federal insurance programs could be presented
as supplemental information in the budget in a number of ways.  In
recent years, OMB has provided some risk-assumed cost information on
insurance programs in the Analytical Perspectives volume of the
President's Budget.\3 This presentation could be continued and
enhanced by developing a consistent format for reporting the risk
assumed by each program. 

Useful information for this type of presentation, some of which has
been presented in previous budgets, includes (1) the annual
risk-assumed cost for each program, (2) summaries of the
methodologies used to generate cost estimates, and (3) explanations
of any changes in estimates from year to year.  Two key advantages of
this type of presentation are (1) a more detailed narrative
discussion is possible than in the account-level presentation of the
budget appendix and (2) all federal insurance programs are discussed
in one place. 

Such a discussion could be further supplemented if accrual-based cost
information was also displayed at the insurance program account level
in the budget appendix even though it would not be included in the
budget totals.  For example, the cost of insurance programs could be
shown on an accrual basis in one table for display purposes and on a
cash basis in another table for the same account.  Although only the
cash-based amount would be included in the budget totals, this
presentation may increase attention paid to risk-assumed cost
estimates at the time budget decisions are made.  Such a display
would highlight the differences in the type of information provided
on a cash basis versus an accrual basis for the various insurance
programs without changing the reporting basis of total budget
authority, net outlays, or the budget deficit. 


--------------------
\3 For example, see chapter 8, "Underwriting Federal Credit and
Insurance," Analytical Perspectives, Budget of the United States
Government, Fiscal Year 1997. 


   AGGREGATE BUDGET AUTHORITY
   APPROACH WOULD INTRODUCE SOME
   ACCRUAL AMOUNTS INTO THE BUDGET
---------------------------------------------------------- Chapter 6:3

The aggregate budget authority approach moves further along the
continuum from cash-based budgeting to full accrual-based budgeting. 
This would incorporate accrual-based cost measures into budget
authority but would stop short of adopting the full credit reform
approach.  The full cost of the risk assumed by the government would
be recognized in the budget authority for the insurance program and
the aggregate budget authority totals.  Net outlays and the budget
deficit would continue to be reported on a cash basis.  Budget
authority would be obligated at the time an insurance commitment was
made and would be held as a reserve in the program account earning
interest.  Future claims would be paid from the authority in these
reserves. 

A key advantage of the aggregate budget authority approach is that it
provides earlier recognition of insurance costs directly in the
budget (in budget authority) while preserving cash-based reporting
for net outlays and the deficit.  Recognizing accrual cost estimates
in budget authority may increase attention to these costs without
potentially subjecting outlays and the deficit to estimation
uncertainty.  This increased attention may also focus efforts on
improving cost estimates.  However, since the accrual-based cost
would not be reflected in the budget deficit, it is unclear how much
more this approach would affect the budget decision-making process
than the supplemental information approach.  Figure 6.2 presents a
summary of the key advantages and disadvantages of the aggregate
budget authority approach. 

   Figure 6.2:  Advantages and
   Disadvantages of the Aggregate
   Budget Authority Approach

   (See figure in printed
   edition.)

The influence of this approach is likely to be further limited by the
fact that most federal insurance programs are classified as
"mandatory" under BEA.\4 This means that only increases in budget
authority caused by changes in legislation must be addressed by the
Congress.  Although this is also true for automatic increases in
outlays, the fact that changes in outlays increase the deficit could
prompt action.  The budget authority only approach does not offer
this incentive.  The potential impact of this approach on the
decision-making process would likely be greater for insurance
programs that are classified as discretionary spending because
accrued costs would be included under the discretionary budget
authority spending caps.  As discretionary budget authority totals
near the discretionary spending limits, this approach may prompt the
Congress to specifically address the costs of these programs.\5


--------------------
\4 As noted in chapter 2, claim payments for 10 of the 13 programs
reviewed are classified as mandatory spending. 

\5 CBO estimates that the President's fiscal year 1998 discretionary
spending proposals will be approximately $4 billion below the budget
authority spending caps. 


      A DISCRETIONARY FEATURE
      COMBINED WITH AGGREGATE
      BUDGET AUTHORITY APPROACH
      COULD PROMPT ACTION TO
      ADDRESS COSTS
-------------------------------------------------------- Chapter 6:3.1

Actions to control costs under the aggregate budget authority
approach could be prompted by requiring a discretionary appropriation
for the government's subsidy cost.  For insurance programs classified
as mandatory, a separate discretionary account would be created to
record the government's subsidy costs.  A general fund appropriation
to the discretionary account would be required to cover any subsidy
costs in the year the insurance is extended, unless alternative
actions were taken to reduce the government's cost, such as
increasing program collections or reducing future program costs. 
Amounts appropriated to the discretionary account would then be paid
to a mandatory program account.  The mandatory program account would
handle all other cash flows including premium collections and claim
payments.  This account would also hold the program's reserves for
future claims.  As a result, the government's accrual-based costs
would be reported in the budget authority and net outlays for the
program's discretionary account and in the budget authority totals
for the government.  Total net outlays and the budget deficit would
continue to be recorded on a cash basis. 

Figure 6.3 demonstrates how the cost of a hypothetical insurance
program would be recorded under this approach, assuming that a
funding deficiency exists on an accrual basis.  The insurance program
is assumed to have the following activity:\6

(1) premium collections of $5 billion;

(2) claim payments of $3 billion; and

(3) an accrual-based subsidy cost to the government of $4 billion,
i.e., an estimated funding shortfall between the risk assumed and
estimated collections. 

   Figure 6.3:  Reporting Flow for
   the Budget Authority Approach
   with Discretionary Option

   (See figure in printed
   edition.)

As shown in Figure 6.3, the mandatory program account would record
cash flows, including $5 billion in premium collections and cash
outlays of $3 billion for claim payments.  The accrual-based subsidy
cost of $4 billion would be appropriated to the discretionary account
and then transferred to the mandatory account.  As a result, the
subsidy cost of $4 billion would be scored against the discretionary
spending caps and, negative outlays of $2 billion would be recorded
in the mandatory program account and included in the deficit. 
Reserves of $6 billion would be held in the program account as
obligated budget authority invested in Treasury securities. 

A key advantage of this approach is that it prompts budget
decisionmakers to address explicitly the government's cost while
preserving the more straightforward cash-based reporting for net
outlays and the budget deficit.  Since the appropriation would be
discretionary, and thus subject to BEA caps (assuming their
extension), decisionmakers would have an incentive to reduce the
government's costs.  Despite this advantage, however, there are
broader policy considerations involved with creating a discretionary
aspect for programs originally funded as mandatory spending.\7

In most cases, such a step would go beyond simply changing the
reporting of program costs in the budget by shifting the locus of
decisions to the annual appropriation process thereby possibly
changing program operations.  Figure 6.4 summarizes the key
advantages and disadvantages of this feature. 

   Figure 6.4:  Advantages and
   Disadvantages of the Aggregate
   Budget Authority Approach With
   Discretionary Outlay Feature

   (See figure in printed
   edition.)


--------------------
\6 Administrative costs have been omitted for simplicity.  In
practice, some insurance programs are designed to cover
administrative costs while the administrative costs of others are
paid from general fund appropriations.  The treatment of
administrative costs should be considered in the development and
implementation of accrual-based budgeting for these programs.  For
example, if the intention is for the program to be completely
self-supporting, then administrative costs should be covered by
program collections and could be paid from the financing account to
the program account.  Alternatively, if administrative costs are
covered by general funds, they could be separately appropriated to
the program account. 

\7 Claim payments for 10 of the 13 insurance programs included in
this study are classified as mandatory spending under BEA. 


   AGGREGATE OUTLAY APPROACH
   INCORPORATES ACCRUALS INTO THE
   PRIMARY BUDGET DATA
---------------------------------------------------------- Chapter 6:4

The aggregate outlay approach is the most far-reaching of the three
general approaches outlined.  This approach is similar to both the
treatment of credit programs under credit reform and the approach
proposed by OMB for federal insurance programs in the fiscal year
1993 budget.  Under this approach, accrual-based costs would be
recorded in both budget authority and outlays for the program and in
the aggregate budget totals, including the budget deficit. 

Like credit reform, two key features of this approach include (1) the
use of a program account and a financing account\8 and (2) the
separation of insurance program activities into transactions that
represent a cost to the government and transactions that are merely
cash flows with no cost to the government, such as working capital
transactions.  The government's accrual-based subsidy cost for an
insurance activity would be recorded in the program account while all
other cash flows would be handled in a separate financing account. 
Similar to credit reform, the program account would be budgetary and
thus included in the calculation of the budget deficit.  The
financing account, on the other hand, would be a nonbudgetary
account\9 and thus not included in the deficit calculation.  Table
6.1 illustrates the relationship between these accounts, the budget
deficit, and the government's borrowing needs. 



                               Table 6.1
                
                   Relationship Between Budgetary and
                         Nonbudgetary Accounts


Budgetary accounts  Total Governmental Receipts

                    minus

                    Net Outlays

                    equals

                    Deficit (or Surplus)

Nonbudgetary        minus
accounts

                    Means of Financing

                    equals

                    Borrowing From the Public
----------------------------------------------------------------------
Source:  Budgetary Treatment of Deposit Insurance:  A Framework for
Reform, Congressional Budget Office, May 1991. 

Under this approach, the accrued cost to the government would first
be recognized in the program account and then outlayed to the
nonbudgetary financing account.  This transaction will cause the
government's accrual-based subsidy cost to be included in the deficit
at the time the insurance is extended.  Therefore, the measurement
basis of outlays for the program account and deficit would be changed
from the current cash basis to an accrual-based estimate of the
government's subsidy cost for an insurance activity.  An
appropriation would cover the government's cost unless other actions
were taken to eliminate the funding shortfall, such as increasing
collections or reducing program costs.  Since most insurance programs
are mandatory, this appropriation\10 would occur automatically unless
additional control mechanisms, as discussed previously, were adopted. 
Even without this additional feature, the fact that increases in
outlays would be reflected in the deficit could prompt action to
address the causes of such increases.  Key factors involved in
implementing this general approach, including reestimation, funding
sources, and reserve levels, are discussed in chapter 7. 

Figures 6.5 and 6.6 compare how the cost for the hypothetical
insurance program outlined above would be recorded under the current
cash-based approach versus the aggregate outlay approach. 

   Figure 6.5:  Reporting Flow for
   the Current Cash-based Approach

   (See figure in printed
   edition.)

As shown in figure 6.5, the current cash-based budget would record
cash inflows of $5 billion and cash payments of $3 billion, resulting
in negative net outlays (income) of $2 billion.  Total net outlays
and the budget deficit would be reduced by this amount in the current
budget period.  Conversely, as shown in figure 6.6, under the
aggregate outlay approach, the insurance program would receive an
appropriation to reflect the subsidy on a risk-assumed basis, unless
some alternative actions are taken to eliminate the funding
shortfall.  An appropriation of $4 billion would be received by the
program account and outlayed to the financing account.\11 This amount
would be included in total net outlays and the budget deficit.  The
financing account would also record nonbudgetary cash flows,
including premium income of $5 billion and claim payments of $3
billion.  As a result, under this approach, the insurance program
would increase the budget deficit by $4 billion rather than reducing
it by $2 billion, as was the case on a cash basis. 

   Figure 6.6:  Reporting Flow for
   the Aggregate Outlay Approach

   (See figure in printed
   edition.)

Without fundamentally changing the nature of most federal insurance
spending, the aggregate outlay approach is the most comprehensive of
the three approaches outlined and has the greatest potential to
achieve many of the conceptual benefits of accrual-based budgeting. 
The isolation and recognition of the government's full cost when
budget decisions are being made would permit more fully informed
resource allocation decisions.  By recognizing the government's cost
in the budget deficit at the time decisions are made, the incentives
for managing insurance costs may be improved.  Furthermore,
recognizing costs in net outlays and the deficit at the time
insurance commitments are made would better reflect their fiscal
effects.  In some cases, such as for the deposit insurance programs,
accrual-based budgeting using the aggregate outlay approach may
smooth spending patterns and reduce cost distortions created by
temporary or sporadic cash flows. 

Despite these potential advantages, the aggregate outlay approach has
several disadvantages.  A primary concern is the uncertainty
surrounding the estimates of the risk assumed by the federal
government for federal insurance programs.  To the extent that
estimates are unreliable, resource allocation may be distorted and
the potential for manipulation increased.  As discussed in chapter 5,
risk-assumed cost estimates for most insurance programs are either
currently unavailable or not fully accepted and thus the uncertainty
surrounding these estimates presents a key obstacle to the successful
implementation of this approach.  In addition, the aggregate outlay
approach adds a layer of complexity to an already complex budget
process.  As was the case with credit reform, the use of
accrual-based budgeting for federal insurance programs will result in
new complexities and implementation challenges. 

Further, unlike the majority of programs covered under credit reform,
most federal insurance spending is classified as mandatory under BEA. 
As discussed above, under BEA for mandatory programs, only legislated
changes that increase the level of the government's commitment would
have to be offset by spending reductions or revenue increases. 
Increases in existing spending for mandatory federal insurance
programs would not require action to address these costs, but the
inclusion of accrued costs in the deficit calculation may provide
more incentive to address them than the aggregate budget authority
approach.  Figure 6.7 summarizes the key advantages and disadvantages
of the aggregate outlay approach.  If additional budget control is
desirable, a discretionary appropriation could be required to fund
the government's accrued subsidy cost unless other corrective action
is taken.  But doing so would go beyond merely changing the reporting
of program costs in the budget by shifting the locus of decisions to
the annual appropriation process, thereby possibly changing program
operations. 

   Figure 6.7:  Advantages and
   Disadvantages of the Aggregate
   Outlay Approach

   (See figure in printed
   edition.)


--------------------
\8 A third account, a liquidating account, may be used for some
insurance programs to handle transactions that occur prior to
changing to the new budget treatment.  Chapter 7 discusses the use of
a liquidating account. 

\9 Nonbudgetary accounts appear in the budget document for
information purposes but are not included in the budget totals for
budget authority or outlays.  They account for transactions of the
government that do not belong within the budget because they are
means of financing and do not represent a cost to the government. 
Nonbudgetary transactions include deposit funds such as state and
local income taxes withheld from federal employee salaries, direct
and guaranteed loan financing accounts, and seigniorage. 

\10 For mandatory programs, this would be a permanent appropriation. 

\11 In the past, OMB and CBO have differed on the treatment of
premiums in discussing a credit reform approach for deposit
insurance.  Under OMB's approach, premium collections flow into the
financing account as described above.  This approach is justified
because these funds are a means of financing deposit insurance and
are not available to fund other federal programs.  Under CBO's
approach, premiums would be credited to the program account and then
transferred to the financing account.  CBO argues that this approach
adheres more closely with the budget's current treatment of insurance
premiums as offsetting collections.  In addition, CBO notes that if
premiums are credited to the financing account, then the program
account would only report activity when it received an appropriation
for accrual-based losses in excess of program funding sources.  We
believe that if the aggregate outlay approach were adopted, the
current treatment of premiums would need to be changed, similar to
the changes made by credit reform for the reporting of loan
repayments. 


   DIFFERENT APPROACHES TO
   ACCRUAL-BASED BUDGETING HAVE
   DIFFERENT IMPACT ON INFORMATION
   FOR BUDGET DECISIONS
---------------------------------------------------------- Chapter 6:5

The various approaches to incorporating accrual-based information
into the budget have different effects on the information and
incentives provided to decisionmakers.  As noted previously, earlier
cost recognition under any of the three general approaches would not
necessarily mean that action would be taken to address costs.  In
fact, the supplemental approach may have little, if any, effect on
budget decision-making.  The extent to which earlier cost recognition
under the other two general approaches prompts action to address
accruing cost depends primarily on whether the program has permanent
budget authority\12 and is classified as mandatory spending.  Since
most insurance programs have permanent budget authority and are
classified as mandatory, even the most comprehensive general
approach--the aggregate outlay approach--would not necessarily
require any action to be taken without the adoption of additional
budget control mechanisms.  It would, however, make more visible any
increase in costs. 

While earlier reporting of accrual-based costs in net outlays and the
budget deficit might prompt deficit reduction efforts, nothing in the
current budget process would require that the cost of insurance
programs specifically be addressed as long as permanent authority was
available to cover these costs.  The earlier recognition would,
however, increase control over legislated changes that increase
future costs because, under PAYGO, legislation enacted during a
session of the Congress affecting mandatory programs must be at least
deficit neutral in the aggregate.  Under both the budget authority
and outlay approach, mechanisms could be developed to increase the
link between earlier cost recognition and budget control.  For
example, requiring the accrued cost to be funded by discretionary
appropriation\13 would increase budget control because these costs
would be forced to compete for limited resources under the
discretionary spending caps.  Alternatively, mechanisms that link
funding shortfalls to premium increases or program coverage
reductions could also be adopted. 

Whether it is desirable for cost recognition automatically to trigger
congressional action is a policy question.  Consideration of the
varied purposes and characteristics of these programs should inform
the discussion on whether to adopt a trigger mechanism, and if so,
how to design it.  For example, requiring a discretionary
appropriation would result in a fundamental change in the nature and
operation of the majority of federal insurance programs that were
originally classified as mandatory spending.  The various approaches
reflect trade-offs between changing budget incentives and other
policy considerations.  Table 6.2 provides an assessment of the
relative potential of each approach to influence budget
decision-making. 



                                    Table 6.2
                     
                      Assessment of Accrual-Based Budgeting
                     Approaches to Influence Budget Decision-
                                      making

                                 Approaches to accrual-based budgeting
                        --------------------------------------------------------
                                              Aggregate
                                              budget
                                              authority              Aggregate
                                              with                   outlay with
                                    Aggregat  discretiona            discretiona
Influence of change in              e budget  ry                     ry
budget treatment on     Supplement  authorit  appropriati  Aggregat  appropriati
budgeting               al          y         on           e outlay  on
----------------------  ----------  --------  -----------  --------  -----------
Better cost             X           X         X            X         X
information would be
available for
potential use

Reserves would be                   X         X            X         X
established for future
costs

Recognition would                             X            X\a       X
prompt action to
address accruing
program costs
--------------------------------------------------------------------------------
\a Unlike the approaches requiring a discretionary appropriation,
this approach would not achieve direct budget control for mandatory
insurance programs but rather would influence budget decision-making
through its impact on the deficit. 

In summary, the supplemental approach would improve and provide more
consistent disclosure of estimates of risk-assumed costs in the
budget documents than is currently the case and might cause
discussion, but it would not directly influence the budget incentives
for these programs.  The aggregate budget authority approach goes a
step further and begins to incorporate accrual-based costs into the
budget process by requiring the provision of budget authority at the
time decisions are made.  However, because accrual-based costs do not
affect the bottom line or the budget deficit, the impact of this
approach on budget decision-making is unclear.  The aggregate outlay
approach goes even further by incorporating costs directly into the
deficit calculation and therefore is more likely to influence budget
decisions than the aggregate budget authority approach.  But direct
budget control is not achieved for the majority of federal insurance
programs, which are classified as mandatory spending.  Under either
the aggregate budget authority or the aggregate outlay approach,
requiring a discretionary appropriation for the government's subsidy
costs would provide direct budget control. 

As shown in table 6.2, the aggregate outlay approach and the two
approaches using a discretionary control option are most likely to
prompt action to address accruing program costs.  However, under
current budget rules, the incentives provided by each differ. 
Requiring a discretionary appropriation under the aggregate outlay
approach would have the most influence on budget decision-making by
affecting both the deficit and the discretionary spending caps. 
Under the aggregate budget authority approach with a discretionary
control option, cost would have to be included under the
discretionary spending caps but would not affect the deficit.  As
discussed earlier, if changing the locus of budget decision-making
for these programs and thereby possibly affecting program operations
is undesirable, then the aggregate outlay approach, through its
impact on the deficit, has the greatest potential to influence budget
decision-making. 


--------------------
\12 Permanent budget authority refers to authority derived from
previous authorizing legislation rather than annual appropriation
acts. 

\13 Under BEA requirements, discretionary spending is subject to
spending limitations, referred to as "caps." As noted earlier, claim
payments for only three of the insurance programs reviewed--Aviation
War-Risk, Maritime War-Risk, and OPIC's political risk insurance--are
classified as discretionary spending.  Claim payments for all of the
other federal insurance programs are mandatory. 


   APPROACHES REFLECT DIFFERING
   VIEWS ON THE USE OF
   ACCRUAL-BASED INFORMATION IN
   BUDGETING FOR FEDERAL INSURANCE
   PROGRAMS
---------------------------------------------------------- Chapter 6:6

Within the budget community there exists a range of views about the
appropriate balance between the need to change budget information and
incentives for federal insurance programs and the increased
uncertainty and complexity introduced by the use of accrual-based
estimates directly in the budget.  The various approaches to
incorporating accrual-based information in the budget discussed above
represent a spectrum of views about the uses of the federal budget
and the trade-offs faced in using accrual-based information in
budgeting for federal insurance programs. 

The aggregate outlay approach reflects the opinion of some budget
experts that the only way to influence budget decision-making
significantly is to have a direct impact on the "bottom line" or the
budget deficit.  The key argument is that since the primary focus of
the budget debate is the deficit, accrual-based reporting will not
significantly influence budget decisions unless these costs are part
of the deficit calculation.  The use of a financing account to
separate costly transactions and noncostly cash flows\14

focuses reporting on the government's subsidy cost.  And, in the
opinion of some budget experts, this increases the difficulty of
diverting to other uses the funding accumulated as reserves. 

The aggregate budget authority approach reflects both general
concerns about the use of the nonbudgetary financing mechanisms and
specific concerns that the aggregate outlay approach may not be
necessary for some federal insurance programs.  Despite the potential
benefits of accrual-based information, a key reservation surrounding
the adoption of the aggregate outlay approach is its use of
nonbudgetary financing accounts.  As a general point, some experts
believe that all cash transactions should be included in the budget
totals and that cash is a superior measure for the deficit because it
is understandable and relatively transparent.  Specifically, concerns
have been expressed that the use of nonbudgetary financing accounts
introduces new risks and may increase the incentive for cost
manipulation.  For example, estimates could be manipulated to obscure
the potential program costs or the deficit if estimation
methodologies are not widely accepted and documented.  The
uncertainty surrounding the risk-assumed cost estimates for some
insurance programs increases these concerns.  Thus, until estimation
techniques are developed sufficiently to allay most of these
concerns, the aggregate budget authority approach may be the most
appropriate way to implement accrual-based budgeting for federal
insurance programs. 

In addition, some agency officials, budget experts, and analysts
expressed concerns that the use of the aggregate outlay approach
would increase the complexity of the budget reporting.  According to
some budget experts, budget authority and obligations are more
appropriate than outlays for recognizing potential costs that have
not yet materialized and that for some programs the costs of
implementing the aggregate outlay approach would outweigh the
potential benefits.  Budget experts also differ on the use of
discretionary spending mechanisms to increase budget control.  One
budget expert emphasized that requiring a discretionary appropriation
for subsidy costs under the budget authority approach would increase
budget control while preserving cash-based reporting of outlays and
the deficit.  However, other budget experts cautioned against
changing the fundamental nature of mandatory federal insurance
spending by requiring a discretionary appropriation under either the
budget authority or outlay approach.  For example, OMB has stressed
that the goal of its previous proposal was not to change the nature
of the spending but rather to improve the budget reporting for these
programs. 

Finally, the supplemental approach reflects the view that cash is the
superior measure for budget decision-making or that the shift to
accrual-based budgeting for federal insurance programs is premature
or unnecessary.  A key argument is that supplemental information can
be used to improve budget decisions without subjecting the budget to
any additional uncertainty or complexity.  However, with this
approach, there is no guarantee that the information will be used
since it is not a part of the formal budget process. 

Any choice among these approaches or variants of them is further
complicated by the fact that the relative implementation
difficulties--and the benefits achieved--vary across federal
insurance programs.  The trade-offs and implementation challenges
associated with adopting accrual measures in the budget are discussed
in the next chapter. 


--------------------
\14 As discussed in chapter 3, most federal insurance programs have
costly and noncostly transactions.  For example, while claim payments
in excess of program collections for an insurance activity represent
a cost to the government, other cash flow imbalances may be temporary
and net out over time. 


IMPLEMENTING ACCRUAL BUDGETING FOR
FEDERAL INSURANCE PROGRAMS
============================================================ Chapter 7

Although accrual-based budgeting for federal insurance programs has
the potential to improve the information available for resource
allocation and fiscal policy decisions, implementing an accrual-based
approach will present policymakers, budget professionals, and agency
managers with many challenges.  As discussed in chapter 5, the key
implementation issue is whether reasonable, unbiased risk-assumed
cost estimates can be developed for the insurance programs.  However,
generating cost estimates is only the first step in implementing
accrual-based budgeting.  Other significant challenges exist that
would need to be addressed in implementing an accrual-based budgeting
approach.  These challenges fall into three broad categories:  (1)
issues inherent in the use of risk-assumed estimates in the budget,
(2) short-term implementation issues that may be reduced or
eliminated over time, and (3) issues related to the design and
structure of an accrual-based budgeting system. 


   ISSUES INHERENT IN THE USE OF
   ACCRUAL ESTIMATES IN THE BUDGET
---------------------------------------------------------- Chapter 7:1

One of the major benefits of accrual-based budgeting is the
recognition--when programmatic and funding decisions are being
made--of the cost of future insurance claims related to the
government's insurance commitment.  This earlier recognition of costs
improves the information available to policymakers about their
decisions and may improve the ability and incentives to manage these
costs.  However, as discussed, this earlier recognition of program
cost is dependent upon reasonable, unbiased estimates of the risk
assumed by the government in undertaking the insurance commitment. 
Because insurance program costs are dependent upon many economic,
behavioral, and environmental variables, which cannot be known with
certainty in advance of the insured loss, there will always be
uncertainty in the reported accrual-based estimates.  In addition,
the use of risk-assumed cost estimation methodologies that attempt to
capture the effects of these variables and new budget mechanisms to
report estimates and reestimates will add complexity to budget
reporting for these programs. 


      EARLIER COST RECOGNITION
      INCREASES ESTIMATION
      UNCERTAINTY
-------------------------------------------------------- Chapter 7:1.1

While budgeting based on estimates of the full cost of the risk
assumed by the government for federal insurance programs has the
potential to improve the information available to policymakers at the
time budget decisions are being made, actual claims paid in any one
year will differ from the estimated cost of the commitments reported
in the budget.  This is an expected condition of using risk-assumed
accrual cost estimates in the budget for insurance programs. 
Although estimates may get more accurate over time due to
improvements in estimation methodologies, available data, and
assumption specification, some error will always remain. 
Policymakers need to understand the nature and extent of the
uncertainty in risk-assumed cost estimates and have assurances that
the estimates are unbiased and based on the best available
information and estimation methodologies. 

The uncertainty embedded in estimates of the risk assumed by federal
insurance programs is unavoidable.  As discussed in earlier chapters,
the nature of the risks covered by some federal insurance programs
require that the risks be pooled over time.  As a result, the
expected long-term cost of the program reflected in the risk-assumed
cost estimates will differ from the cash paid out in any given year. 
Reestimates will probably also be required over time if the program's
claim experience differs significantly from the previously calculated
expected long-term cost.  Improved program data could also lead to
reestimates. 

The uncertain nature of risk-assumed cost estimates must be weighed
against potential improvements in budget reporting and cost control. 
A similar trade-off was made in budgeting for credit programs under
the Federal Credit Reform Act of 1990.  Although the accrual-based
cost estimates of some loan and loan guarantee programs have
significantly changed and their actual cost may not be known for 20
or more years, most budget experts believe that the budgeting for
these programs has been improved.  Specifically, by improving
information and the recognition of program costs, accrual-based
budgeting for credit programs has increased control over credit
program costs, improved comparisons of the costs of credit program
with that of other programs, and subjected credit programs to the
competitive allocation of resources in the budget process.  An
accrual-based budgeting approach for insurance programs also has the
potential to provide an opportunity to consider the appropriate or
desired amount of government funding--or subsidy--provided to a
particular program.  Risk-assumed cost estimates would also allow
policymakers, oversight agencies, and program managers to monitor the
government's risk exposure and to take timely steps to control
program costs. 

Uncertainty in the estimation of insurance program costs must be
evaluated in terms of the direction and magnitude of the estimation
errors.  For budgeting purposes, decisionmakers would be better
served by information that is more approximately correct on an
accrual basis, than they are by cash-based numbers that are exactly
correct but misleading.  For example, industry analysts estimated
that the accruing liabilities of insolvent thrift institutions
exceeded the resources of the insurance fund in the early 1980s,
years before the full magnitude of losses began to be recognized on a
cash basis in the budget.  Although estimates of the growing cost of
the savings and loan crisis were not exact, the magnitude of the
estimated losses proved to be correct.  At the same time, the
President's budget request for the insurance fund prior to its
collapse in 1989 estimated that cash collections would exceed cash
losses in all but one year in the 1980s.  Despite the uncertainty in
the estimates of the government's accruing cost--the exact cost of
the savings and loan crisis is still not known with complete
certainty--policymakers would have had better budgetary information
and incentives for decision-making if the budget had reported such
accrual-based estimates. 

A key implementation challenge in adopting accrual-based budgeting
for insurance programs is the difficulty in producing risk-assumed
cost estimates.  Although analogous to the implementation of credit
reform, the estimation challenges for some insurance programs may be
greater than those faced for most credit programs.  For example, the
cost of the government's deposit and pension insurance commitments is
dependent upon the ability to model complex interrelationships among
highly uncertain variables such as interest rates, market risks, and
the solvency of private companies.  Estimation uncertainty will
dictate continual evaluation of the risk estimation methodologies
used to generate risk-assumed cost estimates for federal insurance
programs. 

For two programs in our study--Aviation War-Risk and Maritime
War-Risk insurance--the uncertainty in the risk-assumed cost
estimates and other implementation complexities probably outweigh the
potential benefit from an accrual-based budget treatment.  Given the
emergency or stand-by nature of these programs, it is difficult to
even know when they will be activated.  As a practical matter, the
infrequent and sporadic issuance of insurance, the resulting lack of
historical experience, and the extraordinary circumstances
surrounding activation of the programs may make the development of
reliable risk-assumed estimates and the use of accrual-based
budgeting for these programs infeasible. 


      ACCRUAL BUDGETING WILL
      INCREASE THE COMPLEXITY OF
      BUDGET TREATMENT
-------------------------------------------------------- Chapter 7:1.2

Earlier recognition of insurance program costs under an accrual-based
budgeting approach will add to the complexity of the budget treatment
of these programs compared with the current cash-based reporting. 
Complexity is increased through the use of (1) sophisticated
estimation models, (2) multiple budget accounts and/or presentations,
and (3) procedures for reestimating costs reported as budget
authority and/or outlays.  Although recognition of insurance program
costs may be improved under an accrual-based budgeting approach,
general understanding of budget data and the budget process may
decline.  All of this must be assessed in relation to the adequacy
and often misleading nature of cash budgeting for insurance programs. 

As discussed in chapter 3, cash-based budgeting for insurance
programs generally does not provide adequate information for resource
allocation and fiscal policy decision-making.  Although cash-based
budgeting is readily understandable to policymakers and the public,
it generally does not provide full information on insurance program
costs at the time the government's commitment is extended and thus
may impair resource allocation and fiscal policy decision-making. 
Under credit reform, many budget experts agree that despite the
complexity of credit reporting, decisions regarding a program's
structure--direct loans versus loan guarantees versus grants--and
funding have been improved.  A similar increase in complexity may be
a necessary element to improving the budget information on the cost
of insurance programs.  Discomfort with and skepticism of these new
measures could be alleviated by complete documentation of the
estimation and reestimation procedures. 

The complexity of the budget treatment of credit programs was
significantly increased under the Federal Credit Reform Act of 1990. 
Very few people really understand the details of budgeting and
accounting for credit programs.  Although policymakers generally
understand the concept behind budgeting for credit programs--setting
aside funds for future losses--many still consider estimates of such
costs as coming from a "black box." Such lack of understanding of the
estimation and reporting processes risks a loss of confidence in
budget data.  An accrual-based approach to budgeting for federal
insurance would entail many of the same complexities, such as
prospective cost estimation, multiple budget accounts, and periodic
reestimation of reported costs.  To provide confidence in the budget
data, documentation and clear reporting are crucial. 


      APPROACHES FOR INCORPORATING
      ACCRUAL CONCEPTS INTO THE
      BUDGET REFLECT TRADE- OFF
-------------------------------------------------------- Chapter 7:1.3

The three general approaches for incorporating accrual concepts into
the budget for insurance programs discussed in the previous chapter
illustrate the fundamental trade-off between earlier cost recognition
on the one hand and increased uncertainty and complexity of budget
reporting on the other.  The degree of integration of accrual
estimates in the budget--whether in budget authority alone or also in
outlays--will determine the impact of this information on
decision-making.  While supplemental reporting of accrual-based costs
would improve the information available for resource allocation and
fiscal policy decisions, the actual impact on budget decisions is
uncertain since the primary budget data would be unaffected. 
However, integration of accrual estimates into the budget beyond the
supplemental approach also increases the complexity of the budget
treatment and the uncertainty in the budget numbers. 

The inherent uncertainty and complexity of accrual-based budgeting
approaches for insurance programs heightens the need for careful
consideration in the design and implementation of accrual-based
budgeting for these programs.  Policymakers face a trade-off between
the need to improve information and incentives for decision-making
and the acceptable level of uncertainty and complexity in budget
reporting.  Some budget experts believe that to have the most
influence on budget decisions, accrued costs should be recognized in
budget authority and outlays so that costs are reflected in the
deficit.  Others expressed concern about increased complexity and the
use of nonbudgetary financing mechanisms such an approach would
entail.  This concern was heightened by the uncertainty surrounding
risk-assumed estimates for some insurance programs.  Further, some
budget users stated that accrual-based information is already
available to policymakers--in supplemental budget schedules and
financial statements--and could be used in budget decision-making
without the added complexity of putting accrual estimates into the
budget numbers.  The design and implementation of accrual-based
budgeting needs to address these concerns if the potential benefits
of accrual-based budgeting are to be achieved. 


   SHORT-TERM IMPLEMENTATION
   ISSUES
---------------------------------------------------------- Chapter 7:2

In implementing any of the three general approaches for accrual-based
budgeting for insurance programs, several short-term transitional
issues would need to be addressed.  First, as discussed in detail in
chapter 5, the current capacity to generate reliable risk-assumed
estimates varies considerably across insurance agencies.  Difficulty
in developing risk-assumed cost estimates should be anticipated. 
Second, many agencies expressed concern about the skills and
resources necessary to implement accrual-based budgeting and comply
with new reporting requirements.  Experience gained in implementing
an accrual-based budgeting approach for credit programs could help
guide the transition to accrual-based budgeting for insurance
programs.  Supplemental reporting of risk-assumed estimates would
provide additional information for policymakers while providing time
to evaluate a more comprehensive approach. 


      DIFFICULTY IN DEVELOPING
      RISK-ASSUMED COST ESTIMATES
      SHOULD BE ANTICIPATED
-------------------------------------------------------- Chapter 7:2.1

Agency capacity to generate reasonably reliable risk-assumed cost
estimates for budget purposes varies considerably across programs. 
Indeed, the ability to generate reasonable cost estimates of the risk
assumed by the government was a primary concern expressed by the
insurance program agency officials and budget experts we spoke with. 
At present, risk-assumed estimates of insurance losses related to
coverage extended in a budget year do not exist for all programs and
are not reported on a regular basis.  Time and experience in
developing these estimates will be required.  Credit agencies have
had difficulties in calculating reasonably accurate accrual cost
estimates; similar and in some cases greater difficulties can be
anticipated for insurance programs.  However, experience also
indicates that the focus placed on these estimates in the budget has
led to their improvement. 

Agency capacity to generate risk-assumed cost estimates for insurance
programs will take time to develop.  To implement accrual-based
budgeting for insurance programs would require refining and adapting
the models discussed in chapter 5.  For example, an amortization
process must still be developed and tested to take the total
estimated cost to the government of pension insurance commitments
generated by OMB's model and convert it to an annual basis for
budgeting.  Modifications that are likely to be necessary to adapt
the flood and crop insurance premium rate-setting models for use in
generating risk-assumed cost estimates for the budget will also
require time and resources.  Agencies will require specialized
professional staff such as actuaries, economists, and statisticians
to develop and refine estimation models and produce the accrual cost
estimates on a regular basis.  Some agency officials we spoke with
expressed concern about their ability to generate such estimates
given current staff resources. 

In implementing credit reform, we found that agencies experienced
difficulty accurately estimating accrual-based cost estimates for
three principle reasons:  (1) future economic conditions are
uncertain, (2) the government often is the lender of last resort,
making it difficult to judge the risk, and (3) agencies' historical
data were nonexistent or unreliable.\1 These same factors will
complicate estimating risk-assumed cost estimates for insurance
programs.  In addition, differences between insurance commitments and
loan guarantees will add to the complexity of generating accrual cost
estimates for insurance programs.  Unlike most credit programs which
are limited by discretionary appropriations, the majority of
insurance programs are mandatory and thus are not limited to a
specific amount of insurance.  Estimation is therefore complicated by
the need to forecast demand for insurance--for example, the number of
farmers opting for crop insurance coverage or the amount of deposits
flowing into banks as opposed to other investment opportunities. 

Improvements in the estimation of the government's cost of insurance
extended can be expected if accrual-based budgeting is adopted for
these programs.  Advocates of accrual-based budgeting point to the
credit reform experience and argue that requiring estimates of the
full cost of insurance programs in the budget would provide the
incentive necessary to improve the quality of the estimates.  For
example, in response to credit reform reporting requirements, the
Small Business Administration (SBA), in conjunction with OMB,
recently completed an extensive analysis of its loan records dating
back to fiscal year 1983.  Prior to this study, SBA had been unable
to validate its subsidy estimates or provide reestimates as required
by credit reform.  As a part of this analysis, SBA developed a model
which allows it to take into account various loan and borrower
characteristics in its subsidy estimates.  Refined estimates of
historical default and recovery rates were used to generate fiscal
year 1997 budget estimates and reestimate prior year subsidy
estimates. 

The accuracy and reliability of estimates will also improve as a
result of refinements in methodologies and the collection of
additional data on program experience.  For example, OMB has made
several refinements to its deposit insurance model since it first
applied options pricing theory to estimate the accruing costs of the
program in 1990.  The availability of data necessary for estimating
accrual costs for some programs should improve over time, which in
turn should improve the reliability of the risk-assumed cost
estimates.  For example, in its model for estimating costs of thrift
deposit insurance, OMB has used data from small commercial banks to
estimate various parameters.  Although significant differences exist
between bank and thrift institutions, the data OMB needed for its
model do not exist for thrifts before 1990.  Data available since
1990 are biased due to the recovery of the thrift industry during
this period after the savings and loan crisis.  As thrift data
encompassing periods of both economic growth and contraction become
available, OMB will be able to incorporate them into its thrift
deposit insurance model.  Similarly, risk assessment for the National
Vaccine Injury Compensation program has been hampered by the
relatively short existence of the program.  Claims from 1989, the
program's first year of operation, have not all been settled.  The
reliability of estimates of the program's potential costs will be
difficult to judge until data from at least several years of program
operation are available. 


--------------------
\1 For additional information see Federal Credit Programs:  Agencies
Had Serious Problems Meeting Credit Reform Accounting Requirements
(GAO/AFMD-93-17, January 6, 1993). 


      AGENCIES RAISED CONCERNS
      ABOUT THEIR ABILITY TO
      IMPLEMENT ACCRUAL-BASED
      BUDGETING
-------------------------------------------------------- Chapter 7:2.2

Agencies expressed concern about the staff and system resources
necessary to implement an accrual-based budgeting approach similar to
the outlay approach described in the previous chapter.  They stated
that additional resources would likely be necessary to generate cost
estimates, collect the necessary data, and comply with new reporting
requirements.  Some agencies question whether the benefits of this
type of an approach would be worth the resources required.  Smaller
agencies expressed concern that new requirements under an accrual
system could divert resources from program operations and management. 
For example, the National Vaccine Injury Compensation program is
administered by a staff of 25, 12 of whom are medical examiners. 
According to the agency, an increase in resources would be necessary
to develop risk-assumed estimates for the budget. 

As with credit reform, agencies would be faced with significant
implementation challenges.  While the characteristics of insurance
programs may add additional complexity, experience gained from
implementing credit reform could mitigate some of the agencies'
concerns.  Some agencies that administer insurance programs also
administer credit programs.  Officials and staff at these agencies
expressed strong concerns about the expanded reporting and data
requirements of accrual-based budgeting, which were required under
credit reform.  For example, officials at the Veterans Benefits
Administration (VBA), which oversees veterans' life insurance
programs as well as several loan guaranty programs, were generally
supportive of an accrual-based budgeting approach for insurance. 
However, they stated that they would not be supportive of an
accrual-based approach modeled after the treatment of loan guarantees
under credit reform.  They said that they did not believe that the
outcome would be worth the cost needed to achieve it.  In particular,
they cited greatly expanded reporting requirements for VBA's loan
guaranty programs that required a large increase in resources needed
to prepare the budget and track all the necessary data.  Officials
and staff of the Overseas Private Investment Corporation expressed
similarly strong concerns.  These concerns were based on their
experience implementing credit reform for OPIC's loan and loan
guarantee programs. 

Some agency officials suggested that accrual-based information is
already available in the budget.  Officials at OPIC expressed the
opinion that the agency uses accrual-based estimates in the budget to
the extent appropriate.  OPIC currently obligates funds as a general
reserve for the risk inherent in its insurance activities in the
period it is estimated.  OPIC officials stressed that the tools
necessary to recognize insurance program costs already exist within
the current budget process.  Improved recognition of insurance
program costs could be achieved by requiring agencies to obligate
budget authority as a reserve when costs are estimated.  This is
essentially the aggregate budget authority approach outlined in the
previous chapter.  OPIC officials said that such an approach would
improve cost recognition for federal insurance programs without
adding to the complexity and burden of a system similar to credit
reform.  Officials of the Office of Personnel Management Retirement
and Insurance Service also stated that currently, information on the
assets and accrued liabilities of the Federal Employees' Group Life
Insurance fund is provided as part of the budget presentation.  They
said that they would be uncomfortable using accrual concepts more
extensively in the budget due to the many factors involved in
estimation and the uncertainty of such estimates. 

Lessons learned from the implementation of credit reform could help
address agency concerns about accrual-based budgeting for federal
insurance programs.  In the 5 years since credit budgeting and
accounting reforms were implemented, OMB and the Department of the
Treasury have been working with credit agencies to simplify
requirements.  Several interagency working groups have been formed to
identify ways to comply with credit reform at the lowest possible
cost, improve and standardize audit requirements, and utilize credit
reform data and concepts for internal management purposes.\2
Recommendations to streamline a number of data reporting and
reestimation requirements have been partially implemented.  The
experience and recommendations of these work groups could aid in the
development of rational procedures and reporting requirements for an
accrual-based budgeting approach for insurance programs. 


--------------------
\2 These groups include the Federal Credit Policy Working Group,
Credit Reform Committee of the Chief Financial Officers Council, and
a GAO, OMB, and Treasury task force on auditing guidance. 


      SUPPLEMENTAL REPORTING OF
      RISK-ASSUMED ESTIMATES WOULD
      ALLOW TIME TO EVALUATE A
      MORE COMPREHENSIVE
      ACCRUAL-BASED BUDGETING
      APPROACH
-------------------------------------------------------- Chapter 7:2.3

The potential for accrual-based budgeting based on estimates of the
risk assumed by the government to improve budget information and
incentives for federal insurance programs argues for its
implementation.  However, the need to build capacity to generate
risk-assumed cost estimates and the complexity of the implementation
issues involved indicate that it is not feasible at this time to
integrate risk-assumed cost estimates directly into the budget. 
Since risk-assumed estimates have not been produced and reported on a
regular basis for most insurance programs, supplemental reporting of
these estimates over a number of years could help policymakers
understand the extent and nature of the estimation uncertainty and
allow time to evaluate the feasibility of adopting a more
comprehensive accrual-based budgeting approach. 

If evaluation of the risk-assumed estimates demonstrates that
estimation has developed sufficiently so that use of risk-assumed
data in the budget will not introduce an unacceptable level of
uncertainty, policymakers could consider a second phase of
implementation--incorporating risk-assumed estimates into budget
authority.  The final phase would be the use of risk-assumed
estimates in budget authority, outlays, and the deficit. 

Supplemental reporting of risk-assumed cost estimates in the budget
would allow time to: 

  -- develop and refine estimation methodologies,

  -- assess the reliability of risk-assumed estimates,

  -- gain experience and confidence in cost measures for budget
     purposes,

  -- evaluate the feasibility of a more comprehensive accrual-based
     budgeting approach, and

  -- formulate cost-effective reporting procedures and requirements. 

During this period, policymakers should continue to draw on
information provided in audited financial statements.  As noted in
the report, financial statements can provide earlier recognition of
accruing liabilities than does the cash-based budget for insurance
commitments. 

The Government Performance and Results Act (GPRA) of 1993, which laid
out a series of steps to better integrate performance measures into
the budget, could be used as a model for incorporating accrual cost
measures in the budget.  Statutorily-required evaluation of
risk-assumed estimates would focus attention on improving cost
estimation and provide an opportunity to assess the practicality of
incorporating such estimates directly into budget authority, outlays,
and the deficit.  In the case of credit programs, estimates of
interest rate subsidies were reported in the budget for 20 years
prior to the implementation of accrual-based budgeting for those
programs.  The current capacity to generate risk-assumed estimates
for insurance programs suggests that the additional focus and time
allowed under a phased-in approach is warranted.  Experience gained
from this period would also be helpful in evaluating whether
additional control mechanisms, such as discretionary funding of
subsidies, are needed or desirable. 

An alternate strategy would be to implement accrual-based budgeting
on a program-by-program basis with consistent treatment of all
insurance programs as the ultimate goal.  Programs which have
well-developed or established estimation methodologies would
immediately be switched to an aggregate outlay accrual approach. 
Programs for which estimation methodologies do not exist or are not
widely accepted would be required to develop or refine models.  This
would allow for the benefits of accrual-based budgeting to be
realized immediately for some programs while other programs develop
the necessary estimation methodologies and expertise. 

Such a program-by-program approach has several drawbacks.  First, it
would introduce a lack of comparability among insurance programs in
the budget--perhaps even skewing their apparent relative costs--and
increase confusion about the information provided on insurance
program costs.  Second, a program-by-program approach fails to
establish a standard for new insurance programs.  Without such a
standard, the long-term expected cost of any new insurance program
may not be fully considered when the decision is made to establish it
since only the program's initial years' cash flows would be reported
in the budget.  Finally, programs for which accrual-based budgeting
holds the greatest benefits, such as deposit insurance and pension
guarantees, are the ones for which implementation will be most
difficult.  Focusing time and resources on implementing accrual-based
budgeting where the potential benefits are greatest offers the
greatest potential for improved information.  Implementing
accrual-based budgeting for those programs where the benefits are low
but not for other programs may lead to a situation in which efforts
exceed the benefits and this could make it more difficult to sustain
the effort necessary to proceed where potential benefits are
greatest. 


   TECHNICAL DESIGN ISSUES
---------------------------------------------------------- Chapter 7:3

The design and structure of an accrual-based budgeting approach for
insurance programs will be a critical factor in its acceptance and
effectiveness.  Although accrual-based budgeting for these programs
has the potential to improve budget information and incentives,
individual program characteristics, differences in the government's
commitment, and the ability to generate reasonably reliable accrual
cost estimates will require considerable effort in the design of
processes and reporting requirements.  The increased uncertainty and
complexity involved in incorporating accrual measures into the budget
heightens the need for careful consideration of technical design
issues before moving to a more comprehensive accrual-based budgeting
approach.  These issues include the treatment of loss reserves,
reestimation and funding shortfalls, previously accumulated program
deficits, and administrative costs. 


      ESTABLISHING AND MAINTAINING
      INSURANCE RESERVES
-------------------------------------------------------- Chapter 7:3.1

Under a risk-assumed accrual-based budgeting approach for insurance
programs, premium income in some years will exceed claim payments,
while in other years income will be lower than claims.  Because the
insured risks cannot be diversified or pooled over a large enough
number of participants with different potential for losses, reserves
cannot be tied to commitments made in a given year.  Instead, a
general reserve would be established based on the risk inherent in
the type of insurance provided.  This would be a major difference
between reserves for insurance programs and credit programs.  In
general, for credit programs, the large volume of loans or guarantees
issued in any single year allows for sufficient diversification of
risk and permits reserves to be set aside for each annual "book of
business." These reserves are reestimated annually over the life of
each book of business. 

Establishment of program reserves sufficient to cover the long-term
cost of the insurance extended will take time and involve significant
program funds.  If premium rates were set to cover the long-term
expected cost of the insurance extended, sufficient reserves could be
established over time.  However, until such reserve levels are
reached, appropriations or borrowing authority may be necessary to
cover claims in high loss years.  Assuming that the program's risk is
adequately estimated, premium income would be sufficient in the
long-term to repay any borrowing or appropriation and build reserves. 

Maintaining funds set aside for insurance program reserves was a
concern raised by several budget professionals.  Because insurance
reserves must be accumulated over several years and since the
reserves are not tied to any specific year's insurance commitments,
funds could potentially be diverted to fund other program priorities,
particularly given current budget constraints.  This may be more of
an issue under the budget authority approach than the outlay approach
described in the previous chapter.  Reserves held in nonbudgetary
financing accounts under credit reform have thus far been maintained
for their intended purpose.  On the other hand, officials at the
Federal Emergency Management Agency (FEMA) stated that when the flood
insurance program began to accumulate reserves in the late 1980s, the
Congress used the surplus to fund flood studies, flood plain
management, and program salaries. 


      REESTIMATION AND FUNDING
      SHORTFALLS
-------------------------------------------------------- Chapter 7:3.2

Periodic reestimation of the expected cost of the government's
insurance commitments will be necessary.  Upward reestimates of the
cost of the risk insured should be reflected in premium rates for new
insurance commitments and/or the government's subsidy.  More
complicated will be the funding of increases in the estimated costs
of outstanding insurance commitments.  Under credit reform, agencies
are given permanent, indefinite authority to cover upward reestimates
of the government's costs related to credit commitments made in prior
years.  The architects of credit reform contend that this authority
is necessary to encourage unbiased cost estimates and because some
factors that affect costs--such as the economy--are beyond an
agency's control.  Agencies are required to incorporate the factors
that prompted a reestimation into the estimates of future subsidy
costs.  Conversely, some budget experts contend that the provision of
permanent authority has created the potential for bias in original
estimates since funding for any additional cost is provided
automatically outside the appropriation process. 

The nature of the government's insurance commitment and the
sensitivity of the largest insurance programs--deposit and pension
insurance--to fluctuations in interest rates and general business
conditions may make limiting the costs of reestimates and funding
shortfalls difficult.  Most federal insurance programs are
open-ended, providing as much insurance as demanded.  Unlike most
federal credit programs in which the number of loans or loan
guarantees can be specified and funding provided, it is neither
practical nor desirable to directly limit insurance coverage--for
example, by limiting the number of children vaccinated or the vesting
of pension benefits.  Thus, in changing the budget treatment of these
programs, consideration must be given to the impact changes may have
on the programs' operations. 

The Bush administration's 1992 accrual budgeting proposal would have
required the Congress to provide a mandatory appropriation when an
insurance program's costs exceeded its available resources on an
accrual basis.  The administration argued that, given the nature of
the insurance commitments and their current budget treatment, this
would have only explicitly authorized what was implicit under
existing law.  Other methods of handling shortfalls in a program's
funding could be considered.  For example, a funding shortfall could
trigger a premium increase unless the Congress acted to implement
program reforms aimed at reducing program costs.  Alternatively,
premium increases or program coverage reductions could be implemented
if reserves fell below certain specified levels.  The impact on
program participants could be mitigated by spreading the premium
increase over several years. 

Additional control mechanisms must be carefully designed or they
could risk increasing overall costs to the government due to program
interactions.  For example, if a funding shortfall were to develop in
the flood insurance program leading to a premium increase, this could
cause participation in the program to fall.  Ultimately, diminished
participation could potentially lead to increased future costs to the
government in the form of disaster relief. 


      PREVIOUSLY ACCUMULATED
      PROGRAM DEFICITS
-------------------------------------------------------- Chapter 7:3.3

Some federal insurance programs that provide coverage for an extended
or indefinite period of time, such as the Federal Employees' Group
Life Insurance program, currently report program deficits as measured
under traditional accounting standards.  The deficit for FEGLI at the
end of 1996 was $3.4 billion.  How costs incurred prior to conversion
to accrual-based budgeting should be treated would have to be
determined. 

Several options exist for the reporting and funding of these costs. 
If estimates of the accrued costs at conversion can be made, under an
accrual outlay approach these costs could be reported as a separate
line in the program account.  If the information necessary to
estimate the future cash flows resulting from the previously accrued
costs is unavailable or if the population insured changes
significantly from year to year, a separate liquidating account could
be used.  If a liquidating account is used, funding of accrued costs
could remain on a cash basis and simply be paid as claims come due. 
Alternatively, accumulated deficits could be amortized over a
reasonable period of time and funded through appropriations or
premium increases.  These funds would be outlayed to the financing
account and paid out for claims as necessary. 


      ADMINISTRATIVE COSTS
-------------------------------------------------------- Chapter 7:3.4

The treatment of insurance programs' administrative costs will need
to take into account the intended financing of such expenses. 
Currently, most programs fund administrative costs out of premium
income, although some receive appropriated funds to cover these
expenses.  If a program is intended to be self-supporting, then an
amount to cover administrative costs should be included in the
risk-based premiums charged to participants.  Under an accrual outlay
approach, premium income would flow into the financing account and an
amount would be transferred to the program account to cover
administrative costs.  The reported cost to the government would be
zero.  If a program is not self-supporting, an appropriation to the
program account to cover administrative costs would be required. 
Administrative costs would be charged to the program account along
with any premium subsidy.  Outlays from this account would equal the
total cost to the government for the insurance extended. 


CONCLUSIONS, RECOMMENDATION, AND
AGENCY COMMENTS
============================================================ Chapter 8

To support current and future resource allocation decisions and be
useful in the formulation of fiscal policy, the federal budget needs
to be a forward-looking document that enables and encourages users to
consider the future consequences of current decisions.  As such, the
budget should clearly reflect the financial consequences of decisions
made and provide the information and incentives necessary to assess
the future implications of these choices.  The current cash-based
budget, however, generally provides incomplete and misleading
information on the cost and fiscal impact of federal insurance
programs.  The use of accrual concepts in the budget for these
programs has the potential to better inform budget choices.  However,
technical and practical challenges exist which will require careful
and deliberate consideration in the design and implementation of an
accrual-based budgeting approach for insurance programs. 

Cash-based budgeting for federal insurance programs is limited for
resource allocation and fiscal policy decisions because its focus on
single-period cash flows does not usually reflect the government's
cost at the time the decisions are made to provide insurance
coverage.  The cash-based budget may misstate the cost of the
government's insurance commitments in any particular year because the
time between receipt of program collections, the occurrence of an
insured event, and the final payment of a claim can extend over
several budget periods.  As a result, current and future resource
allocations may be distorted.  Cash budgeting also is generally not
an accurate gauge of the economic impact of federal insurance.  While
these shortcomings of cash-based budgeting exist for all insurance
programs, the degree to which cash-based information is misleading
varies significantly across programs. 

The use of accrual-based budgeting for federal insurance programs has
the potential to overcome a number of the deficiencies of cash-based
budgeting.  Accrual-based reporting would recognize the cost of the
insurance commitment when the decision is made to provide the
insurance, regardless of when cash flows occur.  This earlier
recognition of the cost of the government's commitment would (1)
allow for more accurate cost comparisons with other programs, (2)
provide an opportunity to control costs before the government is
committed to making payments, (3) build budget reserves for future
claims, and (4) better capture the timing and magnitude of the impact
of the government's actions on private economic behavior.  The degree
to which accrual-based measures would improve cost recognition in the
budget for insurance programs will vary based on the size and length
of the government's commitment, the nature of the insured risks, and
the extent to which costs are currently captured in the budget. 
Further, whatever the conceptual benefits of risk- assumed cost
measurement, the effective implementation of accrual-based budgeting
on this basis is dependent on the ability to generate reasonable
unbiased estimates of these costs. 

In the past, concerns over the limitations of cash-based budgeting
and the benefits of a shift to accrual-based budgeting have been
driven by the financial condition of the two largest
programs--deposit and pension insurance.  These two programs remain
central to the argument for accrual-based budgeting for insurance
programs.  The size of these programs in relation to total federal
spending, and therefore their potential to distort resource
allocation and fiscal policy, make the limitations of cash-based
budgeting and the benefits of accrual-based budgeting more
pronounced.  The case for using accrual-based budgeting for other
federal insurance programs varies in strength.  Their smaller size
and the degree to which cost information is currently considered by
policymakers reduce to a varying degree the extent to which
information and incentives would be improved under an accrual-based
budgeting approach. 

The ability to generate reasonable, unbiased estimates of the risk
assumed by the government is critical to the successful
implementation of accrual-based budgeting for insurance programs.  As
described in this report, the development and acceptance of
estimation methodologies varies considerably across programs.  The
characteristics of the risks insured by the federal government,
frequent program modifications, and the absence of sufficient data on
possible losses have hampered the development of risk-assumed
estimates.  The use of risk-assumed estimates in the budget will
require the refinement and adaptation of existing models and, in some
cases, the development of new methodologies.  Because risk-assumed
estimates for the various insurance programs have not been produced
and reported on a regular basis, it should be expected that agencies
will need time to develop the capacity to generate these estimates
for the budget.  During this time period, the information on
insurance losses contained in the programs' financial statements,
which are included in the budget appendix, provide policymakers with
a valuable resource in monitoring these programs. 

Improvements in estimation methodologies, available data, and
assumption specifications may, over time, lead to more accurate cost
estimates, but because insurance program costs are dependent upon
many variables, some uncertainty in the reported accrual estimates is
unavoidable.  The use of sophisticated estimation models, new budget
presentations, and the need for periodic reestimates will add
complexity to the budget process.  As a result, understanding of
budget data and the budget process may decline.  However, this
increased complexity should be assessed in relation to the adequacy
of cash-based budgeting for insurance programs.  Although cash-based
budgeting is readily understandable to policymakers and the public,
it generally provides incomplete or misleading information on
insurance program costs and thus may impair resource allocation and
fiscal policy decision-making. 

We believe that the potential benefits of an accrual-based budgeting
approach for federal insurance programs warrant continued effort in
the development of risk-assumed cost estimates.  The complexity of
the issues involved and the need to build agency capacity to generate
risk-assumed cost estimates suggest that it is not feasible to
integrate accrual-based costs directly into the budget at this time. 
Supplemental reporting of these estimates in the budget over a number
of years could help policymakers understand the extent and nature of
the estimation uncertainty and permit an evaluation of the
desirability and feasibility of adopting a more comprehensive
accrual-based approach.  The value of reporting risk-assumed
estimates was also endorsed by FASAB in accounting standards it
developed, which require disclosure of risk-assumed cost estimates as
supplemental information for insurance programs beginning with
financial statements for fiscal year 1997.  However, the Board also
recognized the difficulty of preparing reliable risk-assumed
estimates and, therefore, did not require their recognition on the
financial statements as a liability. 

Supplemental reporting of risk-assumed cost estimates in the budget
has several attractive features.  It would allow time to (1) develop
and refine estimation methodologies, (2) assess the reliability of
risk-assumed estimates, (3) formulate cost-effective reporting
procedures and requirements, (4) evaluate the feasibility of a more
comprehensive accrual-based budgeting approach, and (5) gain
experience and confidence in risk-assumed estimates.  At the same
time, the Congress and the executive branch will have had several
years of experience with credit reform, which can help inform their
efforts to apply accrual-based budgeting to insurance.  During this
period, policymakers should continue to draw on information provided
in audited financial statements. 

If risk-assumed estimates develop sufficiently so that their use in
the budget will not introduce an unacceptable level of uncertainty,
policymakers could consider incorporating risk-assumed estimates
directly into the budget.  While supplemental reporting of
risk-assumed estimates would improve the information on the cost of
insurance commitments, the actual impact on budget decisions is
uncertain since the primary budget data--budget authority and
outlays--would be unaffected.  Directly incorporating accrual-based
cost estimates in both budget authority and outlays would have the
greatest impact on the incentives provided to decisionmakers but
would also significantly increase reporting complexity and introduce
new uncertainty in reported budget data.  Between these two
approaches is one of incorporating accrual-based costs in budget
authority alone, which has fewer of the disadvantages of the full
accrual approach but also less impact on decision-making incentives. 
If an action-causing budget mechanism is desired, requiring a
discretionary appropriation for the accrual-based cost of the
government's subsidy could provide additional incentive to control
the government's cost but--by changing the locus of decisions to the
annual appropriation process--would go beyond merely changing the
reporting of program costs and change the nature of federal
insurance. 


   MATTER FOR CONGRESSIONAL
   CONSIDERATION
---------------------------------------------------------- Chapter 8:1

The Congress may wish to consider encouraging the development and
subsequent reporting of annual risk-assumed cost estimates in
conjunction with the cash-based estimates for all federal insurance
programs in the President's budget.  The Congress may also wish to
consider periodically overseeing and assessing the reliability and
usefulness of these estimates, making adjustments, and determining
whether to move toward a more comprehensive accrual-based budgeting
approach for insurance programs. 


   RECOMMENDATION
---------------------------------------------------------- Chapter 8:2

We recommend that the Director of the Office of Management and Budget
develop risk-assumed cost estimation methods for federal insurance
programs and encourage similar efforts at agencies with insurance
programs.  As they become available, the risk-assumed estimates
should be reported annually in a standardized format for all
insurance programs as supplemental information along with the
cash-based estimates.  A description of the estimation methodologies
used and significant assumptions made should be provided.  To promote
confidence in risk-assumed cost measures, the estimation models and
data should be available to all parties involved in making budget
estimates and should be subject to periodic external review.  As data
become available, OMB should undertake and report on evaluations of
the validity and reliability of the reported estimates. 


   AGENCY COMMENTS AND OUR
   EVALUATION
---------------------------------------------------------- Chapter 8:3

Officials from OMB agreed with this report's conclusion that
budgeting for insurance programs should be based on the government's
long-term expected cost of the insurance extended--the risk assumed
by the government.  Furthermore, OMB agreed that the challenges
involved in bringing risk-assumed estimates into the budget are
significant and that additional effort to improve estimation methods
is required.  OMB officials noted that they would like to pursue such
improvements but are not doing so because they do not currently have
the expertise that would be required. 

OMB officials expressed concern about GAO's use of the terms "cash"
and "accrual" in this report to describe different approaches to
budgeting for insurance programs.  GAO chose to use the term
"cash-based" because cash is the measurement basis for the amounts
shown in the budget for budget authority, obligations, outlays, and
receipts.  The estimates for these amounts generally are made in
terms of cash payments to be made or received.  Under current budget
concepts, these amounts reflect the cash flows associated with the
insurance program activities--paying claims for events that have
already occurred and collecting premiums for new commitments.  GAO
uses the term "accrual-based" to describe the use of risk-assumed
cost estimates as the basis for reporting an insurance program's
budget authority, obligations, and outlays.  Although, as OMB noted
and discussed in chapter 4 of this report, the term "accrual" can be
applied to a range of concepts and measures, GAO uses the term in the
report because it is generally understood as a basis of measuring
cost rather than cash flows. 

OMB officials also suggested that the current federal budget system
can be thought of as commitment-based or obligation-based budgeting
and that the use of risk-assumed cost estimates is consistent with
this concept.  GAO agrees that this is a useful way of thinking about
the potential changes in budgeting for insurance programs described
in this report.  As discussed in the report, using accrual-based cost
information rather than cash-based information for reporting budget
authority, obligations, and outlays could improve the recognition of
the cost of the government's commitments at the time it makes them. 
OMB officials made this same point saying that "cash does not carry
out the principle of recognizing the cost of commitments at the time
they are made."

GAO modified relevant sections of the report to clarify its
explanation of OMB's views on the budget treatment of deposit
insurance under an accrual-based approach.  According to OMB
officials, it was not OMB's intent to treat deposit insurance
differently from other insurance programs under the Bush
administration's 1992 insurance budgeting proposal.  OMB agrees with
GAO that for all programs what should be measured is the long-term
expected cost of loss-generating events less premiums collected. 
However, given the nature and complexity of deposit insurance, the
extent to which the OMB model--or any model--would be able to capture
the full long-term expected cost of the government's commitments is
open to debate.  This is due, in part, as OMB acknowledges, to the
very difficult conceptual and measurement problems associated with
accounting for rare catastrophic events, such as the savings and loan
crisis, in a risk-assessment model. 

Based on OMB officials' suggestions, GAO dropped from chapter 1 a
brief discussion of early budget commissions' recommendations
regarding accrual accounting in the federal government which was not
necessary to convey our message that the current system of budgeting
for insurance programs is deficient and may be improved with the use
of risk-assumed measures. 

OMB officials also provided a number of technical comments, which
were incorporated into the report as appropriate. 


BUDGET ACCOUNT STRUCTURE AND
REPORTING FLOWS FOR CREDIT
PROGRAMS
=========================================================== Appendix I

The Credit Reform Act set up a special budget accounting system to
record the budget information necessary to implement credit reform. 
It provides for three types of accounts--program, financing, and
liquidating--to handle credit transactions. 

Credit obligations and commitments made on or after October 1,
1991--the effective date of credit reform--use only the program and
financing accounts.  The program account receives separate
appropriations for the administrative and subsidy costs of a credit
activity and is included in budget totals.  When a direct or
guaranteed loan is disbursed, the program account pays the associated
subsidy cost for that loan to the financing account.  The financing
account, which is nonbudgetary,\1 is used to record the cash flows
associated with direct loans or loan guarantees over their lives.  It
finances loan disbursements and the payments for loan guarantee
defaults with (1) the subsidy cost payment from the program account,
(2) borrowing from the Department of the Treasury, and (3)
collections received by the government.  If subsidy cost calculations
are accurate, the financing account will break even over time as it
uses its collections to repay its Treasury borrowing.  Figure I.1
diagrams this cash flow. 

   Figure I.1:  Simplified Credit
   Reform Cash Flow

   (See figure in printed
   edition.)

Direct loans and loan guarantees made before October 1, 1991, are
reported on a cash basis in the liquidating account.  This account
continues the cash budgetary treatment used before credit reform and
has permanent, indefinite budget authority\2 to cover any losses. 
Excess balances are transferred periodically--at least annually--to
the Treasury. 

In addition to the three accounts specified in the Credit Reform Act,
OMB has directed that credit programs or activities with negative
subsidies must have special fund receipt accounts to hold receipts
generated when the program or activity shows a profit. 


--------------------
\1 Nonbudgetary accounts may appear in the budget document for
information purposes but are not included in the budget totals for
budget authority or outlays.  They do not belong in the budget
because they show only how something is financed and do not represent
the use of resources. 

\2 Permanent budget authority is available as a result of permanent
legislation and does not require annual appropriation.  Indefinite
budget authority is budget authority of an unspecified amount of
money. 


DEPOSIT INSURANCE
========================================================== Appendix II


   PURPOSE
-------------------------------------------------------- Appendix II:1

Federal deposit insurance was initiated in the 1930s to help restore
confidence in the nation's banking system after thousands of
financial institutions failed and millions of dollars in deposits
were lost during the Great Depression.  The Banking Act of 1933
established the Federal Deposit Insurance Corporation (FDIC) to
provide protection for bank depositors and to foster sound banking
practices.  A year later, the Housing Act of 1934 extended federal
deposit insurance to thrift institutions.  Federal insurance for
credit unions was established in 1970.  Federally insured deposits
are explicitly backed by the full faith and credit of the U.S. 
government. 

Federal deposit insurance is administered by two federal
agencies--FDIC and the National Credit Union Administration (NCUA). 
Pursuant to the Financial Institutions Reform, Recovery, and
Enforcement Act of 1989 (FIRREA), FDIC oversees both the Bank
Insurance Fund (BIF), which insures deposits at commercial banks and
some savings banks,\1 and the Savings Association Insurance Fund
(SAIF), which insures deposits at savings and loan institutions and
savings banks not covered by BIF.\2 The Deposit Insurance Funds Act
of 1996 (Title II, Subtitle G of Public Law 104-208) makes provisions
for the merger of BIF and SAIF into a single deposit insurance fund
effective January 1, 1999, provided that the Congress enacts
legislation to merge the bank and thrift charters and to eliminate
differences in powers and ownership structures between banks and
savings associations.  NCUA administers the Credit Union Share
Insurance Fund (CUSIF), which insures credit union accounts. 



--------------------
\1 State-chartered mutual savings banks--those owned by depositors
rather than shareholders--were included among the institutions
eligible for deposit insurance from FDIC when it was established in
1933.  In recent years, many of these savings banks have converted
from mutual ownership to stock ownership and are simply referred to
as savings banks.  Historically, these savings banks have operated
like savings and loan institutions in that they channelled savings
from individuals to make residential mortgages, but have had broader
lending and investment powers than savings and loans. 

\2 Prior to August 9, 1989, federal deposit insurance for thrift
institutions was provided through the Federal Savings and Loan
Insurance Corporation (FSLIC).  FIRREA abolished FSLIC and
transferred its assets, liabilities, and contracts to a newly created
FSLIC Resolution Fund and established SAIF as the new thrift
insurance fund.  FDIC was designated the administrator of both funds. 
In addition, FIRREA created the Resolution Trust Corporation to
resolve all troubled institutions placed into conservatorship or
receivership from January 1, 1989, through August 8, 1992.  This
period was later extended to June 30, 1995. 


   BANK AND THRIFT DEPOSIT
   INSURANCE
-------------------------------------------------------- Appendix II:2

Budget Accounts:  Bank Insurance Fund (BIF)
(51-4064-0-3-373)
Savings Association Insurance Fund (SAIF)
(51-4066-0-3-373)

Agency:  Federal Deposit Insurance Corporation (FDIC)


      COVERAGE
------------------------------------------------------ Appendix II:2.1

Domestic deposits in commercial banks, savings banks, savings
associations, and other thrift institutions are insured up to
$100,000 per account.  The insured amount has been raised six times
since 1934 with the current limit of $100,000 set by the Depository
Institutions Deregulation and Monetary Control Act of 1980.  At the
end of 1995, over $1.9 trillion in deposits at approximately 10,000
commercial banks and savings banks were insured by BIF, while SAIF
insured more than $700 billion in deposits at approximately 1,700
thrift institutions. 


      ELIGIBILITY REQUIREMENTS
------------------------------------------------------ Appendix II:2.2

Banks and savings institutions can only conduct business if they
obtain a charter (license to operate) from either the federal or
state government.  The laws and regulations underlying charters
specify the activities in which institutions may engage and the
supervisory requirements they must meet.  Federal charters are
granted by two offices within the Department of the Treasury.  The
Office of the Comptroller of the Currency (OCC) is responsible for
chartering federal (national) banks and the Office of Thrift
Supervision (OTS) approves charters for federal savings associations. 

In evaluating an application to organize a new bank, OCC considers
the institution's earning prospects, the adequacy of its capital, its
anticipated community services, the ability of its management, and
the safety and soundness of intended operations.  In issuing charters
to operate thrift institutions, OTS is required to give primary
consideration to the best practices of thrift institutions in the
United States, which generally means the same factors applied by OCC
for banks.  Chartering requirements for state banks and savings
associations vary by state.  However, most if not all states now
require that new banks and thrifts obtain federal deposit insurance,
which effectively provides FDIC with veto power over the granting of
state charters. 

In extending deposit insurance, FDIC is required by law to consider
(1) the financial history and condition of the depository
institution, (2) the adequacy of its capital, (3) the future earnings
prospects of the institution, (4) the general character and fitness
of its management, (5) the risk presented by the institution to the
insurance fund, (6) the needs of the community to be served, and (7)
whether the institution's corporate powers are consistent with the
purposes of the Federal Deposit Insurance Act.  Successful applicants
for national charters qualify immediately for federal deposit
insurance. 

All federally insured banks and thrifts are subject to federal
supervision and examination whether they are state or federally
chartered.  Federal oversight is split among four regulatory agencies
based on the type of institution.  OCC supervises all national banks. 
OTS serves as the primary regulator for thrift institutions and
thrift holding companies.  The Federal Reserve Board has oversight
authority for state-chartered banks that are members of the Federal
Reserve System (FRS) and bank holding companies.  FDIC is the primary
federal regulator of state-chartered banks that are not members of
FRS.  By law, federal regulators are required to conduct annual
on-site examinations of all federally insured institutions except for
certain well-managed and financially strong institutions with assets
of less than $250 million, which must be examined every 18 months. 


      PROGRAM FINANCING
------------------------------------------------------ Appendix II:2.3

Federal deposit insurance for banks and thrift institutions is
financed from annual premium assessments.  Other sources of funds
include interest earned on investments in U.S.  Treasury obligations,
income from the management and disposition of assets acquired from
failed institutions, and U.S.  Treasury and Federal Financing Bank
(FFB) borrowing.\3 Specifically, under the Federal Deposit Insurance
Corporation Improvement Act of 1991 (FDICIA), FDIC is authorized to
borrow up to $30 billion from the Treasury to cover BIF and SAIF
losses.  The Omnibus Budget Reconciliation Act of 1990 (OBRA'90)
authorized FDIC to borrow funds from FFB to finance the acquisition
of failed bank and thrift assets.  Additional sources of financing
available to SAIF include (1) borrowing from the Federal Home Loan
Banks, (2) up to $8 billion in Treasury funds for losses sustained by
SAIF in fiscal years 1994 through 1998, contingent upon
appropriations, and (3) unused funds appropriated to the Resolution
Trust Corporation (RTC) for 2 years following the termination of the
RTC. 

OBRA'90 removed the caps on deposit insurance premium rate increases
and authorized FDIC to set BIF and SAIF premium rates semiannually. 
In 1991, a provision of FDICIA required FDIC to implement a
risk-related premium system and to build BIF and SAIF reserves to a
minimum of 1.25 percent of total insured deposits within 15 years. 
BIF reached the statutorily required reserve level in May of 1995.  A
special one-time assessment of 68 cents per $100 of deposits in
SAIF-insured institutions was mandated by the Deposit Insurance Funds
Act of 1996 to fund SAIF to the required reserve level.\4

Since January 1993, FDIC has assessed risk-related insurance
premiums.  FDIC calculates risk-related assessments for individual
banks and thrifts by placing each institution in one of nine risk
categories based on capital ratios and supervisory ratings.  Under
this system, institutions will pay assessment rates in 1997 of
between 0 and 27 cents per $100 of insured domestic deposits based on
risk category.  The average annual assessment rate for BIF-insured
institutions will be 0.17 cents per $100 insured deposits and 0.6
cents per $100 insured deposits for SAIF members. 


--------------------
\3 Funding to resolve the savings and loan crisis was provided
primarily from taxpayers in the form of general fund appropriations. 
We estimate the total direct and indirect cost of resolving the
savings and loan crisis to be $160 billion, of which approximately
$132 billion (83 percent) will have been paid using taxpayer funding
sources.  For a detailed analysis, see Financial Audit:  Resolution
Trust Corporation's 1995 and 1994 Financial Statements
(GAO/AIMD-96-123, July 2, 1996). 

\4 The act also spread responsibility for interest payments on bonds
held by the Financing Corporation (FICO) that were issued in 1987
through 1988 to finance the resolution of failed thrift institutions. 
Previously, FICO interest payments were borne entirely by SAIF
assessments.  Beginning January 1, 1997, BIF-insured institutions
will pay 1.29 cents and SAIF-insured institutions will pay 6.44 cents
per $100 of covered deposits.  In the year 2000, all banks and
thrifts will pay 2.43 cents per $100 of deposits. 


      CURRENT BUDGET TREATMENT
------------------------------------------------------ Appendix II:2.4

BIF and SAIF are reported separately but treated similarly in the
budget.  All administrative and insurance expenses as well as revenue
from premium assessments, interest earnings, asset sales, and other
fees flow through a single budget account for each fund.  All
expenses of these accounts, including administrative expenses and the
expenses of FDIC's Office of Inspector General, are classified as
mandatory under the Budget Enforcement Act of 1990 (BEA).  However,
deposit insurance revenue and spending are exempt from BEA
pay-as-you-go restrictions.  As such, any additional spending
necessary to maintain the safety and soundness of the government's
deposit insurance commitment does not need to be offset by tax
increases or spending cuts in other direct spending.  Similarly,
increases in insurance premiums or other deposit insurance
collections cannot be used to offset increased spending for other
mandatory programs.  Budgeted and actual bank insurance outlays for
fiscal years 1973 through 1996 are shown in figure II.1.  Figure II.2
displays comparable information for thrift insurance. 

   Figure II.1:  Bank Deposit
   Insurance Budget Estimates
   Versus Actual Outlays, Fiscal
   Years 1973-1996

   (See figure in printed
   edition.)

   Figure II.2:  Thrift Deposit
   Insurance Budget Estimates
   Versus Actual Outlays, Fiscal
   Years 1973-1996

   (See figure in printed
   edition.)


   CREDIT UNION SHARE INSURANCE
-------------------------------------------------------- Appendix II:3

Budget Account:  Credit Union Share Insurance Fund (CUSIF)
(25-4468-0-3-373)

Agency:  National Credit Union Administration (NCUA)


      COVERAGE
------------------------------------------------------ Appendix II:3.1

The National Credit Union Administration insures members' shares
(deposits) up to $100,000 per shareholder account in federal and
state-chartered credit unions that qualify for insurance.  In fiscal
year 1995, NCUA insured $266 billion in deposits in approximately
12,000 credit unions. 


      ELIGIBILITY REQUIREMENTS
------------------------------------------------------ Appendix II:3.2

All federally chartered credit unions are required to be federally
insured.  Most states prohibit credit unions from operating without
any insurance but allow nonfederally backed private insurance in lieu
of federal insurance.  To be eligible for federal insurance, each
applicant must be approved by the NCUA board and agree to comply with
all statutory and regulatory requirements.  These requirements
include:  the reporting of financial and statistical information on a
quarterly or semiannually basis to NCUA, periodic examination as
determined by NCUA, and the payment of premium assessments.  The NCUA
board assesses each application for share insurance based on (1) the
history, financial condition, and management policies of the
applicant; (2) the economic advisability of insuring the applicant
without undue risk to the fund; (3) the general character and fitness
of the applicant's management; (4) the convenience and needs of the
credit union's members to be served; and (5) whether the applicant is
a cooperative association organized for the purpose of promoting
thrift among its members and creating a source of credit for
provident or productive purposes. 


      PROGRAM FINANCING
------------------------------------------------------ Appendix II:3.3

CUSIF is structured to be entirely self-supporting through monies
provided by member credit unions.  The insurance program is financed
primarily from insurance premiums that may be assessed annually and
from mandatory credit union deposits in the insurance fund.  The
assessment rate is set in statute (Public Law 91-468) at one-twelfth
of 1 percent of a credit union's total member share accounts.  Title
VIII of Public Law 98-369 (July 18, 1984), which provided for the
capitalization of the insurance fund, requires each insured credit
union to deposit and maintain in the insurance fund an amount equal
to 1 percent of its insured member accounts.  Other sources of funds
include income generated from the investment of monies received from
the insured credit unions and funds received from the management and
disposition of assets acquired from failed institutions.  CUSIF is
authorized to borrow $100 million from the Treasury at any time for
the purpose of carrying out the insurance program. 

Public Law 98-369 also requires that the CUSIF balance be maintained
at a normal operating level to be determined by the NCUA board.  The
board has determined this level to range from 1.25 percent to 1.3
percent of insured shares.  Since the recapitalization of the
insurance fund in 1985, credit unions have been assessed premiums in
only 1 year, 1992.  In 1996, CUSIF paid a $106 million dividend to
federally insured credit unions because the fund balance exceeded the
1.3 percent reserve requirement. 


      CURRENT BUDGET TREATMENT
------------------------------------------------------ Appendix II:3.4

All administrative and insurance expenses as well as revenue from
assessments, investment earnings, asset sales, and other fees flow
through a single budget account, the Credit Union Share Insurance
Fund.  All expenses of this account including administrative expenses
are classified as mandatory under BEA.  Like the funding provided
through BIF and SAIF, cost resulting from the government's current
insurance guarantee is exempt from BEA controls.  The CUSIF account
reimburses NCUA's Operating Fund budget account for its share of the
agency's administrative costs.  The reimbursement percentage, which
is reviewed and adjusted periodically, is currently 50 percent. 
Budgeted and actual credit union insurance outlays for fiscal years
1973 through 1996 are displayed in figure II.3. 

   Figure II.3:  Credit Union
   Share Insurance Budget
   Estimates Versus Actual
   Outlays, Fiscal Years 1973-1996

   (See figure in printed
   edition.)


   METHODS FOR ASSESSING RISK
   ASSUMED UNDER DEPOSIT INSURANCE
-------------------------------------------------------- Appendix II:4

Numerous methodologies have been developed that attempt to forecast
the future financial condition of the bank and thrift industries
which affects the condition of the deposit insurance funds. 
Subsequent sections summarize the following major types of models: 

  -- OMB's options pricing,

  -- actuarial,

  -- transition matrix,

  -- asset markdown,

  -- proportional hazards, and

  -- pro forma projection. 

Only one of the methodologies--options pricing--provides accrued cost
estimates for the funds in its present form.  In general, the
alternative models provide forecasts of future bank and thrift
insolvencies which can be used to estimate accrued costs.  However,
the options pricing approach developed by OMB estimates the accruing
cost to the insurance funds by drawing the analogy between the price
of a put option and actuarially fair insurance premiums. 

The methodologies described in this section have been developed
and/or used by various federal agencies and private forecasters.  For
example, OMB has been using its options-based methodology for several
years to project the condition of the bank and thrift insurance funds
and to develop budget estimates.  OMB does not rely exclusively on
the estimates generated by its options approach.  It consults with
FDIC, Treasury, and the Federal Reserve on their near and long-term
projections.  FDIC, in turn, relies on several methodologies in
preparing loss estimates for purposes of updating the
recapitalization schedules of the insurance funds and the agency's
financial statements.  Methodologies used by FDIC include actuarial
models, pro forma analyses, and proportional hazard models. 

While the methodologies differ considerably in their approach and
assumptions, they all rely to some extent on financial data supplied
by institutions to their federal regulator (call report data).  We
have reported on several occasions that these data cannot always be
depended upon to provide an accurate picture of the value of an
institution's assets.\5

Because of this, most loss estimation methodologies adjust call
report data to attempt to approximate the economic or market value of
an institution's assets.  Based on this estimated asset value and the
value of an institution's deposit and other liabilities, the solvency
of an institution is calculated as are resulting losses to the
government insurance funds from failed institutions.  In a prior
review of loss estimation methodologies, we emphasized that
long-range estimates of bank failures and their impact on the
insurance fund is a highly subjective process dependent on many
variables, such as interest rates, which are extremely difficult to
predict.\6


--------------------
\5 For additional discussion see Failed Banks:  Accounting Reforms
Urgently Needed (GAO/AFMD-91-43, April 22, 1991), Credit Unions: 
Reforms for Ensuring Future Soundness (GAO/GGD-91-85, July 10, 1991),
and Credit Unions:  Both Industry and Insurance Fund Appear
Financially Sound (GAO/T-GGD-94-142, September 29, 1994). 

\6 Bank Insurance Fund:  Review of Loss Estimation Methodologies
(GAO/AIMD-94-48, December 9, 1993). 


      OMB'S OPTIONS PRICING MODEL
------------------------------------------------------ Appendix II:4.1

OMB has developed a model based on options pricing theory\7 to
estimate the government's cost of deposit insurance.  In the options
pricing framework, the government's cost of deposit insurance is
dependent on the probability that a financial institution will
exercise its option to transfer its deposit liabilities to the
government.  This occurs only if the value of the institution's
assets is lower than the value of its liabilities, at which point the
institution is technically insolvent.  In other words, the cost of
the government's deposit insurance commitment is directly related to
the probability that an institution will go bankrupt.  Using options
pricing techniques, OMB is able to estimate the full risk-based
premium of deposit insurance based on a relatively few variables. 
These are (1) the current value of the institutions' assets, (2) the
volatility in the value of the assets, (3) the ratio of the
institutions' assets to liabilities in the future--the exercise price
of the option, (4) the time to expiration of the option, and (5) the
risk-free interest rate of corresponding duration.  The government's
cost of the deposit insurance commitment is then calculated as the
difference between the actuarially fair premium (price of the option)
and the actual premiums collected.\8

The general application of options pricing theory presumes that
current asset values can be readily observed and measured.  In
practice, most common applications of options pricing models rely on
asset market prices to compute the value of the option.  However,
there is no active market for the assets--loans--of financial
institutions.  Some researchers have used stock prices of large
publicly traded institutions, however, the stock of many insured
institutions is not actively traded.  Therefore, in order to estimate
the government's cost of insuring all institutions, OMB must first
infer the market value of individual institutions from call report
data. 

OMB uses call report data to estimate the market value of each
depository institution with assets over $100 million and subsidiaries
of bank and thrift holding companies with assets over $500 million. 
The market value of each institution's assets is estimated by
capitalizing\9 its adjusted gross earnings net of taxes and interest
earned but not collected.  OMB makes two adjustments to reported
earnings before estimating the market value.  First, because an
institution's provision for loan losses tends to be erratic and
subject to lags, OMB substitutes an estimated loss provision for each
bank based on recent loss experience of similar institutions. 
Second, earnings are adjusted to account for the tendency of very
high or low reported earnings to revert toward the industry's
long-term mean rate of return.  OMB uses a simple regression model,
with estimated coefficients for four classes of banks, to forecast
future cash flows.  The resulting estimate of an institution's
current asset value, less the face value of its deposits and other
liabilities, provides an estimate of net worth. 

OMB uses the estimated net worth of approximately 3,000 insured
institutions as input to its options pricing model.  As mentioned
above, the estimated current and future net worth of an institution
are key variables in determining the value of the option--and
ultimately the government's cost of deposit insurance.  For future
periods, OMB uses a stochastic\10 simulation process to forecast the
future net worth of individual institutions.  The simulation does not
project the performance of actual institutions.  Instead, OMB takes
the actual measured distribution of economic net worth of the banking
or thrift industries and breaks up the distribution into 8,000
equal-sized fictitious institutions.  This is the statistical
equivalent of averaging many independent projections. 

Over a chosen simulation period, such as a 5- or 7-year budget
horizon, the financial condition of some institutions improves and
others declines.  Assumptions about the volatility of asset earnings
and expected trends in average industry earnings are significant
determinants of the value of the option and simulated flow of costs. 
OMB assumes constant volatility of assets across banks of the same
size and stability over the simulation period based on the experience
of a sample of banks from 1984 through 1990.  The simulation yields
annual estimates of the volume of financial institution assets that
will be closed if the regulators continue to follow recent or other
specified closure behavior.  Closure is defined in terms of the asset
to liability ratio of an institution.  The government's cost of
deposit insurance is calculated as losses resulting from newly
insolvent institutions, additional deterioration in previously
identified insolvent institutions, and increases in the risk of
failure of solvent firms--offset by improvements in the financial
condition of any institutions.  These costs less premiums collected
constitute the net cost to the government. 


--------------------
\7 See figure 5.1 for a description of options pricing theory. 

\8 For a detailed description of the model, see Richard L. 
Cooperstein, George G.  Pennacchi and F.  Stevens Redburn, "The
Aggregate Cost of Deposit Insurance:  A Multiperiod Analysis,"
Journal of Financial Intermediation, vol.  4, no.  3 (July 1995), pp. 
242-271. 

\9 Capitalization refers to a method of valuing a firm based on the
cash it generates.  First, future earnings over a reasonable number
of years must be estimated.  Second, estimated earnings are adjusted
for non-cash consuming or generating items, such as depreciation, to
determine annual cash flows.  Third, capital outlays required to
support the current level of earnings are estimated.  The resulting
cash flows are discounted at an appropriate interest rate.  The
resulting net present value represents an estimate of the value of
the firm. 

\10 A stochastic or random process is one in which only chance
factors determine the particular outcome of a single run through the
process or trial.  The possible outcomes are known in advance, but
not the exact outcome of any one trial.  The process does have some
regularity, which allows a probability to be assigned to possible
outcomes. 


      ACTUARIAL MODELS
------------------------------------------------------ Appendix II:4.2

Actuarial models use historical frequencies of resolution for
categories of banks as an estimate of the probability of resolution
for the current population of banks in some future period.  This
approach assumes that recent failure rates will continue in the
future.  Actuarial models do not identify specific banks that are
likely to fail nor do they provide the specific timing of
resolutions.  These models are a top-down descriptive statistical
approach to estimating future resolutions. 

Using an actuarial approach, a table is constructed to categorize the
entire population of banks or thrifts based on characteristics such
as size (value of assets), capitalization ratios, nonperforming
assets, loan loss reserves, and geographic location.  The probability
of resolution over a finite time period for each category of
institutions is estimated from the incidence of resolution during a
historical period of similar length.  A separate loss distribution
function is used to make annual projections of resolved bank or
thrift assets. 


      TRANSITION MATRIX MODELS
------------------------------------------------------ Appendix II:4.3

Transition matrix models are a variation of actuarial models that use
the relative incidence of the movement of the number and assets of
banks or thrifts across CAMEL\11

categories to determine subsequent year CAMEL ratings and
resolutions.  Like actuarial models, transition matrix models do not
identify individual bank or thrift failures but simply project a
future distribution of failed institutions. 


--------------------
\11 CAMEL ratings are a numerical index of financial condition used
by bank regulators based on on-site examinations and examiners'
assessment of risk.  The five components of a CAMEL rating are
capital adequacy, asset quality, management practices, earnings, and
liquidity.  CAMEL ratings range from 1 for financially sound banks to
5 for unsound banks. 


      ASSET MARKDOWN MODELS
------------------------------------------------------ Appendix II:4.4

Asset markdown models attempt to estimate the net worth of every
institution in the industry.  This is accomplished by (1) using asset
deflators to explicitly value assets and equity of individual banks
or (2) discounting cash flows after adjusting for potential loan
losses, nonperforming loans, expenses, and reserves.  Banks with
negative net worth based on the estimated market value of assets and
equity are identified.  The results are used to produce an estimate
of embedded (future) losses to the insurance fund.  The
data-intensive nature of this approach limits its application. 


      PROPORTIONAL HAZARDS MODELS
------------------------------------------------------ Appendix II:4.5

Proportional hazards models are based on the premise that certain
financial and economic variables determine a financial institution's
risk of failure and thus affect its time-to-failure.  This type of an
approach attempts to estimate the time-to-failure using bank
attributes and other variables in a regression model.  The model
generates the probability that a bank will survive beyond any given
time period.  A probability distribution of an institution's expected
life can be plotted for a range of future time periods.  A
proportional hazards model developed by FDIC analysts used seven bank
and economic variables.\12 These variables were two regulatory
indicators of a bank's financial stress and measures of a bank's
capital level, riskiness of assets, asset quality relative to
reserves, profitability, and a leading economic indicator--the annual
percentage change in housing permits at the state level. 


--------------------
\12 For more information, see Gary S.  Fissel, "Risk Measurement,
Actuarially-Fair Deposit Insurance Premiums and the FDIC's
Risk-Related Premium System," FDIC Banking Review, vol.  7, no.  1
(Spring/Summer 1994), pp.  16-27. 


      PRO FORMA PROJECTION MODELS
------------------------------------------------------ Appendix II:4.6

This type of model projects individual bank or thrift income and
capital based on current conditions.  An institution's earnings are
based on returns to current earning assets.  Assumptions are made
regarding the movement of nonperforming assets based on the expected
macroeconomic conditions.  Liabilities can be modeled under
optimistic or pessimistic scenarios.  Assumptions are also made about
earnings retention and nonperforming loans and charge-offs. 


   IMPLEMENTATION CONSIDERATIONS
   FOR DEPOSIT INSURANCE
-------------------------------------------------------- Appendix II:5


      ADEQUACY OF CURRENT BUDGET
      REPORTING
------------------------------------------------------ Appendix II:5.1

The cash-based budget's focus on cash flows can make deposit
insurance look profitable when costs are rising or look more costly
when there has been no change in actual costs.  This is because: 

  -- Cash-based budget reporting does not recognize the cost of a
     failed institution until cash is required to pay off depositors,
     which may not occur until months or years after an institution
     becomes insolvent. 

  -- The cost of new or growing deposit insurance commitments are not
     recognized in the budget when they occur. 

  -- The government's cost for deposit insurance is obscured by the
     recording of financing (working capital) transactions.  When the
     government closes an institution and pays its depositors, it
     acquires assets which are subsequently sold.  The cash-based
     budget records the cash needed to acquire the assets as an
     outlay and the proceeds of asset sales as offsetting
     collections.  The government's cost--the difference between what
     it paid for the assets and what it was able to sell them for--is
     not shown in the budget. 


      ISSUES IN IMPLEMENTING AN
      ACCRUAL-BASED BUDGETING
      APPROACH FOR DEPOSIT
      INSURANCE
------------------------------------------------------ Appendix II:5.2

  -- Estimates of future bank failures and their impact on the
     deposit insurance funds are inherently uncertain due to their
     dependence on uncertain economic conditions, firm behavior, and
     industry changes. 

  -- Methodologies currently available for estimating the accrual
     cost of deposit insurance are generally based on recent program
     experience and may not capture the long-term risk to the
     government. 

  -- The uncertainty inherent in accrual cost estimates and
     reestimates for deposit insurance could potentially introduce
     new volatility in the annual reported program cost and the
     budget deficit. 


PENSION INSURANCE
========================================================= Appendix III

Budget Account:  Pension Benefit Guaranty Corporation Fund
(16-4204-0-3-601)

Agency:  Pension Benefit Guaranty Corporation (PBGC)


      PURPOSE
----------------------------------------------------- Appendix III:0.1

The Pension Benefit Guaranty Corporation (PBGC) was established by
Title IV of the Employee Retirement Income Security Act of 1974
(ERISA) to protect the retirement income of participants and
beneficiaries covered by private sector, defined benefit pension
plans.  These plans provide a specified monthly benefit at
retirement, usually based on salary or a stated dollar amount and
years of service.  PBGC usually assumes responsibility for paying
insured retirement benefits when a plan sponsor experiences severe
financial stress and cannot pay all promised benefits.  This
generally occurs only when an employer is being liquidated or, if
after filing for bankruptcy protection, it is determined that
termination of the pension plan is necessary for the company's
survival.  Under certain circumstances, PBGC can also terminate a
plan and assume responsibility for plan benefits if, for example, the
plan fails to meet minimum funding requirements or cannot pay current
benefits. 


      COVERAGE
----------------------------------------------------- Appendix III:0.2

At the end of fiscal year 1996, PBGC insured the pension benefits of
nearly 42 million workers and retirees in approximately 50,000
private sector, defined benefit pension plans.  Defined contribution
plans, such as 401(k) plans, are not insured.  PBGC administers two
legally distinct programs, one for pension plans sponsored by a
single employer and one for plans to which several companies make
payments.  These multiemployer plans are collectively bargained by
industry or trade groups and generally cover a particular
geographical area.  Multiemployer plans account for approximately
2,000 of the 50,000 plans insured by PBGC. 

PBGC guarantees the basic monthly retirement benefit of insured
workers.  The guarantee includes benefits beginning at normal
retirement age (usually 65), certain early retirement and disability
benefits, and benefits for survivors of deceased plan participants. 
PBGC guarantees only vested benefits.\1 In addition, ERISA sets a
limit for guaranteed benefits based on a formula which is adjusted
periodically for growth in wages.  For pension plans taken over by
PBGC in 1997, the maximum annual pension guarantee is $33,136.  Once
the insured benefit amount is determined, it is not subsequently
adjusted for inflation.  In fiscal year 1996, PBGC paid benefits of
$792 million to approximately 200,000 retirees. 


--------------------
\1 Vested benefits are those to which an employee is entitled as the
result of having met certain requirements, such as length of
employment, even if the employee ceases employment prior to
retirement. 


      ELIGIBILITY REQUIREMENTS
----------------------------------------------------- Appendix III:0.3

Under ERISA, employers who provide defined benefit pension plans must
meet minimum standards and provide prudent management of pension
funds.  The standards specify who must be covered, how long a person
has to work to be eligible for benefits, and how much money must be
contributed annually by the employer to the plan.  ERISA excludes
certain defined benefit plans from coverage.  These include
professional service plans that cover fewer than 26 participants,
plans of fraternal societies financed entirely by member
contributions, and plans maintained exclusively for substantial
owners of a business. 


      PROGRAM FINANCING
----------------------------------------------------- Appendix III:0.4

PBGC is required by ERISA to be self-financing on an actuarial basis. 
PBGC receives funds from premiums collected from ongoing pension
plans, investment income, terminated plan assets, and recoveries from
sponsors of terminated plans.  PBGC is also authorized to borrow up
to $100 million from the U.S.  Treasury in the event that its
resources are insufficient to pay guaranteed benefits.  At the end of
fiscal year 1996, PBGC's financial statements reflected a surplus
$993 million.  This was the first time since it was created that the
agency has posted a surplus. 

Annual premiums for the single-employer programs are $19 per
participant for a fully funded plan.  Underfunded single-employer
plans pay an additional variable rate of $9 per participant for each
$1,000 of unfunded vested benefits.  Prior to passage of the
Retirement Protection Act of 1994 (RPA) the variable rate was capped
at $53 per participant.  The cap is being phased out under RPA over a
3-year period which began July 1, 1994.  The multiemployer plan
premium is $2.60 per participant. 


      CURRENT BUDGET TREATMENT
----------------------------------------------------- Appendix III:0.5

Prior to 1981, PBGC was treated as an off-budget federal entity and
as such its transactions were excluded from the budget totals. 
Beginning in 1981, Public Law 96-364 required that PBGC's receipts
and disbursements be included in the budget.  The on-budget
activities of PBGC are reported in a single account.  Outlays from
this account are classified as mandatory under BEA with the exception
of administrative expenses, which are discretionary.  Figure III.1
shows the budgeted and actual outlays of PBGC's on-budget account
since 1981. 

   Figure III.1:  Pension Benefit
   Guaranty Corporation Budget
   Estimates Versus Actual
   Outlays, Fiscal Years 1981-1996

   (See figure in printed
   edition.)

The budget treatment of PBGC is complicated by the use of a second
account for some activities which is not included in the federal
budget.  This account records the assets and liabilities that PBGC
acquires from terminated plans.  As a result, the budget only reports
PBGC's net annual cash flows between its on-budget account and all
other entities--including the other PBGC account.  It does not
provide information on liabilities PBGC incurs when it takes over an
underfunded plan or other changes in PBGC's assets and liabilities. 


   METHODS OF ASSESSING RISK
   ASSUMED FOR PENSION INSURANCE
------------------------------------------------------- Appendix III:1

Calculation of the risk-assumed cost of the government's pension
insurance has focused on two methods.  The first, developed by OMB
staff, uses a mathematical model based on options pricing theory. 
The second method is a simulation approach based on the research of
two economists at the Federal Reserve Bank of New York and developed
and refined by PBGC staff. 


      OMB'S OPTIONS PRICING MODEL
----------------------------------------------------- Appendix III:1.1

As part of the Bush administration's 1992 proposal to implement
accrual-based budgeting for federal insurance programs, OMB staff
developed an options pricing approach to estimate PBGC's risk-assumed
liability and annual accruing cost.  Options pricing is commonly used
by financial markets to estimate the future value of various types of
assets.  A brief overview of options pricing theory is provided in
figure 5.1.  In OMB's model, the pension guarantee is treated as
giving the owners of a firm the option to transfer their pension plan
liabilities to PBGC when the firm becomes insolvent.  However, since
the cost to the government is contingent on the financial conditions
of both the pension plan and the plan's sponsoring firm, OMB's model
specifies stochastic processes for projecting the value of the assets
and liabilities of the pension plan and the sponsoring firm.\2

OMB's model uses actual stock and financial data on sponsoring firms
and their pension plans.  Data used include company stock price
information and pension plan assets and liabilities.  Data from
approximately 1,800 individual companies representing approximately
70 percent of the single-employer pension liability insured by PBGC
is used.  OMB's model is limited to publicly traded firms that
sponsor defined benefit pension plans.  Assumptions are made about
the growth rates of pension assets and liabilities, PBGC recovery
rates from bankrupt firms, and plan participation rates.  In
addition, a number of parameters are estimated based on recent
experience to characterize changes in the asset-to-liability ratios
of the firms and pension plans. 

Using data on the current financial conditions of pension plans, plan
sponsors, and information on recently observed changes, OMB's model
solves a series of simultaneous differential equations to estimate
the probability of bankruptcy and plan underfunding for individual
insured pension plans.  The model then calculates the cost of PBGC's
potential losses resulting from the projected terminated underfunded
plans and the value of the potential insurance premiums that PBGC
will collect.  Together, these calculations provide the net cost to
the government of the pension guarantee or subsidy extended to the
pension plans in the model.  A separate amortization schedule is used
to spread this cost on an annual basis based on the increase in
vested guaranteed benefits. 


--------------------
\2 For a detailed description of the model, see George G.  Pennacchi
and Christopher M.  Lewis, "The Value of Pension Benefit Guaranty
Corporation Insurance," Journal of Money, Credit, and Banking, vol. 
26, no.  3 (August 1994, part 2), pp.  735-753. 


      PBGC'S PENSION INSURANCE
      MODELING SYSTEM
----------------------------------------------------- Appendix III:1.2

Around the time that OMB unveiled its model, PBGC began developing a
simulation approach to forecast its exposure to future claims under a
wide range of possible future economic conditions.  PBGC's efforts
built upon earlier research of economists at the Federal Reserve Bank
of New York.\3 The model, which PBGC calls the Pension Insurance
Modeling System (PIMS), is designed to simulate pension funding and
bankruptcy rates over a 30-year period.  The model, which is still
under development, generates estimates of average expected claims and
probability measures of the uncertainty surrounding the estimates
under various economic and policy scenarios.\4

PBGC expects to use this information to analyze its exposure to
future losses and evaluate various legislative changes in the pension
insurance program and related laws. 

The heart of PIMS is the simulation of a series of dynamic
relationships that characterize the growth of firm assets and
liabilities, the number of plan participants, the assets and
liabilities of the pension plan, and the normal cost associated with
the plan.  The pension plan and the sponsoring firm are treated as
separate but related entities.  The future financial condition of the
firm and plan are interdependent and also dependent on current
financial conditions, legal and regulatory restrictions, and
uncertainty of future economic conditions.  Stochastic variables are
used to model this uncertainty.  The model simulates these dynamic
relationships over a specified period of time.  In order to forecast
future expected claims, the model is run a large number of times to
produce a distribution of possible outcomes.  This provides an
estimate of the average expected future claims and a measure of the
probability that actual claims will be within a certain range around
the estimate. 

PIMS is data-intensive, using numerous attributes of individual
pension plans and sponsoring firms.  The model is run using a
stratified sample of firms.  PBGC currently has data on 266 plans
representing approximately 50 percent of the government's liability
and 50 percent of plan underfunding in PIMS.  Model results can be
extrapolated to account for the entire population of plan sponsors. 
For each plan in PIMS, Internal Revenue Service funding requirements
are modeled.  The probability of firm bankruptcy is also modeled and
is dependent upon several factors, including company size, industry,
and firm characteristics.  All parameters in the model are
empirically based.  PBGC, working with outside reviewers, has been
conducting extensive testing of PIMS over the past year. 


--------------------
\3 Arturo Estrella and Beverly Hirtle, "Estimating the Funding Gap of
the Pension Benefit Guaranty Corporation," Federal Reserve Bank of
New York Quarterly Review, vol.  13, no.  3 (Autumn 1988), pp. 
45-59. 

\4 For a detailed discussion, see Richard Ippolito and William Ross
eds., Pension Insurance Modeling System, draft report presented at
the Pension Research Council, PIMS Technical Review Panel, at the
Wharton School of the University of Pennsylvania, November 1996. 


   IMPLEMENTATION CONSIDERATIONS
   FOR PENSION INSURANCE
------------------------------------------------------- Appendix III:2


      ADEQUACY OF CURRENT BUDGET
      REPORTING
----------------------------------------------------- Appendix III:2.1

  -- PBGC's annual net cash flows reported in the budget reduce the
     annual budget deficit while its growing long-term commitment to
     pay pension benefits has no effect on the deficit. 

  -- Liabilities from terminated, underfunded pension plans taken
     over by PBGC are not recognized in the budget. 

  -- The government's exposure to future claims from insuring
     currently healthy firms--the risk assumed by the government--is
     not recognized in the budget. 

  -- Changes in the government's exposure to future claims due to the
     annual growth in insured benefits or program changes are not
     recognized in the budget as they occur. 


      ISSUES IN IMPLEMENTING AN
      ACCRUAL-BASED BUDGETING
      APPROACH
----------------------------------------------------- Appendix III:2.2

  -- Estimates of PBGC's exposure to the future costs of pension
     benefits are inherently uncertain due to its sensitivity to
     changes in interest rates and the difficulty of projecting firm
     bankruptcies. 

  -- Potentially large annual swings in the accruing cost of pension
     guarantees due to changes in economic conditions could introduce
     new volatility in the annual budget deficit. 

  -- Development, testing, and documentation of both the OMB model
     and PBGC's PIMS is not yet complete. 


OTHER INSURANCE PROGRAMS
========================================================== Appendix IV


   FEDERAL LIFE INSURANCE
-------------------------------------------------------- Appendix IV:1

The federal government provides life insurance coverage to employees,
retirees, and veterans.\1 The following sections provide an overview
of the three life insurance programs included in our study: 

  -- Federal Employees' Group Life Insurance

  -- Service-Disabled Veterans Insurance

  -- Veterans Mortgage Life Insurance


--------------------
\1 Only veterans' life insurance programs underwritten by the federal
government and still open to new participants were included in this
study. 


      FEDERAL EMPLOYEES' GROUP
      LIFE INSURANCE
------------------------------------------------------ Appendix IV:1.1

Budget Account:  Employees' Life Insurance Fund
(24-8424-0-8-602)

Agency:  Office of Personnel Management (OPM)

Bureau:  Retirement and Insurance Service (RIS)


         PURPOSE
---------------------------------------------------- Appendix IV:1.1.1

The Federal Employees' Group Life Insurance (FEGLI) program was
established in 1954 (Public Law 83-598) to provide federal employees
the opportunity to obtain low-cost term life insurance comparable to
that widely offered by private sector employers.  The establishment
of the program was seen as an essential element of a well-rounded
personnel program for the federal government.  The Office of
Personnel Management (OPM) manages FEGLI, sets and collects insurance
premiums, and invests program funds.  The Metropolitan Life Insurance
Company, under contract with OPM, settles and pays insurance claims. 
Prior to the establishment of FEGLI, life insurance coverage was
offered to groups of federal employees by beneficial associations. 
With the creation of FEGLI, membership in these associations was
closed.\2


--------------------
\2 In 1955, the Congress authorized OPM to purchase a qualified
insurance policy to insure agreements assumed from the beneficial
associations.  This very small program is underwritten by the
Shenandoah Life Insurance Company. 


         COVERAGE
---------------------------------------------------- Appendix IV:1.1.2

FEGLI covers 90 percent of eligible employees and retirees of the
executive, legislative, and judicial branches of the federal
government as well as many of their family members.  Basic coverage
is automatic upon eligibility unless declined by the employee.  At
the end of fiscal year 1996, $484 billion\3 in life insurance
coverage was provided under FEGLI to 2.4 million active employees and
about 1.6 million annuitants.  Total insurance in force is projected
to increase to $496 billion by the end of fiscal year 1998. 

The FEGLI program provides basic life insurance coverage equal to the
employee's annual salary rounded to the next higher $1,000, plus
$2,000.  The minimum coverage is $10,000 and the maximum is limited
to the amount based on the Level II Executive Schedule salary.  For
accidental death, the amount is doubled.  One-half the basic benefit
is payable for accidental dismemberment--the loss of one hand, one
foot, or one eye--while the full benefit is paid for the loss of two
or more such members.  Employees age 35 or under receive insurance
coverage equal to twice the basic amount at no additional cost to
them.  This extra benefit decreases by 10 percent each year until at
age 45 there is no extra benefit.  This extra benefit does not apply
to the accidental death and dismemberment benefit.  Effective July
25, 1995, the FEGLI Living Benefits Act of 1994 (Public Law 103-409)
established a new provision allowing terminally ill enrollees with
life expectancies of 9 months or less to elect to receive a lump-sum
payment equal to their basic insurance amount with some adjustments. 
Employees with basic FEGLI coverage are eligible to purchase
additional optional insurance coverage. 

If certain conditions are met, full basic coverage is provided to
retirees until age 65.  After age 65, three levels of coverage are
available.  If no action is taken, the basic coverage amount is
reduced by 2 percent each month until 25 percent of the original
coverage remains.  However, retirees may elect to purchase one of two
alternatives for post-age 65 coverage.  They can elect (1) coverage
that is reduced by 1 percent each month after age 65 until it reaches
50 percent of the original amount or (2) no reduction in coverage
after age 65.  If basic life insurance coverage is continued into
retirement, the optional insurance coverage may also be continued at
an additional cost to the retiree.  Accidental death and
dismemberment coverage stops at retirement. 


--------------------
\3 Includes $113 billion for accidental death and dismemberment
(AD&D) coverage.  In prior years, OMB excluded AD&D coverage from the
reported FEGLI face value. 


         ELIGIBILITY REQUIREMENTS
---------------------------------------------------- Appendix IV:1.1.3

Most civilian employees of the federal government (and individuals
first employed by the District of Columbia government before October
1, 1987) are eligible to participate in the FEGLI program.  Basic
life insurance coverage is effective on the first day of pay and duty
status unless waived by the employee.  Employees are also eligible
for optional coverage at this point, but it is not effective until
elected by the employee.  Employees working under temporary
appointments are not eligible. 


         PROGRAM FINANCING
---------------------------------------------------- Appendix IV:1.1.4

The FEGLI program is financed by insurance premiums and interest
earned on Treasury securities held by the insurance fund.  Employees
pay two-thirds of the insurance premium for basic insurance coverage
and agencies pay the remaining third except for the Postal Service
which pays the full premium for its employees.  The cost of optional
insurance is paid entirely by the employee or annuitant.  Federal
retirees, including Postal Service retirees under age 65 who retired
after 1989, also pay two-thirds of the basic premium.  After age 65,
retirees pay nothing for coverage equal to 25 percent of the original
basic benefit.  The retiree pays premiums to continue coverage at the
full basic benefit level or at the 50-percent level. 

In fiscal year 1996, the Employees' Life Insurance Fund collected
premiums of approximately $1.5 billion and had investment income of
over $1.2 billion.  During this period, the program paid
approximately $1.6 billion in insurance benefits.  Although the FEGLI
program is expected to continue to have a positive cash flow for the
next 15 years, the program reported a $3.4 billion unfunded liability
at the end of fiscal year 1996. 


         CURRENT BUDGET TREATMENT
---------------------------------------------------- Appendix IV:1.1.5

All administrative and insurance outlays as well as collections from
insurance premiums and earnings on invested funds are reported on a
cash basis in the Employees' Life Insurance Fund--a trust revolving
fund.  Associated with this fund is a payment account:  Government
Payment for Annuitants, Employees' Life Insurance.  This payment
account is used to transfer to the fund appropriations received from
the Congress to cover the government's share (one-third of the cost)
of basic life insurance premiums for certain federal annuitants.\4
All FEGLI program costs are classified as mandatory spending and all
administrative costs are classified as discretionary.  Figure IV.1
shows budgeted and actual outlays for the fund since 1973. 

   Figure IV.1:  Employees' Life
   Insurance Fund Budget Estimates
   Versus Actual Outlays, Fiscal
   Years 1973-1996

   (See figure in printed
   edition.)


--------------------
\4 Annuitants under age 65 retiring after December 31, 1989. 


      SERVICE-DISABLED VETERANS
      INSURANCE
------------------------------------------------------ Appendix IV:1.2

Budget Account:  Service-Disabled Veterans Insurance Fund
(36-4012-0-3-701)

Agency:  Department of Veterans Affairs (VA)

Bureau:  Veterans Benefits Administration (VBA)


         PURPOSE
---------------------------------------------------- Appendix IV:1.2.1

Service-Disabled Veterans Insurance (SDVI) was established in 1951
under the Serviceman's Indemnity Act to provide life insurance
coverage to veterans having service-connected disabilities at the
same rates available to nondisabled veterans. 


         COVERAGE
---------------------------------------------------- Appendix IV:1.2.2

Under the SDVI program, life insurance is available to
service-disabled veterans in multiples of $500 with minimum coverage
set at $1,000 and the maximum set at $10,000.  Under Public Law
102-568, totally disabled veterans may purchase supplemental
insurance coverage not to exceed $20,000.  Policyholders may borrow
up to 94 percent of the cash value of their policies.  Insurance in
force at the end of fiscal year 1996 totaled approximately $1.5
billion covering 163,053 veterans. 


         ELIGIBILITY REQUIREMENTS
---------------------------------------------------- Appendix IV:1.2.3

Any person who is released from active military service, under
conditions other than dishonorable, on or after April 25, 1951, and
is found by the Secretary of Veterans Affairs to be suffering from a
service-connected disability or disabilities, is eligible to apply
for coverage.  A disabled veteran must complete a written application
within 2 years from his or her discharge date to be granted coverage. 
Veterans who are determined to be totally disabled are eligible to
apply for supplemental insurance. 


         PROGRAM FINANCING
---------------------------------------------------- Appendix IV:1.2.4

The program is financed from premiums, interest on policy loans, and
general funds received by transfer from the Veterans Insurance and
Indemnities appropriation.  The premiums charged for this coverage
are based on standard rates for nondisabled individuals.  Premiums
for totally disabled veterans are waived.  Totally disabled veterans
who apply for supplemental insurance pay premiums for the additional
coverage.  Because of these provisions, premiums are not actuarially
sound and the program is not self-sufficient.  At the end of fiscal
year 1996, the program's liability for future benefits exceeded
available assets by $457 million.  This deficit is expected to remain
approximately at this level through the end of fiscal year 1997. 


         CURRENT BUDGET TREATMENT
---------------------------------------------------- Appendix IV:1.2.5

All cash flows of the program with the exception of administrative
expenses are reported on a cash basis in the Service-Disabled
Veterans Insurance Fund.  These cash flows include premium
collections, payment of death claims, payment of cash value of
policies surrendered, disbursement of policy loans, interest on
loans, and repayment of loans.  This account also receives a transfer
of funds from the Veterans Insurance and Indemnities appropriation
account as needed to cover outlays.  All activity of this account is
classified as mandatory under BEA.  The program's administrative
expenses are paid out of the Department of Veterans Affairs (VA)
General Operating Expenses appropriation and are discretionary. 
Figure IV.2 shows budgeted and actual outlays for the SDVI fund since
1973. 

   Figure IV.2:  Service-Disabled
   Veterans Insurance Fund Budget
   Estimates Versus Actual
   Outlays, Fiscal Years 1973-1996

   (See figure in printed
   edition.)


      VETERANS MORTGAGE LIFE
      INSURANCE
------------------------------------------------------ Appendix IV:1.3

Budget Account:  Veterans Insurance and Indemnities
(36-0120-0-1-701)

Agency:  Department of Veterans Affairs (VA)

Bureau:  Veterans Benefits Administration (VBA)


         PURPOSE
---------------------------------------------------- Appendix IV:1.3.1

Veterans Mortgage Life Insurance (VMLI) was established in 1971
(Public Law 92-95) to provide mortgage protection life insurance to
severely disabled veterans who are granted VA assistance in securing
specially adapted housing. 


         COVERAGE
---------------------------------------------------- Appendix IV:1.3.2

The amount of insurance provided to a veteran under this program is
the lesser of $90,000 or the amount of the loan outstanding on the
housing unit.  The amount of insurance is reduced according to the
amortization schedule of the loan and may not at any time exceed the
amount of the outstanding loan with interest.  If there is no loan
outstanding on the housing unit, no insurance is available under this
program. 


         ELIGIBILITY REQUIREMENTS
---------------------------------------------------- Appendix IV:1.3.3

Severely disabled veterans who have received VA grants for specially
adapted housing are automatically insured against death unless the
veteran declines coverage in writing to the Secretary of the VA or
fails to provide the VA with the necessary information on which to
calculate the insurance premium.  A veteran who elects not to be
insured can subsequently obtain insurance upon submission of an
application. 


         PROGRAM FINANCING
---------------------------------------------------- Appendix IV:1.3.4

The program is financed by premiums paid by policy holders and
general fund appropriations.  Under law, the premium rates charged to
eligible veterans are based on mortality data that are appropriate to
cover only the cost of insuring nondisabled persons.  As a result,
the program is not self-supporting and requires appropriated funds to
pay claims to mortgage holders.  At the end of fiscal year 1996, VA
estimated that the VMLI program's liability for future benefits
exceed available assets by $93 million. 


         CURRENT BUDGET TREATMENT
---------------------------------------------------- Appendix IV:1.3.5

All activities including premium collections and claim disbursements
of the VMLI program are recorded on a cash basis in the Veterans
Insurance and Indemnities appropriation account.  The program has
permanent authority and appropriations are made as needed to cover
claims.  This account is classified as mandatory under BEA.  The
program's administrative expenses are paid out of the VA's General
Operating Expenses appropriation and are discretionary. 


      METHODS OF ASSESSING RISK
      ASSUMED FOR LIFE INSURANCE
------------------------------------------------------ Appendix IV:1.4

Officials at VA and OPM currently use actuarial approaches that are
the standard practice of the life insurance industry.  Measurement of
the risk assumed in insuring lives is well established in actuarial
science.  Mortality tables, which are mathematical models based on
the laws of probability and mortality statistics, provide the basis
for estimating the occurrence of future deaths.  This information
together with assumptions about interest rates and contractual policy
benefits allow for the calculation of expected insurance claims. 

A basic principle of actuarial science holds that, by studying the
rate of death within any large group of people and gathering
information on all factors that may affect that rate, it is valid to
anticipate that any future group of persons with approximately the
same factors will experience the same rate of death.  Mortality
tables are constructed to reflect probabilities of death at each age. 
The accuracy with which the estimated future claims approximates the
actual experience depends upon two factors:  (1) the accuracy and
appropriateness of the underlying mortality statistics and (2) the
number of observations the estimate is based on and the number of
individuals insured.  Most mortality tables in use today are based
upon the experience of insured individuals because of the accuracy
and completeness of data on these lives. 

In the construction of mortality tables, adjustments are generally
made to the observed mortality rates.  For example, actuaries have
derived mathematical formulas that attempt to explain mortality
rates.  These formulas, which have gained general acceptance in the
field, are used to smooth unexplained deviations in observed
mortality data.  These formulas are also used where data are limited,
such as for very young or old lives.  Adjustments are also made to
mortality tables to provide a margin of financial safety in insurance
contracts and are considered sound practice in the insurance
industry. 

Mortality tables used for the FEGLI program have been developed
internally by OPM actuaries based on the demographic composition of
the federal civilian workforce and the historic mortality rates of
insured employees.  According to the OPM actuaries, this is done
because the characteristics of the federal civilian workforce appear
to be different from the population at large.  OPM has constructed
separate mortality tables for male and female employees, active
employees, retired employees, and disabled persons.  For SDVI, VA is
required to use the Commissioners 1941 Standard Ordinary Table of
Mortality.  For VMLI, VA is directed by law to use mortality data
appropriate to cover only the cost of insuring nondisabled lives. 

An estimate of the expected cost of future insurance benefits can be
derived based on the expected rates of death, assumed rate of
interest, and policy benefits.  This information is used by insurance
companies to establish premium rates such that the present value of
the future premiums less operating expenses equals the present value
of future benefits.  If the present value of future benefits exceeds
the present value of future premiums plus any previously accumulated
premiums held in reserve, the program would have an unfunded
liability.  As such, mortality tables and interest rate assumptions
are generally fairly conservative to ensure sufficient resources to
pay future benefits.  In the SDVI and VMLI programs, the Congress has
chosen to subsidize the premiums of disabled veterans through the use
of mortality assumptions for nondisabled individuals.  Premiums
collected are not sufficient to cover expected future benefits and an
unfunded liability exists. 


      IMPLEMENTATION
      CONSIDERATIONS FOR LIFE
      INSURANCE PROGRAMS
------------------------------------------------------ Appendix IV:1.5


         ADEQUACY OF CURRENT
         BUDGET REPORTING
---------------------------------------------------- Appendix IV:1.5.1

  -- Increases in life insurance obligations due to program changes
     or growth are not reflected in the year in which they occur
     since cash payments may not be required for many years. 

  -- Program income is not matched with program expenses.  Premium
     and investment income necessary to pay future claims is recorded
     in the budget as negative outlays (income), while the future
     expense is not recorded.  As a result, the relative cost of the
     program may be understated and sufficient funds may not be
     available to pay future claims. 


         ISSUES IN IMPLEMENTING AN
         ACCRUAL-BASED BUDGETING
         APPROACH
---------------------------------------------------- Appendix IV:1.5.2

  -- Although methodology for estimating the risk-assumed cost of
     extending life insurance is well established in actuarial
     science, estimates are highly sensitive to assumptions such as
     interest rates. 


   NATIONAL FLOOD INSURANCE
   PROGRAM
-------------------------------------------------------- Appendix IV:2

Budget Account:  National Flood Insurance Fund
(58-4236-0-3-453)

Agency:  Federal Emergency Management Agency (FEMA)

Bureau:  Federal Insurance Administration (FIA)


      PURPOSE
------------------------------------------------------ Appendix IV:2.1

The National Flood Insurance Program (NFIP) was established by the
National Flood Insurance Act of 1968 (Public Law 90-448) to (1)
identify flood prone areas, (2) make flood insurance available to
property owners living in communities that joined the program, (3)
encourage floodplain management efforts to mitigate flood hazards,
and (4) reduce federal spending on disaster assistance.  Some of the
key factors leading to the program's establishment were the
reluctance of private insurers to sell flood coverage, increasing
losses caused by floods because of floodplain encroachment, and
higher federal expenditures for relief and flood control.\5 Since its
establishment, NFIP has been expanded and modified several times. 


--------------------
\5 Federal Disaster Assistance, Bipartisan Task Force on Funding
Disaster Relief, S.  Doc.  No.  4, 104th Cong., 1st Sess.  (1995). 


      COVERAGE
------------------------------------------------------ Appendix IV:2.2

Federal flood insurance is available in the 50 states, the Virgin
Islands, Puerto Rico, Guam, the District of Columbia, and American
Samoa.  In fiscal year 1995, NFIP had about 3.3 million policies,
totaling over $325 billion, in force in over 18,000 communities
nationwide.  As of January 1997, there was approximately $380 billion
of insurance in force. 

NFIP has two principal components:  emergency and regular.  The
emergency program is available in communities before detailed
mapping\6 is completed.  Under the emergency program, structures
identified in flood-prone areas are eligible for limited amounts of
coverage at subsidized rates.  However, according to FIA, flood
insurance rate maps (FIRMs) have been completed for nearly all
communities considered to be flood-prone, and only a very few
communities remain in the emergency program. 

After mapping is completed, the communities enter the regular
program.  Under the regular program, there are basically two
classifications of properties (1) pre-FIRM properties--those built
before the initial mapping studies were completed and (2) post-FIRM
properties--those built after the mapping studies were completed. 
After a community joins the regular program, the rates for the
pre-FIRM properties may still be subsidized, but post-FIRM properties
are to be charged actuarially-based rates.  In fiscal year 1996,
subsidized policies accounted for approximately 38 percent of the
total policies in force.  Under the regular program, coverage is
available for virtually all types of buildings and their contents
with coverage limits of up to $350,000 for residential properties and
$1 million for other properties. 


--------------------
\6 Flood insurance rate maps (FIRMs) provide information such as
elevation and flood zone that are necessary for classifying
properties according to flood risk. 


      ELIGIBILITY REQUIREMENTS
------------------------------------------------------ Appendix IV:2.3

To be eligible for federal insurance, communities must adopt and
enforce floodplain management ordinances that meet or exceed the
minimum standards of the program.  Communities must join the program
within
1 year of the time they are identified as flood-prone. 

The purchase of flood insurance was voluntary until the adoption of
the Flood Disaster Protection Act of 1973.  The 1973 Act required the
purchase of flood insurance to cover structures in special flood
hazard areas of communities participating in the program if (1) any
federal loans or grants were used to acquire or build the structures
and (2) loans were secured by improved properties and the loans were
made by lending institutions that are regulated or insured by the
federal government.  In 1994, the Congress amended NFIP to, among
other things (1) prohibit federal disaster relief in flood disaster
areas to persons who failed to obtain and maintain required flood
insurance and (2) establish civil monetary penalties for regulated
lenders who fail to ensure that their borrowers obtain required flood
insurance. 


      PROGRAM FINANCING
------------------------------------------------------ Appendix IV:2.4

The program is financed primarily through premiums, fees, and
interest income.  As noted above, owners of post-FIRM structures pay
actuarially-based rates.  By contrast, subsidized rates are available
for owners of older, generally less flood-worthy, pre-FIRM
structures.  FIA is authorized to borrow up to $500 million from the
Treasury without approval of the President and up to $1 billion with
approval of the President.\7 In addition, the Congress has
appropriated funds for NFIP from time to time over the program's
history.  The program has not received a general fund appropriation
since 1986. 

By design, NFIP is not actuarially sound.  The Congress authorized
FIA to subsidize a significant portion of the total policies in force
but did not provide annual appropriations to cover the implicit
subsidy costs.  Although the program has achieved a goal of becoming
self-supporting for the average historical loss year,\8 it may not
have sufficient resources to meet potential catastrophic losses.\9
This is the case because the historical experience period used does
not include any loss years considered to be of a catastrophic
level.\10

Figure IV.3 shows the program's premium income and loss and loss
adjustment expenses since the program's inception.  The volatility in
the program experience demonstrates the uncertainty surrounding the
average loss and loss adjustment costs. 

   Figure IV.3:  NFIP Premium
   Income Versus Loss and Loss
   Adjustment Expenses, Fiscal
   Years 1969-1996

   (See figure in printed
   edition.)

Source:  FIA unaudited data. 

The program has had to borrow from the Treasury several times in
recent years.  For example, in fiscal year 1993, the nation
experienced severe flood damage resulting in flood insurance claims
more than double the historical average loss.  As a result, the
program borrowed and since repaid funds from the Treasury to pay
excess claims.  Similarly, in fiscal year 1995, the program
experienced losses that were much greater than the historical average
loss.  As a result, the program again exercised its borrowing
authority.  According to FIA, as of March 31, 1997, NFIP owed the
Treasury $818 million. 


--------------------
\7 FIA's borrowing authority was increased to $1.5 billion for fiscal
year 1997 only. 

\8 Premium rates for NFIP are established so that total premium
revenue is sufficient to cover the average historical loss year. 
According to FIA, the rate review typically first determines whether
the actuarial rates need to be adjusted.  The effects of any such
adjustments on maintaining the overall target level are then
projected.  Should there be a shortfall, adjustments to policy
coverage or premiums for pre-FIRM policies will likely be proposed to
make up the difference so that the combination of actuarial and
subsidized policies would be generating written premiums at least to
the level of NFIP's self-supporting target. 

\9 Flood Insurance:  Financial Resources May Not Be Sufficient to
Meet Future Expected Losses (GAO/RCED-94-80, March 21, 1994). 

\10 According to FIA, the probable maximum loss resulting in $4.5
billion to $5 billion in claim losses has a 1 in 1,000 chance of
occurring.  For comparison purposes, Hurricane Hugo resulted in
claims of $0.4 billion. 


      CURRENT BUDGET TREATMENT
------------------------------------------------------ Appendix IV:2.5

The National Flood Insurance Fund, a revolving fund, was established
to carry out NFIP.  The program includes both mandatory and
discretionary spending.  Funding for expenses other than costs
incurred in the adjustment and payment of claims is available only to
the extent provided in appropriation acts.  Figure IV.4 shows the
program's budgeted versus actual outlays from fiscal years 1973
through 1996. 

   Figure IV.4:  National Flood
   Insurance Fund Budget Estimates
   Versus Actual Outlays, Fiscal
   Years 1973-1996

   (See figure in printed
   edition.)


      RISK ASSESSMENT METHODS
------------------------------------------------------ Appendix IV:2.6

According to FIA, its actuarial rate-setting method for unsubsidized
policies (post-FIRM) could be used to generate reasonable estimates
of the costs of the long-term expected risk for the entire program. 
The difference between the costs of the long-term expected risk and
the actual premium rates could be used to provide an estimate of the
government's subsidy costs. 

FIA uses a class-rating system to establish actuarial rates.  That
is, FIA classifies properties according to key characteristics of
flood risk.  All owners of properties in the same risk group are then
charged the same rates.  Even though individual risk may vary among
properties within each risk group classification, these rates are
actuarially-based in the sense that risk exposure for like properties
is considered when setting the group's rates. 

Information about the risk of flooding is essential to establishing
actuarially sound rates.  Two key characteristics that are used to
classify properties according to flood risks include (1) the flood
zone and (2) the elevation of the structure relative to the base
flood elevation (BFE).\11 Information about flood zones and BFEs is
obtained from FIRMS. 

The basic method for establishing actuarial rates on post-FIRM
structures lying within the 100-year floodplain follows the
hydrologic model described in a 1966 Department of Housing and Urban
Development (HUD) report, Insurance and Other Programs for Financial
Assistance to Flood Victims.\12 The basic logic of the model is to
set rates for a property according to its risk of being flooded. 
According to an FIA actuary, the model is basically an expected value
calculation based on measures of the frequency and severity of
floods.\13

Thus, a key data element is an estimate of the probabilities that
floods of different severities, relative to BFE, will occur in a
given year.  FIA calls these data probability of elevation (PELV)
values.  Although within any zone there is a 1-percent chance that
flood waters will exceed the BFE, the degree to which flood waters
will reach above or below that level will vary across zones.  PELV
tables provide detailed information, by zone, about the frequency
with which floods of all possible water surface elevations can be
expected to occur.  These data were generated on the basis of
detailed engineering studies, available flood insurance data,
simulations, and professional judgments and were established for each
flood-hazard zone to meet generally accepted scientific parameters
and legal considerations.\14

Another key data element is the structural damage that will be
suffered when a flood occurs.  For a variety of depths of floods, and
the associated depth of water in a structure, FIA has data that
provide estimates of the percent of the value of a structure that is
expected to be damaged.  FIA calls these data the
depth-percent-damage relationship or the damage by elevation (DELV)
values.  Information is presented by 1-foot increments of flood level
within the structure and expressed as the average percentage of the
property's value that will be damaged due to a flood of that
elevation.  For example, according to 1987 DELV information, a
one-story, no basement structure located in the AE zone would sustain
damage equal to 21 percent of the property's value if flood water
reached a depth of 2 feet.  As with the PELV data, information used
in establishing DELV values was obtained primarily from engineering
studies.  In 1973, data for DELVs were selected on the basis of
studies done by the U.S.  Army Corps of Engineers and available flood
claims at the time.  According to FIA, DELVs are compared to actual
experience and updated when sufficient data exist.\15

Knowledge of the elevation-frequency relationship and the
depth-damage allows a summation of the range of flood probabilities
and their associated damage to property and contents.  Each possible
flood is multiplied by the expected damage should such a flood occur,
and then each of these is added together.  The total of each possible
flood's damage provides an expected per annum percent of the value of
property damage due to flooding.  This expected damage can then be
converted to an expected loss per $100 of property value covered by
insurance.  This per annum expected loss provides the fundamental
component of rate-setting. 

Several other factors important for modifying expected losses or for
building additional expense items into the rates are also considered. 
These variables include the following: 

  -- Loss adjustment factor:  Rates are "loaded" or adjusted upward
     to account for costs associated with claims and loss
     adjustments. 

  -- Deductible offset factor:  Rates are adjusted downward to take
     into account that some portion of each claim will not be covered
     because of the policy deductible. 

  -- Underinsurance factor:  Rates are adjusted to take into account
     that the full value of the property may not be insured. 

  -- Expense items factor:  Rates are loaded for certain expenses,
     such as agents' commissions. 


--------------------
\11 BFE is the elevation relative to mean sea level at which there is
a 1-percent chance of flood waters exceeding that level in a given
year.  The level of BFE within a community can change throughout the
floodplain.  These changes are delineated on FIRMs. 

\12 Rates for post-FIRM properties that are outside the 100-year
floodplain are set primarily through an analysis of previous years'
claims. 

\13 The HUD report describes the "hydrologic method" of rate-making
as a method which "uses available data on the occurrence of floods
and damage, but is considerably more sophisticated than merely
averaging losses over a period of time."

\14 As noted in GAO/RCED-94-80, March 21, 1994, one of the problems
in originally establishing the PELV tables was that the flood
histories on which these studies were based were generally not very
long.  Statistical literature has shown that this may cause a bias
toward establishing frequency probabilities that are too low. 
Consequently, the original PELV values have been modified to account
for this bias. 

\15 FIA determines whether it has sufficient data on actual floods of
different severities since 1978 to replace the original DELV.  If
data are sufficient then there is "full credibility" and the original
DELVs are replaced.  If insufficient claims data exist for full
credibility, DELVs are based on a weighted average of the original
base table values and the actual experience since 1978, where the
weighting is determined by the ratio of actual experience claims to
the number of experience claims necessary for full credibility.  This
would mean that over time the original, theoretical DELV will have
less weight in determining actual DELV used for rate-setting. 


      IMPLEMENTATION
      CONSIDERATIONS FOR THE
      NATIONAL FLOOD INSURANCE
      PROGRAM
------------------------------------------------------ Appendix IV:2.7


         ADEQUACY OF CURRENT
         BUDGET REPORTING
---------------------------------------------------- Appendix IV:2.7.1

  -- The current cash-based budget does not recognize or fund the
     subsidy cost implicit in the government's flood insurance
     commitment for losses in excess of the historical loss year.  As
     a result, the relative budgetary cost of the program may be
     understated and the program may not have sufficient funds to
     cover future claims.  FIA estimates the annual "missing
     premium"--the government's subsidy--at about $520 million. 

  -- The sporadic nature of floods may cause fluctuations in the
     deficit unrelated to the budget's long-term structural balance. 


         ISSUES IN IMPLEMENTING AN
         ACCRUAL-BASED BUDGETING
         APPROACH
---------------------------------------------------- Appendix IV:2.7.2

  -- Although FIA's risk assessment experience and established
     rate-setting methodology will provide a useful foundation for
     generating risk-assumed estimates, some additional work may be
     required to adapt these estimates for use in an accrual-based
     budget. 

  -- The appropriate level of reserves and the basis for reestimation
     will have to be determined.  FIA has done some work developing
     estimates of catastrophic reserve levels but additional
     refinements and modifications will likely be required. 

  -- The appropriate basis for accrual cost measurement--the average
     historical loss year or the program's long-term expected loss
     (including rare catastrophic loss years)--will have to be
     determined. 

  -- FIA officials and staff expressed concern about the amount of
     staff resources required to update and adapt estimates and to
     comply with the requirements of an accrual-based budgeting
     approach. 


   FEDERAL CROP INSURANCE PROGRAM
-------------------------------------------------------- Appendix IV:3

Budget Account:  Federal Crop Insurance Corporation Fund
(12-4085-0-3-351)

Agency:  Department of Agriculture (USDA)

Bureau:  Risk Management Agency (RMA)\16


--------------------
\16 The Federal Crop Insurance program is administrated by the
Federal Crop Insurance Corporation (FCIC), a wholly owned government
corporation.  The 1996 Farm Bill established the Risk Management
Agency within USDA and it has jurisdiction over FCIC. 


      PURPOSE
------------------------------------------------------ Appendix IV:3.1

The Federal Crop Insurance program was established in 1938 by the
Federal Crop Insurance Act\17 to protect crop farmers from
unavoidable risks associated with adverse weather, plant diseases,
and insect infestations.  The program has been amended numerous times
during its history.  Extensive amendments were adopted in the Federal
Crop Insurance Act of 1980\18 (Public Law 96-365) and the Federal
Crop Insurance Reform Act of 1994\19 (Title I of Public Law 103-354). 
Most recently, several changes were made to the program by provisions
of the Federal Agricultural Improvement and Reform Act of 1996
(Public Law 104-127, the Farm Bill). 


--------------------
\17 Title V of the Agricultural Adjustment Act of 1938 (Public Law
75-430, 7 U.S.C., 1501-1520). 

\18 The Federal Crop Insurance Act of 1980 authorized FCIC to expand
coverage to include all agricultural crops in all agricultural
counties and to subsidize producer premiums. 

\19 Among other things, the 1994 Act repealed ad hoc disaster
authority and authorized FCIC to offer catastrophic risk protection. 


      COVERAGE
------------------------------------------------------ Appendix IV:3.2

Crop insurance is available in all 50 states and Puerto Rico.  In
crop year 1996, there was about $27 billion of insurance in force
written in over 3,000 counties.  These policies provided coverage for
approximately 200 million acres. 

Under the changes made by the Federal Crop Insurance Act of 1994, two
types of coverage--catastrophic and additional--are available for
most major crops.  Catastrophic coverage provides producers a minimum
level of protection for a small processing fee.  This coverage
compensates farmers for crop yield losses greater than 50 percent at
a payment rate of 60 percent of the expected market price.  Premiums
for this coverage are fully subsidized by the government. 

Farmers can also purchase additional coverage from participating
private insurance companies.\20 Farmers who purchase this additional
insurance must choose both the coverage level (the proportion of the
crop to be insured) and the unit price (e.g., per bushel) at which
any loss is calculated.  Farmers can choose to insure as much as 75
percent of normal production or as little as 50 percent of normal
production at different price levels.\21 With respect to unit price,
farmers can choose to value their production at USDA's full price or
a percentage of the full price.  The government pays part of the
farmer's premium for this additional coverage. 

The Crop Insurance Reform Act of 1994 also created a new program, the
Noninsured Assistance Program (NAP), for producers of most crops not
currently covered by the crop insurance program.  For a farmer to
become eligible for payment, area-wide losses must be at least 35
percent of normal yields and the farmer must experience an individual
minimum loss of at least 50 percent.  Coverage levels are similar to
those under the catastrophic coverage level once the trigger is
activated. 


--------------------
\20 According to FCIC, additional coverage can be made available
through USDA if private insurance providers do not service an area. 
The 1996 Farm Bill allows USDA to continue offering catastrophic risk
protection through its local offices but only in states where there
are too few approved private insurance providers. 

\21 According to FCIC, the 1994 Act authorizes 85-percent coverage. 
This coverage may be implemented on a limited basis in 1997. 


      ELIGIBILITY REQUIREMENTS
------------------------------------------------------ Appendix IV:3.3

The 1996 Farm Bill eliminated mandatory participation\22 in the
federal crop insurance program to qualify for assistance under other
farm programs.  This applies to farmers who provide a written waiver
to the Secretary of Agriculture agreeing to forgo eligibility for
disaster payments in connection with a crop.\23 If a farmer does not
sign a waiver, catastrophic coverage is required for receipt of a
Conservation Reserve Program (CRP) payment, a USDA loan, or the
7-year market transition payment for eligible wheat, feed grain,
cotton, or rice growers. 


--------------------
\22 The Federal Crop Insurance Act of 1994 required producers to
obtain at least the catastrophic level of insurance to be eligible
for benefits under the price support or production adjustment
program, the conservation reserve program, or farm credit programs. 

\23 Effective for spring planted 1996 crops and all subsequent crops. 


      PROGRAM FINANCING
------------------------------------------------------ Appendix IV:3.4

The crop insurance program is financed primarily through general fund
appropriations and farmer-paid premiums.  FCIC is authorized under
the Federal Crop Insurance Act, as amended, to use funds from the
issuance of capital stock which provides working capital for FCIC.\24

FCIC does not earn interest on cash maintained in U.S.  Treasury
accounts. 

Under the Federal Crop Insurance Reform Act of 1994, FCIC is required
to set insurance premiums at rates that are actuarially sufficient to
attain an expected loss ratio\25 of not greater than 1.1 through
September 1998, and not greater than 1.075 thereafter.  In addition
to the premiums paid by producers, FCIC receives a mandatory
indefinite appropriation to the Insurance Fund to provide funds for
the program's premium subsidy costs, excess losses, and delivery
expenses.  For fiscal year 1996, net insurance premium revenue from
farmers was approximately $600 million and the appropriation for the
government's premium subsidy was approximately $990 million.  Total
appropriations received to cover operating costs in fiscal year 1996
were approximately $1.6 billion.  In addition, farmers are required
to pay an administrative fee.\26

The program has a history of financial losses.\27 For example, since
the program was expanded, it has paid out approximately $3.4 billion
more in claims than it has received premiums from farmers and the
federal government between crop years 1980 and 1996.  Since losses in
excess of premium income are a cost to the government, they represent
additional federal subsidies.  Figure IV.5 shows the total
premiums--both producer and government--and claim payments from
fiscal years 1980 through 1996. 

   Figure IV.5:  Federal Crop
   Insurance Premiums Versus Loss
   and Loss Adjustments, Fiscal
   Years 1980-1996

   (See figure in printed
   edition.)

Source:  USDA. 


--------------------
\24 The act authorizes capital stock of $500 million subscribed by
the United States.  There has been no change in the level of capital
stock issued since August 1985. 

\25 The loss ratio represents insurance claims expense divided by
premium revenues. 

\26 Producers are required to pay a $50 processing fee per covered
crop per county upon enrollment in the program for catastrophic and
limited additional coverage up to 65 percent of production at full
price.  The total fees cannot exceed $200 per producer per county, up
to a maximum of $600 per producer for all counties in which a
producer has insured crops.  According to FCIC, the fee for
additional coverage greater than 65 percent of production at full
price is $10 per crop per county without limitation.  USDA can waive
processing fees for financial hardship cases. 

\27 See Crop Insurance:  Additional Actions Could Further Improve
Program's Financial Condition (GAO/RCED-95-269, September 28, 1995). 


      CURRENT BUDGET TREATMENT
------------------------------------------------------ Appendix IV:3.5

Budget reporting for the Federal Crop Insurance program has undergone
several changes in recent years.\28 Under the most recent changes
included in the 1996 Farm Bill, the expenses of the Federal Crop
Insurance program will be handled in two budget accounts.  Beginning
in fiscal year 1998, the Federal Crop Insurance Fund will handle
insurance premiums, the government's premium subsidy, claim losses,
and a portion of the program's insurance sales and claims processing
administrative costs.  These costs will be covered by a mandatory
indefinite appropriation.\29 Salaries, general governmental
administrative costs, and agent commissions will be handled in a
separate administrative and operating account of the newly
established RMA.  These costs will be classified as discretionary. 

Figure IV.6 shows the budget estimates versus actual outlays for the
Federal Crop Insurance Corporation Fund from fiscal years 1973
through 1996.  This figure shows both that budget estimates are
usually lower than actual outlays and the sometimes erratic nature of
actual outlays. 

   Figure IV.6:  FCIC Fund Budget
   Estimates Versus Actual
   Outlays, Fiscal Years 1973-1996

   (See figure in printed
   edition.)


--------------------
\28 Both the Federal Crop Insurance Reform Act of 1994 and the
Federal Agricultural Improvement and Reform Act of 1996 (the Farm
Bill) made changes in the budget reporting for the Federal Crop
Insurance program.  Prior to the 1994 reforms, FCIC maintained two
funds:  (1) the Crop Insurance Fund--primarily for insurance premiums
and claim losses and (2) a separate administrative and operating
account which handled salary and general administrative expenses as
well as insurance sale and claim processing costs.  Under the 1994
reforms, FCIC's salary and general administrative expenses were
shifted to the Farm Service Administration's administrative and
operating account.  Claim losses and all other program expenses were
handled in the Crop Insurance Fund. 

\29 In the past, FCIC relied on Commodity Credit Corporation (CCC)
funding for losses that exceeded premiums.  Although this authority
to use CCC funding still exists, FCIC is also authorized to draw
necessary funds directly from the Treasury. 


      RISK ASSESSMENT METHODS
------------------------------------------------------ Appendix IV:3.6

FCIC has an established rate-setting methodology which, according to
agency officials, could serve as the foundation for estimating the
risk assumed by the government for the crop insurance program.  For
example, the difference between the "pure" or "full-risk" premium and
actual premium rates could be used to provide an estimate of the
government's subsidy costs for policies issued in a given year. 
However, as discussed in chapter 5, the rate-setting methodology is
complex because the risk of growing a particular crop varies by
county, farm, and farmer.  Because of all the possible combinations
involved, hundreds of thousands of rates are in place.  Thus, a
number of implementation issues would have to be resolved.  The
following discussion provides a brief overview of the premium
rate-setting process.\30

Each year, FCIC follows a multistep process to establish rates for
each crop included in the program.  The process involves establishing
base rates for each county crop combination and adjusting these basic
rates for a number of factors, such as coverage and production
levels.  In addition, rates are adjusted to account for the
legislated limitations in price increases. 

For each crop, FCIC begins the process by extracting data on
counties' crop experiences for all years available (up to 20) from
its historical database.  The data elements for each crop, crop year,
and county include (1) the dollar amount of the insurance in force
(coverage sold), (2) the dollar amount of the claims paid
(indemnities), and (3) the average coverage level.  Data for farmers
who incur frequent and severe losses relative to other farmers are
removed from the resulting database to avoid setting rates that are
higher than necessary for the risk represented by the farmers who are
not considered high risk.  The premium rates for high-risk producers
are established separately under the high-risk program. 

The historical data are then adjusted to the 65-percent coverage
level.  Using the adjusted data, FCIC computes the loss-cost ratio
for each crop in each county.  The loss-cost ratio is calculated by
dividing the total claim payments by the total insurance in force;
the result is stated as a percentage.\31 The loss-cost ratios are
calculated using the latest available data, which are for the period
ending 2 years before the year for which the rates are being
established.  For example, the crop year 1995 rates were established
in 1994 at which time the most recent 20-year record was for crop
years 1974 through 1993. 

A loss-cost ratio is calculated for each of the 20 years and then
these data are divided into two segments--the 4 years with the
highest loss-cost ratios and the 16 years with the lowest loss-cost
ratios.\32 For the 4 years with the highest loss-cost ratios, the
ratios are capped at the loss-cost ratio for the highest loss year in
the 16-year segment.  To establish the county unloaded rate, the
average for all 20 years is calculated, using the capped loss-cost
ratios for each of the 4 years.  Thus, the 20-year average loss-cost
ratio consists of actual loss-cost ratios for the 16 lowest loss
years and the capped loss-cost ratios for the 4 highest loss
years.\33

The county unloaded rates are then adjusted to minimize the
differences in rates among counties using a weighting process.  A
surcharge for catastrophic coverage for each crop in each state is
then developed and added to the adjusted unloaded rate for each
county in the state.\34 The result of this process is a basic rate
for the county for the 65-percent coverage level and the average
production level in that county.\35

Next, the rates at the 65-percent coverage level are adjusted for
each farming practice, such as whether the insured acreage is
irrigated or dryland, and for each crop type, such as winter or
spring wheat, using factors based on historical data.  Field
underwriters review the rates\36 for reasonableness on the basis of
their knowledge and continuing research on farmers' experiences with
the particular crop in the county and the surrounding area and
recommend changes when they believe they are warranted.  On the basis
of these recommendations, FCIC analysts make adjustments. 

Following this review and any resulting adjustments, rates are
adjusted upward for the risk represented when farmers choose to
subdivide their farming operation for a given crop into multiple
units for crop insurance purposes.  According to FCIC, this is done
because USDA's historical data show that farming operations insured
on a multiple unit basis are more likely to make claims than those
insured as one unit. 

As noted above, the calculated rates are for farmers whose historic
production level (yield) is about equal to the average for all
producers in the county.  However, many farmers' average production
levels are above or below the county's average.  According to USDA,
farmers' chances of having a loss decreases as production increases. 
Therefore, rates are adjusted using a mathematical model to account
for production levels that differ from the average production level. 

The calculated full rates are reduced as needed to ensure that they
do not exceed the maximum 20-percent increase per year allowed by
law.  As a final step, discounts are developed for farmers who buy
hail and fire protection from private insurance companies.\37


--------------------
\30 The rate-setting for the crop insurance program is discussed in
Crop Insurance:  Additional Actions Could Further Improve Program's
Financial Condition (GAO/RCED-95-269, September 28, 1995). 

\31 For example, if the claims paid in 1 year totaled $7.36 and the
insurance in force was $100, the loss-cost ratio is 7.36 percent. 
The percentage represents the rate that would need to be charged per
$100 of insurance coverage if total premiums are to equal the total
claim payments for that year.  In this example, the 7.36 percent
indicates that a rate of $7.36 was required per $100 of insurance
coverage sold. 

\32 According to FCIC's senior actuary, the procedures described
herein will be modified beginning with 1998 crops to incorporate the
results of an analysis conducted by an actuarial firm and USDA. 

\33 The excess of losses above the capped amount is pooled at the
state level and reallocated to the counties.  According to FCIC, this
procedure is intended to reduce the variation of rates from one year
to the next. 

\34 The surcharge is established by pooling the amount of insurance
in force and the claim payments for the 4 years with the highest
loss-cost ratios in each county that were not factored into the
county unloaded rates at the state level.  These data are used to
calculate a statewide surcharge for catastrophic coverage (pooled
claims payments divided by pooled insurance in force).  If the pooled
losses at the state level exceed five points, the excess is returned
to the counties and included in the county unloaded rate. 

\35 Rates for the 50-percent and 75-percent coverage levels are also
established by applying factors of 0.72 and 1.54 to the rates
established for the 65-percent coverage level. 

\36 Because the regional service office underwriters are more
familiar with the 75-percent coverage level, the basic rates are
adjusted to the 75-percent level to facilitate review.  According to
FCIC, in the 1980s, when much of the crop insurance business was at
the 75-percent coverage level, rates were calculated on that basis. 
Today, although most business is at the 65-percent level and rates
are calculated on that basis, many underwriters remain more
comfortable performing comparisons at the 75-percent level. 

\37 By law, this option is only offered to farmers who purchase at or
above the 65-percent and full-price coverage. 


      IMPLEMENTATION
      CONSIDERATIONS FOR THE
      FEDERAL CROP INSURANCE
      PROGRAM
------------------------------------------------------ Appendix IV:3.7


         ADEQUACY OF CURRENT
         BUDGET REPORTING
---------------------------------------------------- Appendix IV:3.7.1

Since the costs associated with a normal loss year are included in
the budget year estimates for the crop insurance program,
policymakers receive some signals about the program's potential costs
at the time decisions are made.  However: 

  -- On a cash basis, the program sustained significant losses
     without prompting recognition of funding deficiencies until
     claims had occurred.  Between crop years 1980 and 1996, the
     program's claims exceeded premiums by approximately $3.4
     billion. 

  -- The need and cost of establishing reserves over time is not
     explicitly recognized in outlays or in the deficit in the year
     the insurance is extended. 

  -- The sporadic nature of crop losses will cause fluctuations in
     the deficit unrelated to the budget's long-term structural
     balance. 


      ISSUES IN IMPLEMENTING AN
      ACCRUAL-BASED BUDGETING
      APPROACH
------------------------------------------------------ Appendix IV:3.8

  -- Although FCIC's risk assessment experience and established
     rate-setting methodology will provide a foundation for
     generating risk-assumed estimates, additional work will be
     required to determine how to adapt these estimates for
     accrual-based budgeting purposes. 

  -- Due to the complexity of the risk assessment process and the
     timing differences between the detailed rate-setting process and
     the budget cycle, a number of implementation challenges will
     have to be resolved.  For example, an appropriate and feasible
     aggregation level for risk factors will have to be determined. 

  -- Additional work would be required to determine the basis for
     reestimation and reserve levels. 


   POLITICAL RISK INSURANCE
-------------------------------------------------------- Appendix IV:4

Budget Account:  Noncredit Account
(71-4184-0-3-151)

Agency:  Overseas Private Investment Corporation (OPIC)


      PURPOSE
------------------------------------------------------ Appendix IV:4.1

The Overseas Private Investment Corporation (OPIC), which began
operations in 1971, was established to facilitate U.S.  private
investment in developing countries and countries with emerging
markets.  OPIC's insurance programs reduce the risk of U.S.  private
investment in these countries by offering protection against several
political risks.  All valid claims arising from OPIC's investment
insurance are explicitly backed by the full faith and credit of the
United States.  In general, the coverage offered by OPIC is more
comprehensive--both in scope and duration--than that currently
available from private sector insurers. 

OPIC operates as a self-financing governmental agency.  In addition
to its insurance activities, OPIC provides project financing and
makes equity capital available by guaranteeing long-term loans to
private equity investment funds. 


      COVERAGE
------------------------------------------------------ Appendix IV:4.2

OPIC insures against three types of political risks: 

  -- Currency inconvertibility:  The deterioration of an investor's
     ability to convert and transfer profits, debt service, and
     similar remittances related to insured investments from local
     currency to U.S.  dollars due to new currency restrictions. 
     OPIC does not protect against currency devaluation. 

  -- Expropriation:  The loss of investment due to nationalization,
     confiscation, or expropriation by a foreign government including
     "creeping" expropriation--government actions that deprive an
     investor of fundamental rights in a project for a period of at
     least 6 months.  Losses due to lawful regulatory or revenue
     actions by host governments as well as actions deemed to be
     provoked by the investor or foreign enterprise are excluded. 

  -- Political violence:  The loss of assets or income due to war,
     revolution, insurrection, or politically motivated civil strife,
     terrorism, and sabotage.  Actions, such as strikes, undertaken
     primarily to achieve labor or student objectives are not
     covered. 

In addition to these three areas, OPIC has specialized insurance
programs for financial institutions, leasing arrangements, natural
resource projects, oil and gas projects, and contractors and
exporters.  With limited exception, OPIC's insurance policies cover a
maximum of 90 percent of an eligible investment.  Policy terms can
extend up to 20 years\38 and are generally only cancelable by OPIC in
the event of default. 

OPIC operates in approximately 140 counties, including countries in
central and eastern Europe.  OPIC's outstanding exposure as of
September 30, 1996, totaled $13.4 billion.  This exposure is governed
by OPIC's statutory limitation and represents the amount for which
OPIC is contingently liable.  An adjustment of outstanding exposure
for standby coverage, for which OPIC is committed but not currently
at risk, yields OPIC's reported Current Exposure to Claims (CEC),
which was $6.4 billion for fiscal year 1996.  The face value of
aggregate insurance outstanding at the end of fiscal year 1996 was
$31.4 billion.  This represents the sum of all current and standby
coverage elected by investors. 

OPIC's insurance exposure has grown significantly in recent years. 
Between 1990 and 1996, OPIC's CEC almost doubled from $3.3 billion to
$6.4 billion.  About 45 percent of this 6-year increase occurred in
fiscal year 1995.  Demand for OPIC's insurance is expected to
continue to grow due to expanding international investment
opportunities.  The projected face value of insurance outstanding at
the end of fiscal year 1997 is $33.7 billion. 


--------------------
\38 The insurance term for loans, leases, and transactions is
generally equal to the duration of the underlying contract or
agreement. 


      ELIGIBILITY REQUIREMENTS
------------------------------------------------------ Appendix IV:4.3

OPIC's political risk insurance is available to U.S.  investors,
contractors, exporters, and financial institutions involved in
international transactions.  Specifically, OPIC's insurance program
covers U.S.  citizens, U.S.  companies that are more than 50 percent
owned by U.S.  citizens, foreign corporations that are at least 95
percent U.S.-owned, and other foreign entities that are 100 percent
U.S.-owned.  OPIC's insurance coverage is available for new
investments or expansion of existing enterprises. 

According to OPIC officials, they have discretion in determining the
insurability of projects and certain coverage may be unavailable or
limited due to underwriting or other reasons.  For example, coverage
amounts may be limited for investments in countries where OPIC has a
high portfolio concentration and for highly sensitive projects. 


      PROGRAM FINANCING
------------------------------------------------------ Appendix IV:4.4

OPIC's income is derived primarily from (1) interest earnings on
invested assets, (2) premiums, (3) recoveries and (4) fees.  In
addition, OPIC has the authority to borrow up to $100 million from
the U.S.  Treasury.  Premium rates for OPIC's insurance are based on
a standard pricing table for four different sectors with adjustments
made for project-specific risks.  Actual premiums may be increased or
decreased, generally by up to one-third of the base rate, depending
upon an insured project's risk profile.\39

For fiscal year 1996, interest earnings on funds invested in U.S. 
Treasury securities were OPIC's largest source of revenue, accounting
for 55 percent of OPIC's total revenue of $300 million.  Insurance
revenues--premiums ($80.5 million) and miscellaneous income ($1.0
million)--accounted for about 27 percent of OPIC's total revenues. 
The majority of the remainder of OPIC's revenues stemmed from its
investment financing activities. 

As a whole, OPIC has been self-sustaining with positive net income in
each year since its inception.  As shown in figure IV.7, since 1972
insurance premium collections have exceeded claim payments in all but
3 years, which were in OPIC's early years of operation.  As of
September 30, 1996, OPIC's insurance program collected premium
revenue totaling $922.5 million, paid cash settlements of just over
$288.8 million, and collected cash recoveries of $277.9 million,
resulting in total premium income net of claims of $911.5 million. 
In addition to cash claim payments, OPIC negotiated noncash
settlements of approximately $227 million.  At the end of fiscal year
1996, OPIC had $1.8 billion in insurance loss reserves and retained
earnings available for insurance losses. 

   Figure IV.7:  OPIC Insurance
   Premium Collections Versus
   Claim Payments, Fiscal Years
   1972-1996

   (See figure in printed
   edition.)

Source:  OPIC. 


--------------------
\39 Actual premiums for natural resource projects and projects with
investments of $50 million may vary by more than one-third. 


      CURRENT BUDGET TREATMENT
------------------------------------------------------ Appendix IV:4.5

OPIC's insurance activities are currently handled in one budget
account, the "Noncredit Account," a revolving fund that is a
discretionary account under BEA.  Although outlays are reported on a
cash basis, OPIC uses accrual concepts to obligate funding for claim
reserves.\40 According to OPIC officials, this reserve is used to
recognize losses that are probable and can be reasonably estimated in
accordance with private sector accounting standards as required by
the Government Corporation Control Act.  When a claim occurs, cash
payments are made from these reserves. 


--------------------
\40 To the extent that these reserves are a sufficient measure of the
risk assumed, OPIC is already using the aggregate budget authority
approach to accrual-based budgeting discussed in chapter 6. 


      RISK ASSESSMENT METHODS
------------------------------------------------------ Appendix IV:4.6

The risk assessment methods used by OPIC to establish insurance
reserves and set premium rates rely heavily on expert judgment and
are not highly quantitative.  According to OPIC officials, no
standard actuarial model exists for quantifying political risks. 
Although OMB has suggested using options pricing or other econometric
modeling approaches to assess the risk assumed by OPIC's insurance
program, these models have not been developed and some analysts we
spoke with expressed concerns about their cost-effectiveness and
usefulness for this purpose. 

In order to establish insurance reserves, OPIC analyzes, on a
quarterly basis, the losses inherent in the outstanding insurance
portfolio.  OPIC officials told us that a general nonspecific reserve
based on its entire insurance portfolio is used because
project-specific losses cannot be reasonably estimated.  Reserves are
established in consultation with OPIC's external auditors and are
based on OPIC management's evaluation of historical loss experience,
the composition and volume of current insurance commitments, and
anticipated worldwide political and economic conditions. 

As a starting point, OPIC uses a calculation based on historical
experience.  According to OPIC officials, the program's entire
historical experience is used to determine reserve levels because
claims have been sporadic over the life of the program and no
discernible pattern exists.  An OPIC official stressed that while
this historical-based calculation provides a useful starting point,
management's judgment is a key factor in determining the appropriate
reserve levels and additional adjustments are made to account for
OPIC's new business and other factors that affect the level of risk
assumed. 

OPIC also uses risk assessment to adjust rates from the standard
pricing tables for project-specific risk.  In determining the risk
associated with a particular project, OPIC considers (1)
project-specific risk, such as the structure of the project and the
experience of the project's sponsors and (2) country-based risk, such
as projections of the country's general economic conditions,
including balance of payments and foreign exchange reserve levels. 
According to OPIC officials, the level of risk for each project is
assessed individually during the rate-setting process.  However, OPIC
officials stressed that overall portfolio management is important in
controlling its overall risk exposure precisely because predicting
political risk for particular projects over long periods of time is
so difficult. 


      IMPLEMENTATION
      CONSIDERATIONS FOR OPIC'S
      POLITICAL RISK INSURANCE
------------------------------------------------------ Appendix IV:4.7


         ADEQUACY OF CURRENT
         BUDGET REPORTING
---------------------------------------------------- Appendix IV:4.7.1

OPIC's multiyear, long-term contracts commit the federal government
to pay future claims for extended periods.  As noted above, OPIC
currently uses accrual concepts to obligate funding for claims
inherent in insurance coverage outstanding.  Thus, to the extent that
these reserves represent the risk assumed by the federal government,
OPIC's current budgetary reporting is similar to the budget authority
approach used for accrual-based budgeting outlined in this report. 
However, a number of limitations to this approach have been
identified as follows: 

  -- Although budgetary reserves are obligated when they are
     realized, claim payments are not recognized in net outlays or
     the deficit until they come due.  As a result, the future cost
     of new or growing commitments may not be isolated or fully
     recognized at the time they are extended. 

  -- OMB raised concerns that general reserves based on historical
     experience and management judgment may not fully focus attention
     on the risk assumed for new insurance commitments at the time
     they are extended. 

  -- As a result, changes in the composition and riskiness of OPIC's
     insurance activities may not be fully recognized in the budget
     at the time the insurance is extended. 


      ISSUES IN IMPLEMENTING AN
      ACCRUAL-BASED BUDGETING
      APPROACH
------------------------------------------------------ Appendix IV:4.8

  -- Estimates of political risk and future claims are uncertain due
     to their dependence on variables that are inherently difficult
     to predict, such as the political stability of governments and
     long-term economic conditions. 

  -- Risk assessment is complicated by (1) the individualistic nature
     of the risk covered, (2) the lack of relevant historical data,
     and (3) the constant volatility of the international political
     and economic environment. 

  -- The nature of OPIC's insurance operations may not fit well with
     a credit reform model (an aggregate outlay approach using annual
     cohorts). 

According to OPIC officials, estimating the net cost for a particular
project would be complicated by (1) the uncertainties surrounding the
magnitude and timing of loss recoveries and (2) the difficulty of
allocating the benefits of overlapping contract provisions such as
agreements that limit total losses for the same company. 

OPIC insurance activities deal with a small number of diverse
projects with individually negotiated terms and thus implementation
challenges similar to those faced by international credit programs
under the Credit Reform Act are likely. 

  -- Based on their experience with credit reform, OPIC officials
     expressed concerns that the additional reporting requirements to
     comply with an aggregate outlay approach would divert staff
     resources from portfolio management and loss recovery activities
     that are critical to mitigating total losses. 


   FEDERAL WAR-RISK INSURANCE
   PROGRAMS
-------------------------------------------------------- Appendix IV:5

The two federal war-risk insurance programs--aviation and
maritime--provide insurance to commercial airlines and ship owners
during extraordinary circumstances, such as war and other
hostilities, in order to support the foreign policy interests of the
United States.  The Aviation War-Risk Insurance program was
established in 1951.\41 The War-Risk Insurance program for vessels
was established under Title XII of the Merchant Marine Act of 1936. 
Because of their similar purposes, these programs would likely face
common risk assessment and implementation challenges under an
accrual-based budgeting approach. 


--------------------
\41 49 U.S.C.   44301 et.  seq. 


      AVIATION WAR-RISK INSURANCE
------------------------------------------------------ Appendix IV:5.1

Budget Account:  Aviation Insurance Revolving Fund
(69-4120-0-3-402)

Department:  Department of Transportation (DOT)

Bureau:  Federal Aviation Administration (FAA)


         PURPOSE
---------------------------------------------------- Appendix IV:5.1.1

The Aviation War-Risk Insurance program was established to insure
commercial aircraft that provide essential air service during
extraordinary circumstances--such as war and other hostilities--when
such insurance is not available commercially or is only available on
unreasonable terms and conditions.\42


--------------------
\42 In November 1977, the Congress expanded FAA's authority to
provide all-risk insurance, but none has been issued to date. 


         COVERAGE
---------------------------------------------------- Appendix IV:5.1.2

FAA issues both hull and liability war-risk insurance.  Hull
insurance covers the aircraft itself.  Liability insurance covers
bodily injury to or the death of the crew and passengers and the loss
of or damage to cargo, property, and people on the ground.  The
maximum amount of hull and liability coverage provided under FAA's
war-risk insurance is limited to the amounts insured by an airline's
commercial policy.  The insured value of the hull cannot exceed the
fair and reasonable value of the aircraft. 

FAA issues two forms of war-risk insurance:  (1) nonpremium
insurance, which is provided at no cost to the airlines other than a
one-time registration fee and (2) premium insurance for which
airlines pay a risk-related premium.  Generally speaking, FAA's
nonpremium insurance covers flights performed directly for the
government and premium insurance covers other flights that are
considered necessary to support the foreign policy interests of the
United States.  According to FAA officials, coverage under both types
of insurance is issued sporadically and may remain active for only
limited durations. 

FAA registers aircraft for nonpremium insurance when the carriers
perform contract services for federal agencies that have
indemnification agreements with DOT.  Under the indemnification
agreement, these federal agencies reimburse FAA for the insurance
claims they pay to the airlines.  Over 99 percent of all war-risk
insurance issued has been nonpremium insurance for flights sponsored
by the Department of Defense (DOD).  According to FAA, the program
has issued nonpremium coverage several times since 1975.\43

Premium insurance is only provided when the President makes a
determination that flights to a specific location are necessary to
carry out the foreign policy interest of the United States. 
Authority for this type of insurance is provided for an initial
period of 60 days, with an additional 60-day extension granted when
it is considered necessary by the President.  According to FAA
officials, premium policies have only been issued during one period
since 1975 when 36 premium policies were in force during the Persian
Gulf conflict.  No premium policies have been issued since March
1991. 


--------------------
\43 According to FAA, the nonpremium aviation insurance has been
activated since 1975 as follows:  50 flights to Honduras were covered
between 1983 and 1984; approximately 5,000 flights into the Middle
East were covered from mid-1990 through mid-1991; one flight was
covered from Oman to Frankfurt in January 1991; 20 flights to Kuwait
were covered from November 1992 to April 1993; 155 flights into
Somalia were covered between late 1992 and early 1994; three flights
to Georgia were covered in early 1994; 32 flights to Haiti were
covered between September 1994 and October 1994; and 111 flights to
Bosnia were covered between April 15 and September 30, 1996. 


         PROGRAM FINANCING
---------------------------------------------------- Appendix IV:5.1.3

The program is financed primarily through interest on Treasury
securities, insurance premiums for flights insured by premium
insurance, and registration fees for flights insured with nonpremium
insurance.  From its inception through fiscal year 1996, the
program's revolving fund accumulated approximately $67 million in
revenues--including interest earnings--and paid out net claims
totaling only about $151,000. 

Despite the fund's positive position, the accumulated balance may be
insufficient to pay potential claims.  For example, in 1994, we
reported that claims for the loss of one aircraft--such as a Boeing
747-400 which can cost over $100 million--could liquidate the fund's
entire balance and still leave a substantial portion of the claim
unpaid.\44 If claims exceed the fund balance, FAA would have to seek
a supplemental appropriation to cover losses.  These potential
funding shortfalls would not only subject the government to
unexpected costs but, as we previously reported, may also reduce the
effectiveness of the program.\45


--------------------
\44 See Aviation Insurance:  Federal Insurance Program Needs
Improvements to Ensure Success (GAO/RCED-94-151, July 15, 1994). 

\45 GAO/RCED-94-151. 


         CURRENT BUDGET TREATMENT
---------------------------------------------------- Appendix IV:5.1.4

The Aviation War-Risk Insurance program is handled in one account,
the Aviation Insurance Revolving Fund.  Figure IV.8 shows the
program's budgeted versus actual outlays from fiscal years 1973
through 1996.  Negative outlays mean that the program's receipts
exceeded its outlays. 

   Figure IV.8:  Aviation
   Insurance Fund Budget Estimates
   Versus Actual Outlays, Fiscal
   Years 1973-1996

   (See figure in printed
   edition.)


      MARITIME WAR-RISK INSURANCE
------------------------------------------------------ Appendix IV:5.2

Budget Account:  War-Risk Insurance Revolving Fund
(69-4302-0-3-403)

Department:  Department of Transportation (DOT)

Bureau:  Maritime Administration (MARAD)


         PURPOSE
---------------------------------------------------- Appendix IV:5.2.1

The War-Risk Insurance program provides protection against loss or
damage from marine war risks in order to provide for the availability
of merchant vessels for national defense or to protect the continued
flow of U.S.  foreign commerce during periods when commercial
insurance cannot be obtained on reasonable terms and conditions. 


         COVERAGE
---------------------------------------------------- Appendix IV:5.2.2

Three types of war-risk insurance coverage are available:  (1)
interim binder, (2) section 1202, and (3) section 1205. 

  -- Interim binder:  Interim binder insurance is a standby emergency
     program which provides insurance coverage for 30 days for
     eligible vessels when their commercial war-risk insurance is
     terminated under automatic termination and cancellation clauses
     included in commercial policies.  According to MARAD, commercial
     insurance automatically terminates upon outbreak of war,
     declared or undeclared, among any of the five major powers--the
     United States, United Kingdom, France, the Russian Republic
     (formerly the Soviet Union), or The Peoples' Republic of China. 
     The binder policy provides immediate coverage so that the
     covered vessels can complete their voyages without interruption. 

  -- Section 1202:  Under Section 1202, the Secretary, with the
     approval of the President, can offer insurance and reinsurance
     against loss or damage caused by war risks to commercial vessels
     when commercial coverage cannot be obtained on reasonable terms
     and conditions.  Premiums are charged for this type of
     insurance. 

  -- Section 1205:  Under Section 1205, any United States department
     or agency may obtain from MARAD war risk insurance.  Insurance
     is provided without premiums in consideration of the insured
     agency's agreement to indemnify MARAD for all losses covered by
     such insurance. 


         PROGRAM FINANCING
---------------------------------------------------- Appendix IV:5.2.3

As noted above, the financing mechanisms vary among the different
types of war-risk coverage.  Under interim binder insurance, premiums
are to be established to cover losses.  If the original premiums do
not meet the losses, then a retroactive premium is to be assessed to
cover the losses without limit.  Therefore, under this agreement, the
fund should be self-supporting; however, MARAD officials noted that
this financing mechanism has never been tested.  For section 1202
coverage, risk-related premiums are charged.  In this case, the
government bears the full risk of losses on policies it issues. 
Under section 1205, MARAD is reimbursed by the insured agency or
department for losses covered by such insurance; thus the insured
agency or department bears the risk.  In addition, the program earns
interest on funds invested in Treasury securities. 

At the end of fiscal year 1996, the War-Risk Insurance Revolving Fund
had a balance of approximately $26 million.  However, despite this
positive balance, the fund may not have sufficient funds to cover
potential claims when the program is activated.  According to a MARAD
official, the values of covered vessels generally range from
approximately $2 million to $50 million. 


         CURRENT BUDGET TREATMENT
---------------------------------------------------- Appendix IV:5.2.4

The War-Risk Insurance program is reported in the War-Risk Insurance
Revolving Fund.  The account has permanent authority from offsetting
collections.  Program costs are mandatory but administrative expenses
are discretionary.  Figure IV.9 shows the program's budgeted versus
actual outlays since 1973.  Negative outlays mean that the program's
offsetting collections exceeded its outlays. 

   Figure IV.9:  War-Risk
   Insurance Revolving Fund Budget
   Estimates Versus Actual
   Outlays, Fiscal Years 1973-1996

   (See figure in printed
   edition.)


      RISK ASSESSMENT FOR THE
      WAR-RISK PROGRAMS
------------------------------------------------------ Appendix IV:5.3

The unique role of the maritime and aviation war-risk insurance
programs complicates risk assessment.  As noted above, the war-risk
insurance programs provide insurance to commercial airlines and ship
owners during extraordinary circumstances, such as war and other
hostilities, in order to protect the interests of the United States. 
Both programs provide coverage only when commercial insurance is not
available or is available only on unreasonable terms.  This unique
role complicates risk assessment because (1) the insured risks tend
to be case-specific and highly variable, (2) historical program data
are limited and (3) commercial sector war-risk insurance data are
unlikely to be directly applicable to the risk assumed by these
federal programs.  Currently, risk assessment for both programs
relies heavily on expert judgment.  Neither program uses quantitative
modeling or standard risk assessment procedures. 

The risks assumed by the federal war-risk insurance programs tend to
be case-specific and highly variable.  MARAD officials stressed that
each conflict is different and involves numerous factors.  FAA
officials told us that at the point commercial sector insurers leave
the market--and federal war-risk insurance is activated--the
calculation of risk becomes very difficult and subjective.  According
to FAA officials, it is not practical to develop a mathematical
formula to calculate appropriate premium rates due to the uniqueness
of each case. 

Officials from both agencies noted that the level of risk assumed by
the government can also be highly variable.  Coverage may remain
active only for short durations such as a few months, days, or even
hours.  For example, according to FAA officials, insured flights can
be in operation for only a few hours and only a portion of a
flight--as the plane flies through zones excluded by commercial
policies--may be covered.  In addition, given the extraordinary
conditions surrounding the issuance of federal war-risk insurance,
the level of risk assumed can change rapidly.  For these reasons,
MARAD officials also stressed that risk assessment is an on-going
process, requiring continuous reassessment. 

Both MARAD and FAA officials told us that because of the two
programs' infrequent activation and extremely rare losses, there is a
lack of historical program data for risk assessment.  As noted above,
according to FAA officials, aviation war-risk insurance has only been
issued during a few periods since 1975.  MARAD officials also stated
that its war-risk insurance is activated infrequently and remains
active for short durations, usually less than a year.  Further, not
only is the issuance of federal war-risk insurance infrequent, but
claims under the programs have also been extremely rare.  According
to agency officials, the Aviation War-Risk Insurance program has paid
out only four claims totaling about $151,000.  According to an agency
official, the only claims to date under MARAD's war-risk insurance
program occurred during the Vietnam Era and totaled approximately
$110,000 and were reimbursed by the Navy under Section 1205. 

Further, officials from both agencies told us that historical
information from commercial war-risk insurance may not be useful in
assessing the risk undertaken by their programs because commercial
information often is not readily available or applicable.  For
example, officials from both agencies stated that the basis for
setting commercial premiums generally is not released by private
sector companies.  In addition, FAA officials described the goals and
operations of the Aviation War-Risk Insurance program as
significantly different from those of commercial aviation insurance
because federal war-risk insurance is activated only infrequently,
for very short durations and under extreme conditions.  MARAD
officials added that historical loss information from commercial
policies is of limited use for projecting future losses of its
insurance programs because commercial policies also cover events
which are not war-related.  Nevertheless, MARAD officials told us
that, when available, they do consider the quoted commercial sector
rate for a particular voyage as a starting point in the risk
assessment process. 

Because of the above limitations, risk assessment for the federal
war-risk programs currently relies heavily on expert judgment. 
Premiums for both programs are set in consideration of the risk
involved, U.S.  policy interests, and to encourage the participation
of commercial insurers.  In general, risk assessment for the programs
involves the subjective evaluation of numerous factors associated
with a particular flight or voyage.  According to FAA officials, they
consider factors such as (1) the hull value, (2) the potential
liability for passengers, crew, cargo, and losses on the ground, and
(3) the apparent danger associated with flights into the area(s)
excluded by commercial insurers.  They told us that, in assessing the
risks associated with a particular area, they consider available
information on potential dangers, such as intelligence information on
terrorist groups, and the types of weapons involved in the conflict. 
If available, they consider historical losses in the area. 

FAA officials also noted that although they do not currently use a
standard risk assessment model, they are looking for ways to improve
risk assessment techniques.  For example, they have developed a
database of war-risk incidents since 1980 containing (1) the type of
incident, (2) the region where the incident occurred, (3) a text
section describing the incident, and (4) the value of the aircraft. 
According to officials, this database will be used as a reference and
training tool.  In addition, the agency is studying the actuarial
process used by the private sector.  Although not directly applicable
to their programs, they said they are interested in what might be
learned from the private sector methods. 

MARAD officials also described their risk assessment process as ad
hoc and judgmental.  For example, during the Persian Gulf conflict, a
committee was established to determine premium rates.  They said that
a number of factors were considered in assessing risk, such as (1)
the destination of the vessels, (2) the extent of the military
threat, (3) the current commercial rates, and (4) the value of the
vessels.  Although in a few cases, historical information can be
useful in assessing conditions, such as the military threat in a
particular area, MARAD officials noted that the number of cases in
which historical information is available and useful is very limited. 
According to agency officials, an outside consultant, the American
War-Risk Agency, has provided advice on risk assessment. 

Overall, agency officials for the war-risk insurance programs
expressed concerns that accrual-based budgeting may not be feasible
or useful for their programs.  They described the infrequent and
limited issuance of insurance and the resulting lack of historical
experience as a key obstacle to developing risk-assumed estimates and
using accrual-based budgeting for these programs.  Because of the
programs' unique roles, it is difficult to effectively pool risk or
to develop discernible loss patterns.  Further, the emergency
(standby) nature of the programs makes it difficult to know in
advance when the programs will be activated and limits the time
available for risk assessment.  FAA officials stated that, in their
opinion, it was not feasible to generate a reliable risk-assumed
estimate for the budget.  MARAD officials provided a similar
assessment for their program, stating that given the nature of the
program, reliable estimates of the risk assumed could not be
developed. 


      IMPLEMENTATION
      CONSIDERATIONS FOR THE
      WAR-RISK INSURANCE PROGRAMS
------------------------------------------------------ Appendix IV:5.4


         ADEQUACY OF CURRENT
         BUDGET REPORTING
---------------------------------------------------- Appendix IV:5.4.1

  -- Although infrequently activated, when in force the war-risk
     insurance programs expose the federal government to potential
     unfunded claims without recognizing these potential funding
     shortfalls at the time the insurance is extended.  For each of
     the war-risk programs, one major loss could deplete the
     program's fund balance and leave a portion of the claim unpaid. 

  -- The amount of risk assumed by the federal government is not
     explicitly recognized in the budget process. 

  -- However, the government's budgetary exposure may be limited
     because of the war-risk programs' infrequent activation and
     limited coverage. 


         ISSUES IN IMPLEMENTING AN
         ACCRUAL-BASED BUDGETING
         APPROACH
---------------------------------------------------- Appendix IV:5.4.2

  -- Significant uncertainty will surround the risk-assumed estimates
     for the war-risk insurance programs because of the volatile
     nature of the risk and the lack of relevant historical data. 

  -- The unique role of war-risk programs may make the use of
     accrual-based budgeting difficult because the need for coverage
     may not be apparent during the normal budget cycle. 

  -- A decision would have to be made as to whether accrual-based
     budgeting should be applied only to the premium portion of the
     war-risk programs or to the nonpremium portions as well.  If
     applied to the nonpremium portions, additional implementation
     issues would have to be resolved.  For example, for insurance
     provided under indemnification agreements, it would have to be
     determined whether the insured agency or FAA should report the
     accrued cost for the risk assumed by the government. 

  -- The combination of the catastrophic nature of war-risk losses
     and the limited number of policies issued may impede the
     establishment of sufficient reserves, even if costs were
     measured on an accrual basis. 

  -- The use of accrual-based budgeting may not lead to significant
     policy changes because of the war-risk programs' limited
     activation and unique roles. 


   NATIONAL VACCINE INJURY
   COMPENSATION PROGRAM
   (POST-1988)
-------------------------------------------------------- Appendix IV:6

Budget Account:  Vaccine Injury Compensation Program Trust Fund
(20-8175-0-7-551)

Agency:  Department of Health and Human Services (HHS)

Bureau:  Health Resources and Services Administration (HRSA)


--------------------
\46 Claims resulting from vaccines administered prior to October 1,
1988, are treated separately (pre-1988 program) and paid with general
fund appropriations.  The deadline for filing claims under the
pre-1988 program has expired.  This appendix provides a general
summary of the ongoing post-1988 program. 


      PURPOSE
------------------------------------------------------ Appendix IV:6.1

The National Vaccine Injury Compensation Program (VICP) was
established by the National Childhood Vaccine Injury Act of 1986
(Public Law 99-660).  VICP, which went into effect in October 1988,
is a no-fault alternative to state tort law and private liability
insurance systems for compensating individuals, including adults, who
have been injured by vaccines routinely administered to children. 
The program was intended to improve the stability of the childhood
vaccine market by reducing the adverse impact of the tort system on
vaccine supply, cost, and innovation.\47 VICP is administered jointly
by the United States Court of Federal Claims, the Department of
Health and Human Services (HHS), and the Department of Justice (DOJ). 


--------------------
\47 The intent of the Congress is reflected in H.  Rept.  908, 99th
Cong., 2nd Sess.  (1986) which states "manufacturers have become
concerned .  .  .  with the availability of affordable product
liability insurance that is used to cover losses related to vaccine
injury cases .  .  .  there is little doubt that vaccine
manufacturers face great difficulty in obtaining insurance." Agency
officials, however, contend that VICP is not an insurance program
because (1) there is no insurance contract between VICP and
manufacturers, (2) the program is funded through an excise tax on
manufacturers and not premiums, and (3) a lawsuit must be filed to
receive compensation.  As noted in chapter 1, there is not universal
agreement on which programs constitute federal insurance, but the
factors cited by the agency do not necessarily preclude classifying
VICP as insurance.  Explicit insurance contracts do not exist for
many federal insurance programs and a program's financing mechanism
does not affect the government's liability.  The requirement that
injured persons begin compensation proceedings in claims court has no
bearing on whether to classify VICP as insurance since the party
being provided something akin to liability insurance is the
manufacturer. 


      COVERAGE
------------------------------------------------------ Appendix IV:6.2

VICP compensates individuals or families of individuals who have been
injured by childhood vaccines, whether administered in the private or
public sector.\48 Compensation for petitioners alleging
vaccine-related injuries is provided through a no-fault
administrative hearing process conducted by Special Masters of the
U.S.  Court of Federal Claims.  The vaccine manufacturer and whoever
administered the vaccine are not involved as parties to the
proceedings.  Awards for vaccine-related deaths are limited to
$250,000 plus attorney's fees and costs.  There is no limitation on
the amount of an award in a vaccine-related injury; however, the law
does contain certain restrictions. 

Petitioners may not obtain compensation from both the program and
litigation.  Claims arising from post-1988 claims in excess of $1,000
or of an unspecified amount must be processed through VICP before a
civil suit may be brought against a vaccine manufacturer or
administrator. 


--------------------
\48 VICP compensation is secondary to all insurance coverage except
Medicaid. 


      ELIGIBILITY REQUIREMENTS
------------------------------------------------------ Appendix IV:6.3

An individual claiming injury from a vaccine must file a petition for
compensation with the claims court.  In order to qualify for
compensation a petitioner must: 

  -- show that an injury found on the Vaccine Injury Table occurred
     within a specified period of time after receiving a vaccination,
     or

  -- prove that the vaccine caused the condition, or

  -- prove that the vaccine significantly aggravated a pre-existing
     condition. 

The Vaccine Injury Table lists specific injuries or conditions and
the time frames in which they must occur after a vaccine is
administered.  According to HRSA, the Injury Table is a legal
mechanism for defining complex medical conditions and allows a
statutory "presumption of causation."\49


--------------------
\49 Most claims allege that a Table Injury occurred because it is
much easier to demonstrate a Table Injury than to prove that a
vaccine caused a condition.  However, compensation is not awarded if
the court determines that the injury or death was due to a cause
unrelated to the vaccine. 


      PROGRAM FINANCING
------------------------------------------------------ Appendix IV:6.4

The Vaccine Injury Compensation Program Trust Fund is supported by
revenues from an excise tax on vaccine manufacturers and interest
earned on fund balances invested in Treasury securities.  Each
vaccine has a predetermined per dose excise tax rate.\50

According to HRSA officials, rates are related to "perceived" risk
but are not based on empirical risk assessment.  Gross excise tax
receipts are reduced by 25 percent before being transferred from the
General Fund to the Vaccine Trust Fund.\51

The Vaccine Trust Fund has a significant and growing balance.  The
fund balance as of the end of fiscal year 1996 was $1.0 billion. 
Although significant uncertainty surrounds future claims, the risk of
injury or death due to vaccination is considered extremely small. 
Figure IV.10 shows the excise tax receipts and budget obligations for
claim payments for the Vaccine Injury Compensation Trust Fund since
fiscal year 1988. 

   Figure IV.10:  VICP Excise Tax
   Receipts Versus Obligations for
   Claim Compensation, Fiscal
   Years 1988-1996

   (See figure in printed
   edition.)

Note:  Excise tax receipts for fiscal year 1990 reflect the
termination of taxes on December 31, 1992 and reenactment of taxes
effective August 10, 1993. 

Source:  Budget of the United States Government, Appendix. 


--------------------
\50 Exported vaccines are not subject to these excise taxes, except
where the export is to a U.S.  possession. 

\51 The Omnibus Budget Reconciliation Act of 1987 requires that net
revenues be transferred from the general fund to the VICP Trust Fund. 


      CURRENT BUDGET TREATMENT
------------------------------------------------------ Appendix IV:6.5

The VICP (post-1988) is reported in the budget in a single account,
the Vaccine Injury Compensation Trust Fund. The majority of the
program's spending is mandatory.  Claim payments and the
administrative expenses of the public health service are mandatory
while the administrative costs for the Court of Federal Claims and
DOJ are discretionary.  Figure IV.11 shows the program's budget
estimates versus actual outlays from fiscal years 1989 through 1996. 

   Figure IV.11:  Vaccine Injury
   Compensation Program Budget
   Estimates Versus Actual
   Outlays, Fiscal Years 1989-1996

   (See figure in printed
   edition.)


      RISK ASSESSMENT METHODS
------------------------------------------------------ Appendix IV:6.6

Risk assessment for the post-1988 VICP is complicated by several
factors including (1) the program's limited historical experience,
(2) the lack of scientific evidence linking adverse events to
vaccines, and (3) the dynamic or subjective nature of some variables
such as injury coverage and settlement amounts.  These factors
increase the difficulty of assessing the many variables required to
estimate the aggregate awards that are likely to be paid for
vaccinations administered in a particular year, such as (1) the
number of vaccines administered, (2) the frequency of adverse
reactions, (3) the probability that petitions will be filed following
an adverse reaction,\52 (4) the filing of petitions for adverse
reactions that are not attributable to vaccinations,\53 and (5) the
probability and amount of awards. 

VICP's limited historical experience is a key factor in the
uncertainty surrounding estimates of the program's future costs. 
Although the VICP has been in existence since 1988, a full cohort of
cases has not yet been resolved.  As a result, a 1994 Department of
Treasury study concluded that VICP had not been in existence long
enough to project future outlays with confidence.\54 Both a
Department of Treasury analyst who worked on the report and HRSA
officials reiterated that, until more historical data on closed cases
become available, estimates of the program's future outlays will be
uncertain.  Specifically, HRSA officials stressed that, in their
opinion, there is not sufficient historical evidence on the cost of
claims to produce meaningful estimates of the program's future costs. 

According to HRSA officials, scientific data also may not be useful
in predicting the program's future claims.  HRSA officials told us
that although the Food and Drug Administration (FDA) tests vaccines
to prove safety and identify potential side effects, these studies
would not be very useful in determining future claims under VICP
because they do not estimate the number of injuries that are likely
to occur over a long period of time.  The Treasury report confirms
that although new vaccines "may be proven to be completely acceptable
in clinical trials involving thousands of doses, a few adverse
reactions may still occur when doses are administered routinely to
millions of children."\55

HRSA officials added that calculating the risk associated with
vaccines is becoming increasingly difficult with the use of combined
antigens in single vaccinations. 

In addition, HRSA officials expressed concern that the dynamic or
subjective nature of some variables makes it difficult, if not
impossible, to generate reasonable projections of the program's
future claims.  For example, agency officials noted that changes in
the injuries covered by the program make it more difficult to assess
the amount of risk associated with the program because a change in
the injury table means a change in the risk involved.  Further, HRSA
officials described award amounts as case-specific and subjective. 
According to an HRSA official, the program's obligations are derived
from court judgments which vary from year to year and are not
"susceptible to the type of actuarial analysis that is an integral
part of insurance schemes." Additional factors, such as the
introduction of new vaccinations and changes in the recommended
vaccination schedules, may also complicate risk assessment. 

Overall, HRSA officials expressed serious reservations about their
ability to produce reasonable projections of the program's future
costs and the use of accrual-based budgeting for the program.  HRSA
officials stressed that risk assessment has never been and is not
currently used for VICP.  They stated that, in their opinion, there
currently is no meaningful way to quantify the program's risk. 

The Treasury report concurred that until the program matures, program
outlays cannot be estimated with confidence, but noted that "as the
program matures sufficient program data will become available to
permit more sophisticated methods of estimating future outlays to be
used."\56 A Treasury analyst noted that it may not be necessary to
establish causation between the vaccine and the adverse event in
order to establish an estimate of the program's future outlays.  For
example, as more cases are closed, it may be possible to establish a
pattern between adverse events and award amounts based on historical
data.  However, changes in variables over time, such as injury
coverage and the introduction of new vaccines, will have an impact on
the usefulness of estimates based on historical data. 


--------------------
\52 Not all individuals who are eligible for compensation under the
program will file claims.  For example, a 1994 Department of the
Treasury report, entitled National Vaccine Injury Compensation
Program:  Financing the Post-1988 Program and Other Issues, points
out that some victims of adverse reactions may be compensated through
ordinary health insurance. 

\53 A petition does not require evidence proving a causal
relationship between administration of a vaccine and an adverse
event. 

\54 Department of the Treasury, National Vaccine Injury Compensation
Program, p.  v. 

\55 Department of the Treasury, National Vaccine Injury Compensation
Program, p.  21. 

\56 Department of the Treasury, National Vaccine Injury Compensation
Program, p.  v. 


      IMPLEMENTATION
      CONSIDERATIONS FOR VICP
------------------------------------------------------ Appendix IV:6.7


         ADEQUACY OF CURRENT
         BUDGET REPORTING
---------------------------------------------------- Appendix IV:6.7.1

  -- The current cash-based budget does not recognize the program's
     future claims costs.  Because tax receipts are not matched with
     potential claim payments, policymakers may not receive timely
     signals of the reasonableness of the program's financing levels. 
     However, there does not currently appear to be a funding
     deficiency. 

  -- The current cash-based budget may not prompt decisionmakers to
     consider the implications of changes in the level of risk
     assumed by the government at the time the changes are made. 


         ISSUES ASSOCIATED WITH
         IMPLEMENTING AN
         ACCRUAL-BASED BUDGETING
         APPROACH
---------------------------------------------------- Appendix IV:6.7.2

  -- Both the magnitude and timing of the cost of VICP (post-1988)
     future claims are uncertain. 

  -- HRSA officials expressed concern about the staff resources
     required to implement an accrual-based budgeting approach. 




(See figure in printed edition.)Appendix V
COMMENTS FROM THE OFFICE OF
MANAGEMENT AND BUDGET
========================================================== Appendix IV

See comment 1. 

See comment 2. 

On p.  5. 



(See figure in printed edition.)



(See figure in printed edition.)

See comment 2. 

Now on p.  22
and p.  23. 

See comment 3. 

Now on p.  89. 



(See figure in printed edition.)



(See figure in printed edition.)

See comment 4. 


The following are GAO's comments on the Office of Management and
Budget's letter dated July 1, 1997. 


   GAO COMMENTS
-------------------------------------------------------- Appendix IV:7

1.  We have incorporated OMB's technical comments in the report as
appropriate but have not reprinted them in this appendix. 

2.  Section omitted. 

3.  Discussed in the Executive Summary and the "Agency Comments and
Our Evaluation" section of chapter 8. 

4.  Statements omitted. 


MAJOR CONTRIBUTORS TO THIS REPORT
========================================================== Appendix VI


   ACCOUNTING AND INFORMATION
   MANAGEMENT DIVISION,
   WASHINGTON, D.C. 
-------------------------------------------------------- Appendix VI:1

Christine E.  Bonham, Assistant Director
James R.  McTigue, Jr., Evaluator-in-Charge
Elizabeth A.  McClarin, Senior Evaluator
Eileen T.  McGowan, Intern


   OFFICE OF THE GENERAL COUNSEL
-------------------------------------------------------- Appendix VI:2

Carlos E.  Diz, Attorney

*** End of document. ***