oversight

Farmer Mac: Some Progress Made, but Greater Attention to Risk Management, Mission, and Corporate Governance Is Needed

Published by the Government Accountability Office on 2003-10-16.

Below is a raw (and likely hideous) rendition of the original report. (PDF)

               United States General Accounting Office

GAO            Report to Congressional Requesters




October 2003
               FARMER MAC
               Some Progress Made,
               but Greater Attention
               to Risk Management,
               Mission, and
               Corporate Governance
               Is Needed




GAO-04-116

               a

                                                October 2003


                                                FARMER MAC

                                                Some Progress Made, but Greater
Highlights of GAO-04-116, a report to           Attention to Risk Management, Mission,
congressional requesters
                                                and Corporate Governance Is Needed



In the late 1990s, GAO found that               Farmer Mac’s income has increased since 1999, and its capital continues to
the Federal Agricultural Mortgage               exceed required levels. At the same time, we identified trends that indicated
Corporation (Farmer Mac), a                     a more complex risk profile. For example, its off-balance sheet standby
federal government-sponsored                    agreements have grown 350 percent in 3 years and comprise nearly 50
enterprise, had significant assets in           percent of Farmer Mac’s total loans. To date, the underlying loans have been
nonmission investments and
analyzed its long-term viability.
                                                performing better than the on-balance sheet loans. However, if rapid growth
Recently, Congress asked GAO to                 in standby agreements continues, and Farmer Mac were to undergo stressful
report on Farmer Mac’s (1)                      economic conditions, it could face substantial funding liquidity risk. Farmer
financial condition, (2) mission, (3)           Mac has risk management systems in place, but certain aspects of its risk
corporate governance, and (4)                   management capacity have not kept pace with its increasingly complex
oversight provided by the Farm                  portfolio. For example, the loans used in the loss estimation model have
Credit Administration (FCA).                    characteristics that differ from Farmer Mac’s portfolio both with respect to
                                                geographic distribution and interest rate terms. In addition, although Farmer
                                                Mac has maintained sufficient liquidity to support its loan purchase and
GAO recommends that Farmer Mac                  guarantee activity, it has lacked a formal contingency plan to address
improve its risk management and                 potential liquidity needs under stressful agricultural economic conditions.
corporate governance practices,
including improve its loan loss                 Since our 1999 report, Farmer Mac has significantly reduced the ratio of
estimation model, and develop a                 nonmission investments and correspondingly increased its mission
contingency funding liquidity plan.             activities—providing long-term credit to farmers and ranchers at stable
GAO also recommends that FCA                    interest rates. These activities include loan purchases, guarantees, and
improve the model that analyzes                 commitments related to agricultural mortgages. However, there is
Farmer Mac’s credit risk and assess             geographic and lender concentration in the loan and guarantee portfolio, and
Farmer Mac’s impact on the                      the overall impact of the activities on the agricultural real estate market is
agricultural real estate market.                unclear. Farmer Mac’s enabling legislation lacks specific or measurable
                                                mission-related criteria that would allow for a meaningful assessment of its
GAO suggests that Congress
consider legislative changes that               mission achievement. In addition, the depth and liquidity of the current
would establish clearer,                        market for agricultural mortgage backed securities (AMBS) is unknown
measurable mission goals for                    because Farmer Mac’s strategy of holding AMBS has been a contributing
Farmer Mac; amend Farmer Mac’s                  factor in limiting the development of a liquid, secondary market for these
board structure; and allow FCA to               securities.
adjust capital standards for Farmer
Mac.                                            The Sarbanes-Oxley Act of 2002 and the proposed New York Stock
Farmer Mac agreed with some                     Exchange (NYSE) listing standards are both applicable to Farmer Mac
GAO findings and                                because its securities are publicly traded and listed on the NYSE. However,
recommendations and did not                     Farmer Mac’s efforts to meet the new standards regarding an independent
address others. FCA agreed with                 board could be limited by its statutory board structure. Under its statute,
GAO’s findings and                              two-thirds of the board’s directors are elected by institutions that have a
recommendations.                                business relationship with Farmer Mac and own the only two classes of
                                                voting stock. Since 2002, FCA enhanced oversight of Farmer Mac by
www.gao.gov/cgi-bin/getrpt?GAO-04-116           performing a more thorough annual safety and soundness examination, and
To view the full product, including the scope
                                                by proposing liquidity standards and regulatory limits for nonmission
and methodology, click on the link above. For   investments. However, FCA still faces challenges, including limitations in its
more information, contact Ms. Davi M.           tools to analyze capital and credit risk, as well as the lack of criteria and
D’Agostino or Ms. Jeanette Franzel at (202)
512-8678.                                       procedures to assess and report on Farmer Mac’s mission achievement.
Contents




Letter
                                                                                                 1
                              Background                                                                2

                              Results in Brief 
                                                        7
                              Farmer Mac’s Financial Condition Has Improved, but Risk 

                                Management Practices Have Not Kept Pace with Its More
                                Complex Risk Profile                                                   12
                              Mission-Related Activities Have Increased, but Impact of Activities
                                on Agricultural Real Estate Market Is Unclear                          30
                              Farmer Mac’s Statutory Governance Structure Does Not Reflect
                                Interests of All Shareholders and Some Corporate Governance
                                Practices Need to Be Updated                                           38
                              FCA Has Taken Steps to Enhance Oversight of Farmer Mac, but
                                Faces Challenges That Could Limit the Effectiveness of Its
                                Oversight                                                              47
                              Conclusions                                                              54
                                                                                                       

                              Recommendations
                                                         57
                              Matters for Congressional Consideration 
                                59
                              Agency Comments and Our Evaluation 
                                     59


Appendixes
               Appendix I:    Objectives, Scope, and Methodology                                       63
               Appendix II:   Farmer Mac’s Programs and Products                                       67
              Appendix III:   Financial Trends and Comparisons with Other Entities                     69
                              Revenue Has Increased, but Some Financial Performance Indicators
                                Lag Comparative Entities                                               72
              Appendix IV:    Farmer Mac’s Underwriting Standards                                      78
               Appendix V:    Interest Rate Risk                                                       81
                              Asset-Liability Management                                               81
                              Prepayment Model                                                         82
                              Farmer Mac’s IRR Measurement Process                                     84
              Appendix VI:    Farm Credit Administration Credit Risk Model                             86

                              Data Limitations                                                         87

                              Model Limitations                                                        91

             Appendix VII:    Comments from the Federal Agricultural Mortgage 

                              Corporation                                                              95

                              GAO Comments                                                            100

             Appendix VIII:   Comments from the Farm Credit Administration                            102




                              Page i                                                 GAO-04-116 Farmer Mac
                          Contents




                          GAO Comments                                                          107
             Appendix IX:	 GAO Contacts and Staff Acknowledgments                               112
                           GAO Contacts                                                         112
                           Acknowledgments                                                      112


Glossary of Terms                                                                               113


Tables                    Table 1: Loans Covered by Standby Agreements                           13
                          Table 2: Comparison of Total Minimum Capital Levels Per $100 of
                                   Loans                                                         36
                          Table 3: Annual Compensation and Options Granted for CEO’s of
                                   Farmer Mac and Housing GSEs                                   46


Figures                   Figure 1: Percentage of Outstanding Balance of Loans, AMBS and 

                                     Standby Agreements, as of December 31, 2002
                 6
                          Figure 2: Long-term Interest Rates on Loans Secured by 

                                     Agricultural Real Estate
                                   32
                          Figure 3: Securitization Status of Farmer Mac I Portfolio, as of 

                                     December 31, 2002
                                          33
                          Figure 4: Farmer Mac Portfolio Exposure by Loan Origination 

                                     Type
                                                       35
                          Figure 5: Farmer Mac I Geographic Concentration of Exposure by 

                                     Region, as of December 31, 2002
                            37
                          Figure 6: Farmer Mac’s Impaired Loans from 1997 to 2002 
              70
                          Figure 7: Farmer Mac’s Nonperforming to Total Loans Compared

                                     to Other Entities, as of December 31, 2002
                 71
                          Figure 8: Income by Program Assets 
                                   72
                          Figure 9: Farmer Mac’s ROA Compared to Other Entities
                 74
                          Figure 10: Farmer Mac’s ROE Compared to Other Entities                 75
                          Figure 11: Farmer Mac’s Capital to Asset Ratios Compared to Other
                                     Entities                                                    76




                          Page ii                                              GAO-04-116 Farmer Mac
Contents




Abbreviations

ACA         Agricultural Credit Association

AMBS        agricultural mortgage-backed securities

CEO         chief executive officer

Fannie Mae  Federal National Mortgage Association

Farmer Mac  Federal Agricultural Mortgage Corporation

FCA         Farm Credit Administration

FCBT        Farm Credit Bank of Texas

FCS         Farm Credit System

FHA         Federal Housing Administration

FHESSA      Federal Housing Enterprises Financial Safety and 

            Soundness Act
FHFB        Federal Housing Finance Board
FLCA        Federal Land Credit Association
Freddie Mac Federal Home Loan Mortgage Corporation
GAAP        generally accepted accounting principles
GSE         government-sponsored enterprise
HUD         Housing and Urban Development
IRR         interest rate risk
LIBOR       London Interbank Offered Rate
LTV         loan-to-value
MTN         medium term-notes
MVE         market value of equity
NII         net interest income
NYSE        New York Stock Exchange
OFHEO       Office of Federal Housing Enterprises Oversight
OSMO        Office of Secondary Market Oversight
QRM         Quantitative Risk Management
ROA         return on assets
ROE         return on equity
SAB         staff accounting bulletin
SEC         Security Exchange Commission
SFAS        Statement of Financial Accounting Standard
SPI         subordinated participation interest
USDA        U.S. Department of Agriculture



Page iii                                             GAO-04-116 Farmer Mac
Contents




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




Page iv                                                           GAO-04-116 Farmer Mac
A

United States General Accounting Office
Washington, D.C. 20548



                                    October 16, 2003


                                    The Honorable Thad Cochran

                                    Chairman

                                    The Honorable Tom Harkin

                                    Ranking Minority Member

                                    Committee on Agriculture, Nutrition, and Forestry

                                    United States Senate


                                    The Honorable Richard G. Lugar

                                    United States Senate


                                    Farmer Mac is a government-sponsored enterprise (GSE)1 established by 

                                    Congress to create a secondary market in agricultural real estate and rural 

                                    housing loans, and improve the availability of agricultural mortgage credit. 

                                    In 1998 and 1999, we found that a significant amount of Farmer Mac’s 

                                    assets were in nonmission investments and we discussed issues 

                                    surrounding the long-term viability of Farmer Mac.2 Recently, you asked us 

                                    to conduct a comprehensive review of Farmer Mac. This report discusses 

                                    (1) Farmer Mac’s current financial condition and risk management 

                                    practices; (2) the extent to which Farmer Mac has achieved its statutory 

                                    mission; (3) Farmer Mac’s corporate governance as it pertains to board 

                                    structure and oversight, and executive compensation; and (4) the Farm 

                                    Credit Administration’s (FCA) oversight of Farmer Mac.


                                    To address these objectives, we analyzed trends in Farmer Mac’s key

                                    indicators of financial performance and condition for fiscal year 2002—

                                    including measures of earnings and profitability, capital, liquidity, and its 

                                    asset and liability mix—and determined how Farmer Mac has managed and





                                    1
                                     As used in this report, a GSE is a federally chartered, privately owned corporation
                                    established by Congress to provide a continuing source of credit nationwide to a specific
                                    economic sector.
                                    2
                                     U.S. General Accounting Office, Government Sponsored Enterprises: Federal Oversight
                                    Needed for Nonmortgage Investments, GAO/GGD-98-48 (Washington, D.C.: Mar. 11, 1998).
                                    U.S. General Accounting Office, Farmer Mac: Revised Charter Enhances Secondary Market
                                    Activity, but Growth Depends on Various Factors, GAO/GGD-99-85 (Washington, D.C.: May
                                    21, 1999). In this report, we reviewed the progress that Farmer Mac had made in achieving
                                    its statutory mission and examined its future viability.




                                    Page 1                                                             GAO-04-116 Farmer Mac
              measured the risks it faces—credit, liquidity, and interest rate risk.3 We
              reviewed documents and interviewed representatives from Farmer Mac,
              FCA, other market participants, and individuals with expertise in the
              agricultural real estate market. We analyzed Farmer Mac’s loan portfolio
              growth. We obtained and reviewed FCA’s previous examinations and its
              most recent examination of Farmer Mac and other consultants’ studies
              related to Farmer Mac. We did not report specific details of Farmer Mac’s
              investment and loan portfolio nor details of reports of auditors,
              consultants, and examiners because of the proprietary nature of such
              information.

              We conducted our work in California, Indiana, New York, Virginia, and
              Washington, D.C., between August 2002 and May 2003 in accordance with
              generally accepted government auditing standards. Appendix I contains a
              detailed description of the scope and methodology of our work.



Background	   Farmer Mac, a GSE, was chartered by Congress in the Farm Credit Act of
              1971, as amended by the Agricultural Credit Act of 1987 (the 1987 Act).4 It
              is a federally chartered and privately operated corporation that is publicly
              traded on the New York Stock Exchange. Farmer Mac is also an
              independent entity within the Farm Credit System (FCS), which is another
              GSE. As an FCS institution, Farmer Mac is subject to the regulatory
              authority of FCA. FCA, through its Office of Secondary Market Oversight
              (OSMO), has general regulatory and enforcement authority over Farmer
              Mac, including the authority to promulgate rules and regulations governing
              the activities of Farmer Mac and to apply its general enforcement powers
              to Farmer Mac and its activities. According to the 1987 Act, Farmer Mac, in
              extreme circumstances, may borrow up to $1.5 billion from the U.S.
              Treasury to guarantee timely payment of any guarantee obligations of the
              corporation.5

              Under the 1987 Act, Farmer Mac’s mission is to provide for a secondary
              marketing arrangement for agricultural real estate and rural housing loans
              subject to its underwriting standards. A secondary market is a financial


              3
              See Background section for definitions.

              4
              Pub. L. No. 100-233.

              5
              Id.





              Page 2                                                  GAO-04-116 Farmer Mac
market for buying and selling loans, individually or by securitizing them.
When loans are securitized, they are repackaged into a “pool” by a trust in
order to be sold to investors. By returning cash to primary lenders in
exchange for their loans, theoretically, a secondary market would generate
additional funds for the lenders to lend and enhance the lenders’ ability to
manage credit and interest rate risk. Ideally, a Farmer Mac-sponsored
secondary market would increase liquidity to lenders by providing the
lenders access to national capital markets. This in turn would reduce
regional imbalances in loanable funds and possibly increase the overall
availability of credit to the primary agricultural real estate market and
lower interest rates for borrowers.

To relieve structural impediments that had limited Farmer Mac’s ability to
function efficiently, Congress passed the Farm Credit System Reform Act
of 1996 (the 1996 Act), which significantly revised Farmer Mac’s statutory
authority and had significant impact on Farmer Mac’s operations.6 Among
other things, the 1996 Act allowed Farmer Mac to (1) purchase agricultural
mortgage loans directly from lenders, “pool” the loans, and issue and sell
securities that are backed by these pools to investors and (2) eliminate the
mandatory requirement for loan originators and poolers to retain 10
percent, first-loss subordinated participation interest (SPI) with each
securitized loan pool. 7

Farmer Mac operates a cash window program where Farmer Mac
purchases mortgages directly from lenders for cash and purchases bonds
from agricultural lenders. (See app. II for Farmer Mac’s programs and
products.) Periodically, Farmer Mac transfers its purchased loans into
trusts that it uses as vehicles for the securitization of those loans.
Securitization is the transfer of assets (in this case, loans) to a third party
or trust. In turn, the third party or trust issues certificates to investors.
Farmer Mac refers to the certificates sold to investors as “guaranteed
securities” or as agricultural mortgage-backed securities (AMBS). The cash
flow from the transferred loans supports repayment of the AMBS. Farmer
Mac guarantees timely payments to investors holding the certificates,



6
Pub. L. No. 104-105.
7
 SPI is the right to receive a portion of the principal and interest payments on a loan or pool
of loans, but only after other investors in the Farmer Mac-guaranteed securities backed by
these pools have received all payments due to them. Originators could have retained SPIs in
the loans they sold to Farmer Mac or they could have sold SPIs to a pooler.




Page 3                                                                GAO-04-116 Farmer Mac
regardless of whether the trust has actually received such scheduled loan
payments.

Farmer Mac loan programs are divided into two main groups referred to as
Farmer Mac I and Farmer Mac II. Farmer Mac I consists of agricultural and
rural housing mortgage loans that do not contain federally provided
primary mortgage insurance. For loans underlying pre-1996 Act Farmer
Mac I AMBS, 10-percent first-loss subordinated interests mitigate Farmer
Mac’s credit risk exposure. Before Farmer Mac incurs a credit loss, losses
are first absorbed by the poolers’ or originators’ subordinated interest. As
of December 31, 2002, Farmer Mac had not experienced any credit losses
related to the pre-1996 Act Farmer Mac I AMBS, and the first-loss
subordinated interests are expected to exceed the estimated credit losses
on those loans. Current risks in Farmer Mac’s loan and guarantee portfolio,
such as those discussed later in this report, are generated primarily by post-
1996 guaranteed securities.

Farmer Mac receives an annual guarantee fee from the third party or trust
involved based on the outstanding balance of the Farmer Mac I post-1996
guaranteed securities. During 2002, all AMBS sold were to Zions Bank, a
related party of Farmer Mac, and totaled $47.7 million. Guarantee fees
earned from Zions Bank were $1.0 million in 2002. 8

Farmer Mac II consists of agricultural mortgage loans containing primary
mortgage insurance provided by the U.S. Department of Agriculture
(USDA). USDA-guaranteed loans collateralizing Farmer Mac II AMBS are
backed by the full faith and credit of the United States. Similar to the pre-
1996 Act securities, as of December 31, 2002, Farmer Mac had experienced
no credit losses on any Farmer Mac II AMBS and did not expect to incur
any such losses in the future.

Farmer Mac’s long-term standby purchase commitments (standby
agreements), introduced in 1999, represent a commitment by Farmer Mac
to purchase eligible loans from financial institutions at an undetermined


8
 Zions Bank, a national bank chartered by the Office of the Comptroller of Currency, is
referred to as a related party of Farmer Mac’s because it is the largest holder of Farmer
Mac’s Class A voting Common Stock and a major holder of Class C nonvoting Common
Stock. In addition, Zions Bank’s Executive Vice President is on Farmer Mac’s Board of
Directors, and Zions Bank sells loans to Farmer Mac and serves as a Central Servicer of
loans for Farmer Mac. Zions Bank also acted as an underwriter, agent, and dealer regarding
Farmer Mac’s discount and medium-term notes.




Page 4                                                            GAO-04-116 Farmer Mac
future date when a specific event occurs. This commitment represents a
potential obligation of Farmer Mac that does not have to be funded until
such time as Farmer Mac is required to purchase a loan. The specific
events or circumstances that would require Farmer Mac to purchase loans
under a standby agreement include when (1) an institution determines it
will sell some or all of the loans under the agreement to Farmer Mac or (2)
a borrower fails to make installment payments for 120 days on a loan
covered by a standby agreement. Financial institutions effectively transfer
the credit risk on the loans covered by a standby agreement to Farmer Mac.
Consequently, these institutions’ regulatory capital requirements and loss
reserve requirements would then be reduced. To date, FCS institutions
have been the only participants in standby agreements. In exchange for
Farmer Mac’s commitment under the standby agreement, Farmer Mac
receives an annual commitment fee from institutions entering into these
agreements, based on the outstanding balance of the loans covered by the
standby agreement. In 2002, these fees represent a significant portion of
Farmer Mac’s total revenues.

Farmer Mac funds its loan purchases primarily by issuing debt obligations
of various maturities. As of December 31, 2002, Farmer Mac had
outstanding $2.9 billion of short-term discount notes and $1.0 billion of
medium-term notes. To the extent the proceeds of the debt issuances
exceed Farmer Mac’s need to fund program assets, those proceeds are used
to purchase assets for the nonmission investment portfolio.

As of December 31, 2002, loans held by Farmer Mac and loans that either
back Farmer Mac AMBS or are subject to standby agreements totaled $5.5
billion. Nearly $3 billion of the $5.5 billion loan and guarantee portfolio is
not on Farmer Mac’s balance sheet. See figure 1 for a breakdown of the $5.5
billion loan and guarantee portfolio. As of December 31, 2002, Farmer Mac
employed 33 persons.




Page 5                                                   GAO-04-116 Farmer Mac
Figure 1: Percentage of Outstanding Balance of Loans, AMBS and Standby
Agreements, as of December 31, 2002
                                                            1% Pre 1996 Act loans

                                                            Farmer Mac II
                                12%



           48%
                                      39%                   Loans and AMBS




                                                            Standby agreements
Source: GAO analysis of data from Farmer Mac 2002 SEC 10-K filing.




Like any other private financial firm, Farmer Mac faces credit, liquidity,
interest rate, and operations risks when conducting its secondary market
operations. Farmer Mac is exposed to the following risks:

•	 Credit risk—the possibility of financial loss resulting from default by
   borrowers on farming assets that have lost value or other parties’ failing
   to meet their obligations. Credit risk occurs when Farmer Mac holds
   mortgages in portfolio and when it guarantees principal and interest
   payment to investors in the AMBS it issues. Farmer Mac is also exposed
   to credit risk for the approximately $2.7 billion of loans under Farmer
   Mac standby agreements, which represent unconditional commitments
   to purchase performing loans at a market price, and to purchase120 day
   delinquent loans at par.

•	 Liquidity risk—the possibility or the perception that Farmer Mac will
   be unable to meet its obligations as they come due because of an
   inability to liquidate assets or obtain adequate funding (referred to as
   “funding liquidity risk”) or will not be able to easily unwind or offset
   specific exposures without significantly lowering market prices because
   of inadequate market depth or market disruptions (“market liquidity
   risk”).




Page 6                                                                           GAO-04-116 Farmer Mac
                    •	 Interest rate risk—the potential that changes in prevailing interest
                       rates will adversely affect on-balance sheet assets, liabilities, capital,
                       income or expenses at different times in different amounts.

                    •	 Operations risk—the possibility of financial loss resulting from
                       inadequate or failed internal processes, people and systems, or from
                       external events.

                    As a GSE, the structure of Farmer Mac’s board of directors was
                    congressionally established. Its 15-member board of directors includes 5
                    members elected by Class A stockholders that are banks, insurance
                    companies, and other financial institutions, 5 members elected by Class B
                    stockholders that are FCS institutions, and 5 members appointed by the
                    president of the United States. Farmer Mac has a third class of common
                    stock that is held by the general public, Class C, which does not have voting
                    rights.

                    The federal government’s creation and continued relationship with Farmer
                    Mac has created the perception in financial markets that the government
                    will not allow the GSE to default on its debt and AMBS obligations,
                    although no such legal requirement exists. As a result, Farmer Mac can
                    borrow money in the capital markets at lower interest rates than
                    comparably creditworthy private corporations that do not enjoy federal
                    sponsorship. During the 1980s, the federal government did provide limited
                    regulatory and financial relief to Fannie Mae when the GSE was
                    experiencing financial difficulties; and in 1987, Congress provided financial
                    assistance to FCS.



Results in Brief	   Since 1999, Farmer Mac’s financial condition has improved, but its risk
                    management practices have not kept pace with its more complex risk
                    profile. Farmer Mac’s net income has steadily increased from $4.6 million
                    in 1997 to $22.8 million in 2002, for a total increase of 392 percent. On the
                    other hand, Farmer Mac’s off-balance sheet standby agreements, which are
                    commitments to purchase loans under specific circumstances, such as
                    when a loan becomes 120 days delinquent, have grown 350 percent in 3
                    years to $2.7 billion and represent nearly 50 percent of the total loans
                    included in Farmer Mac’s programs. Regarding the credit quality of the
                    loans underlying current standby agreements, those loans have been
                    performing better than the loans on Farmer Mac’s balance sheet. While
                    these standby agreements have fueled revenue growth, going forward, if
                    this rapid growth continues, standby agreements could generate



                    Page 7                                                     GAO-04-116 Farmer Mac
substantial funding liquidity risk under stressful economic conditions.
Further, nonperforming, or impaired, loans have been increasing for
Farmer Mac and totaled $75.3 million at the end of 2002 as compared to
zero at the end of 1997. While Farmer Mac has substantially increased its
allowance for loan losses and reserve for losses (loan loss allowance), the
ratio of its allowance to its impaired loans has gone down by over 50
percent since December 31, 1998. This indicates that Farmer Mac’s
impaired loans have increased at a faster rate than the increases in its loan
loss allowance and may also indicate increasing credit risk. Nevertheless,
forensic accountants retained by Farmer Mac Board’s outside counsel
concurred with Farmer Mac’s methodology for estimating loan loss
allowance.

Farmer Mac has risk management systems in place, such as underwriting
standards for purchasing and guaranteeing loans (including loans
underlying standby agreements), and has generally sound processes in
place for estimating credit losses. However, Farmer Mac has not (1)
consistently well documented the exceptions made to its loan underwriting
and servicing procedures, (2) included the current characteristics of its
loan portfolio in the loan loss estimation model, and (3) adequately
documented the results of the model compared to actual portfolio and
economic conditions, resulting in the increased possibility that
management’s objectives of minimizing credit risk have not been met. We
make recommendations to Farmer Mac designed to enhance its loan loss
estimation model and to improve its documentation of policies and
procedures, and management’s actions that relate to reducing credit risk.
Regarding liquidity risk, Farmer Mac has maintained sufficient liquidity to
support its loan purchase and guarantee activity through continued access
to the capital markets. However, Farmer Mac lacks a formalized
contingency plan to address its potential liquidity needs that could
potentially be created by the standby agreements under stressful
agricultural economic conditions. Although Farmer Mac issues debt
securities for liquidity purposes, it is not required and it has decided not to
obtain a credit rating from a nationally recognized statistical rating agency.
As for interest rate risk, the methods employed by Farmer Mac to measure
interest rate sensitivity appeared reasonable but we identified limitations
with some elements of its prepayment methodology. In terms of capital,
Farmer Mac exceeded the capital levels required by its statute and
regulator but could improve its plan for capital adequacy. Specifically, it
lacked a test for sufficiency in assessing its capital adequacy, other than its
stated goal of meeting its statutory minimum and regulatory risk-based
capital requirements. We make recommendations to Farmer Mac to



Page 8                                                    GAO-04-116 Farmer Mac
develop a contingency funding liquidity plan, improve the quality of its
prepayment model, and enhance its analysis of capital adequacy. Finally,
Farmer Mac also faces some uncertainty involving its line of credit with the
Department of the Treasury (Treasury). Specifically, while the legal
opinion of Farmer Mac’s outside counsel disagrees, Treasury has taken the
position that it is not obligated to cover losses on AMBS held in Farmer
Mac’s portfolio.

Farmer Mac has increased its agricultural mortgage loan purchase and
guarantee activity since our 1999 report, and has reduced the relative size
of its nonmission portfolio. Nevertheless, its enabling legislation contains
broad mission purpose statements and lacks specific or measurable
mission-related criteria that would allow for a meaningful assessment of
whether Farmer Mac had achieved its public policy goals. Farmer Mac’s
strategy of holding AMBS for profitability reasons has been a contributing
factor in limiting the development of a liquid secondary market for these
securities. As a result, the depth and liquidity of the demand for AMBS in
the current market are unknown. Farmer Mac introduced the standby
agreement program to provide greater lending capacity for agricultural real
estate lenders. However, FCS institutions’ increased use of standby
agreements potentially reduces the sum of capital required to be held by
FCS and Farmer Mac. Such a reduction in capital could be consistent with
a reduction in risk if there were diversification at the secondary market
level. However, as of December 31, 2002, 10 financial institutions
generated 90 percent of Farmer Mac’s business, and over 70 percent of the
outstanding balance of Farmer Mac’s loan portfolio was located in the
Southwest and Northwest. Finally, the size of Farmer Mac’s nonmission
investment portfolio has decreased as a percentage of its total on- and off-
balance sheet portfolio. Still, the composition and criteria for nonmission
investments could potentially lead to investments that are excessive in
relation to Farmer Mac’s financial operating needs or otherwise would be
inappropriate to the statutory purpose of Farmer Mac. We make
recommendations to Farmer Mac to reevaluate its current strategy of
holding AMBS in its portfolio and issuing debt to obtain funding. We also
suggest that Congress consider legislative changes to establish clearer
mission goals for Farmer Mac.

Like other publicly traded companies, Farmer Mac is in the process of
taking actions to ensure that it complies with provisions of the Sarbanes-
Oxley Act of 2002 (Sarbanes-Oxley) requirements, the Security Exchange
Commission (SEC) rules, and proposed changes in the New York Stock
Exchange (NYSE) listing standards. In accordance with these new



Page 9                                                  GAO-04-116 Farmer Mac
requirements, Farmer Mac has reaffirmed its audit committee charter and
has hired internal and external auditors that are from different firms.
Sarbanes-Oxley requires that members of audit committees of listed
companies be independent and proposed NYSE listing standards require
that a majority of the board of directors of listed companies be
independent. Since Farmer Mac’s Class A and Class C stock are listed on
the NYSE, Farmer Mac is currently subject to the auditor independence
requirements of Sarbanes-Oxley and, unless waived, the listing standards.
Farmer Mac has taken steps to update its corporate governance practices,
but its statutory board structure, which is set by law, could make it difficult
to comply with the board independence requirements proposed in the
NYSE listing standards. Moreover, since Farmer Mac shareholders include
both institutions that utilize its services (Class A and Class B common
stock) and public investors (Class C common stock), and because all
members of the board of directors are chosen by the cooperative investors
or by the President of the United States, the board may face difficulties in
representing the interests of all shareholders. Additionally, since Farmer
Mac is a publicly traded company, the nonvoting structure of Farmer Mac’s
Class C common stock may not be appropriate in today’s corporate
governance environment. In most respects, Farmer Mac’s board policies
and processes appear reasonable, but the process to identify and select
nominees, director training, and succession planning could be further
developed and formalized. Finally, Farmer Mac’s total executive
compensation was within its consultants’ recommended parameters;
however, its vesting program appears more generous than industry
practices, given Farmer Mac’s maturity. We make recommendations to
Farmer Mac designed to provide more transparency to the nomination
process and succession planning and more consistency in training for
directors. We also recommend that Farmer Mac reevaluate the vesting
period for stock options. We further suggest that Congress consider
legislative changes to amend the structure of the Farmer Mac board and the
structure of Farmer Mac’s Class C common stock.

Since 2002, FCA took several steps to enhance supervisory oversight of
Farmer Mac but faces significant challenges that could limit the
effectiveness of its oversight. FCA’s June 2002 annual safety and soundness
examination was more comprehensive than previous examinations. FCA
also has taken some actions to improve its regulatory framework for
Farmer Mac by developing proposed regulations regarding liquidity
standards and nonmission investments. Although FCA has increased its
efforts to help oversee and examine Farmer Mac’s operations, our review
identified weaknesses in FCA’s off-site monitoring process regarding call



Page 10                                                   GAO-04-116 Farmer Mac
reports. As it continues to oversee Farmer Mac, FCA faces five significant
challenges related to Farmer Mac’s risk-based capital model as well as
regulatory management. First, limitations exist in FCA’s model used to
estimate Farmer Mac’s credit risk for calculation of the risk-based capital
requirement. Individually, each limitation may under or over estimate the
risk-based capital for Farmer Mac’s credit risk, but overall, the relative
magnitude of these effects is unclear. Second, FCA’s risk-based capital
regulation does not capture credit risk on Farmer Mac’s liquidity
investments, AgVantage bonds, and counterparty risk on derivatives. Third,
FCA’s market risk and income models may understate estimated levels of
required risk-based capital. Fourth, FCA does not have criteria and
procedures to assess and report on the relationship of Farmer Mac’s
activities to the achievement of its mission. Finally, being the single
regulator for both FCS institutions and Farmer Mac could cause potential
conflicts of interest because FCA may, in times of stress, attempt to
support one type of participant at the expense of the other. We make
recommendations to FCA designed to enhance the risk-based capital
model, improve off-site monitoring of Farmer Mac, and assess and report
on how Farmer Mac is achieving its mission. We also suggest that Congress
consider a legislative change to allow FCA more flexibility in setting
minimum capital requirements for Farmer Mac.

We provided a draft of this report to the heads or their designees of the
Farmer Mac, FCA, SEC, and Treasury. We received written comments from
Farmer Mac and FCA that are reprinted in appendixes VII and VIII
respectively. SEC did not provide comments. Farmer Mac, FCA, and
Treasury also provided technical comments that we have incorporated as
appropriate. Farmer Mac stated that it agreed with the report’s findings
and conclusions on Farmer Mac’s risk management practices and has taken
a number of steps toward implementing the majority of the
recommendations. While Farmer Mac seemed to agree with the report’s
recommendations to improve its analysis of capital adequacy, develop a
contingency funding plan, and improve documentation of management
exceptions to its eight major underwriting standards, it did not address the
rest of our recommendations. Farmer Mac commented that the uncertainty
regarding the Treasury line of credit is a moot point because a legal opinion
by its outside counsel stated that the Treasury line of credit would be
available in the circumstances noted. Our position is that this issue may
remain unresolved until Farmer Mac approaches Treasury for assistance.
Farmer Mac appeared to disagree with our concern about funding liquidity
risk that might arise from standby agreements. However, we noted that
Farmer Mac seems to believe that liquidity funding risk is captured and



Page 11                                                 GAO-04-116 Farmer Mac
                                 accounted for in the risk-based capital model, whereas it is not. Moreover,
                                 Farmer Mac has not fully recognized all loan amounts that could be
                                 presented to Farmer Mac for funding as part of its liquidity funding needs.
                                 FCA generally concurred with the report’s findings and conclusions that
                                 are focused on FCA’s work to oversee the safety and soundness of Farmer
                                 Mac and agreed to implement the report’s recommendations. FCA does not
                                 agree that additional data and modeling would add value to the risk-based
                                 capital model, although FCA stated that it is studying the possibility of
                                 updating the data used in the model. We discuss Farmer Mac and FCA’s
                                 comments and our response in greater detail at the end of this letter.



Farmer Mac’s Financial           Farmer Mac’s net income has steadily increased from $4.6 million in 1997 to
                                 $22.8 million in 2002, for a total increase of 392 percent. At the same time,
Condition Has                    we identified trends that increased the complexity of Farmer Mac’s risk
Improved, but Risk               profile, such as rapid growth in its standby agreement program, certain
                                 weaknesses in its risk management practices, and an uncertainty involving
Management Practices             Treasury’s line of credit.
Have Not Kept Pace
with Its More Complex
Risk Profile

Farmer Mac’s Income Has          Two primary revenue sources contributed to the growth of Farmer Mac’s
Increased, and Risk Profile      net income—interest income and commitment fees. Interest income
                                 earned on Farmer Mac’s portfolio of loans, guaranteed securities, and
Has Become More Complex          investments has more than doubled due to substantial growth in Farmer
                                 Mac’s portfolios over the same period. Interest income was Farmer Mac’s
                                 principal source of revenue in 1997. But recently, a new source of
                                 revenue—commitment fees earned on standby agreements— has grown
                                 since the product’s inception in early 1999, amounting to over 25 percent of
                                 Farmer Mac’s total revenues for 2002. See appendix III for further
                                 discussion of trends and comparisons of Farmer Mac’s financial condition.

High Growth in Standby           Although Farmer Mac’s net income has been increasing since 1997, there
Agreements Fuels Revenue, but    could be indicators of increasing funding liquidity risk due to high levels of
Could Generate Funding           growth in Farmer Mac’s standby agreement program. Farmer Mac’s
Liquidity Risk Under Stressful   earnings growth has principally been driven by its off-balance sheet
Conditions                       standby agreements. First offered in early 1999, the standby agreement
                                 program grew rapidly. As shown in table 1, the balance of loans covered by



                                 Page 12                                                  GAO-04-116 Farmer Mac
standby agreements grew from zero in 1998 to $2.7 billion at December 31,
2002, with high rates of growth in recent years.



Table 1: Loans Covered by Standby Agreements

                                          1998       1999          2000           2001         2002
Loans covered by

standby agreements                         $0 $0.6 billion   $0.9 billion   $1.9 billion $2.7 billion

Percentage increase 

from previous year                        N/A         N/A          50%           111%           42%

Source: Farmer Mac 2002 SEC 10K filing.


Corresponding with the growth in loans covered under the standby
agreement program is an increase in Farmer Mac’s revenues. For instance,
revenues from commitment fees were $1.6 million in 1999 or 7 percent of
Farmer Mac’s total revenues that year. By 2002, revenues from
commitment fees had increased to $11.0 million, representing 22 percent of
total revenues. During economic times when agricultural land values have
been rising and interest rates have been relatively low, Farmer Mac has, in
2002, purchased about $3.3 million of the eligible agricultural mortgage
loans placed under standby agreements.

Farmer Mac stated that since the program began in 1999, the relatively few
defaulted loans they have had to purchase reflect the credit quality of the
loans underlying standby agreements. Further, those loans are
underwritten and required to be serviced to the same standards used for all
other loans backing Farmer Mac’s AMBS. Between 1999 and 2002, the
standby agreements had no net credit losses. At December 31, 2002, loans
that were at least 90 days delinquent under standby agreements were $3.5
million or 0.13 percent of the total amount of loans under standby
agreements. This was well below the $54.7 million or 2.21 percent of
Farmer Mac’s on-balance-sheet loans and guarantees. The lower
delinquencies and losses under the standby agreement program indicate
that the program through December 31, 2002, experienced lower credit risk
than Farmer Mac’s other programs.

However, guidance from financial regulators indicates rapid growth of
programs or assets is thought to be an increased risk factor. Many financial
institution failures of past decades were blamed, in part, on unchecked
growth particularly in new and innovative products with complicated risk
characteristics. The rapid growth of the standby agreements could result



Page 13                                                                     GAO-04-116 Farmer Mac
                                 in increased funding liquidity risk to Farmer Mac because Farmer Mac’s
                                 commitment under these agreements differ from its off-balance sheet
                                 AMBS. For AMBS, if an underlying loan becomes 90 days delinquent,
                                 Farmer Mac has the option of purchasing the loan, or just making
                                 installment payments. Under its standby agreements, if an underlying loan
                                 is 120 days delinquent, the lender can require Farmer Mac to buy the loan.
                                 Therefore, standby agreements represent a potential future obligation of
                                 Farmer Mac, which does not have to be funded until such time as Farmer
                                 Mac is required to purchase an impaired loan.9 In other words, going
                                 forward, if the rapid growth of standby agreements continue, at a time
                                 when either the agricultural sector is severely depressed or interest rates
                                 are adversely changing, Farmer Mac could be required to purchase large
                                 amounts of impaired or defaulted loans under the standby agreements,
                                 thus subjecting Farmer Mac to increased funding liquidity risks and the
                                 potential for reduced earnings. (See liquidity section for further
                                 discussion.)

                                 Additionally, because of its rapid growth and recent implementation, there
                                 is limited historical information to project the number of loans covered by
                                 standby agreements that Farmer Mac may need to purchase in the future.
                                 As a result, management has limited quantitative data on which to base risk
                                 management and other operating decisions.

                                 Similar to Farmer Mac’s other guaranteed obligations, when Farmer Mac is
                                 required to purchase an impaired or defaulted loan under its standby
                                 agreement obligation, it may adversely affect its earnings in four ways: (1)
                                 it requires an earning asset to be sold or a liability to be incurred in
                                 exchange for an asset that might not be an earning asset; (2) it increases
                                 administrative expenses for monitoring, collection, and recovery efforts;
                                 (3) the annual commitment fees Farmer Mac receives on the loan would
                                 cease; and (4) under economically stressful conditions, Farmer Mac could
                                 incur losses on the disposal of impaired loans it is required to purchase
                                 under the standby agreements if the net proceeds from the sale of collateral
                                 on the loan is insufficient.

Increase in Impaired Loans and   Farmer Mac has established an allowance for loan losses and reserve for
Charge-offs May Indicate         losses (a loan loss allowance) to cover estimated, probable loan losses for
Increasing Credit Risk	          its current portfolio of loans, commitments, and guarantees. The loan loss
                                 allowance has increased substantially from $1.6 million at December 31,

                                 9
                                 These off-balance sheet obligations were disclosed in Farmer Mac’s SEC filings.




                                 Page 14                                                           GAO-04-116 Farmer Mac
                             1997, to $19.4 million at December 31, 2002. On the other hand, the ratio of
                             loan loss allowance to impaired loans has decreased from a high of 59.1
                             percent at December 31, 1998, to 27.8 percent at December 31, 2002. This
                             ratio, a primary credit risk indicator, shows that the balance of Farmer
                             Mac’s impaired assets has increased at a faster rate than the increases in its
                             loan loss allowance. (We further discuss management’s monitoring and
                             assessment of credit risk in a later section of this report.)

                             Farmer Mac’s increase in impaired loans, real estate owned, and write offs
                             of bad loans as well as the growth in its on- and off-balance sheet loans,
                             guarantees, and standby agreements is indicative of increasing credit risk.
                             Impaired loans totaled $75.3 million at December 31, 2002, compared to
                             zero at December 31, 1997.10 Since Farmer Mac only began purchasing
                             loans after the 1996 Act was passed, the loan and guarantee portfolio is
                             relatively new. As loan portfolios age, delinquencies typically increase,
                             eventually peak, and then taper off, establishing a track record of
                             performance often referred to in the industry as a “seasoned” portfolio.
                             According to Farmer Mac, its loans are just becoming seasoned, so the
                             losses and delinquencies are increasing. Farmer Mac’s write offs of
                             impaired loans have been limited to date but delinquencies are increasing.
                             During 2002, Farmer Mac wrote off $4.1 million of bad loans, or 8 basis
                             points11 of post-1996 Act loans and guarantees, which was a significant
                             increase over the $2.2 million, or 6 basis points, written off in 2001.12



Farmer Mac’s Controls Over   Although Farmer Mac has underwriting standards for purchasing and
Credit Risk Were Generally   guaranteeing loans (including loans underlying standby agreements), and
                             has processes for estimating credit losses, Farmer Mac’s implementation of
Sound but Had Certain
                             its standards and its processes need improvement to enhance its overall
Weaknesses                   controls over credit risk. One of its underwriting standards permits


                             10
                               Per Farmer Mac’s 2002 Annual Report, impaired assets are loans that are 90 days or more
                             past due, in foreclosure, loans performing in bankruptcy, either under their original loans
                             terms or a court-approved bankruptcy plan, and real estate owned, which is real estate
                             acquired through foreclosure.
                             11
                                  A basis point is equal to one hundredth of a percent.
                             12
                                Loans written off are losses on the outstanding balance of the loan, any interest payments
                             previously accrued or advanced, and expected collateral liquidation costs. The post-1996
                             Act loans and guarantees are post-1996 Act loans held and loans underlying the guaranteed
                             securities and standby agreements, which represent the credit risk on loans and guarantees
                             assumed by Farmer Mac.




                             Page 15                                                             GAO-04-116 Farmer Mac
                                 management to override one or more of the other eight standards when, in
                                 management’s opinion, other factors compensate for certain loan
                                 weaknesses. Farmer Mac has made use of this provision without
                                 consistently and thoroughly documenting the basis for the exceptions
                                 made. For loans it has purchased, including loans under standby
                                 agreements, Farmer Mac’s process for estimating credit losses is generally
                                 sound but has certain weaknesses. In estimating losses, Farmer Mac uses a
                                 risk model based on loans that differ from those in its own portfolios and
                                 under its standby agreements with respect to geographic distribution and
                                 interest rate terms. This lack of comparability and other limitations of the
                                 model may affect the reasonableness and accuracy of Farmer Mac’s
                                 estimated losses resulting from credit risk either upward or downward. In
                                 estimating the credit risk of loans under standby agreements, a
                                 complicating factor is that Farmer Mac lacks the historical experience with
                                 the long term standby agreements needed to accurately estimate the types
                                 of loans and amount of loans it may ultimately be obligated to purchase,
                                 along with any associated losses. In addition, for estimating and allocating
                                 loan losses, Farmer Mac reverses the order of the methods called for in
                                 accounting guidance and does so without quantifying the effects of its
                                 approach. Finally, recent reviews have shown weak documentation
                                 describing Farmer Mac’s use of its loan loss estimation model, its
                                 quantification process, management’s judgment and key decisions, and the
                                 summary results of the loss estimation process.

Farmer Mac’s Loan Underwriting   Farmer Mac uses underwriting standards and processes for monitoring the
and Servicing Procedures Were    loans it purchases and guarantees (including those loans under its standby
Clear, but Exceptions Were Not   agreements). It also has standards for “sellers” and loan “servicers.”
Consistently Well Documented     Farmer Mac’s underwriting process includes identifying and analyzing
                                 potential risks of loss associated with its loan purchases and guarantees
                                 prior to entering into such agreements. A key element of Farmer Mac’s
                                 system of internal control in underwriting is the use of established, written
                                 standards (for both internal use and for external loan sellers and servicers)
                                 that require analysis of numerous qualitative and quantitative borrower and
                                 property characteristics for loans, prior to purchase or prior to inclusion in
                                 a standby agreement. These standards help streamline the process for
                                 buying and guaranteeing loans, lower transaction costs, and increase
                                 efficiency while providing criteria and controls over the process of
                                 accepting loans for purchase or for inclusion in standby agreements. For
                                 example, Farmer Mac has underwriting standards as documented in its
                                 Seller/Servicer Guide (the guide) to (1) assess whether a borrower has
                                 sufficient income and a good credit history and (2) set a maximum loan-to-
                                 value ratio (LTV) limit. Farmer Mac monitors its credit risk through



                                 Page 16                                                  GAO-04-116 Farmer Mac
                                  periodic monitoring of the borrower’s and seller/servicer’s performance by
                                  reviewing the payment history, visiting borrower and servicers’ facilities,
                                  and in the case of seriously delinquent loans with expected loss in
                                  collateral value, obtaining updated property inspections and valuations.

                                  As shown in appendix IV, Farmer Mac has nine underwriting standards to
                                  which all loans must conform, in order for Farmer Mac to purchase or
                                  guarantee the loans. Underwriting standard number nine allows
                                  management to override one or more specific underwriting criteria when,
                                  in management’s opinion, other factors compensate for certain loan
                                  weaknesses. For example, in cases when the borrower’s debt-to-asset ratio
                                  may not meet standards but the LTV ratio is better than requirements, then
                                  credit risk could be balanced by the LTV ratio. As of December 31, 2002, a
                                  total of $1.4 billion (30 percent) of the outstanding balance of loans held
                                  and loans underlying standby agreements and post-1996 Farmer Mac I
                                  Guaranteed Securities were approved based upon compensating strengths.
                                  Further, during 2002, $327.7 million (28 percent) of the loans purchased or
                                  added under standby agreements were approved based upon compensating
                                  strengths.

                                  However, recent reviews noted that management’s assessments supporting
                                  the override of underwriting criteria, including quantification and
                                  evaluation of compensating risk factors, was often not well-documented.
                                  Without consistently well-documented reasons for exceptions to the
                                  underwriting standards, Farmer Mac increases the risk that management’s
                                  objectives of balancing risk have not been met. During 2003, Farmer Mac
                                  has begun gathering related data, but has not yet developed a process for
                                  fully utilizing the data in its management decision process for making
                                  future overrides and for estimating credit risk and allowance for losses on
                                  those specific loans.

Weaknesses Exist in Farmer        As part of the financial monitoring and reporting process, Farmer Mac’s
Mac’s Monitoring and              management is responsible for assessing the current level of risk
Assessment of Changes in Credit   associated with individual loans and loan portfolios that have been
Risk	                             purchased or guaranteed by Farmer Mac, including loans under its standby
                                  agreements that are off-balance sheet, and estimating credit losses on
                                  those loans for financial reporting purposes. Farmer Mac records its
                                  estimated losses on loans held in an “allowance for loan losses” account,
                                  which serves to reduce the balance of Farmer Mac’s loans. Farmer Mac
                                  estimates credit losses on loans backing its guaranteed securities and loans
                                  covered by its standby agreements and records those losses in “reserve for
                                  losses,” which appears as a liability on Farmer Mac’s balance sheet. When



                                  Page 17                                                 GAO-04-116 Farmer Mac
Farmer Mac records estimated losses in the allowance for loan loss and
reserve for losses accounts, Farmer Mac’s pretax income, and therefore its
core capital, is reduced.

Farmer Mac uses a credit risk modeling tool called the Loan Pool
Simulation and Guarantee Fee Model (the model) as a basis for estimating
loan losses each quarter. This model, developed by an outside consultant,
uses equations to estimate the probability, amounts, and distribution of
losses over a period of time based upon loss experience from the Farm
Credit Bank of Texas (FCBT) from 1979 to 1992. Because Farmer Mac does
not have adequate historical experience from its own portfolio for
estimating losses on loans, data from FCBT are used as a proxy. According
to Farmer Mac management, this was the best data available for estimating
Farmer Mac’s future losses on loans. The resulting projections of losses
were reviewed by Farmer Mac management prior to being recorded to the
financial statements. While this was the best data available, we did find a
number of limitations in Farmer Mac’s loan loss estimation model, its data,
and application of the results to estimate losses, which may impact the
reasonableness of the allowance and reserve amounts, and related losses
recorded in the company’s financial statements. We further discuss the
data limitation in the FCA oversight section of this report since FCA also
used FCBT data in its model to estimate Farmer Mac’s credit risk.

The model used by Farmer Mac to estimate credit risk has some
limitations. The primary limitation of the model is that Farmer Mac’s loan
and guarantee portfolios and the loans included under standby agreements
have different characteristics from the loan characteristics of FCBT loans
used in the model. Although the loans used in the model have similar
characteristics with respect to key underwriting variables, they differ from
Farmer Mac’s portfolio both with respect to geographic distribution and
interest rate terms. Specifically, the data supporting Farmer Mac’s loan
loss estimation process include loans issued in the 1970s and 1980s by
FCBT, which were adjustable-rate mortgages, tied to a farm credit cost of
funds index that changed slowly over time. In contrast, the loans now held
and guaranteed by Farmer Mac are either rapidly changing adjustable-rate
mortgages, or fixed-rate mortgages with financial penalties to the
borrowers that eliminate the incentive to refinance when interest rates
drop. Additionally, the FCBT loans were limited to Texas, while Farmer
Mac may purchase loans in any state.

There are other complicating factors. First, Farmer Mac’s current portfolio
has a high geographic concentration in the Western part of the United



Page 18                                                 GAO-04-116 Farmer Mac
                                 States and is dominated by three lenders. Moreover, Farmer Mac’s
                                 estimation of credit risk for the loans under standby agreements is limited
                                 by Farmer Mac’s lack of historical experience for estimating the amount of
                                 loans it may ultimately be obligated to purchase under the standby
                                 agreements.

                                 Farmer Mac has not quantified the impact of its current approach for
                                 estimating and allocating loan losses versus the approach set forth in
                                 accounting standards as the preferred methodology. SEC Staff Accounting
                                 Bulletin (SAB) No. 102 (July 6, 2001), states that “A registrant’s loan loss
                                 allowance methodology generally should…identify loans to be evaluated
                                 for impairment on an individual basis under SFAS No. 114 and segment the
                                 remainder of the portfolio into groups of loans with similar risk
                                 characteristics for evaluation and analysis under SFAS No. 5.”13 This same
                                 approach is also set forth in a current American Institute of Certified Public
                                 Accountants proposed Statement of Position on Allowance for Credit
                                 Losses dated June 19, 2003.

                                 Farmer Mac’s calculation of its estimated loan loss allowances uses the
                                 reverse order of the approach set forth in the accounting standards as
                                 clarified in SAB 102. Farmer Mac’s model calculates an overall loss result,
                                 from which management allocates portions to the allowance for losses
                                 (related to loans held by Farmer Mac) and the reserve for losses (related to
                                 loans guaranteed by Farmer Mac and included in its standby agreements).
                                 From the overall loss amounts calculated, Farmer Mac deducts specifically
                                 identified loan loss estimates and considers the remaining amount to be
                                 sufficient to cover the remainder of the portfolio. Farmer Mac’s
                                 management stated that its methodology does not result in a materially
                                 different loss estimate than if it followed the preferred methodology of the
                                 accounting standards. However, Farmer Mac has not quantified the effects
                                 of using this methodology.

Documentation on the Loan Loss   Reviews of Farmer Mac conducted in 2002 concluded that Farmer Mac had
Estimation Model Was Weak	       weak documentation describing (1) how its loan loss estimation model
                                 works, (2) its quantification process, (3) management’s judgment and key
                                 decisions, and (4) the summary results of the loss estimation process.
                                 Although Farmer Mac received an unqualified (“clean”) opinion on its 2002


                                 13
                                  Statement of Financial Accounting Standard 5: Accounting for Contingencies, issued
                                 March 1975. Statement of Financial Accounting Standard 114: Accounting by Creditors for
                                 Impairment of a Loan, an amendment of FASB Statements No. 5 and 15, issued May 1993.




                                 Page 19                                                         GAO-04-116 Farmer Mac
                             annual financial statements, Farmer Mac received several
                             recommendations as a result of recent reviews to improve the loan loss
                             estimation process, such as applying the model’s results consistently with
                             management’s policies and improving documentation. During 2002,
                             management took a number of actions in response to these
                             recommendations to improve the data used for estimating losses as well as
                             the disclosure of the risks inherent in its portfolio. In addition, to assess
                             the reliability of Farmer Mac’s estimated losses on loans, the Board of
                             Directors’ outside counsel retained a forensic accounting firm in 2002 to
                             review management’s processes and controls for estimating these losses.
                             Nevertheless, it suggested improvements for Farmer Mac’s SEC annual and
                             quarterly filings and for internal documentation. Similarly, reports of
                             recent reviews noted that management should document (1) the
                             similarities and differences of using the model for both loans and
                             guarantees recorded on the balance sheet as well as standby agreements
                             that were not recorded on the balance sheet; (2) management’s
                             reconciliation of the model’s loss projections to actual amounts recorded in
                             the financial statements; and (3) the results of updated collateral
                             evaluations and reviews of impaired loans, and the results’ effect on the
                             recorded allowance and reserve amounts.



Farmer Mac Maintained        Farmer Mac maintained access to the capital markets, which are its
Access to Capital Markets,   primary source of liquidity, to support its loan purchase and guarantee
                             activity, despite the lack of a credit rating that would make Farmer Mac’s
Its Primary Source of
                             debt more comparable to other firms’ debt issuances. Farmer Mac’s
Liquidity, but It Lacked a   reserve of liquid assets was a secondary source of liquidity, which as of
Formal Liquidity             September 30, 2002, was adequate to pay off current on-balance-sheet
Contingency Plan             liabilities for close to 30 days.14 However, Farmer Mac lacked a formal
                             contingency plan for potential liquidity funding needs under stressful
                             agricultural economic conditions, including unexpected demands for
                             additional liquidity that the standby agreements may create.




                             14
                                Liquid assets are primarily cash and cash equivalents on the balance sheet. Farmer Mac
                             refers to these as the Liquidity Investment Portfolio.




                             Page 20                                                            GAO-04-116 Farmer Mac
Farmer Mac Issued Debt              Our analysis indicated that Farmer Mac had been able to maintain access,
Securities for Liquidity, but Has   at stable interest rates, to the discount note market, even during several
Not Pursued a Credit Rating To      periods of market stress and company exposure to public criticism in 2001
Date                                and 2002. 15 However, these events temporarily affected the interest rates
                                    on medium-term notes.16 Farmer Mac obtained cash for its loan purchase
                                    activities primarily through periodic sales of debt securities at varying
                                    maturities. Referring to publicly traded firms, Moody’s Investors Service
                                    (Moody’s) said that Farmer Mac was the largest issuer of unrated debt in
                                    the United States.17 Yet, Farmer Mac has issued discount notes at virtually
                                    the same interest rates as Fannie Mae, which obtains an annual “risk to the
                                    government” or financial strength rating from a nationally recognized
                                    rating agency.18 Broker-dealers who trade agency securities said that a
                                    cause was that (1) Farmer Mac has a GSE charter just as Fannie Mae and
                                    Freddie Mac do and, therefore, investors tend to conclude that they have a
                                    similar risk profile and (2) investors purchase Farmer Mac’s discount notes
                                    to diversify portfolios that also held Fannie Mae and Freddie Mac short-
                                    term debt.19 Farmer Mac officials noted that the spreads on debt issuances
                                    are driven by the relatively small size of Farmer Mac issuances relative to
                                    the other GSEs, and at this time, the financial and human resources
                                    required to obtain a rating would not be justifiable. While having a credit
                                    rating may not have an effect on the interest rates on Farmer Mac’s debt,
                                    such a rating would provide investors and creditors with information to
                                    assess Farmer Mac’s financial soundness without government backing.
                                    This would facilitate investors and creditors comparing Farmer Mac with
                                    other entities and might also broaden the population of potential
                                    purchasers of Farmer Mac’s debt securities, in particular municipalities,

                                    15
                                     Discount notes are unsecured general corporate obligations that are issued at a discount
                                    but mature at face value. Their maturities range from overnight to 1 year.
                                    16
                                      Medium-term notes (MTN) are debt securities that may be issued with floating or fixed
                                    interest rates with maturities ranging from 9 months to 30 years or longer. An advantage of
                                    MTNs over corporate bonds is that they tend to be more flexible in terms of maturities and
                                    interest rates.
                                    17
                                         There is no statutory or regulatory requirement for Farmer Mac to obtain a credit rating.
                                    18
                                     A rating agency, such as Moody’s or Standard and Poors, provides its opinion on the
                                    creditworthiness of an entity and the financial obligations issued by an entity, using a credit
                                    rating system. The ratings may range from AAA (high quality) to D (in default). Bonds rated
                                    “BBB” or higher are widely considered “investment grade.” This means the quality of the
                                    securities is high enough for a prudent investor to purchase them.
                                    19
                                       Agency papers are short-term debt securities that are predominantly issued by GSE and
                                    federal agencies.




                                    Page 21                                                                GAO-04-116 Farmer Mac
                                 who purchase debt securities, due to internal policies that prohibit
                                 purchasing unrated financial instruments.

Farmer Mac Has Maintained        Farmer Mac’s liquidity investment portfolio was a secondary source of
Close to 30 Days of Liquidity	   liquidity and provided for close to 30 days of funds should access to capital
                                 markets be temporarily impaired. As a comparison, Farmer Mac’s reserve
                                 was larger than FCA’s requirement that FCS institutions maintain a liquidity
                                 reserve of at least 15 days, although FCA officials said that they were
                                 evaluating the adequacy of a 15-day liquidity reserve.20 On the other hand,
                                 Farmer Mac’s liquidity reserve of 15 days is considerably less than the
                                 stated liquidity goals of Fannie Mae, which maintained 3 months of liquidity
                                 to ensure that it could meet all of its obligations in any period of time in
                                 which it did not have access to the capital markets.21 As of September 30,
                                 2002, Farmer Mac’s liquidity portfolio was worth $1.4 billion and consisted
                                 primarily of high-quality, short-term investments. However, according to
                                 our review of SEC filings, the range of permissible investments set by the
                                 board has expanded to include investments that do not have characteristics
                                 of traditional liquidity investments. For example, Farmer Mac’s investment
                                 in a significant amount of unrated preferred stock of two FCS institutions
                                 represents fixed-rate investments that carry the potential for increased
                                 return, but also increased risk.

Farmer Mac Lacked a Formal       Farmer Mac does not yet have a formal contingency plan to maintain
Contingency Plan for Liquidity   liquidity should its access to the capital markets be impaired, although as
                                 previously discussed, it does maintain a large liquidity portfolio to
                                 temporarily meet liquidity needs. In addition, management has standard
                                 written repurchase agreements with large investment banks, which it could
                                 use to pledge or sell its assets as a temporary source of liquidity.22 As of
                                 early 2003, Farmer Mac was in the process of developing a liquidity policy.
                                 Because Farmer Mac primarily relies on external sources of funds, Farmer
                                 Mac is exposed to funding liquidity risk and its access to these external



                                 20
                                    For purposes of this report, we define liquidity as both the capacity and the perceived
                                 capacity to meet obligations as they come due without a material increase in the cost to the
                                 institution.
                                 21
                                    We did not include Freddie Mac’s liquidity reserve since at the time of this report, Freddie
                                 Mac was in the process of restating its financial position.
                                 22
                                    A repurchase agreement is a form of secured, short-term borrowing in which a security is
                                 sold with a simultaneous agreement to buy it back from the purchaser at a future date.




                                 Page 22                                                               GAO-04-116 Farmer Mac
                                funds could potentially be impaired by external or internal events. 23 For
                                example, in 2002, Farmer Mac increasingly relied on issuing discount notes
                                for liquidity, as discount notes in combination with interest rate swaps
                                would provide the lowest interest costs.24 According to financial regulatory
                                guidance, for safety and soundness purposes, an effective plan for
                                managing liquidity risk should not necessarily employ the cheapest source
                                of funding. In addition, each institution’s liquidity policy should include a
                                contingency plan for liquidity, which would address alternative funding
                                sources if initial projections of funding sources and uses were incorrect.
                                The contingency plan would clearly identify any back-up facilities (lines of
                                credit), and note the conditions where they might be used.

Off-balance Sheet Standby       In addition to meeting liquidity demands from expected obligations,
Agreements Can Potentially      Farmer Mac may face unexpected demands on funding liquidity should
Create Unexpected Demands for   lenders that participate in the standby agreements exercise their contracts.
Additional Funding Liquidity	   To date, Farmer Mac has not experienced material demands for additional
                                liquidity that might arise from standby agreements and under current
                                circumstances, Farmer Mac appears to have adequate liquidity to fund
                                purchases of those underlying loans. However, the risk exists that if
                                standby agreements continue to grow and their risks are not closely
                                managed, during an economic downturn, Farmer Mac could experience a
                                large and sudden increase in the exercise of standby agreements by
                                lenders. In the event that Farmer Mac would be required to purchase large
                                amounts of impaired or defaulted loans underlying the standby
                                agreements, Farmer Mac management said that its strategy would be to
                                rely on the capital markets for additional cash by either issuing more debt
                                or selling its AMBS. However, since Farmer Mac did not sell AMBS to
                                independent third party investors in 2002, the depth and liquidity of the
                                demand for these securities in the current market are unknown.25 Broker-
                                dealers with whom we spoke, stated that a Farmer Mac entrance into the
                                debt markets to sell a significant amount of debt (in addition to what they



                                23
                                   Funding liquidity risk is the potential that an institution would be unable to meet its
                                obligations as they come due because of an inability to liquidate a sufficient quantity of
                                assets or to obtain a sufficient quantity of new liabilities.
                                24
                                     See Interest rate risk section and appendix III for further discussion of interest rate swaps.
                                25
                                 Farmer Mac noted that between 1996 and 2000, $553 million of AMBS were sold. In
                                addition, Farmer Mac noted that traders advised management that they believed Farmer
                                Mac could re-enter the AMBS market and achieve pricing relative to comparable Fannie Mae
                                securities at least as favorable as that achieved in 1996–1998.




                                Page 23                                                                  GAO-04-116 Farmer Mac
                             currently issue) would require substantial investor education by Farmer
                             Mac to generate additional interest in their debt securities.



Farmer Mac Managed Its       Our discussions with Farmer Mac officials, reviews of Farmer Mac and
Interest Rate Risk, but      FCA documents, and analysis of data from SEC filings indicated that as of
                             December 31, 2002, Farmer Mac effectively managed its interest rate risk
Elements of Its Prepayment   through a combination of yield maintenance clauses in loan contracts and
Model Have Limitations       through asset-liability matching; however, we found that prepayment
                             model limitations could affect Farmer Mac’s interest rate risk
                             measurement.26 We observed that Farmer Mac has placed reliance on its
                             ability to issue discount notes matched to interest rate swap transactions.
                             Because discount notes are short-term liabilities and the majority of
                             Farmer Mac’s assets are longer term, a potential mismatch of interest rates
                             could occur. Moreover, the retained portfolio strategy has increased the
                             amount of interest rate risk that Farmer Mac must manage.27 By holding
                             AMBS on its balance sheet, Farmer Mac retains and therefore must manage
                             the interest rate risk in addition to the credit risk associated with AMBS. If
                             Farmer Mac sold the AMBS to investors, it would only retain and have to
                             manage the credit risk associated with AMBS. However, much of the
                             concern relating to interest rate risk is mitigated through Farmer Mac’s use
                             of callable debt and interest rate swaps, which have the effect of adjusting
                             the net interest payments to closely match the interest characteristics of
                             Farmer Mac’s assets. (See appendix V for further discussion.)

                             Farmer Mac measured and reported interest rate risk based on parameters
                             set by board policy as follows. Farmer Mac’s principal metrics for
                             analyzing interest rate risk are

                             •	 market value of equity (MVE)-at-risk calculation, which represents the
                                current economic value of the firm;28


                             26
                                Yield maintenance is a penalty paid by borrowers to lenders when a loan is paid off before
                             its scheduled maturity. It is calculated so that the lender is at least made whole in a time of
                             falling interest rates.
                             27
                               As discussed later in this report, Farmer Mac has chosen to retain the majority of the loans
                             it has purchased and securitized as AMBS.
                             28
                              The MVE is the difference between the present values of cash flows associated with assets,
                             minus the present value of cash flows associated with liabilities and obligations. MVE
                             represents the current economic or financial value of the firm as opposed to the accounting-
                             based value represented on the balance sheet.




                             Page 24                                                               GAO-04-116 Farmer Mac
•	 net interest income (NII) forecast, which represents the change in
   earnings relative to changes in interest rates; and

•	 duration gap calculation, which measures the interest rate mismatch
   between Farmer Mac’ assets and liabilities. 29

For further discussion of Farmer Mac’s interest rate risk measurement
process, see appendix V.

During 2002 Farmer Mac managed its MVE within board-approved limits,
with one exception. NII was also managed within the board-approved
range. The duration gap, which is measured in months, widened from—0.8
months in December 2001 to—3.6 months in December 2002, as loan
prepayments increased as interest rates declined, and these figures were
still within the range of the board-approved parameters.

We found that Farmer Mac had a reasonable process and tools to measure
interest rate risk, but the quality of its risk measurement is potentially
limited by elements of its prepayment model. Prepayment models are an
important component of interest rate risk measurement. Since Farmer
Mac has prepayment penalties or yield maintenance terms on 57 percent of
its outstanding balance of loans and guarantees, including 91 percent of its
loans with fixed interest rates, Farmer Mac’s exposure to interest rate risk
stemming from prepayments is limited. But, Farmer Mac does hold some
loans that are subject to interest rate risk caused by prepayments, such as
fixed-rate loans with less than full yield maintenance acquired through bulk
purchase transactions, or Part Time Farm loans, which generally allow
prepayment without penalty. Farmer Mac’s prepayment risk model was
developed internally based on models that predict prepayment behavior for
residential (housing) mortgage borrowers. But, agricultural real estate
borrowers may behave differently than residential mortgage borrowers.
Farmer Mac management said that they followed this approach due to the
unavailability of external data on agricultural mortgage prepayments. They
also said that Farmer Mac backtests, that is, compares its prepayment
model’s prediction to the prepayment rates actually observed in the recent
past, and finds a close correspondence between the model's predictions
and the experience of its portfolio. A consultant to Farmer Mac has


29
   Duration gap is the difference between the average timing of the cash flows of the assets
and the average timing of the cash flows of the liabilities. For a further description of
duration, see the Glossary.




Page 25                                                              GAO-04-116 Farmer Mac
                             indicated that Farmer Mac’s current practice of incorporating proportional
                             adjustment factors in single family prepayment models is consistent with
                             practices at other agricultural lenders. Farmer Mac has begun the process
                             of estimating prepayment functions based directly on agricultural real
                             estate mortgages. Farmer Mac management noted that they are currently
                             working with the consultant to develop an agriculture mortgage
                             prepayment model so that it can better model prepayment risk. For further
                             information regarding prepayment risk, see appendix V.



Farmer Mac Exceeded          As of December 31, 2002, Farmer Mac had capital in excess of its statutory
Statutory and Regulatory     and regulatory requirements. Its core capital was $184 million, exceeding
                             its statutory minimum capital requirement of $137.1 million. Its regulatory
Capital Requirements, but
                             capital was $204 million, compared to the regulatory risk-based capital
Could Improve Its Planning   requirement of $73.4 million. Although Farmer Mac met statutory and
for Capital Adequacy         regulatory capital requirements, Farmer Mac’s analysis of capital adequacy
                             could be improved.30

                             Pursuant to Farmer Mac’s risk-based capital regulation, it is the
                             responsibility of the Farmer Mac board to ensure that Farmer Mac
                             maintains total capital at a level sufficient for continued financial viability
                             and to provide for growth, in addition to ensuring sufficient capital to meet
                             statutory and regulatory capital requirements.31 In projecting Farmer Mac’s
                             capital needs in the 2002 Business Plan, the Farmer Mac board established
                             a capital goal, based on Farmer Mac’s current circumstances and needs, at
                             a certain fixed amount above the higher of the statutory leverage minimum
                             capital requirement or the required risk-based capital level. In doing so,
                             Farmer Mac has not performed a test of sufficiency for financial viability
                             and growth other than exceeding the statutory and regulatory
                             requirements. Farmer Mac officials said that, in their view, FCA’s regulatory
                             risk-based capital requirement was set at a very conservative level and
                             noted that the statutory minimum is higher than the risk-based capital
                             requirement. However, regulatory requirements are only minimums and
                             financial institutions often find it prudent to keep capital in excess of
                             minimum requirements. Moreover, Farmer Mac’s minimum statutory



                             30
                              Farmer Mac is required to comply with the higher of the minimum capital requirement or
                             the risk-based capital requirement.
                             31
                                  12 C.F.R.§650.22(a).




                             Page 26                                                         GAO-04-116 Farmer Mac
capital requirement,32 which is not risk-based, is set in law and may not be
sufficiently responsive to Farmer Mac’s emerging risks to serve as a proxy
for capital sufficiency. In particular, the statutory minimum requirement of
0.75 percent capital for off-balance-sheet obligations applies to Farmer
Mac’s $2.7 billion of standby agreements, a program that did not exist when
the statute was enacted. Whenever Farmer Mac is obligated under a
standby agreement to purchase a delinquent loan, it must also increase the
capital held against the loan from 0.75 to 2.75 percent, nearly a 270 percent
increase. As noted in our discussion of liquidity risk, Farmer Mac’s
potential problem is that multiple loans would likely be sold to Farmer Mac
during times of agricultural economic stress or under other adverse
conditions. Bringing these loans onto Farmer Mac’s balance sheet would
increase Farmer Mac’s required capital level, and in the current
environment, Farmer Mac’s current capital is able to absorb this increase.
However, if standby agreements or off-balance-sheet assets continue to
grow, Farmer Mac may need to raise capital to withstand such a shock
under stressful economic conditions. By comparison, for capital
requirement purposes, bank regulators’ risk-based capital standards treat
similarly structured, off-balance-sheet financial standby arrangements,
such as guarantees, financial letters of credit, and other direct credit
substitutes, as if they were on the balance sheet.

Moreover, Farmer Mac’s annual filings with SEC illustrate the limitations of
using the regulatory and/or statutory minimum capital as a proxy for having
an internal capital adequacy standard. According to Farmer Mac’s 2002
annual filing with SEC, based on the minimum capital requirements,
Farmer Mac’s current capital surplus of $46.9 million could ultimately allow
Farmer Mac to carry the risk of an additional $15.2 billion of off-balance-
sheet guarantees through a combination of selling on-balance sheet
program assets and adding guarantees.




32
 The minimum capital requirement is an amount of core capital equal to the sum of 2.75
percent of Farmer Mac’s aggregate on-balance-sheet assets, as calculated for regulatory
purposes, plus 0.75 percent of the aggregate off-balance-sheet obligations of Farmer Mac.




Page 27                                                            GAO-04-116 Farmer Mac
Disagreements about the      We identified an issue involving Farmer Mac’s $1.5 billion line of credit with
Extent of Coverage of        Treasury that could impact Farmer Mac’s long-term financial condition.
                             Treasury has expressed serious questions about whether Treasury is
Treasury’s Line of Credit    required to purchase Farmer Mac obligations to meet Farmer Mac-
Could Generate Uncertainty   guaranteed liabilities on AMBS that Farmer Mac or its affiliates hold.33 On
                             the other hand, a legal opinion from Farmer Mac’s outside counsel states
                             that Treasury would be required to purchase the debt obligations whether
                             the obligations are held by a subsidiary of Farmer Mac or by an unrelated
                             third party. This disagreement could create uncertainty as to whether
                             Treasury would purchase obligations held in Farmer Mac’s portfolio in
                             times of economic stress. This uncertainty also relates to statements made
                             by Farmer Mac to investors concerning Treasury’s obligation to Farmer
                             Mac, which in turn, could affect Farmer Mac’s ability to issue debt at
                             favorable rates. Ultimately, this uncertainty could impact its long-term
                             financial condition.

                             Farmer Mac’s subsidiary, Farmer Mac Mortgage Securities Corporation,
                             holds the majority of AMBS that Farmer Mac issued. Farmer Mac’s charter
                             (the 1987 Act) gives it the authority to issue obligations to the Secretary of
                             the Treasury to fulfill its guarantee obligations. According to the 1987 Act,
                             the Secretary of the Treasury may purchase Farmer Mac’s obligations only
                             if Farmer Mac certifies that (1) its reserves against losses arising out of its
                             guarantee activities have been exhausted and (2) the proceeds of the
                             obligations are needed to fulfill Farmer Mac’s obligations under any of its
                             guarantees.34 In addition, Treasury is required to purchase obligations
                             issued by Farmer Mac in an amount determined by Farmer Mac to be
                             sufficient to meet its guarantee liabilities not later than 10 business days
                             after receipt of the certification. However, Treasury has indicated that the
                             requirement to purchase Farmer Mac obligations may extend only to
                             obligations issued and sold to outside investors.




                             33
                                Both Treasury and Farmer Mac are in agreement that the authority of Treasury to purchase
                             obligations to enable Farmer Mac to fulfill its guarantee obligations does not extend to the
                             standby agreements because they do not involve Farmer Mac’s guarantee liabilities.
                             34
                                  12 U.S.C.2279aa-13.




                             Page 28                                                             GAO-04-116 Farmer Mac
In a comment letter dated June 13, 1997, and submitted to FCA in
connection with a proposed regulation on conservatorship and
receivership for Farmer Mac (1997 Treasury letter),35 Treasury stated “…we
have ‘serious questions’ as to whether the Treasury would be obligated to
make advances to Farmer Mac to allow it to perform on its guarantee with
respect to securities held in its own portfolio---that is, where the Farmer
Mac guarantee essentially runs to Farmer Mac itself.” The 1997 Treasury
letter indicated that if the purchase of obligations extended to guaranteed
securities held by Farmer Mac this would belie the fact that the securities
are not backed by the full faith and credit of the United States, since a loan
to Farmer Mac to fulfill the guarantee would benefit holders of Farmer
Mac’s general debt obligations. The 1997 Treasury letter stated “Treasury’s
obligation extends to Farmer Mac only in the prescribed circumstances,
and is not a blanket guarantee protecting Farmer Mac’s guaranteed
securities holders from loss. Nor is the purpose of the Treasury’s obligation
to protect Farmer Mac shareholders or general creditors.” According to
Treasury, the 1997 letter remains its position concerning Farmer Mac’s line
of credit.

Meanwhile, the opinion of Farmer Mac’s outside counsel is that the
guarantee is enforceable whether AMBS are held by a subsidiary of Farmer
Mac or by an unrelated third party. Farmer Mac’s legal opinion also states
that Treasury could not decline to purchase the debt obligations issued by
Farmer Mac merely because the proceeds of the obligations are to be used
to satisfy Farmer Mac’s guarantee with respect to AMBS held by a
subsidiary. According to Farmer Mac, if the conditions set forth in the 1987
Act are met—required certification and a limitation on the amount of
obligations of $1.5 billion—then there is no exception in the 1997 Act that
authorizes Treasury to decline to purchase the obligations. Farmer Mac
states that discriminating among Farmer Mac guaranteed securities based
on the identity of the holder in determining whether Farmer Mac could
fulfill its guarantee obligations would lead to an anomalous situation in the
marketplace and thereby hinder the achievement of Congress’ mandate to
establish a secondary market for agricultural loans.




35
   Letter dated April 13, 1997, from then-Under Secretary for Domestic Finance, John D.
Hawke, Jr., to Marsha P. Martin, then-Chairman of the Farm Credit Administration.




Page 29                                                            GAO-04-116 Farmer Mac
Mission-Related            Our analysis of Farmer Mac’s impact on the agricultural real estate loan
                           market indicated that Farmer Mac has increased its agricultural mortgage
Activities Have            loan purchase and guarantee activity since our last report in 1999. At the
Increased, but Impact      same time, its enabling legislation contains broad mission purpose
                           statements and lacks specific or measurable mission-related criteria that
of Activities on           would allow for a meaningful assessment of whether Farmer Mac had
Agricultural Real          achieved its public policy goals. For example, the statute does not contain
Estate Market Is           specific mission criteria for Farmer Mac to make credit available for
                           specific clientele such as small, beginning, and disadvantaged farmers. In
Unclear                    assessing whether Farmer Mac has made available long-term credit to
                           farmers and ranchers at stable interest rates, we found that its long-term
                           interest rates were similar to the rates of agricultural real estate lenders. In
                           addition, Farmer Mac’s strategy of holding AMBS to lower funding costs
                           and increase profitability may have limited the development of a secondary
                           market for these securities. Farmer Mac introduced the standby agreement
                           program to provide greater lending capacity for agricultural real estate
                           lenders, but growth in standby agreements, as with other guarantee
                           obligations, could potentially result in reducing the sum of capital required
                           to be held by the Farm Credit System and Farmer Mac without
                           corresponding mitigating factors such as lender and geographic
                           diversification. We found that Farmer Mac’s business activities are largely
                           concentrated among a small number of business partners and its portfolio
                           is concentrated in the West. Finally, the size of Farmer Mac’s nonmission
                           investment portfolio has decreased as a percentage of its total on- and off-
                           balance sheet portfolio. Still the composition and criteria for nonmission
                           investment could potentially lead to investments that are excessive in
                           relation to Farmer Mac’s financial operating needs or otherwise be
                           inappropriate to the statutory purpose of Farmer Mac.



Farmer Mac Has Continued   Farmer Mac’s loan and guarantee portfolio has continued to grow since
to Grow, but Mission       1999, but purchase activity notwithstanding, the extent to which Farmer
                           Mac has met its public policy mission is difficult to measure. Farmer Mac’s
Criteria Are Lacking
                           enabling legislation contains only broad mission related guidance;
                           therefore, measurable criteria are not available. The 1987 Act stated that
                           Farmer Mac was to provide for a secondary marketing arrangement for
                           agricultural real estate mortgages in order to (1) increase the availability of
                           long-term credit to farmers and ranchers at stable interest rates; (2)
                           provide greater liquidity and lending capacity in extending credit to farmers
                           and ranchers; and (3) provide an arrangement for new lending to facilitate




                           Page 30                                                    GAO-04-116 Farmer Mac
capital market investments in providing long-term agricultural funding,
including funds at fixed rates of interest.

Farmer Mac stated that as a secondary market institution, it faced
significantly lower economic risks than primary lenders, such as FCS
institutions and commercial banks, given its ability to attain geographic
and commodity diversification, access to national and international capital
markets, and the ability to borrow at lower costs due to its agency status. It
also noted that the lower capital requirements provided to primary lenders
through the Farmer Mac I program created the potential for increased
lending capacity, higher profitability, and potentially lower interest rates
for farmers and ranchers. Notwithstanding these claims, and with respect
to the mission related guidance, over the past 2 years, the long-term
interest rates that Farmer Mac offered to agricultural real estate lenders,
through the Farmer Mac I program have decreased along with the rates of
the primary agricultural real estate lenders (see fig. 2). We found that
agricultural mortgage yields have not declined over time relative to 10-year
Treasury securities and that long-term fixed interest rates on Farmer Mac I
loans were similar to those offered by commercial banks and FCS
institutions (see fig. 2).




Page 31                                                  GAO-04-116 Farmer Mac
                              Figure 2: Long-term Interest Rates on Loans Secured by Agricultural Real Estate
                               10




                                   8




                                   6




                                   4




                                   2




                                   0
                                       Q4          Q1           Q2             Q3            Q4            Q1             Q2            Q3               Q4

                                       2000        2001                                                    2002           

                                       Year and quarter


                                               Commercial banks

                                               Farm Credit System

                                               Farmer Mac I

                                               10-year Treasury bond
                               Sources: GAO analysis of Farmer Mac data and USDA's Agricultural Income & Finance Outlook, AIS80, March 11, 2003, p.63.




Farmer Mac’s Strategy of      Farmer Mac’s strategy of holding the loans it purchases and securitizes as
Retaining AMBS Has Been a     AMBS has been a contributing factor in limiting the development of a liquid
                              secondary market for AMBS. This retained portfolio strategy was initially
Contributing Factor in
                              announced in Farmer Mac’s 1998 third quarter filing with SEC. The
Limiting the Development of   explanation given at the time for retaining AMBS was that market volatility
a Liquid Secondary Market     resulted in lower rates on Treasury securities but wider spreads on AMBS.
for AMBS                      These conditions lowered potential gains on issuance of AMBS but
                              facilitated Farmer Mac’s retention of AMBS at favorable spreads; therefore,
                              Farmer Mac would hold the AMBS until market conditions changed. 36
                              According to USDA, holding AMBS has typically been more profitable but


                              36
                                   A spread is the difference between two prices or two rates.




                              Page 32                                                                                         GAO-04-116 Farmer Mac
                         also more risky than selling AMBS to investors. During 2002, Farmer Mac
                         did not make any sales of AMBS to unrelated parties.37 Farmer Mac noted
                         that the economics of retention have proven superior, and Farmer Mac’s
                         growth, profitability, and greater capital market presence should facilitate
                         future AMBS sales. As of December 31, 2002, Farmer Mac had securitized
                         and sold 7 percent of its entire Farmer Mac I portfolio (see fig. 3).



                         Figure 3: Securitization Status of Farmer Mac I Portfolio, as of December 31, 2002
                                                                      Securitized and sold
                                                     7%


                                                                19%   Securitized and retained


                                     55%

                                                              19%     Unsecuritized loans




                                                                      Standby agreements
                         Source: GAO analysis of Farmer Mac data.




Farmer Mac’s Business    With the development of the standby agreement program, Farmer Mac has
Activities Are Largely   continued to provide products to facilitate capital market investments in
                         order to provide long-term agricultural funding, which in turn, could result
Concentrated
                         in additional agricultural lending. This is consistent with its mission to
                         provide an arrangement for new lending to facilitate capital market
                         investments in providing long-term agricultural funding, including funds at
                         fixed rates of interest. The additional lending would be a result of the
                         lower amount of capital that lending institutions would be required to hold,
                         provided their products were guaranteed or in a standby agreement with
                         Farmer Mac. The risks associated with lower capital requirements would


                         37
                            The single AMBS transaction made by Farmer Mac that year was a $47.7 million sale to a
                         related party, which represented only 2 percent of Farmer Mac’s loan purchase, guarantee,
                         and commitment activity for the year.




                         Page 33                                                             GAO-04-116 Farmer Mac
be in part mitigated through sufficient diversification relating to
participating lenders and geography. However, Farmer Mac’s activities
have been largely concentrated in a small number of financial institutions.
According to Farmer Mac’s 2002 annual filing with SEC, Farmer Mac
purchased eligible loans from 63 financial institutions, and provided
standby agreements to 16 entities. During 2002, 10 institutions generated
90 percent of Farmer Mac’s business, and 3 FCS institutions represented 47
percent of the outstanding balance of the standby agreement program as of
December 31, 2002.

Moreover, Farmer Mac’s portfolio does not represent the nationwide
distribution of general farm-related real estate indebtedness across
commercial banks and FCS institutions. As shown in figure 4, FCS
institutions were the source for approximately 2 percent of Farmer Mac I
program loans in 1996, but by December 2002, they accounted for more
than 55 percent. In contrast, commercial banks participation rate has
dropped from 80 percent of Farmer Mac I program loans in 1996 to 22
percent as of December 2002. This compares to FCS institutions holding 36
percent and commercial banks holding 32 percent of nationwide farm-
related real estate debt, as of 2002. Representatives from USDA and a bank
association noted that the banking industry strongly supported the creation
of Farmer Mac in 1987 because they viewed Farmer Mac as a new source of
competitively priced funding. While commercial banks’ relative share of
Farmer Mac’s business has been falling, bank-held farm mortgage volume
has doubled since Farmer Mac was created. Farmer Mac management said
that the decline in the commercial banks’ participation in Farmer Mac’s
programs was due to the falling interest rate environment and a general
desire of the commercial banks’ to retain loans in portfolio. Management
anticipated that when interest rates begin to rise in the near future, as is
forecasted by USDA, commercial banks and mortgage brokers will begin to
take advantage of Farmer Mac’s longer-term products.




Page 34                                                 GAO-04-116 Farmer Mac
Figure 4: Farmer Mac Portfolio Exposure by Loan Origination Type
Percent

100




 80




 60




 40




 20




  0
        1996        1997        1998       1999   2000   2001   2002
      Year


                Commercial banks

                Farm credit system

                Mortgage brokers

                Insurance companies

Source: GAO analysis of Farmer Mac data.



By shifting credit risk exposure from FCS institutions to Farmer Mac,
standby agreements, as with other guarantee obligations, potentially lower
the overall capital required to be held by FCS (see table 2). Whereas the
total capital for an unguaranteed loan is $7, the total capital for a loan
under a standby agreement or swap is only $2.15.




Page 35                                                                GAO-04-116 Farmer Mac
Table 2: Comparison of Total Minimum Capital Levels Per $100 of Loans

Transaction                     FCS institution             Farmer Mac               Total capital
FCS loan                                   $7.00                      N/A                    $7.00
Standby agreement                          $1.40                    $0.75                    $2.15
                                                                          a
Loan sale                                  $0.00                   $2.75                     $2.75
Swap for AMBS                              $1.40                    $0.75                    $2.15
Source: Farm Credit System.
a
 This assumes that Farmer Mac retains the loan or securitizes the loan and holds the AMBS on its
balance sheet.


Farmer Mac’s absorption of FCS institutions’ credit risk through the
standby agreement program might be consistent with a lower capital
requirement if concentration of credit risk was reduced by geographic
diversification. However, Farmer Mac’s risk exposure is concentrated in
the western part of the United States. As of December 31, 2002, over 70
percent of the outstanding balance of Farmer Mac’s loan portfolio was
located in the Southwest and Northwest (see fig. 5). For comparative
purposes, the corresponding percentage of general farm debt in those
regions was only 31 percent of nationwide farm debt.38 Greater geographic
diversification of Farmer Mac loans would lower risks of concentration and
mitigate risks associated with the lower capital requirements.




38
 General farm debt includes more than agricultural real estate mortgages; however, it is a
proxy for the relative proportion of farm borrowing in a region (USDA, Economic Research
Service).




Page 36                                                                  GAO-04-116 Farmer Mac
                            Figure 5: Farmer Mac I Geographic Concentration of Exposure by Region, as of
                            December 31, 2002
                                                                             Mid-south
                                                        5%                   Northeast
                                                               6%
                                                                       7%    Southeast


                                      47%                              11%   Mid-north




                                                            24%              Northwest




                                                                             Southwest
                            Source: GAO analysis of Farmer Mac data.




Proportion of Nonmission    When analyzing Farmer Mac from a mission perspective, an excessively
Investments Has Declined,   large nonmission investment portfolio in relation to Farmer Mac’s business
                            needs could potentially lead to charges that Farmer Mac is misusing its
but Issues Remain about
                            status as a GSE. As of December 31, 2002, the nonmission investments that
Composition and Potential   have $830.4 million combined with $723.8 million in cash and cash
Growth                      equivalents equaled 37 percent of total balance sheet assets at Farmer Mac.
                            This figure is down from 66 percent in 1997, when we last reported on GSE
                            nonmission investments, reflecting an increase in Farmer Mac’s assets
                            resulting from loan purchase and guarantee activity since that time.39
                            Included in Farmer Mac’s nonmission investments, as reported in SEC
                            filings as of December 31, 2002, were $93 million in unrated, preferred
                            stock of CoBank, which is an FCS institution. This investment is one of
                            Farmer Mac’s top five holdings of nonmission assets and represents 6
                            percent of its liquidity portfolio. Also in 2002, Farmer Mac’s board
                            approved a change to the limit of its nonmission investment portfolio. As
                            an alternative to the fixed-dollar amount, the Board approved a percentage


                            39
                             U.S. General Accounting Office, Government Sponsored Enterprises: Federal Oversight
                            Needed for Nonmortgage Investments, GAO/GGD-98-48 (Washington, D.C.: Mar. 11, 1998),
                            p.17.




                            Page 37                                                       GAO-04-116 Farmer Mac
                         limit of 30 percent of the total portfolio, including on-balance sheet assets
                         and off-balance sheet commitments. This is the same maximum that FCA
                         allows its institutions (other than Farmer Mac). The effect of this change
                         was to remove absolute limits on the size of the nonmission investment
                         portfolio. FCA officials with whom we spoke said that FCA neither
                         endorsed nor objected to the policy change. FCA officials noted that they
                         would monitor the portfolio growth to ensure that the potential incentive to
                         growing the nonmission investment portfolio was balanced with
                         appropriate growth in the loan and guarantee portfolio.



Farmer Mac’s Statutory   Similar to other publicly traded companies, Farmer Mac is in the process of
                         taking actions to ensure that it complies with recent legislative and
Governance Structure     regulatory requirements and proposed changes in NYSE listing standards.
Does Not Reflect         In accordance with the new requirements, Farmer Mac has reaffirmed its
                         audit committee charter and has recently hired internal and external
Interests of All         auditors who are from different firms. The Sarbanes-Oxley Act requires
Shareholders and         that members of audit committees of listed companies be independent and
Some Corporate           requires that SEC issue and adopt rules directing the national securities
                         exchanges to prohibit listing any securities of a company that is not in
Governance Practices     compliance with the audit committee requirements. Proposed NYSE listing
Need to Be Updated       standards stress the oversight role of boards of directors and the
                         independence of the directors. Since Farmer Mac’s securities are
                         registered with SEC and Farmer Mac’s Class A and Class C stock are listed
                         on NYSE, Farmer Mac is currently subject to the requirements of Sarbanes-
                         Oxley and implementing SEC rules, and, absent a waiver, the proposed
                         listing standards as they become effective. We noted that Farmer Mac’s
                         board has taken steps to update its corporate governance practices, but its
                         board structure, which is set by law, could make it difficult to comply with
                         the board independence requirements proposed in NYSE listing standards.
                         Moreover, Farmer Mac’s governance structure contains elements of a
                         cooperative and elements of an investor-owned, publicly traded
                         corporation. Because Farmer Mac shareholders include both institutions
                         that utilize its services and public investors, and because all members of
                         the board of directors are chosen by the cooperative investors or by the
                         President of the United States, the board may face difficulties in
                         representing the interests of all shareholders. The interests and loyalties of
                         directors of publicly traded corporations, including publicly traded GSEs,
                         should be clearly focused on serving the interests of all shareholders.
                         However, we found that the statutory structure of Farmer Mac’s board and
                         the voting structure of its common stock hamper Farmer Mac’s ability to
                         have such a focus. In addition, although discussed to some degree in its



                         Page 38                                                  GAO-04-116 Farmer Mac
                               proxy statement, we found from our discussions with Farmer Mac’s 15
                               board members that (1) Farmer Mac’s process for identifying and selecting
                               board nominees was not transparent to them, (2) training for directors was
                               inconsistent, and (3) executive management succession planning was not
                               well documented. When assessing Farmer Mac’s compensation for its
                               executive management, we found that Farmer Mac’s total executive
                               compensation was within its consultants’ recommended parameters;
                               however, its vesting program appears more generous than industry
                               practices, given Farmer Mac’s maturity.



Farmer Mac’s Governance        Like other Farm Credit System institutions, Farmer Mac resembles a
Structure Contains             cooperative controlled by institutions that utilize its services. Under the
                               1987 Act, Farmer Mac has three classes of common stock. Class A voting
Elements of a Cooperative      common stock is owned by banks, insurance companies, and other
and Elements of an Investor-   financial institutions. Class B voting common stock is owned by FCS
Owned Corporation              institutions, but ownership of Class C nonvoting common stock is not
                               restricted. According to the background of Farmer Mac’s charter act, Class
                               C nonvoting common stock was created as a means for Farmer Mac to
                               raise capital and to preserve equal distribution of voting stock between
                               Farm Credit System and non-Farm Credit System Institutions.40 However,
                               unlike ownership interests in the other FCS institutions, but like the
                               common stock of Fannie Mae and Freddie Mac, Farmer Mac’s Class A and
                               Class C stock is publicly traded on the NYSE. Farmer Mac, through the sale
                               of the stock and the issuance of debt securities, depends on the capital
                               markets for funding. Unlike the other GSEs, including Fannie Mae and
                               Freddie Mac, Farmer Mac is subject to the securities laws, and files
                               disclosure documents with respect to its securities issuances. In
                               compliance with the requirements of the securities laws, Farmer Mac files
                               quarterly and annual reports, proxy statements, and other documents that
                               provide information to investors about financial condition and
                               management.




                               40
                                 The authority to issue Class C common, nonvoting stock was added as an amendment to
                               the proposed legislation that became the 1987 Act by Senator Leahy, who explained the
                               purpose of the amendment as follows: “….amendments establish that while the initially
                               issued stock is voting and fairly distributed between the Farm Credit System and non-Farm
                               Credit System participants, the corporation has the authority to issue additional nonvoting
                               common and preferred stock if it is determined by the mortgage corporation that the
                               corporation should raise additional capital.”




                               Page 39                                                            GAO-04-116 Farmer Mac
Farmer Mac’s board of directors is not elected by all of its shareholders.
Under the 1987 Act, Farmer Mac’s board of directors consists of 15
members, 5 of whom are to be elected by holders of the Class A voting
common stock, 5 are to be elected by holders of Class B voting common
stock, and 5 are appointed by the President of the United States, with the
advice and consent of the Senate. The five members appointed by the
President (1) could not be, or have been officers and directors of any
financial institutions or entities and (2) were to be representatives of the
general public—not more than three of whom could be members of the
same political party and at least two were to be experienced in farming or
ranching. According to statements made at the time of consideration of the
1987 Act, this structure was to protect the interests of both the Farm Credit
System and commercial lenders by providing for equal representation on
the board by FCS, commercial lenders, and the public sector.41

Compliance with the disclosure requirements of the 1934 Act provides
investors with information about Farmer Mac, including information that
enables investors to compare Farmer Mac with other publicly traded
companies that participate in the capital markets. However, unlike most
other publicly traded corporations, Farmer Mac is controlled not by
investors but by institutions that have a business relationship with Farmer
Mac. Farmer Mac’s board of directors has a fiduciary responsibility to act
in the best interests of the institution and its shareholders; Farmer Mac
shareholders included businesses that are users of Farmer Mac’s financial
services and investors in nonvoting Class C stock. This structure requires
that directors act in the best interests of shareholders that may have widely
divergent interests. Class A and Class B shareholders are concerned with
the use of Farmer Mac services, while Class C shareholders are generally
investors concerned with maximizing their profits. Good corporate
governance requires that the incentives and loyalties of the board of
directors of publicly traded companies reflect the fact that the directors are
to serve the interests of all the shareholders. Shareholders of public
companies can contribute to the governance of corporate conduct with a
view to enhancing corporate responsibility. Shareholders who exercise the



41
   Statement of Congressman Bereuter, H11869-01, Congressional Record. In addition, the
conference report cites testimony given in hearings held prior to enactment of the bill that
indicate that FCS spokepersons argued that the secondary market mechanism should
operate as an arm of the FCS and private lenders believed that the FCS should have a much
more limited involvement in the secondary market and that additional control over a large
secondary market operation would give the FCS an unfair competitive advantage.




Page 40                                                             GAO-04-116 Farmer Mac
                       power to elect and remove directors can influence corporate policy
                       through governance proposals and nominations to the boards of directors.



Class C Common Stock   Farmer Mac’s Class C shareholders cannot vote on significant matters that
Does Not Have Voting   generally require shareholders’ votes—such as nominating the board of
                       directors, executive compensation policies, and the selection of the
Rights                 independent auditor. We explained Farmer Mac’s nonvoting structure to
                       some shareholder advocacy groups, who stated that shareholders should
                       be able to vote and voice their opinion on governance and management
                       issues. These investor groups advocate “one share, one vote.” According
                       to Farmer Mac management, the provisions of Farmer Mac’s charter
                       intended that the agricultural lending industry control the board and
                       stockholder voting issues while the company developed, which is a process
                       that they believe is still under way. Further, they said that holders of Class
                       C common stock acquire the stock with that information clearly disclosed
                       to them and implicitly accept the representation of their interests by the
                       board and, to a large degree, by Class A and Class B stockholders as
                       surrogates representing their economic interest, since all classes have the
                       same dividend and liquidation rights. However, given Farmer Mac’s rapid
                       growth and today’s corporate governance environment, this nonvoting
                       structure may no longer be appropriate.

                       Eliminating statutory control of the Farmer Mac’s board by Class A and
                       Class B shareholders and providing an equal voice to Class C shareholders,
                       as well as eliminating the statutory requirement that the President appoint
                       members of Farmer Mac’s board would provide for a board elected by all
                       Farmer Mac shareholders. We note however, that holders of Class A and
                       Class B stock also hold a significant proportion of the Class C shares.
                       According to Farmer Mac’s 2003 proxy statement, the company’s executive
                       officers and directors are the “beneficial owners” of 29.8 percent of Farmer
                       Mac’s outstanding nonvoting common stock, as defined by SEC rules.
                       Almost half this amount is shares owned by Zion’s Bancorporation, one of
                       whose officers is on Farmer Mac’s board of directors. SEC’s beneficial
                       owners definition includes stock options that are exercisable within 60
                       days; in Farmer Mac’s case, unexercised options comprise most of the
                       executive officers’ and directors’ beneficially held shares. Consequently,
                       even if Class C shareholders were allowed to vote, the Farmer Mac board
                       of directors would be elected by many shareholders that currently hold the
                       right to vote. In contrast, the executive management and directors of
                       Fannie Mae and Freddie Mac have combined beneficial ownership of less
                       than 1 percent of their respective companies’ outstanding common stock.



                       Page 41                                                  GAO-04-116 Farmer Mac
Farmer Mac Is Subject to    Farmer Mac is subject to NYSE listing requirements, and will be subject to
NYSE Listing Standards on   proposed listing standards on corporate governance, as well as statutory
                            and regulatory requirements. Recent reforms have prompted Farmer Mac’s
Corporate Governance        board to reassess its oversight role of Farmer Mac and take actions to
                            comply with new requirements within the bounds set by its statute. Based
                            on our interviews with Farmer Mac’s 15 board directors, its board
                            committees are taking actions to comply with the provisions of the
                            Sarbanes-Oxley Act of 2002 (Sarbanes-Oxley), SEC rules, and the proposed
                            NYSE listing standards. For example, Farmer Mac’s board has revised its
                            audit committee’s written charter to include the committee’s
                            responsibilities, and has recently hired internal and external auditors who
                            are not from the same firm. However, because Farmer Mac’s board
                            structure is established by its charter act, it may encounter difficulties in
                            complying with the new standards, which require that a majority of the
                            board be independent and that key committees (audit, nominating, and
                            compensation) consist entirely of independent directors.

                            In response to recent corporate scandals, corporate governance
                            policymakers have focused on the importance of an independent board of
                            directors who act in the best interest of the corporation. The Sarbanes-
                            Oxley Act contains new requirements concerning the composition and
                            duties of the audit committee, including a requirement that all audit
                            committee members be independent, which means that the committee
                            member cannot accept any consulting, advisory, or other compensatory
                            fees from the company (other than compensation for serving as director),
                            or be affiliated with the company or any of its subsidiaries. Sarbanes-Oxley
                            also requires that SEC adopt rules requiring national securities exchanges
                            to prohibit listing any company that does not satisfy these requirements.




                            Page 42                                                 GAO-04-116 Farmer Mac
The NYSE has submitted proposed corporate governance listing standards
to SEC. To increase the quality of board oversight and lessen the potential
for conflicts of interest, the proposed listing standards require that a
majority of the board of directors of listed companies be independent. No
director qualifies as independent unless the board of directors affirmatively
determines that the director has “no material relationship” with the listed
company, either directly or as an officer or director of an organization that
has a relationship with a company. Material relationships can include
commercial, banking, consulting, legal, and accounting relationships.42 It
is not clear, however, whether Farmer Mac directors’ business
relationships with Farmer Mac would prevent these individuals from
serving as independent directors under the NYSE proposed rules. Farmer
Mac’s 2002 annual proxy statements indicated that 6 of 15 directors were
listed as having certain relationships or having conducted related
transactions with Farmer Mac. In comparison, in their 2002 annual proxy
statements, Fannie Mae reported that 4 of their 18 directors and Freddie
Mac reported that 3 of their 18 directors as having business relationships.
Because the Class A and Class B directors are from institutions that have
financial relationships of varying degrees with Farmer Mac, they may not
be independent, thus the statutory structure of Farmer Mac’s board could
make it difficult for Farmer Mac to adopt corporate governance practices
and policies that may be required or recommended by authorities on
corporate governance issues. When commenting on our report, Farmer
Mac officials stated that Farmer Mac was in compliance with existing and
proposed NYSE standards. Further, they said that 12 out of the 15 Farmer
Mac directors were “independent” in the opinion of the board’s corporate
governance consultant.




42
   Listed companies are required to disclose these determinations. The proposed standards
also contain descriptions of relationships in which a director is presumed not to be
independent until 5 years after the relationship ceases. These relationships are as follows:
(1) The director or an immediate family member receives more than $100,000 per year in
direct compensation from the listed company, other than director and committee fees and
pensions. (2) The director or an immediate family member is affiliated with or employed in
a professional capacity by a present or former internal or external auditor of the company.
(3) The director or an immediate family member is employed as an executive officer of
another company where any of the listed company’s present executives serves on that
company’s compensation committee. (4) The director or immediate family member is an
executive officer of another company that accounts for (a) at least 2% or $1 million,
whichever is greater, or the listed company’s consolidated gross revenues, or (b) for which
the listed company accounts for at least 2% or $1 million, whichever is greater of the other
company’s gross revenues.




Page 43                                                             GAO-04-116 Farmer Mac
Consistency and            Regarding board processes, we found that Farmer Mac’s board nomination
Transparency of Some       process, director training, and management succession planning were not
                           as concise, formal, or well documented as best practices would suggest.
Board Processes Could Be   For example, during our interviews with existing directors, we received
Improved                   inconsistent responses regarding Farmer Mac’s criteria for identifying and
                           selecting directors and the process for nominating directors, raising
                           concerns about consistency and transparency in the nomination process.
                           To further demonstrate the significance of having a transparent process for
                           nominating directors, SEC has proposed new rules requiring expanded
                           disclosure of companies’ nomination process and specific disclosure of
                           procedures by which shareholders may communicate with directors. The
                           new rules are to enable shareholders to evaluate a company’s board of
                           directors and nominating committee. The proposals include disclosure of
                           the nominating committee’s process for identifying and considering
                           nominees, including criteria used to screen nominees and including the
                           minimum qualifications and standards the nominating committee believes
                           company directors should have.

                           Regarding the training for directors, from our interviews with the directors,
                           we found that some directors were provided with in-depth training, while
                           others were given a brief orientation to Farmer Mac’s operations. Finally,
                           at the time of our review, most directors informed us that they were
                           uncertain if Farmer Mac had an executive management succession plan.
                           Farmer Mac’s corporate governance consultant confirmed that an
                           executive management succession plan did exist, but had not been
                           communicated to the entire board. According to Farmer Mac officials, an
                           executive management succession plan was presented and approved at the
                           June 2003 annual board meeting.



Farmer Mac’s Total         Farmer Mac’s total executive compensation package was within the
Executive Compensation     parameters provided by two compensation consultants, although Farmer
                           Mac is not readily comparable to private companies or GSEs due to its
Was Within Consultants’    small size, business complexity, and cooperative board structure. Farmer
Recommended Parameters,    Mac has considered itself a start-up company—using 1996 as the initial year
but Its Vesting Program    although it has been in business since 1987—and has compared itself to a
Appears Generous           technology company model because of its daily operational risks and
                           demonstrated growth. Generally, start-up companies have aggressive
                           compensation packages to attract highly qualified employees, paying a
                           higher proportion of compensation in the form of equity incentives, such as
                           stock options premised on future growth and earnings. Farmer Mac’s total



                           Page 44                                                 GAO-04-116 Farmer Mac
compensation has included an annual salary, an annual bonus, and stock
options, which are included in its vesting program.

In 1995, the board retained a compensation consultant to establish a
compensation package for its staff. Farmer Mac’s total executive
compensation was based on a number of factors—the compensation
consultant’s suggestions, the board’s business plan targets, and the value of
stock options granted. The consultant assesses the compensation package
annually, and on a multiyear basis, takes into account pay levels and rate of
increase at Farmer Mac and similar private companies and GSEs. In 2002,
FCA retained an independent compensation consultant to determine if
Farmer Mac’s total executive compensation package was reasonable. We
reviewed both Farmer Mac and FCA consultant reports. Both consultants
provided a range of benchmarks to compare Farmer Mac’s compensation,
but used different assumptions that may not be entirely applicable to
Farmer Mac. Specifically, we question whether the “start-up”
assumption—used as an industry benchmark by Farmer Mac’s consultant
to develop its compensation package—was still valid, given the maturity of
Farmer Mac. Further, Farmer Mac’s consultant heavily weighted the
housing GSEs as comparable peer organizations to ensure that Farmer
Mac’s compensation structure was competitive enough to attract and retain
qualified executives. FCA’s consultant also used the housing GSEs as
benchmarks, in addition to mortgage banking organizations and financial
service organizations because the various organizations more closely
represented the positions from which executive management would be
recruited. We question whether putting such heavy emphasis on housing
GSEs as a benchmark is appropriate because they are so much larger and
more complex than Farmer Mac, in terms of size and structure, earnings,
portfolio, and operations. For example, Farmer Mac has 33 employees
compared to Fannie Mae and Freddie Mac’s 4,700 and 3,900 employees,
respectively. As shown in table 3, Farmer Mac’s compensation and options
granted fell below the much larger housing GSEs.




Page 45                                                 GAO-04-116 Farmer Mac
Table 3: Annual Compensation and Options Granted for CEO’s of Farmer Mac and
Housing GSEs

Dollars in millions
                                                                                            Value of

                                                                                       unexercised 

                                                                                      in-the-money 

                                                                 Annual Options grant     options at 

                                                            compensation date present      year-end

                                  Annual salary                  bonus         value    exercisable 

President & CEO,
Farmer Mac

Loan & Guarantee
Portfolio: $5.5 billion

Employees: 33                             $447,480                  $344,195             $1,150,783            $9,511,068
CEO and Chairman,
Fannie Mae

Loan & Guarantee
Portfolio: $1.8 trillion

Employees: 4,700                            992,250                3,300,000               6,680,395             1,441,600
CEO and Chairman
Freddie Mac (A)

Loan & Guarantee
Portfolio: $1.1 trillion

Employees: 3,900                         1,132,500                 2,123,438               3,899,741           17,227,372
Source: Compensation information from Farmer Mac and Fannie Mae proxy statements as of April 2003, and Freddie Mac’s proxy
statement as of April 2002. Portfolio size and employee data from Farmer Mac 2002 Annual Report and Fannie Mae 2002 Annual
Report, and Freddie Mac 2001 Annual Report. Freddie Mac’s 2002 and 2001 financial results subject to restatement.

Notes: Freddie Mac’s loan and guarantee portfolio size is an estimate from its 2001 financial
statements because Freddie Mac was in the process of restating its 2000 to 2002 financial statements.
Freddie Mac was not expected to complete the restatement until November 30, 2003.
This table excludes long-term compensation, restricted stock awards, other annual compensation,
securities underlying options, LTIP payouts, and all other compensation, which includes life insurance
premiums, defined contribution pension plans, and Retirement Savings Plans for Employees.


However, when benchmark issues are set aside, and Farmer Mac is
compared to public companies or GSEs, its total executive compensation
was within the consultants’ recommended parameters but its stock option
vesting program appears generous compared to general industry practices.
Under Farmer Mac’s 1997 Stock Option Plan, Farmer Mac employees and
directors have been granted options in stages, with one-third of the options
vested immediately on the date being granted, one-third vested at the end



Page 46                                                                                       GAO-04-116 Farmer Mac
                           of the following year, and the remainder vested in the second year.
                           According to current practices of public and private companies with a
                           public mission, these companies have average vesting periods of 4 to 5
                           years, with employees vesting 25 percent annually for 4 years after 1 year of
                           employment. Generally, companies structure their vesting schedules to
                           attract and to retain employees. Additionally, according to researchers,
                           start-up companies use vesting programs to attract an important group of
                           intellectual capital employees, and vest sooner to bolster income levels so
                           that the employees can be compensated for their contributions. In these
                           cases, more generous vesting programs serve the need to quickly develop
                           the company to profitability, which may no longer be suitable for Farmer
                           Mac’s needs.



FCA Has Taken Steps        FCA has recently taken several steps to strengthen its oversight of Farmer
                           Mac, including instituting a more comprehensive safety and soundness
to Enhance Oversight       examination and undertaking initiatives to expand its regulatory
of Farmer Mac, but         framework. However, as it continues to improve its oversight of Farmer
                           Mac, FCA faces five major challenges. First, limitations exist in the model
Faces Challenges That      used to estimate Farmer Mac’s credit risk. Second, FCA’s regulation does
Could Limit the            not include a component to measure credit risk on liquidity investments
Effectiveness of Its       held by Farmer Mac. Third, FCA’s market risk and income models may
                           understate estimated levels of required risk-based capital. Fourth, lack of
Oversight                  criteria defining Farmer Mac’s mission limits FCA’s ability to effectively
                           oversee Farmer Mac’s mission achievement. Finally, FCA is challenged by
                           regulating both a primary and secondary market.



FCA Has Expanded Its       FCA’s June 2002 annual safety and soundness examination had a larger
Farmer Mac Examination     scope and employed more resources than its past examinations. The
                           examination included a comprehensive review of Farmer Mac’s financial
and Is Taking Actions to
                           condition, portfolio activity, risk management, and a review of board
Improve Oversight          governance and executive compensation. FCA officials said that this
                           CAMELS-based examination would serve as a guide for future Farmer Mac
                           examinations.43 In addition, FCA was closely monitoring Farmer Mac’s
                           corrective actions to address identified weaknesses.



                           43
                              CAMELS refer to six components of a financial institution’s performance – capital
                           adequacy, asset quality, management, earnings, liquidity, and sensitivity to market risk.




                           Page 47                                                              GAO-04-116 Farmer Mac
As part of its increased focus on Farmer Mac oversight, FCA formed a
working group in 2002 that prepared a white paper on Farmer Mac’s
nonmission investments and liquidity requirements. FCA was developing
regulations in this area to ensure that Farmer Mac’s nonmission
investments would be appropriate in both quality and quantity and that
Farmer Mac’s use of its GSE status to issue debt would be appropriate.
According to FCA representatives, to date, FCA has not placed a regulatory
limit on the level or quality of Farmer Mac’s nonmission investments, nor
has it regulated specific liquidity standards.

FCA also formed another working group in 2002 to study the implications
of regulatory capital arbitrage between FCS institutions and Farmer Mac.
The regulatory capital arbitrage working group provided a white paper to
the FCA board that contained an in-depth analysis of the causes and
sources of capital arbitrage. The white paper presented several options for
how FCA could reduce potential safety and soundness issues that might
arise when FCS institutions and Farmer Mac engaged in capital arbitrage
activities to reduce capital required to be held. FCA said that the agency is
still studying the issue and has made no decisions on any specific actions.

Notwithstanding these positive developments, FCA has not been updating
and reformatting Farmer Mac’s call reports, a tool used for off-site
monitoring. Our review of yearend 2001 and 2002 call report schedules and
corresponding instructions indicated that in some cases, they do not fully
conform to FCA regulations nor have these documents been updated to
reflect recent accounting changes. One of the discrepancies we noted was
FCA’s acceptance of Farmer Mac’s inaccurate reporting of the amount of
one of three categories in which Farmer Mac was required to maintain its
minimum core capital. Although to date the amount of capital affected was
very small, this discrepancy raises questions on FCA’s oversight of this part
of Farmer Mac’s capital requirement. FCA officials responded that they do
not believe this discrepancy weakens FCA’s oversight of Farmer Mac’s
capital requirement. Further, FCA officials recognized that the call report
instructions need to be revised and said that they have plans to update
them but that resources were currently not available due to other priorities
associated with their oversight of Farmer Mac. FCA officials said that
outdated call reports were not a primary concern because they augment
the call report information with various other sources, including SEC
filings and risk-based capital supporting data obtained from Farmer Mac.




Page 48                                                  GAO-04-116 Farmer Mac
FCA Faces Challenges as It       FCA has begun to strengthen its oversight of Farmer Mac, but the agency
Enhances Farmer Mac              still faces a number of technical and supervisory challenges. These include
                                 deficiencies in the estimation and measurement of risk and regulatory
Oversight                        management issues.

Limitations Exist in the Model   The model FCA used to estimate the amount of risk-based capital that is
Used to Estimate Farmer Mac’s    required to cover Farmer Mac’s credit risk, utilizes the same data that are
Credit Risk                      used in Farmer Mac’s loan loss estimation model. As we discussed earlier,
                                 this model is limited by the poor data quality. We identified limitations
                                 related to using FCBT data and issues such as not modeling changes in
                                 interest rates, loan terms, or property values. For example, the model uses
                                 FCBT data to estimate loan losses even though Texas did not have the
                                 highest rates of default and severity of agricultural mortgage losses as
                                 required under Farmer Mac’s statutory risk-based levels.44 Legislation on
                                 Farmer Mac’s risk-based capital requirements requires FCA to stress test
                                 the model, based upon the worst experience for defaults and loss severities
                                 for a period of not less than 2 years for agricultural real estate loans in
                                 contiguous areas comprising at least 5 percent of the U.S. population.45
                                 Analysis by FCA’s consultants indicates that Minnesota, Iowa, and Illinois
                                 experienced the greatest decrease in farmland prices in 1983 and 1984.
                                 However, the loans in FCA’s database are limited to Texas, which
                                 experienced the fourth greatest decrease in farmland prices. FCA’s
                                 consultants found that FCBT had the only usable loan database for the
                                 purpose of building the credit risk model to estimate Farmer Mac’s credit
                                 risk. Since the loan data were limited, it may not provide all data elements
                                 that would be desirable in a stress test. For example, the sample was small
                                 and it did not fully reveal the extent of restructuring of loans that could
                                 affect default estimates and losses. Additionally, the FCBT loan files did
                                 not show the extent to which loan terms had been changed to forestall
                                 foreclosures. Consequently, if some of these loans did have losses, which
                                 were not recorded in the database, the frequency of credit losses may be
                                 understated in the credit risk analysis. As we explained earlier, the loans in
                                 Farmer Mac’s current portfolio tend to adjust for changes in interest rates
                                 more quickly than the loans issued by FCBT in the 1970s and 1980s. As


                                 44
                                      12 U.S.C. §2279bb-1.
                                 45
                                   Stress tests are computer simulations that demonstrate how a firm’s financial holdings and
                                 obligations would perform under adverse economic conditions. Generally, stress tests
                                 simulate an economic environment considered to be a worst-case scenario for the type of
                                 business a firm runs.




                                 Page 49                                                             GAO-04-116 Farmer Mac
                                  such, loans in the current portfolio may be exposed to credit risk longer
                                  than were the FCBT loans used in estimating the credit risk model and
                                  therefore, the FCBT loans would not be representative of Farmer Mac’s
                                  current risks.

                                  We also identified limitations in the structure of FCA’s model. One
                                  limitation is that FCA’s credit risk model was constructed so that the
                                  expected losses in a stressed environment are the same no matter what
                                  appreciation or depreciation in farmland prices occurred over the life of
                                  the loan in any period other than the period of maximum stress. The model
                                  also does not consider the effect that interest rate changes may have on the
                                  probability of default, such as the increased default risk of fixed-rate loans
                                  with yield maintenance in times of falling interest rates, or the increased
                                  risk of adjustable rate loans at times of rising interest rates. Another
                                  limitation of this model is that it does not differentiate between loans with
                                  short- and long-amortization periods, although loans with shorter
                                  amortization periods are likely to have lower credit risk, holding other loan
                                  underwriting terms constant. Because these variables are not included in
                                  the credit risk model, by varying its mix of fixed-rate and adjustable-rate
                                  loans, or short- versus long-amortization loans, Farmer Mac could change
                                  its credit risk profile with no resultant change in the regulatory capital for
                                  credit risk as measured by the FCA model. A more detailed discussion of
                                  the limitations of FCA’s credit risk model is presented in appendix VI.

FCA’s Regulation Does Not         FCA’s regulation for calculating Farmer Mac’s risk-based capital does not
Capture Credit Risk on Farmer     assess the amount of capital that must be held against credit risk
Mac’s Liquidity Investments and   associated with assets in Farmer Mac’s liquidity investment portfolio. As
AgVantage Bonds                   such, there is no credit risk capital charge against approximately 37
                                  percent of Farmer Mac’s total balance sheet assets, which consist of
                                  liquidity investments such as commercial paper or corporate bonds.
                                  Although corporations with investment-grade ratings have relatively high
                                  credit quality, there is a possibility that they will default and fail to make all
                                  interest and principal payments in full and on schedule. In contrast, other
                                  financial regulators, including the Federal Housing Finance Board (FHFB),
                                  Federal Reserve Board, Office of the Comptroller of Currency, Office of
                                  Thrift Supervision, and Office of Federal Housing Enterprises Oversight
                                  (OFHEO), calculate the capital that must be held for the credit risk on
                                  investment securities, loans, and other assets, and also capital for the risk
                                  that a counterparty in a derivative transaction would fail to perform.

                                  In addition, FHFB calculates required risk-based capital for advances that
                                  Federal Home Loan Banks make to their members. Farmer Mac’s



                                  Page 50                                                     GAO-04-116 Farmer Mac
                                Agvantage bond program is structured similarly to the FHLB advances, in
                                that both require overcollateralization using borrower mortgage assets as
                                collateral. 46 However, unlike FHFB, FCA does not include AgVantage
                                bonds in its risk-based capital calculation.

Market Risk and Income Models   FCA uses results from Farmer Mac’s interest rate risk model to measure the
May Understate Estimated        level of market risk to which Farmer Mac is exposed and determine
Levels of Required Risk-based   corresponding levels of risk-based capital. As discussed previously, the
Capital                         Farmer Mac interest rate risk model has limitations with regard to
                                prepayment modeling and the effect of prepayment penalties. These
                                limitations could lead to errors in measuring the prepayment risk to which
                                Farmer Mac is exposed and weaken FCA’s oversight of risk-based capital,
                                in addition to affecting Farmer Mac’s risk management.

                                Farmer Mac uses the estimated behavior of single-family residential
                                mortgage benchmarks to estimate the prepayment risk of commercial
                                agricultural mortgages. Using one type of mortgage as a benchmark for
                                another may lead to an underestimate of the extension risk in Farmer Mac’s
                                commercial agricultural mortgage holdings. Extension risk is the tendency
                                for expected lifetimes of a mortgage to lengthen when interest rates rise.
                                Most single-family residential mortgages have due-on-sale clauses, which
                                compel borrowers to pay off their loan balances when selling their
                                property. However, commercial agricultural mortgages are more easily
                                assumed when it is advantageous to do so, often in the form of a “wrap,” in
                                which the property is sold as part of a long-term contract, so that the title to
                                the property does not formally change hands for several years. The result
                                is, at times of rising interest rates, the average life of commercial
                                agricultural mortgages will increase more than will the average life of
                                residential mortgages.

                                Additionally, FCA has chosen to incorporate an estimate of Farmer Mac’s
                                earnings into its income model, that assumes the level of new business
                                activity and profitability for Farmer Mac per year will be unchanged in a 10-
                                year period (steady state approach). In effect, even in the stress test
                                scenario, by holding new business activity level constant, losses can be
                                compensated for with profits from new business. By not including specific
                                instructions on this issue, the 1991 amendments to the 1971 Act
                                establishing Farmer Mac’s risk-based capital standards gave FCA the


                                46
                                   With the AgVantage program Farmer Mac, in effect, purchases bonds from agricultural
                                lenders with the lenders’ using agricultural mortgages as collateral.




                                Page 51                                                           GAO-04-116 Farmer Mac
choice of including or excluding an estimate of Farmer Mac’s earnings over
a 10-year stress period when calculating Farmer Mac’s risk-based capital
requirements.47 FCA officials said that they made a judgment to use the
steady state approach because it allowed them to treat Farmer Mac as a
going concern business, which they interpreted to be the intent of the
statute. Further, FCA officials said that in developing the model, they found
that using a steady state approach resulted in their use of fewer
assumptions than would have been required by other approaches. In
contrast, we have previously reported on the serious problems involved in
estimating future income for GSEs since it is hard to determine what a
reasonable level of activity, profits, or losses would be during a stressful
period.48

Consistent with our concerns, other regulators such as OFHEO and FHFB
do not use an estimate of earnings on new business when calculating their
regulatory capital requirements.49 In addition, OFHEO assumes that as
Fannie Mae and Freddie Mac refinanced their short-term debt to support
outstanding business, they could face higher interest rates caused by
increasing risks of borrowing during a stressful period. Recently, OFHEO
modified its stress test by increasing the short-term rates, which the model
assumes will be paid by the housing enterprises by 10 basis points.50 If FCA
were to make a similar adjustment to future borrowing costs for Farmer
Mac in a stress environment, the effect would be to reduce the estimated
amount of future income earned by Farmer Mac, hence increasing the level
of capital required to be held.




47
   Pub. L. No. 102-237 states that the risk-based capital test must determine the amount of
regulatory capital for Farmer Mac that is sufficient for Farmer Mac to maintain positive
capital during a 10-year period.
48
   U.S. General Accounting Office, OFHEO’s Risk-based Capital Stress Test: Incorporating
New Business Is Not Advisable, GAO-02-521 (Washington, D.C.: June 28, 2002).
49
   12 U.S.C. § 4611 required the Congressional Budget Office and us to study whether OFEHO
should incorporate new business assumptions into the stress test used to establish risk-
based capital requirements for the housing GSEs. The Director of OFHEO may, after
consideration of these studies, assume that the GSE conducts additional new business
during the stress period.
50
     12 C.F.R. 1750.




Page 52                                                              GAO-04-116 Farmer Mac
Lack of Criteria and Procedures      Although FCA has general regulatory authority over Farmer Mac for both
Limit FCA’s Ability to Effectively   safety and soundness oversight and mission regulation, FCA has focused
Oversee Farmer Mac’s Mission         primarily on safety and soundness.51 We recognize that balancing these two
Achievement                          goals—safety and soundness oversight and mission regulation—is difficult
                                     and could create tensions. However, if FCA is to oversee Farmer Mac’s
                                     mission achievement, a lack of criteria and processes to measure how
                                     Farmer Mac’s activities and products have contributed to mission
                                     achievement will limit its effectiveness. As discussed earlier, Farmer Mac’s
                                     enabling legislation does not establish specific mission obligations that
                                     include specific or measurable goals; rather, Farmer Mac’s mission is
                                     broadly stated. FCA officials said that FCA’s authority to establish specific
                                     and measurable goals is fact specific and would depend on the particular
                                     nature of the proposal. Further, unlike the Department of Housing and
                                     Urban Development (HUD), FCA has received no congressional direction
                                     to undertake an analysis to determine the net public policy benefit of
                                     Farmer Mac’s actions.

                                     FCA officials said that the continued combined effect of FCA’s supervisory
                                     efforts and regulatory development plans would bring greater focus on
                                     Farmer Mac’s accomplishment of its public policy purpose. The officials
                                     also said that FCA has taken various steps to indirectly monitor Farmer
                                     Mac’s mission achievement, including looking at Farmer Mac’s book of
                                     business to see how it has grown over time and to identify inappropriate
                                     activities and products.

Overseeing Both FCS Banks and        FCA’s role as regulator of Farmer Mac and the FCS institutions raises a
Farmer Mac Is a Regulatory           concern about regulatory conflict of interest. FCS is a primary market for
Challenge                            agricultural real estate loans, while Farmer Mac is the secondary market
                                     for these loans. We have previously reported that to carry out oversight
                                     responsibilities effectively, a GSE regulatory structure must separate
                                     regulation of primary and secondary market participants.52 This criterion
                                     posits that a regulator overseeing both a GSE and its primary business
                                     partners could be subject to conflicts of interest. For example, if an FCS
                                     institution was in danger of failing, the regulator might be tempted to


                                     51
                                        Under the Farm Credit Act of 1971, FCA has general authority to examine and supervise
                                     the safety and sound performance of the powers, functions, and duties of Farmer Mac and
                                     its affiliates (12 U.S.C.§2279aa-11(a)).
                                     52
                                      U.S. General Accounting Office, Options for Federal Oversight of GSEs, GAO/GGD-91-90
                                     (Washington D.C.: May 22, 1991). Other criteria for effective oversight are independence and
                                     objectivity, prominence, economy and efficiency, and consistency.




                                     Page 53                                                             GAO-04-116 Farmer Mac
                pressure a healthy GSE into increasing the price it pays the bank for loans.
                Or, if a GSE was in poor financial health, the regulator might be tempted to
                encourage the GSE counterparties to discontinue their relationships. On
                the other hand, we recognize that a single regulator could offer some
                benefits such as knowledge of the market and its participants, and the
                opportunity to observe the transactions and trends between the primary
                and secondary markets.

                Congress recognized this potential regulation problem and it attempted to
                mitigate this by creating OSMO, a separate office within FCA to regulate
                Farmer Mac. As required by the 1987 Act, the director of OSMO is selected
                by and reports to the FCA Board.53 Yet, the 1987 Act directs FCA
                examiners, who also examine FCS institutions, to examine Farmer Mac’s
                financial transactions. The 1987 Act also charges FCA with ensuring that
                OSMO is adequately staffed to supervise Farmer Mac’s secondary market
                activities; although, to the extent practicable, the personnel responsible for
                supervising the powers, functions, and duties of the corporation should not
                also be responsible for supervising the banks and associations of the Farm
                Credit System. While this regulatory structure provides for a degree of
                separation between FCA’s responsibilities for FCS institutions and its
                responsibilities with respect to Farmer Mac, in practice, the FCS
                institutions and Farmer Mac are still subject to oversight by the same FCA
                board and reviewed by some of the same FCA examiners and analysts.
                Consequently, FCA could be subject to potential conflicts of interest. In
                our discussions with FCA officials, they said that they were aware of the
                need to maintain the proper balance in their oversight roles to avoid such
                potential conflicts.



Conclusions 	   Government sponsorship of a financial institution, such as Farmer Mac,
                can generate a number of public benefits and costs, which are difficult to
                quantify. To the degree that lower funding costs and other benefits are
                passed on to borrowers in the affected financial sector, public benefits are
                generated. However, government sponsorship also generates potential
                public costs. One potential cost is the risk that taxpayers will be called
                upon if a GSE is unable to meet its financial obligations. In Farmer Mac’s
                case, it would be the need to draw on its $1.5 billion line of credit with
                Treasury and the possibility that the federal government might appropriate


                53
                     12 U.S.C. §2279aa-(11)(a)(3)(C).




                Page 54                                                  GAO-04-116 Farmer Mac
further funds in the event that Farmer Mac faces financial difficulties. GSE
status inherently weakens market discipline, which heightens the
importance of internal and external oversight by the GSE’s board of
directors, auditors, and regulators, as well as transparency through
financial reporting and credit ratings to the creditors and investors.

Farmer Mac’s financial condition has improved since we last reported in
1999; specifically, its income has increased and its capital continues to
exceed required levels. Farmer Mac’s risk profile has become more
complex as a result of the growth in size and complexity of its loan and
guarantee portfolio. Although the company has made progress over the
past few years to enhance its credit controls, asset management, and
reduction of asset liability mismatch, its efforts to measure and monitor its
risks have not kept pace and could be improved. As its loan and guarantee
portfolio ages and delinquencies increase, it is key for Farmer Mac to
continue to manage its credit risk by improving its loan loss estimation
model and documentation of policies, procedures, and management
judgments related to loan purchases and guarantees. More importantly, the
rapid growth of standby agreements has generated a need for Farmer Mac
to consider a funding strategy that would allow it to meet unexpected
demands to fund purchases of underlying impaired or defaulted loans, in
the event of stressful economic conditions. A funding strategy would entail
a comprehensive contingency funding liquidity plan and a detailed analysis
of capital adequacy. As noted, a strategy that consists of selling AMBS to
obtain funding would potentially be limited by the lack of knowledge of the
depth and liquidity of the secondary market for AMBS.

Farmer Mac has increased its mission-related activities since we last
reported on this in 1999, but it is still not apparent if sufficient public
benefits are derived from these activities. The lack of specific or
measurable mission goals in its statute beyond providing a secondary
market and stable long-term financing does not allow for a meaningful
assessment of whether Farmer Mac’s activities are having the desired
impact on the agricultural real estate market. Further, because Farmer
Mac has elected to retain nearly all its AMBS in portfolio for profitability
reasons, the depth and liquidity of the secondary market for AMBS is
unknown.

Similar to other publicly traded companies, Farmer Mac is faced with the
challenges of updating its corporate governance practices to comply with
Sarbanes-Oxley, SEC rules, and proposed NYSE listing standards as they
become effective. As a GSE, Farmer Mac, however, has a board structure



Page 55                                                  GAO-04-116 Farmer Mac
set in statute, which hampers its efforts to comply with the stricter
independence requirements in proposed NYSE listing standards,
specifically those requirements calling for a fully independent and
competent audit committee. Moreover, Farmer Mac’s statutory governance
structure has elements of a cooperative and investor-owned publicly traded
company, which does not reflect the interest of all shareholders. While we
did not draw conclusions on Farmer Mac’s overall executive compensation,
we would no longer consider Farmer Mac a start-up company and the
assumptions used to set its executive compensation may no longer be
valid. Changes are needed to Farmer Mac’s vesting program for stock
options to bring them more in line with general industry practices and
other GSEs. Farmer Mac could improve its training for directors, provide
more transparency to its directors on the nomination process, and better
document its succession plan for its executive management. These
actions, along with obtaining a credit rating to provide transparency to the
market, could also help Farmer Mac respond to criticisms and increased
expectations in today’s market environment.

In addition to Farmer Mac’s internal management of risks, as a GSE, it is
required to have regulatory oversight to ensure that it operates in a safe and
sound manner. Beginning in 2002, FCA had improved its oversight of
Farmer Mac, but continues to face significant challenges in sustaining and
further enhancing its oversight. While FCA has improved its examination
approach, more remains to be done to improve its assessment of risk-based
capital and mission oversight. We discussed a number of issues related to
the data and structure of FCA’s risk-based capital model, but the overall
impact these issues have on the estimate of risk-based capital for Farmer
Mac’s credit risk is uncertain. Some concerns, such as the potential
undercounting of loans that experienced credit losses, or greater
prepayment of FCBT loans relative to Farmer Mac loans, may result in the
FCA credit risk model underestimating the credit risk capital requirement.
Other issues, such as the lack of a variable to track land price changes for
any but the most stressed year, may cause the model to overestimate the
credit risk capital requirement. Augmented data and more analysis could
better determine the relative magnitudes of these effects. While FCA’s
oversight of Farmer Mac typically has focused on safety and soundness, it
lacks criteria and procedures to effectively oversee how well Farmer Mac
achieves its mission. At the same time, Farmer Mac’s enabling legislation is
broadly stated and does not include any measurable goals or requirements
to assess progress toward meeting its mission. More explicit mission goals
or requirements would help FCA in improving its oversight of Farmer Mac.




Page 56                                                  GAO-04-116 Farmer Mac
Recommendations	   To help ensure that Farmer Mac’s management can properly identify,
                   manage, and control risks, we recommend that Farmer Mac management
                   ensure that it has adequate staff resources and technical skills to oversee
                   the following actions:

                   •	 Address weaknesses in its loan loss estimation model, which could
                      affect the reasonableness and adequacy of the loan loss allowance,
                      through the following actions:

                      •	 Include current data on farm loan payment, delinquency, and
                         valuation for the loans included in the estimation model so that the
                         estimation process reflects current loan and economic conditions;

                      •	 Explore other data sources that are relevant to Farmer Mac’s current
                         portfolio for estimating probability, amounts, and distribution of
                         credit losses in its estimation model; and

                      •	 Improve documentation of the results of the model compared to
                         actual portfolio and economic conditions, and of the reconciliation to
                         the amounts recorded in the financial statements.

                   •	 Continue to reduce its credit risk by improving its documentation of
                      policies and procedures, and management’s actions and judgments
                      through the following actions:

                      • Continue to gather documentation supporting management’s
                        assessment of loans approved using underwriting standard 9,
                        including quantification and evaluation of compensating risk factors,
                        and develop a process for utilizing such information in the
                        management decision process for future exceptions and for
                        estimating credit losses, and

                      • Improve documentation supporting and quantifying the effect of
                        extracting specific loan loss estimates from the overall loss estimate
                        to determine whether this approach differs materially from
                        estimating specific loan losses separately.

                   •	 Reevaluate its current strategy of holding agricultural mortgage-backed
                      securities in portfolio and issuing debt to obtain funding.




                   Page 57                                                 GAO-04-116 Farmer Mac
•	 Develop a contingency funding liquidity plan to address potential
   vulnerabilities in less favorable capital markets conditions and liquidity
   needs arising from the rapid growth of standby agreements.

•	 Improve the quality of its prepayment model to ensure accurate interest
   rate risk measurements.

•	 Improve its analysis of capital adequacy to help ensure that capital
   would meet the needs of increasing and potential credit risks and
   growth.

Although the Farmer Mac board has taken steps to strengthen its corporate
governance practices, we recommend that the Chairman, Farmer Mac,
further enhance those practices by

•	 reevaluating stock option levels and vesting period to ensure that they
   are not excessive in relation to comparable industry standards for
   vesting and waiting period for stock options;

•	 better communicating the criteria for identifying and selecting director
   nominees and the process to nominate directors among the directors;

•	 formalizing executive management succession plan and communicate
   plan with all board members to provide transparency; and

•	 providing consistent training on governance and Farmer Mac related
   topics to all board members to increase directors’ understanding of
   risks facing the corporation.

Finally, to improve the quality and effectiveness of FCA’s oversight of
Farmer Mac, we recommend that FCA implement the following steps:

•	 Continue to obtain more relevant and current data on farm loan
   behavior used in the risk-based capital model and consider more
   flexible modeling approaches to credit risk, such as those used by
   OFHEO for regulatory purposes or the Federal Housing Administration
   (FHA) for evaluating actuarial soundness;

•	 Continue to improve and formalize off-site monitoring of Farmer Mac,
   including reviews of Farmer Mac’s regulatory reporting;




Page 58                                                  GAO-04-116 Farmer Mac
                        •	 Create a plan to implement actions currently under consideration to
                           reduce potential safety and soundness issues that may arise from capital
                           arbitrage activities of Farmer Mac and FCS institutions;

                        •	 Examine how other secondary market regulators developed regulations
                           to require the GSEs to obtain a government risk credit ratings from
                           nationally recognized statistical rating agencies; and

                        •	 Assess and report on the impact Farmer Mac’s activities has on the
                           agricultural real estate lending market.



Matters for             Congress may wish to consider the following legislative change:

Congressional           •	 Establish clearer mission goals for Farmer Mac with respect to the
Consideration              agricultural real estate market to allow for a meaningful assessment of
                           whether Farmer Mac had achieved its public policy goals;

                        •	 Allow FCA more flexibility in establishing capital standards that are
                           commensurate with Farmer Mac’s changing risk profile and in setting
                           minimum capital requirements;

                        •	 If Congress intends for Farmer Mac to operate in a cooperative manner
                           and maintain its current board structure of Class A and Class B stock, it
                           may wish to consider making Farmer Mac a true cooperative entity like
                           the Federal Home Loan Bank System, and rescind Farmer Mac’s
                           authority to issue Class C stock. However, if Congress intends for
                           Farmer Mac to operate as a publicly traded company, it should consider
                           amending (1) Farmer Mac’s board structure to ensure an independent
                           board and independent and competent audit committee and (2) the
                           structure of Farmer Mac’s Class C common stock to include a one share,
                           one vote principle to provide the opportunity to better reflect all
                           shareholder interests.



Agency Comments and 	   We requested comments on a draft of this report from the heads or their
                        designees of the FCA, Farmer Mac, SEC, and Treasury. We received written
Our Evaluation	         comments from Farmer Mac and FCA that are summarized below and
                        reprinted in appendixes VII and VIII, respectively. SEC did not provide
                        comments. FCA, Farmer Mac, and Treasury also provided technical
                        comments that we have incorporated as appropriate.



                        Page 59                                                 GAO-04-116 Farmer Mac
In commenting on this report, Farmer Mac stated that it agreed with the
report’s findings and conclusions on Farmer Mac’s risk management
practices and has taken a number of steps toward implementing the
majority of the recommendations. While Farmer Mac seemed to agree with
the report’s recommendations to improve its analysis of capital adequacy,
develop a contingency funding plan, and improve documentation of
management exceptions to its eight major underwriting standards, Farmer
Mac’s comments did not address our recommendations to reevaluate its
strategy of holding AMBS in its portfolio, improve the quality of its
prepayment model, better communicate the criteria for selecting director
nominees and provide consistent training to the board of directors.

Farmer Mac commented that in discussing the availability of the Treasury
line of credit relative to AMBS that Farmer Mac or its affiliates hold, the
report acknowledged that Farmer Mac has a legal opinion by its outside
counsel stating that the Treasury line of credit would be available in those
circumstances; therefore, the question is moot. In fact, the report
discussed the line of credit because Treasury has expressed serious
questions about whether Treasury is required to purchase Farmer Mac
obligations to meet Farmer Mac-guaranteed liabilities on AMBS that are
held by Farmer Mac or its affiliates, and therefore, this issue remains
unresolved until that time when Farmer Mac approaches Treasury for
assistance. Farmer Mac commented that if it were coming under pressure
to fund its guarantee obligations, it could sell AMBS it held to third parties
long before it needed to use the line of credit. As we stated in the report,
however, the depth and liquidity of the demand for these securities in the
current market is unknown. Therefore, Farmer Mac would be selling
AMBS at the same time that it was coming under pressure to fund its
guarantee obligations, which would most likely affect Farmer Mac’s ability
to sell these securities and the price at which it could sell them.

Farmer Mac seems to disagree with our concern on funding liquidity risk
that might arise from standby agreements. Farmer Mac commented that
the report posits a situation in which loan defaults go far beyond the
default rate peak for agricultural loans within the Farm Credit System in
1986. We do not provide an estimate of the level of default rate at which
Farmer Mac would need additional funding. The report stated that if rapid
growth continues, standby agreements could generate substantial funding
liquidity risk under stressful economic conditions. By using the default
rate peak, Farmer Mac is alluding to the stressful conditions incorporated
in the risk-based capital model. However, this model addresses credit risk,
not liquidity risk. Under standby agreements, Farmer Mac would need to



Page 60                                                  GAO-04-116 Farmer Mac
fund not only the net losses from foreclosures that were used in estimating
the risk-based capital requirement but must fund the gross amount of loans
that enter foreclosure and seriously delinquent loans presented for
purchase to Farmer Mac. Other more technical comments provided by
Farmer Mac and our detailed response is discussed in appendix VII.

Finally, FCA overall concurred with our report’s findings and conclusions
that are focused on FCA’s work to oversee the safety and soundness of
Farmer Mac. FCA also agreed to implement the recommendations for
improving FCA’s oversight of Farmer Mac contained in this report through
current regulatory and examination work that is in process, and as
necessary, new initiatives. In response to our recommendation regarding
the risk-based capital model, FCA does not agree that additional data and
modeling would add value, although FCA is studying the possibility of
updating the data used in its model. As we stated in the report, the data
used by FCA do not include all the components of credit losses, may not
capture all the loans that experienced losses, and the loans used in the
model have different interest rate characteristics than those currently
purchased by Farmer Mac. Also as stated in the report, the key
independent variable used in FCA’s model—land price decline—is defined
in such a way that the model will produce a biased estimate of the impact
of land price declines on credit losses. FCA’s technical comments and our
detailed response are discussed in appendix VIII.


We are sending copies of this report to the Chairmen and Ranking Minority
Members of the Senate Committee on Banking, Housing, and Urban Affairs;
the House Committee on Financial Services; and the House Committee on
Agriculture. We are also sending copies of this report to the President and
Chief Executive Officer of Farmer Mac; the Chairman and Chief Executive
Officer of the Farm Credit Administration, the Chairman of SEC, the
Secretary of Treasury, and other interested parties. This report will also be
available at no charge on GAO’s Internet homepage at http://www.gao.gov.




Page 61                                                 GAO-04-116 Farmer Mac
Please contact us at (202) 512-8678 if you or your staff have any questions
concerning this work. Key contributors are ackknowledged in appendix IX.




Davi M. D’Agostino
Director, Financial Markets and
 Community Investment




Jeannette M. Franzel
Director, Financial Management
  and Assurance




Page 62                                                GAO-04-116 Farmer Mac
Appendix I

Objectives, Scope, and Methodology



              As requested by the Senate Committee on Agriculture, Nutrition, and
              Forestry, we conducted a review of the Federal Agricultural Mortgage
              Corporation (Farmer Mac). Our objectives were to (1) assess Farmer Mac’s
              current financial condition and risk management practices, (2) determine
              the extent to which Farmer Mac has achieved its statutory mission, (3)
              evaluate Farmer Mac’s corporate governance as it pertains to board
              structure and oversight and executive compensation, and (4) evaluate the
              Farm Credit Administration’s (FCA) oversight of Farmer Mac.

              The focus of our review on Farmer Mac’s secondary market activity in
              agricultural mortgages was on the Farmer Mac I Program because it is the
              primary program through which Farmer Mac conducts its secondary
              market activity. However, we included Farmer Mac II Program activity in
              our overall analysis of Farmer Mac’s financial condition. To address our
              objectives overall, we reviewed the legislative history and statutory
              authorities governing Farmer Mac. We also reviewed relevant Farmer Mac
              public filings with the Securities and Exchange Commission (SEC) and
              regulatory reporting to the Farm Credit Administration (FCA), FCA
              regulatory reporting instructions, and examined copies of reports from
              Farmer Mac’s regulator, external auditors, internal auditors, and held
              discussions with its external counsel and forensic accountants. Further, we
              held numerous discussions with Farmer Mac management and staff; FCA
              officials and examiners; and interviewed representatives of the American
              Bankers Association, the Farm Credit Council, and former FCA and Farmer
              Mac management.

              To assess Farmer Mac’s financial condition and risk management practices,
              we performed three major steps. First, we reviewed Farmer Mac’s trends
              for earnings, capital, and asset (credit) quality, including return on average
              assets, return on common stockholders’ equity, capital to assets ratio,
              nonperforming loans as a percentage of total loans, and the ratio of
              allowance for loan losses to nonperforming loans. In performing our trend
              analysis and cost of funds analysis, we did not verify the data provided by
              Farmer Mac. In addition, we did not audit Farmer Mac’s financial statement
              or its loan loss allowance balances nor did we review any transactions or
              loan files.

              Second, we determined how Farmer Mac compares to other entities. To do
              so, we identified appropriate measures of rates of return, capital, and asset
              quality for Farmer Mac and comparable entities. Because of its unique role,
              Farmer Mac does not have any direct peers. However, for purposes of our
              analysis, we determined that the following entities had similar



              Page 63                                                  GAO-04-116 Farmer Mac
Appendix I

Objectives, Scope, and Methodology





characteristics that could be compared to Farmer Mac: Federal National
Mortgage Association (Fannie Mae), Agricultural Credit Association (ACA),
Federal Land Credit Association (FLCA), and commercial agriculture
banks.1 While these organizations share some similar characteristics with
Farmer Mac, distinct differences exist between each of these entities and
Farmer Mac. For instance, while Fannie Mae is a government-sponsored
entity (GSE) and publicly traded like Farmer Mac, Fannie Mae deals
primarily with residential housing mortgages, which are less risky than the
agriculture mortgages held by Farmer Mac. Farmer Mac’s agricultural
mortgages are commercial loans that fund a wide variety of agriculture
activity (for example, poultry farms or orange groves), while Fannie Mae’s
single-family mortgages represent a fairly homogeneous asset. As a result,
in the event of foreclosure, farm properties can be harder to appraise and
more difficult to liquidate than single-family residences.

Like Farmer Mac, ACA and FLCA are both Farm Credit System (FCS)
institutions and their business is farm related. However, unlike Farmer
Mac, they originate loans instead of purchasing loans. Also, Farmer Mac is
a publicly traded institution and therefore subject to SEC oversight,
whereas ACA and FLCA are not publicly traded institutions. Also included
in our comparisons are commercial agriculture banks, which are banks
that have a higher proportion of farm loans to total loans than other
commercial banks. Commercial agriculture banks originate a range of
farm-related loans, unlike Farmer Mac, which buys or guarantees only
agricultural mortgage loans, and does not originate loans. 2 Due to the
significant impact of the 1996 Act on Farmer Mac’s operations, we analyzed
Farmer Mac’s financial performance for calendar years 1997 through 2002
and used that same period for our comparison of Farmer Mac’s financial
measures to other entities.

Third, we assessed Farmer Mac’s risk management practices and exposure
to credit, liquidity, interest rate, and legal risks. We (1) obtained Farmer
Mac’s written and oral responses to questions on measurement, analysis,

1
 Federal Home Loan Mortgage Corporation (Freddie Mac) was not included in the analysis
because it was in the process of restating its 2000 to 2002 financial statements. Freddie Mac
was not expected to complete the restatement until November 30, 2003.
2
 For purposes of this study, Commercial Agriculture Banks reflect the combined
performance of banks “that have a proportion of farm loans to total loans that is greater
than the unweighted average at all banks” and were obtained from the fourth quarter 2002
and first quarter 2003 Board of Governors of the Federal Reserve System Agricultural
Finance Databook.




Page 64                                                             GAO-04-116 Farmer Mac
Appendix I

Objectives, Scope, and Methodology





and mitigation of those risks; (2) reviewed Farmer Mac Board-approved
policies and standards related to those risks; (3) reviewed methodologies
for determining loan loss reserves, examined existing studies of loan
performance and research on agricultural loan performance conducted by
contractors working for FCA and Farmer Mac, and interviewed the
contractors; (4) received a demonstration of the model used by Farmer
Mac to measure market risk; (5) analyzed financial data relating to the
liquidity portfolio, outstanding debt, derivatives and total loan portfolio
(on- and off-balance sheet); (6) interviewed representatives from the
investment community; and (7) examined copies of reports from FCA,
external auditors, and internal auditors and held discussions with external
counsel and forensic accountants.

To assess Farmer Mac mission accomplishment, we gained general
background related to agricultural secondary markets and obtained a
regulatory perspective on Farmer Mac activities from meetings with
representatives from the U.S. Department of Agriculture (USDA) Economic
Research Service, FCA’s Office of Secondary Market Oversight (OSMO),
and the Department of the Treasury’s Office of Financial Institutions. To
gain an understanding of the lenders’ perspective on Farmer Mac’s
programs, we interviewed agricultural real estate lenders and banking
associations. We also compared lending institutions’ market share in the
agricultural real estate market with their percentage of participation in
Farmer Mac’s programs. We measured the amount of Farmer Mac’s
secondary activity by analyzing Farmer Mac’s portfolio growth by
identifying growth by product type and the ratio of retained agricultural
mortgage-backed securities (AMBS) to AMBS that are sold to investors. In
addition, we compared average long-term fixed interest rates offered by
Farmer Mac with average rates offered by agricultural real-estate lenders.
To the extent possible, we relied on publicly available data; therefore, there
could be some inconsistencies with some of the characteristics of the data
sets used to compare interest rates.

To evaluate Farmer Mac’s corporate governance practices, we reviewed
Farmer Mac’s enabling legislation to understand the legal authority,
oversight, and structure of Farmer Mac and its Board of Directors. We
analyzed the Sarbanes-Oxley Act of 2002, the recently proposed New York
Stock Exchange (NYSE) listing standards, and spoke with NYSE
representatives to identify the requirements that Farmer Mac would need
to meet. We reviewed relevant GAO reports and other related literature,
and attended relevant seminars to gain a better understanding of corporate
governance best practices. We conducted structured interviews with all 15



Page 65                                                  GAO-04-116 Farmer Mac
Appendix I

Objectives, Scope, and Methodology





members of Farmer Mac’s current Board of Directors to obtain their
perspectives on board governance and communication with management.
Further, we reviewed selected information packages prepared for board
members and board minutes. To evaluate Farmer Mac’s executive
compensation, we obtained, compared, and analyzed two consultant
reports on Farmer Mac’s compensation and stock option and vesting
program policies. We compared Farmer Mac’s executive packages to the
housing GSEs. We also reviewed compensation policies for senior officers.
In addition, we interviewed the corporate governance consultant retained
by Farmer Mac to obtain her views on Farmer Mac’s governance structure
and practices.

To evaluate FCA’s oversight of Farmer Mac, we reviewed examination
scope and reports on Farmer Mac from 1999 through 2002. We reviewed
Farmer Mac year-end 2001 and 2002 call reports and compared the
instructions to the schedules and its legal requirements. We examined a
copy of the spreadsheet model used by FCA to measure Farmer Mac's
credit risk, examined the computer programs and data, which produced
FCA’s credit risk model, and interviewed the FCA contractors who built the
model. Additionally, we examined regulations promulgated by other GSE
regulators, such as Office of Federal Housing Enterprise Oversight
(OFHEO) and the Federal Housing Finance Board, and we met with
officials from OFHEO and the Department of Housing and Urban
Development (HUD) to understand their examination programs.

We conducted our work in California, Indiana, New York, Virginia, and
Washington, D.C., between August 2002 and May 2003 in accordance with
generally accepted government auditing standards.




Page 66                                               GAO-04-116 Farmer Mac
Appendix II

Farmer Mac’s Programs and Products





Program                 Program description                              Product feature
Farmer Mac I    a


Cash Window Program	    Sellers receive cash by selling 100 percent of   Terms and rates are described below under the Full-Time
                        qualifying first mortgage agricultural real      Farm, Part-Time Farm, and AgVantage Programs.
                        estate loans directly to Farmer Mac.
 Full-Time Farm         Designed for borrowers who live on               Types of agricultural loans offered include:
 Program	               agricultural properties and derive a significant
                        portion of their income from farm employment. • 15-year fixed rate, 15-year maturity with 15- or 25-year
                                                                          amortization and partial open prepayment available
                                                                          (annual, semiannual, or monthly payments);

                                                                         • 10-year fixed rate, 10-year maturity fully amortizing
                                                                           (semiannual or monthly payments);

                                                                         • 5-year reset loan with a 5-year term (renewable twice); 5-,
                                                                           10-, 15-, or 25-year amortization (annual, semiannual, or
                                                                           monthly payments);

                                                                         • 30-day, 1-, 3- and 5-year ARMs (convertible to long-term,
                                                                           fixed rate), 15-year maturity, 15- or 25-year amortization,
                                                                           (semiannual or monthly payments); and

                                                                         • facility loans,10- or 15-year fixed rate maturity, and fully
                                                                           amortized.
 Part-Time Farm         Designed for borrowers who live on               Farmer Mac offers a 15- and 30-year loan for single-family,
 Program                agricultural properties with a valuable          detached residences; 3/1, 5/1, 7/1 and 10/1 ARMs and 15-
                        residence and derive a significant portion of    and 30-year fixed rate mortgages (monthly payments).
                        their income from off-farm employment.
AgVantage Program	      Farmer Mac purchases and guarantees timely AgVantage bonds may range in maturity from short-term to
                        payment of principle and interest on       15 years and have low fixed or variable rates of interest.
                        mortgage-backed bonds.
Swap Program	           Farmer Mac acquires eligible loans from      Security terms, rates, etc., are negotiated with the seller on
                        sellers in exchange for Farmer Mac           the basis of the characteristics of the loan.
                        Guaranteed Securities backed by those loans.
Long-Term Standby       Farmer Mac commits to purchase loans from        Terms are negotiated with institution based on the
Purchase Commitments	   a segregated pool of loans on one or more        characteristics of the underlying loan.
                        undetermined future dates.




                                            Page 67                                                                GAO-04-116 Farmer Mac
                                                   Appendix II

                                                   Farmer Mac’s Programs and Products





(Continued From Previous Page)
Program                        Program description                                Product feature
Farmer Mac II b
Cash Window Program	           Lenders receive cash by selling 100 percent of • 7-year fixed rate and 15-year fixed rate based on full
                               the guaranteed portion of USDA loans directly    amortization;
                               to Farmer Mac.
                                                                              • 5- or 10-year fixed rate based on full amortization with 5- or
                                                                                10-year rate reset periods—which are tied to the Farmer
                                                                                Mac 5- or 10-year Reset Cost of Funds Index Net Yield; and

                                                                                  • floating rate is tied to Farmer Mac 3-month Cost of Funds
                                                                                    Index’s “Net Yield” with calendar quarter rate adjustments
                                                                                    or The Wall Street Journal’s Prime Rate.
Swap Program	                  Lenders receive Farmer Mac-guaranteed              Security terms, rates, etc., are negotiated with the seller on
                               securities in return for the guaranteed portion    the basis of the characteristics of the loan.
                               of USDA loans.
Sources: Farmer Mac and FCA.
                                                   a
                                                    Farmer Mac I operates as a secondary mortgage market for high-quality agricultural real estate and
                                                   rural home mortgages. Participation is limited to financially healthy farmers as established in the
                                                   Agricultural Credit Act of 1987.
                                                   b
                                                    In the 1990 Act, Farmer Mac was authorized to serve as the pooler for secondary sales of agricultural
                                                   and rural development loans that are guaranteed by USDA. This program benefits borrowers who are
                                                   unable to get commercial credit at affordable rates because of financial problems.




                                                   Page 68                                                                    GAO-04-116 Farmer Mac
Appendix III

Financial Trends and Comparisons with Other
Entities

               Farmer Mac’s increase in impaired loans and in write offs of bad loans is
               indicative of increasing credit risk. Farmer Mac’s percentage of impaired
               loans1 to total outstanding post-1996 Act loans, AMBS, and standby
               agreements increased each year from 1997 through 2001, and then
               decreased slightly, by 14 basis points2 from 1.70 percent at December 31,
               2001, to 1.56 percent at December 31, 2002.3 (See fig. 6) On a comparative
               basis, the proportion of Farmer Mac’s nonperforming loans to total loans is
               higher than other comparable entities. For instance, Agricultural Credit
               Associations’ (ACA) and Federal Land Credit Associations’ (FLCA)
               nonperforming loans to total loans at December 31, 2002, were .89 percent
               and .57 percent, respectively. See fig. 7.




               1
                Post-1996 Act loans and guarantees are Farmer Mac I loans and guarantees that Farmer
               Mac acquired or guaranteed after the 1996 Farm Act was passed.
               2
                A basis point is equal to one hundredth of a percent. It is used to measure changes in or
               differences between yields or interest rates.
               3
                Impaired loans are analyzed on a loan-by-loan basis to measure impairment on the current
               value of the collateral for each loan relative to the total amount due from the borrower.
               Farmer Mac specifically determines an allowance for the loan for the difference between
               the recorded amount due and its current collateral value, less estimated costs to liquidate
               the collateral.




               Page 69                                                             GAO-04-116 Farmer Mac
Appendix III

Financial Trends and Comparisons with 

Other Entities





Figure 6: Farmer Mac’s Impaired Loans from 1997 to 2002 

Dollars in millions                                         Percent

80                                                               2.0


70



60                                                               1.5


50



40                                                               1.0


30



20                                                               0.5


10



 0                                                               0.0

       1997       1998         1999        2000   2001   2002

      Year

               Impaired loans to total loans

               Total impaired loans

Source: GAO's analysis of Farmer Mac's data.




Page 70                                                                 GAO-04-116 Farmer Mac
Appendix III

Financial Trends and Comparisons with 

Other Entities





Figure 7: Farmer Mac’s Nonperforming to Total Loans Compared to Other Entities,

as of December 31, 2002

Percent


2.0




1.5




1.0




0.5




0.0
      Farmer       ACA        FLCA
       Mac
Source: GAO's analysis of Farmer Mac and FCA data.



Farmer Mac’s write offs of impaired loans have been limited to date, but
delinquencies are increasing. During 2002, Farmer Mac wrote off $4.1
million of bad loans, or 8 basis points of post-1996 Act loans and
guarantees,4 which was a significant increase over the $2.2 million, or 6
basis points, written off in 2001.




4
 Loans written off are losses on the outstanding balance of the loan, any interest payments
previously accrued or advanced, and expected collateral liquidation costs. The post-1996
Act loans and guarantees are post-1996 Act loans held and loans underlying the guaranteed
securities and standby agreement, which represent the credit risk on loans and guarantees
assumed by Farmer Mac.




Page 71                                                            GAO-04-116 Farmer Mac
                        Appendix III

                        Financial Trends and Comparisons with 

                        Other Entities





Revenue Has             Farmer Mac’s net interest income grew from $7.1 million in 1997 to $35.0
                        million in 2002. Net interest income is interest income generated from
Increased, but Some     Farmer Mac Guaranteed Securities, loans, and investments, less interest
Financial Performance   expense, which Farmer Mac pays on its debt. Interest rates Farmer Mac
                        earned on Farmer Mac Guaranteed Securities and loan products declined
Indicators Lag          177 basis points from 7.41 percent in 1997 to 5.64 percent in 2002. During
Comparative Entities    the same period, the weighted average interest rates that Farmer Mac paid
                        on its debt decreased 216 basis point from 5.75 percent to 3.59 percent.
                        The growth in Farmer Mac Guaranteed Securities and loans from $442
                        million and $47 million at year-end 1997 to $1.6 billion and $966 million at
                        year-end 2002, respectively, caused Farmer Mac’s interest income to
                        increase. See fig. 8.



                        Figure 8: Income by Program Assets 

                        Dollars in millions


                        250





                        200




                        150




                        100




                         50




                          0
                                1997        1998        1999           2000   2001   2002
                               Year


                                         Commitment and guarantee fees


                                         Loan interest income


                                         Farmer Mac guarantee securities

                                         Investment interest income


                        Source: GAO's analysis of Farmer Mac's data.




                        Page 72                                                             GAO-04-116 Farmer Mac
Appendix III

Financial Trends and Comparisons with 

Other Entities





Farmer Mac’s return on assets (ROA) generally increased between 1997
and 2002, but continued to lag behind other comparative entities. During
this period, Farmer Mac’s performance as measured by percentage return
on average assets fluctuated from a low of .31 percent in 1999 to a high of
.60 percent in 2002. The increase in 2002 was driven by continued growth
in the off-balance sheet standby agreement product, which experienced 42
percent growth in 2002 and 118 percent growth in 2001. As previously
mentioned, Farmer Mac earns and recognizes income from the standby
agreements as commitment fees. The standby growth caused Farmer
Mac’s net income growth rate between 2001 and 2002 to exceed its average
asset growth rate.

During the period 1997 to 2002, Farmer Mac’s ROA was consistently lower
than the ROA of the following comparative banking institutions: Fannie
Mae (except in 2002), commercial agriculture banks, ACA, and FLCA. This
indicates that Farmer Mac is using its assets differently than comparative
banking entities. For instance, of its total assets, Farmer Mac had 17.1
percent in cash and 19.7 percent in investments at December 31, 2002,
while Fannie Mae had 0.2 percent in cash and 6.7 percent in investments.
ACA and FLCA held even lower portions of their assets as cash and
investments. See fig. 9.




Page 73                                                GAO-04-116 Farmer Mac
Appendix III

Financial Trends and Comparisons with 

Other Entities





Figure 9: Farmer Mac’s ROA Compared to Other Entities
Percent

2.5




2.0




1.5




1.0




0.5




0.0
      1997                1998                 1999                 2000           2001         2002
      Year

                ACA

                Commercial agriculture banks
                Fannie Mae

                Farmer Mac

                FLCA
Source: GAO's analysis of Farmer Mac, FCA, Federal Reserve, and Fannie Mae data.




Farmer Mac’s return on average common stockholder equity (ROE) of
15.04 percent for 2002 increased steadily from 7.57 percent in 1997.
Between 1997 and 2002, Farmer Mac’s ROE remained well below Fannie
Mae’s ROE, which was 30.2 percent for 2002. However, for 2002, Farmer
Mac’s ROE exceeded the comparative banking institutions of commercial
agriculture banks, ACA, and FLCA. One reason for the difference is that
Farmer Mac’s capital as a percentage of total assets is less than that of the
comparative banking institutions, but greater than Fannie Mae’s capital
ratio. See fig. 10.




Page 74                                                                              GAO-04-116 Farmer Mac
Appendix III

Financial Trends and Comparisons with 

Other Entities





Figure 10: Farmer Mac’s ROE Compared to Other Entities
Percent

40


35


30


25


20


15


10


 5
     1997                 1998                 1999                 2000           2001         2002
     Year

               ACA

               Commercial agriculture banks
               Fannie Mae

               Farmer Mac

               FLCA
Source: GAO's analysis of Farmer Mac, FCA, Federal Reserve, and Fannie Mae data.



Farmer Mac’s total capital (stockholder equity) to total assets of 4.35
percent as of December 31, 2002, is significantly below ACA’s and FLCA’s
ratios of 15.81 percent and 16.46 percent, respectively, but above Fannie
Mae’s ratio of 1.84 percent. See fig. 11. Capital’s primary function is to
support the institution’s operations, act as a cushion to absorb
unanticipated losses and declines in asset values that could otherwise
cause an institution to fail, and provide protection to debt holders in the
event of liquidation. A higher capital to assets ratio, such as ACA’s and
FLCA’s compared to Farmer Mac’s, indicates there is more coverage for
potential financial losses. Because Fannie Mae’s housing loans have
different risks than agriculture loans, it is expected that its capital would be
lower than Farmer Mac’s, ACA’s, and FLCA’s. In general, since 1997, Farmer
Mac has operated in economic times when agriculture land values have



Page 75                                                                              GAO-04-116 Farmer Mac
Appendix III

Financial Trends and Comparisons with 

Other Entities





been rising and interest rates have been relatively low, experienced
minimal credit losses, and has not experienced net income losses, so its
capital has not been stressed and therefore has not demonstrated whether
it can absorb unanticipated losses and declines in asset values.



Figure 11: Farmer Mac’s Capital to Asset Ratios Compared to Other Entities
Percent

20




15




10




 5




 0
     1997                 1998                 1999             2000   2001         2002
     Year

               ACA

               Farmer Mac

               FLCA

               Fannie Mae
Source: GAO analysis of Farmer Mac, FCA, and Fannie Mae data.




As of December 31, 2002, Farmer Mac’s capital was in excess of its
statutory requirements. According to the 1991 Amendment to the
Agricultural Credit Act of 1987 and the 1996 Act, Farmer Mac has the
following capital requirements:

•	 Minimum required capital level is an amount of core capital equal to the
   sum of 2.75 percent of Farmer Mac’s aggregate on-balance sheet assets,
   as calculated in accordance with generally accepted accounting



Page 76                                                                GAO-04-116 Farmer Mac
Appendix III

Financial Trends and Comparisons with 

Other Entities





   principles (GAAP), plus .75 percent of the aggregate off-balance sheet
   obligations of Farmer Mac, specifically including the unpaid principal
   balance of outstanding Farmer Mac AMBS, instruments issued or
   guaranteed by Farmer Mac, and other off-balance sheet obligations.

•	 Critical capital level is an amount of core capital equal to 50 percent of
   the total minimum capital requirement at that time.

•	 Core capital is the sum of par value of common and preferred stock plus
   paid-in capital and retained earnings, determined in accordance with
   GAAP.




Page 77                                                  GAO-04-116 Farmer Mac
Appendix IV

Farmer Mac’s Underwriting Standards



              Underwriting standards are used by Farmer Mac to determine which
              mortgages it will buy, and then choose to either hold in its own portfolios of
              loans, place into mortgage pools to be sold to investors, or place under
              standby agreements. Generally, eligible loans must meet each of the
              underwriting standards. The standards are meant to limit the risk that the
              mortgages will create losses for Farmer Mac or the holders of mortgage
              pools by ensuring that the borrower has the ability to pay, is creditworthy,
              and is likely to meet scheduled payments and, in the event of the default,
              the value of the agriculture real estate limits any losses. Standards are
              tailored to loans depending upon whether the loan is newly originated or
              seasoned, based upon full- or part-time agricultural production, or for
              specialized facilities. Farmer Mac requires lenders originating and selling
              the loans to (1) ensure that loan documentation in each loan file
              conclusively supports determination of each standard and (2) provide
              representations and warranties to help ensure that the qualified loans
              conform to these standards and other requirements of Farmer Mac.

              Farmer Mac has nine underwriting standards for newly originated loans
              that are based on credit ratios such as debt-to-assets, other quantitative
              measures such as loan-to-appraised value (LTV), and qualitative terms such
              as receipts supporting agricultural use of the property. These standards are
              a chapter in Farmer Mac’s Seller/Servicer Guide, and provide guidelines to
              its staff and lenders, supported with detailed examples and explanatory
              comments for each standard. A summary of each of these nine standards is
              condensed below.

              •	 Standard 1: Creditworthiness of Borrowers. A complete and current
                 credit report must be obtained for each applicant and guarantor that
                 includes historical experience, identification of all debts, and other
                 pertinent information.

              •	 Standard 2: Balance Sheet and Income Statements. This standard
                 requires the loan applicant to provide fair market value balance sheets
                 and income statements for at least the last 3 years.

              •	 Standard 3: Debt-to-Asset (or Leverage) Ratio. The entity being
                 financed should have a pro forma debt-to-asset ratio of 50 percent or
                 less on a market value basis. The debt-to-asset ratio is calculated by
                 dividing pro forma liabilities by pro forma assets. A pro forma ratio
                 shows the impact of the amount borrowed on assets and liabilities.




              Page 78                                                  GAO-04-116 Farmer Mac
Appendix IV

Farmer Mac’s Underwriting Standards





•	 Standard 4: Liquidity and Earnings. The entity being financed should
   be able to generate sufficient liquidity and net earnings, after family
   living expenses and taxes, to meet all debt obligations as they come due
   over the term of the loan and provide a reasonable margin for capital
   replacement and contingencies. This standard is achieved by having a
   pro forma current ratio of not less than 1.0 and a pro forma total debt
   service ratio of not less than 1.25, after living expenses and taxes. The
   current ratio is calculated by dividing pro forma current assets by pro
   forma liabilities. Total debt service coverage ratio is calculated by
   dividing net operating income by annual debt service. Net income from
   farm and nonfarm sources may be included.

•	 Standard 5: Loan-to-Value (LTV) Ratio. The LTV should not exceed 70
   percent in the case of a typical Farmer Mac loan secured by agricultural
   real estate, 75 percent in the case of qualified facility loans, 60 percent
   for loans greater than $2.8 million, or 85 percent in the case of part-time
   farm loans with private mortgage insurance coverage required for
   amounts above 70 percent. The LTV ratio is important in determining the
   probability of default and the magnitude of loss.

•	 Standard 6: Minimum Acreage and Annual Receipts Requirement.
   Agricultural real estate must consist of at least 5 acres or be used to
   produce annual receipts of at least $5,000 to be eligible to secure a
   qualified loan.

•	 Standard 7: Loan Conditions. The loan (1) must be at a fixed payment
   level and either fully amortize the principal over a term not to exceed 30
   years or amortize the principal according to a schedule not to exceed 30
   years and (2) mature no earlier than the time at which the remaining
   principal balance (i.e., balloon payment) of the loan equals 50 percent of
   the original appraised value of the property securing the loan. The
   amortization is expected to match the useful life of the mortgaged asset
   and payments should match the earnings cycle of the farm operations.
   For facilities, the amortization schedule should not extend beyond the
   useful agricultural economic life of the facility.

•	 Standard 8: Rural Housing Loans Standards. Farmer Mac has
   adopted the credit underwriting standards applicable to Fannie Mae,
   adjusted to reflect the usual and customary characteristics of rural
   housing. These standards include, among other things, allowing loans
   secured by properties that are subject to unusual easements, having
   larger sites than those for normal residential properties in the area, and



Page 79                                                  GAO-04-116 Farmer Mac
Appendix IV

Farmer Mac’s Underwriting Standards





   having property that is located in areas that are less than 25 percent
   developed.

•	 Standard 9: Nonconforming Loans. On a loan-by-loan determination,
   Farmer Mac may decide to accept loans that do not conform to one or
   more of the underwriting standards or conditions, with the exception of
   standard 5. Farmer Mac may accept those loans that have compensating
   strengths that outweigh their inability to meet all of the standards.
   Examples of compensating strengths include substantial borrower net
   worth or a larger borrower down payment. The granting of standard 9
   exceptions is not intended to provide a basis for waiving or lessening in
   any way Farmer Mac’s focus on buying only high-quality loans.




Page 80                                                 GAO-04-116 Farmer Mac
Appendix V

Interest Rate Risk




Asset-Liability   As of December 31, 2002, over 70 percent of Farmer Mac’s liabilities ($2.9
                  billion) were short-term—maturing in 1 year or less—while most of the
Management        assets it held were agricultural real estate mortgages, which can have
                  maturities of up to 30 years. As most of these longer-term assets are either
                  fixed-interest rate loans or loans with adjustable rates that will adjust more
                  than 1 year in the future, this would result in an asset liability mismatch,
                  which would occur when assets and liabilities do not have the same
                  maturity or interest rate characteristics. Farmer Mac’s use of interest rate
                  swaps substantially reduces this problem. In addition, Farmer Mac uses
                  callable debt to mitigate the risk from prepayable mortgages.

                  Farmer Mac is subject to interest rate risk on its portfolio due to the
                  potential timing differences in the cash-flow patterns of its assets and
                  liabilities. Farmer Mac uses callable debt, derivatives and yield-
                  maintenance terms in its loan contracts to mitigate interest rate risk
                  (IRR).1 Financial institutions often match the cash flow and duration of
                  newly acquired assets with liabilities of equal cash flow and duration. In
                  order to achieve an overall lower cost of funding for the assets it
                  purchases, Farmer Mac relies on short-term discount notes as its primary
                  source of funding. However, since funding longer-term assets with short-
                  term liabilities causes an asset-liability mismatch, Farmer Mac enters into
                  derivative contracts to convert the short-term discount notes into longer-
                  term liabilities, which more closely match the duration of the assets. The
                  majority of Farmer Mac’s interest rate contracts are floating to fixed-
                  interest rate swaps, in which Farmer Mac pays fixed rates of interest to,
                  and receives floating rates from, the derivative counterparty. If interest
                  rates were to rise, Farmer Mac would have to pay higher rates when its
                  discount notes matured and had to be reissued, but the interest it receives
                  from the swaps would also rise, compensating Farmer Mac for the
                  increased funding cost. Farmer Mac also enters into basis swaps in which
                  it pays variable rates of interest based on its discount notes, and receives
                  variable rates of interest based on another index, such as LIBOR.2 Farmer
                  Mac also has prepayment penalties or yield-maintenance terms on 57


                  1
                   Interest rate risk is the potential that changes in prevailing interest rates will adversely
                  affect assets, liabilities, capital, income, or expenses at different times in different amounts.
                  2
                   London Interbank Offered Rate (LIBOR) is the rate that the most creditworthy
                  international banks dealing in Eurodollars charge each other for large loans. The LIBOR rate
                  is usually the base for other large Eurodollar loans to less creditworthy corporate and
                  government borrowers.




                  Page 81                                                                GAO-04-116 Farmer Mac
                    Appendix V
                    Interest Rate Risk




                    percent of its outstanding balance of loans and guarantees (including 91
                    percent of loans with fixed-interest rates), which limits Farmer Mac’s
                    exposure to losses stemming from declines in interest rates. Prepayment
                    penalties and yield-maintenance agreements reduce the borrower’s
                    incentive to refinance into a lower interest rate loan when interest rates
                    drop, and produce additional revenue for the owner of the mortgage if it is
                    refinanced at a time of falling interest rates.



Prepayment Model	   Prepayment models are an important component of interest-rate risk
                    measurement. Approximately 57 percent of Farmer Mac’s loan portfolio
                    has some form of yield-maintenance protection, which mitigates the effects
                    of loan prepayments. The fixed-rate loans that do not have yield
                    maintenance expose Farmer Mac to prepayment risk. This is particularly
                    true for the purchases of large portfolios of loans (bulk purchases) that
                    include loans with characteristics different from the rest of the portfolio.
                    For fixed-rate loans without yield-maintenance agreements, falling interest
                    rates result in a loss for the financial institution if the mortgage is paid off
                    early, as the owner of the mortgage can only reinvest the funds at a lower
                    interest rate if the mortgage is paid off early. For fixed-rate loans with
                    yield-maintenance agreements, falling rates may result in a gain for the
                    financial institution, as any loans that do pay off early will pay a penalty
                    that generally compensates the lender for the lower interest rate received
                    on the reinvested funds.3 Prepayment models predict the number and
                    timing of early payments, hence, the losses or gains that may result from
                    changes in interest rates.

                    Farmer Mac’s prepayment risk model was developed internally based on
                    models that predict prepayment behavior for residential mortgage
                    borrowers. Farmer Mac followed this approach due to the unavailability of
                    external data on agricultural mortgage prepayments. But agricultural real-
                    estate borrowers may behave differently than residential mortgage


                    3
                     Yield maintenance is designed to compensate lenders for loss in market value when loans
                    are paid off early in falling rate environments. The yield-maintenance penalty formula tends
                    to slightly overcompensate lenders for early repayment because the formula does not
                    consider the effect of amortization, and the formula uses the gross spread between the
                    interest rate on the mortgage (net of servicing fees) and a Treasury security of comparable
                    maturity, although some of that spread represents the higher cost of agency debt, and not
                    the net interest margin on the loan. Because yield maintenance is not collected for the last
                    six months of a loan’s life, it may less than fully compensate the lender when a loan is paid
                    off near its maturity date.




                    Page 82                                                             GAO-04-116 Farmer Mac
Appendix V
Interest Rate Risk




borrowers for many reasons. First, Farmer Mac’s fixed-rate agricultural
real-estate loans often have prepayment penalties or yield-maintenance
agreements, which are rare for single-family residential borrowers.
Therefore, at times of falling interest rates, single-family mortgages will
experience waves of refinancing induced prepayment which will be absent
for many types of agricultural mortgage. In addition, single-family
borrowers are influenced by price appreciation on single-family housing,
and agricultural real estate may have significantly different patterns of
price appreciation. Single-family prepayments are also determined in part
by mobility and the sale of owner-occupied housing, and agricultural real
estate may show different patterns of sale-induced prepayment over time.
Farmer Mac makes substantial downward revisions to prepayment speeds
for loans with penalties or yield maintenance, but these adjustments are
not based on a model of borrower behavior. Rather, they are based on
long-run historical averages for prepayments on similar loan types. For
loans that allow open prepayment, Farmer Mac uses a multiplicative
adjustment factor applied to the prepayment speeds of single-family
mortgages. These revisions to prepayment speeds more closely align the
prepayment behavior of single-family mortgages with the loans held or
securitized by Farmer Mac. The adjustment factors are backtested over
several previous quarters to ensure that they fit the recent past and are
revised from time to time. However, because single-family prepayment
rates fluctuate, sometimes substantially, for different reasons than do
prepayment rates on agricultural mortgages, a simple proportional
adjustment factor may be insufficient to capture the differences in
prepayment behavior. For example, if agricultural real-estate prices were
flat or falling while single family homes were appreciating rapidly, single-
family prepayments may rise without a corresponding increase in
agricultural prepayment rates, or vice versa. If the relative rate of
agricultural mortgage prepayments to single-family mortgage prepayments
were different for prepayments caused by property sales (which
predominate at times of flat interest rates) than for prepayments caused by
refinancing, a proportionate adjustment factor calculated at a time of flat
interest rates would not provide a good forecast of agricultural mortgage
behavior when rates are falling.

Loans with prepayment penalties are likely to experience higher default
probabilities at times of falling interest rates. Yield-maintenance penalties
have the effect of increasing the loan’s payoff amount in a falling interest
rate environment. This has an effect similar to an increase in the LTV ratio,




Page 83                                                 GAO-04-116 Farmer Mac
                      Appendix V
                      Interest Rate Risk




                      a prime determinant of default in studies of borrower behavior.4 As a
                      concrete example of this effect, consider a $700,000 loan on a $1,000,000
                      property. If the agricultural market is stressed, and the value of the farm
                      falls to $800,000, a borrower may consider selling the property and using
                      the proceeds to pay off the loan. If interest rates have fallen; however, and
                      the loan payoff additionally includes a $150,000 prepayment penalty, the
                      borrower would be unable to pay off the loan with the proceeds from the
                      sale of the property and would therefore be more likely to default or to
                      negotiate a costly restructuring. Farmer Mac’s IRR model assumes that
                      default behavior does not change when interest rates change, hence does
                      not model an increased probability of failing to collect yield maintenance
                      or prepayment penalties in times of falling rates.



Farmer Mac’s IRR      On a monthly basis, or more frequently if necessary, Farmer Mac measures
                      its IRR using an industry standard package, Quantitative Risk Management
Measurement Process   (QRM).5 The primary IRR metric that is reported to the Farmer Mac board
                      of directors is MVE-at-risk. Farmer Mac calculates MVE by first obtaining
                      the market prices of Farmer Mac’s assets, liabilities, and off-balance sheet
                      obligations. Then Farmer Mac uses QRM to calculate the sensitivity of
                      MVE to parallel changes of the Treasury yield curve of plus and minus 100,
                      200, and 300 basis points.6 In addition, on a quarterly basis, Farmer Mac
                      management analyzes the effect that changes in interest rates have on the
                      financial value of Farmer Mac. Farmer Mac management also managed NII
                      in a similar fashion as MVE. Finally, Farmer Mac also measures the


                      4
                       Numerous studies of the performance of commercial mortgage behavior incorporate this
                      effect. It is generally done by using the market, as opposed to the book value of the
                      mortgage when calculating loan-to-value ratios. The market value is calculated by taking
                      the stream of payments of the mortgage, discounted at the currently prevailing interest rate.
                      The market value of a mortgage rises when interest rates fall, in line with the yield-
                      maintenance payment. Market value and yield maintenance are two different approaches to
                      calculating the same concept, namely, the value of the mortgage to investors. Some
                      examples of papers that use market value of the mortgage as a predictor of loan default
                      include Vandell, Barnes, Hartzell, Kraft, and Wendt (1993) “Commercial Mortgage Defaults:
                      Proportional Hazards Estimation Using Individual Loan Histories,” American Real Estate
                      and Urban Economics Association Journal, V 21 Number 4, pp. 451-480, or Huang, and
                      Ondrich (2002) “Pay, Stay, or Walk Away: A Hazard Rate Analysis of FHA Multifamily
                      Mortgages,” Journal of Housing Research, V 13 Number 1, pp. 85-117.
                      5
                      QRM is a commercial software used to manage IRR.
                      6
                       Yield curve is a graph showing the relationship between the yield on bonds of the same
                      credit quality but different maturities.




                      Page 84                                                             GAO-04-116 Farmer Mac
Appendix V
Interest Rate Risk




duration gap of its assets, liabilities, and off-balance sheet obligations.
Other sensitivity analyses are done on a regular basis, such as examining
the effects of changes in the prepayment speed assumptions for mortgages
underlying the AMBS.7




7
    Prepayment speed is the rate at which mortgages pay off before their scheduled maturity.




Page 85                                                              GAO-04-116 Farmer Mac
Appendix VI

Farm Credit Administration Credit Risk
Model

               FCA measures the credit risk component of Farmer Mac’s risk-based
               capital requirement with a statistical model that relates loan
               characteristics, such as the loan-to-value (LTV) ratio, and changes in
               agricultural real estate prices, to credit losses on loans secured by
               agricultural real estate. The estimated relationship between credit losses
               and the prediction variables is used to forecast the losses expected on
               agricultural real estate mortgages under a severe stress scenario, such as
               that experienced in Minnesota, Iowa, and Illinois in 1983 and 1984.

               The data used to estimate the credit loss model consist of loans from the
               Farm Credit Bank of Texas (FCBT) observed over the period 1979 to 1992.
               This data source was identified by FCA’s consultants who found that FCBT
               had the most reliable loan data for agricultural mortgage losses for the
               purpose of building the credit risk model to estimate Farmer Mac’s credit
               risk. The data include several important underwriting variables: the LTV
               ratio, the ratio of the borrower’s debt to the borrower’s assets, and the ratio
               of the borrower’s debt payments to farm income. The data also contain the
               dollar amount of the loan, the year in which the loan was written, the year
               in which the loan was foreclosed (for those loans that completed
               foreclosure), and the amount that was lost on the foreclosed loan. The data
               files used by the contractors did not contain information on other key
               variables, such as the amortization period of the loan, the interest rate on
               the loan, or an indicator of whether the loan was paid off early.

               The model consists of three equations, estimated sequentially. In the first
               equation, the loss frequency equation, the probability that a loan will
               experience a credit loss at any point over its life is predicted by three
               underwriting variables—the LTV ratio, the debt-to-asset ratio, the debt
               payment to farm income ratio—the dollar amount of the loan (in inflation
               adjusted dollars), and the maximum percentage decline in farmland value
               experienced over the life of the loan. Logistic regression is used to model
               the probability of a credit loss. Several of the explanatory variables are
               modified for use in the regression. The LTV ratio is raised to a power, the
               dollar amount of the loan is modified with an exponential function, and the
               decline in farmland value is adjusted downward with a multiplicative factor
               that varies with the age of the loan. The second equation multiplies the loss
               frequency by a loss severity, assumed to be a constant 20.9 percent. The
               final equation uses a beta function to distribute the product of loss
               frequency and loss severity over time, so that the losses expected over the
               remaining lives of the loans may be isolated.




               Page 86                                                   GAO-04-116 Farmer Mac
                    Appendix VI
                    Farm Credit Administration Credit Risk
                    Model




Data Limitations	   FCA’s contractors and we have identified several shortcomings in the data
                    used to estimate the credit risk model, including: (1) data have not been
                    updated with post-1992 loan information; (2) data may not have captured
                    all the credit losses experienced by the FCBT; (3) the data set consists
                    entirely of loans in Texas; and (4) the data set does not contain information
                    on prepayments.1 These shortcomings were noted by FCA’s contractors in
                    the Federal Register (Final Rule) document that presents the credit risk
                    model. FCA’s contractors told us that despite these flaws they believe this
                    data set represents the best data available for estimating a credit risk model
                    in a stressed time period.

                    We have identified other data shortcomings, which were not indicated in
                    the Federal Register risk-based capital document. These include: (1) the
                    FCBT data systems did not record all the components of loss on foreclosed
                    loans; (2) the loans made by the FCBT from 1979 to 1992 had very different
                    interest rate terms than the most common loans bought by Farmer Mac;
                    and (3) the data set does not include other important predictors of credit
                    loss, such as interest rate or amortization terms. These shortcomings limit
                    the ability of the credit risk model to forecast the credit risk on loans held
                    by Farmer Mac.

                    Restricting the data set to 1979 through 1992 creates the possibility that
                    credit losses on the loans used in the data will be missed. For example, a
                    15-year loan originated in 1990 may experience a credit loss in any year
                    from 1990 to 2005, but only credit losses that occur in 1990 to 1992 will be
                    predicted by the regression.2 Updating the data set with post-1992
                    borrower behavior would allow more credit losses to be observed in the
                    data. Because a longer history is available for older loans, it is likely that
                    fewer credit losses are missed on older loans than on newer loans.
                    Because a key predictor, the greatest decline in land prices, varies with the
                    age of loan, the result is likely to be a biased regression coefficient for this
                    variable.

                    The data systems in use by FCBT did not identify all the loans that resulted
                    in losses to the bank. Some loans that were merged or restructured may


                    1
                     Farmer Mac uses the same data for its credit risk models and therefore faces the same
                    limitations.
                    2
                     This problem is known as “right censoring” in statistical analysis. SeeYamaguchi, Kazuo,
                    1991, Event History Analysis (Sage Publications, Inc., Newbury Park, CA) pp. 3-9.




                    Page 87                                                            GAO-04-116 Farmer Mac
Appendix VI
Farm Credit Administration Credit Risk
Model




have resulted in losses to the bank, but these losses are not captured by the
foreclosure variable used in the credit risk model. Thus, the frequency of
credit losses may be understated in the model’s forecast.

Additionally, the data system did not record the time value of money
foregone during foreclosure, a process that could take 2 years. This has
two implications for the credit risk model. First, some foreclosures, which
appear to have had no credit loss may, in fact, have resulted in credit
losses. Thus, the frequency of a credit loss would be understated in the
model’s forecast. Second, the model’s estimate of severity given default
may be different than the historical average. To the extent that costs are
not captured on loans that resulted in credit losses, the calculated severity
will understate the loss severity actually experienced by the bank. To the
extent that some loans are excluded from the severity analysis because the
database recorded that they had no credit losses, severity may be either
understated or overstated, depending on the magnitude of the severity for
these loans. The data used by FCA’s contractors indicated that 62 percent
of the loans that went through foreclosure had no credit loss recorded.
While the number of loans that may have been misidentified as having no
losses is not known, it is potentially large.

The data set contains only Texas loans. Previous work by FCA’s
contractors indicates that a region consisting of Minnesota, Iowa, and
Illinois was the area that experienced the highest level of stress as legally
defined for FCA’s credit risk test, and that Texas was the fourth most
severely stressed geographic region in the mid-1980s.3 Hence, the model
must extrapolate credit losses to a stress situation beyond that contained in
the data used to estimate the model. The form of extrapolation used in
FCA’s credit risk model assumes that there is a straight line relationship
between land price declines and a function of the probability of credit loss.
Without data on loans that experienced property price declines akin to
those in the most stressed region of the country, it is impossible to know if
the true relationship is linear or nonlinear.4

The data used by FCA’s contractors did not include information on whether
or when the loan was prepaid. This has several consequences for the credit

3
12 CFR Part 650.
4
 Problems relating to extrapolation and the form of the relationship are discussed in
Snedecor, George, and Cochran, William, 1967, Statistical Methods, Sixth Edition (Iowa
State University Press, Ames, IA) p. 144 and p. 456.




Page 88                                                          GAO-04-116 Farmer Mac
Appendix VI
Farm Credit Administration Credit Risk
Model




risk model. The model assigns the land value decline that occurred
between 1985 and 1986 to any loan written between 1979 and 1986 that had
not entered foreclosure by 1986. It is possible that some of these loans had
refinanced by 1985 as interest rates declined, so that these loans would not
have been exposed to the 1985 and 1986 land price change. For these
loans, the regression would be predicting the probability of a credit loss on
a loan using a value for the predictor that occurred after the loan had been
paid off. The lack of information on loan prepayment also precludes the
measurement of the impact of loan duration on the probability of credit
loss. It is likely that a loan that was active for 10 years is more likely to
experience a credit loss than is an otherwise identical loan active for only 2
years, as it is exposed to the potential of adverse events for a longer time.
But the data do not identify which loans were active for only 2 years versus
those active for 10 years. To the extent that loans with lower credit risk as
measured by underwriting variables, such as lower LTV ratios, are more
likely to prepay, the underwriting variables in the regression are likely to
capture both the direct effect of the underwriting variable on the
probability of credit loss, and an indirect effect caused by the tendency of
these higher credit quality loans to prepay more often; hence, be exposed
to risk of a credit loss for a shorter period of time.5

The loans now purchased by Farmer Mac have different interest rate terms
than those used in FCA’s credit risk model. Over the time period covered
by the data, Farm Credit System (FCS) institutions, including FCBT, made
loans with adjustable interest rates, in which the interest rate was tied to
FCS’ cost of funds. The average cost of funds changed more slowly than
did the prevailing rate of interest, as FCS institutions used a mix of short-
and long-term debt, and the average cost of funds was an average of rates
on debt recently incurred and debt incurred over several previous years.
Because of these interest rate terms, when interest rates fell after 1982,
many farm credit borrowers found it advantageous to refinance their debt
with other lenders. The mismatch between fixed rate liabilities and
variable rate, prepayable assets was a cause of the FCS’s financial
problems in the mid-1980s.6 However, the bulk of the loans now purchased
by Farmer Mac are either rapidly adjusting adjustable-rate mortgages, tied


5
 Yamaguchi calls this nonindependent censoring. See Yamaguchi, op. cit., p. 6 and pp.169-
172.
6
 General Accounting Office, Preliminary Analysis of the Financial Condition of
the Farm Credit System, GAO/GGD-86-13-BR (Washington, D.C.: Oct. 4, 1985).




Page 89                                                           GAO-04-116 Farmer Mac
Appendix VI
Farm Credit Administration Credit Risk
Model




to short-term interest rates, or are fixed-rate loans with prepayment
penalties or yield maintenance agreements. For these loans, there is little
or no advantage in refinancing when interest rates drop. In the case of
adjustable-rate loans, the interest rate on the mortgage will drop without
the need to refinance, and in the case of fixed-rate loans the prepayment
penalties or yield maintenance agreements increase the cost of refinancing,
making it less advantageous. As interest rates generally declined over the
period 1979 through1992 (the high point for interest rates was 1982, the low
point was 1992), it is likely that a larger percentage of the loans in the data
set paid off early than would be the case for the loans now purchased by
Farmer Mac. Therefore, the loans purchased by Farmer Mac are likely to
be exposed to adverse events for a longer time period than the loans used
in estimating the credit risk model. This would have the effect of
understating the credit risk capital requirement. The prevalence of yield
maintenance agreements has another effect on the potential for credit
losses in Farmer Mac’s portfolio. As previously discussed, the fixed-rate
loans now purchased by Farmer Mac that have yield maintenance
agreements are likely to experience elevated credit risk in times of falling
interest rates. A borrower in financial distress is more likely to go to
foreclosure, and is more likely to impose a severe credit loss, if the value of
the debt substantially exceeds the value of the collateral. After a fall in
interest rates, fixed-rate loans with yield maintenance agreements will owe
substantial amounts in excess of their unpaid principal balance. Therefore,
these loans are more likely to have total obligations (unpaid principal
balance plus yield maintenance) that exceed the value of the collateral,
than would loans of otherwise similar characteristics that did not have
yield maintenance agreements, such as those used in estimating FCA’s
credit risk model, resulting in an underestimate of credit risk by FCA’s
model.

Because the data set did not contain information on interest rates or
amortization terms, these variables could not be included in the credit risk
model regression analysis. Other studies of credit risk have found these to
be important variables in predicting credit losses.7 Loans which amortize
faster are exposed to adverse events for a shorter period of time, and
accumulate equity more rapidly, which reduces credit risk. Higher interest


7
 General Accounting Office, Mortgage Financing: FHA Has Achieved Its Home Mortgage
Capital Reserve Target, GAO/RCED-96-50 (Washington, D.C.: April 12, 1996) and Office of
Federal Housing Enterprise Oversight, 1999, Risk-Based Capital Regulation: Second Notice
of Proposed Rulemaking, Federal Register: (Volume 64, Number 70) (April 13, 1999).




Page 90                                                          GAO-04-116 Farmer Mac
                     Appendix VI
                     Farm Credit Administration Credit Risk
                     Model




                     rates lead to higher payment burdens, which can put greater stress on
                     borrower’s financial resources. Adjustable rate mortgages are subject to
                     “payment shock” in which defaults increase after a rise in interest rates,
                     which leads to a rise in the mortgage payment.8 Since FCA’s model does
                     not assign higher credit risk to longer amortization loans, or to adjustable-
                     rate loans in times of rising interest rates, Farmer Mac could increase its
                     exposure to credit risk by buying more of these types of loans, without
                     facing a higher risk-based capital requirement.

                     FCA’s ability to estimate a detailed credit risk model was limited by the
                     scarcity of relevant data for agricultural real estate loans. FCA’s
                     consultants identified the Farm Credit Bank of Texas’ data from 1979 to
                     1992 as the only available data set of agricultural loans observed during a
                     stressed period.9 The data file used by the contractors had 19,418 loans,
                     including 180 loans with credit losses. In contrast, the Office of Federal
                     Housing Enterprise Oversight’s (OFHEO) risk-based capital model for
                     Fannie Mae and Freddie Mac thirty year fixed rate single-family loans is
                     based on about 15 million loans, 176,000 of which had credit losses. While
                     none of the loans in FCA’s model were observed during a stress event as
                     severe as that called for in its risk-based capital statute, over 7,000 of the
                     30-year single family fixed rate loans used by OFHEO were observed during
                     the benchmark stress event specified by OFHEO’s risk-based capital
                     legislation.



Model Limitations	   We also have identified several limitations in the form of the credit risk
                     model used by FCA. These limitations include: (1) the methodology chosen
                     by FCA’s contractors; (2) use of an independent variable, greatest land
                     price decline, whose value is a function of the event predicted by the
                     regression; (3) transformations of the independent variables to enhance
                     goodness of fit prior to and independent of the calculation of significance
                     tests; and (4) the use of state averages to model credit risk on the long-term
                     standby agreements.




                     8
                     Price-Waterhouse, 1997, An Actuarial Review for Fiscal Year 1996 of the Federal Housing
                     Administration’s Mutual Mortage Insurance Fund, (Washington, D.C., Feb. 14, 1997).
                     9
                      Data after 1992 were not readily useable, as the Texas Bank changed computer systems and
                     post-1992 data could not be readily linked to earlier loans. FCA noted they are now studying
                     the data to determine if it is possible to link post-1992 data to earlier loans.




                     Page 91                                                             GAO-04-116 Farmer Mac
Appendix VI
Farm Credit Administration Credit Risk
Model




FCA’s credit risk model uses observations on loans, and predicts the
probability that a loan will experience a credit loss at some point in its life.
Many models of mortgage credit risk use a different structure, and predict
the probability that a loan will experience a credit event over a defined time
period, such as a quarter or a year. 10 For example, our model of the
Veteran’s Administration credit subsidy costs and OFHEO’s multifamily
risk-based capital model predict annual probabilities of a credit event,
while OFHEO’s single-family model predicts quarterly probabilities of a
credit event.11 These models have the advantage of accounting for the
different level of risk inherent in loans that are active for longer or shorter
periods, and can readily estimate the effects of predictor variables that
change over time, such as interest rates or the value of collateral. The
ability to incorporate such variables in the measurement of credit risk is
important when the goal is to measure the risk of a pool of loans, some of
which are new, and some of which have been active for a long time. For
example, the credit risk on a loan originated 5 years ago in a state with a 50
percent rise in agricultural real estate prices over that 5-year period is likely
to have less credit risk than an otherwise identical loan originated 5 years
ago in a state where agricultural real estate prices have remained constant.
In order to capture the changing credit risk over time in a portfolio with
seasoned loans, it is necessary to include measures of credit risk
determinants that change over time.

The credit risk model does incorporate a variable, change in the value of
agricultural real estate, which changes over time. However, its inclusion in
FCA’s model, which predicts lifetime credit event probabilities, instead of
annual or quarterly probabilities, leads to biased estimates of the effects of
land price changes on credit risk. The variable is defined as the greatest
annual percentage decline in agricultural land price in Texas from the year
that the loan is originated until either 1992 or the year of loan foreclosure,
whichever comes first. The regression is designed to predict the
probability of foreclosure with credit losses, but the variable’s value is
determined, in part, by whether the loan enters foreclosure. For example, a

10
 Different models use different measures of credit risk, such as a loan terminating in a
claim (such as our study of VA Subsidy Rates, Homeownership: Appropriations Made to
Finance VA’s Housing Program May be Overestimated (GAO-RCED-93-173)) or
delinquency, Calem, P. and Wachter, S. 1991, Community Reinvestment and Credit Risk:
Evidence from an Affordable-Home-Loan Program, Real Estate Economics, V. 27 #1,
pp.105-134. The term credit event is used as a general description of these various
definitions.
11
     Such models are known as hazard models. Yamaguchi, op. cit., p. 9.




Page 92                                                              GAO-04-116 Farmer Mac
Appendix VI
Farm Credit Administration Credit Risk
Model




1979 loan that does not enter foreclosure will be assigned a value of—17
percent—that being the maximum decline in land prices from 1979 to 1992.
However, a loan that enters foreclosure in 1980 would be assigned a value
of 7 percent, as land prices rose over the short time that the loan survived.12
Thus, the definition of the land price variable is not independent of the
event being estimated, and its estimated coefficient is likely to be biased.13
It can be shown that the structure of estimating a lifetime probability,
combined with the definition of land price change used in FCA’s credit risk
model, can produce a statistically significant coefficient for the land price
change variable, even if there were no effect of land price changes on credit
risk.14 The land price change variable is the key variable used to
extrapolate the stress scenario called for in the Farm Credit Act of 1971, as
amended, and the regression may be estimating its impact in a biased
fashion, which would lead to a misestimate of the credit risk in Farmer
Mac’s portfolio.

Significance tests reported with the regression results in the Federal
Register document, which describes the credit-risk model, are likely to be
biased in favor of a finding of significance. Several variables are
transformed in various ways before their coefficients are estimated in the
logistic regression. The greatest decline in land value variable is modified
by a “dampening factor,” which proportionately decreases the absolute
value of the variable, the longer the loan is observed to have not defaulted.
The LTV ratio is raised to a power. The loan size (dollar amount) variable is
defined as 1 minus the exponential of the product of the loan size in
thousands and the number –0.00538178, so that the variable is close to 0
when loan size is small and rises towards 1 as loan size increases. These
transformations were not estimated as part of the logistic regression.
Instead, several values were tried for each of these parameters, and the
values giving the best goodness-of-fit measurements were used to



12
   The land price change variable is then modified by a “dampening factor” before entering
the regression, but the value of the transformed variable is still determined, in part, by
whether and when the loan enters foreclosure.
13
   Working, E “What do Statistical Demand Curves Show?", 1927, Quarterly Journal of
Economics V 41 #1.
14
   A similar example is cited in Yamaguchi, op. cit., pp. 26-27. Yamaguchi concludes that
independent variables, which are determined by life course characteristics, can only be used
if their value is determined prior to the observation entering the period in which they are
subject to the risk of experiencing the event to be modeled.




Page 93                                                            GAO-04-116 Farmer Mac
Appendix VI
Farm Credit Administration Credit Risk
Model




transform the predictor variables prior to estimating the logistic regression.
Such pre-testing leads to inflated tests of significance.15

The estimated credit risk model is not used directly to produce an estimate
of the credit risk inherent in Farmer Mac’s standby agreements. Instead,
the average credit risk for loans in each state is used as an estimate for the
credit risk in the standby agreements. The regression-based model cannot
be used because key underwriting variables for standby agreements are not
reported to FCA. The use of state averages in place of credit risk
calculations based on underwriting variables would allow Farmer Mac to
purchase standby agreements with higher loan-to-value or debt-to-asset
ratios; hence, higher credit risk, than is contained in their loan portfolio,
without a commensurate increase in their risk-based capital requirements.
Although Farmer Mac’s current portfolio of standby agreements has, on
average, lower LTV ratios than its on-balance sheet portfolio, the structure
of the credit risk model provides Farmer Mac with the incentive to shift
risk into standby agreements should the risk-based capital constraint
become binding.




15
 See Kennedy, P, 1987, A Guide to Econometrics, 2nd Edition, (MIT Press, Cambridge, MA),
p. 164.




Page 94                                                          GAO-04-116 Farmer Mac
Appendix VII

Comments from the Federal Agricultural
Mortgage Corporation

Note: GAO comments
supplementing those in
the report text appear
at the end of this
appendix.




                         Page 95   GAO-04-116 Farmer Mac
Appendix VII

Comments from the Federal Agricultural

Mortgage Corporation





Page 96                                   GAO-04-116 Farmer Mac
                 Appendix VII

                 Comments from the Federal Agricultural

                 Mortgage Corporation





See comment 1.




See comment 2.




                 Page 97                                   GAO-04-116 Farmer Mac
                 Appendix VII

                 Comments from the Federal Agricultural

                 Mortgage Corporation





See comment 3.




                 Page 98                                   GAO-04-116 Farmer Mac
Appendix VII

Comments from the Federal Agricultural

Mortgage Corporation





Page 99                                   GAO-04-116 Farmer Mac
               Appendix VII

               Comments from the Federal Agricultural

               Mortgage Corporation





               The following are GAO’s comments on the Federal Agricultural Mortgage
               Corporation's letter dated September 19, 2003.



GAO Comments   1. Farmer Mac commented that its own loan portfolio database is
                  relatively new and that the U.S. agriculture has not been through a
                  significant downturn during the period it covers. Further, Farmer Mac
                  expects to continue to use the FCBT database as a conservative
                  benchmark for evaluating credit risk. While it may be true that during
                  the period of time Farmer Mac has been accumulating information to
                  develop its own loan database, the U.S. agricultural industry has not
                  faced a similar catastrophic decline as that experienced during the
                  1980's as captured in the Texas data, we disagree with Farmer Mac's
                  inference that its portfolio is too new to provide loan loss experience
                  from which to estimate credit losses. Farmer Mac has been buying and
                  retaining its portfolio of loans for over 7 years, and has been executing
                  its guarantees under standby commitments for over 3 years.
                  Accounting industry guidance suggests, “Two to three years of lending
                  experience normally would provide data that is more relevant than peer
                  group experience.” Further, because Farmer Mac's loan portfolio has
                  characteristics, which differ from the FCBT data used in the model, and
                  quantification of the effect of these differences-whether it would
                  increase, decrease, or have no material impact to the allowance-has not
                  been made by Farmer Mac, we believe that Farmer Mac should use the
                  more relevant data. Farmer Mac asserts, however, that the FCBT is a
                  more conservative tool to benchmark the allowance because it includes
                  an economically depressed time period. In fact, the loan loss
                  allowance should reflect current environmental factors and conditions
                  that could cause probable future losses rather than the most severe loss
                  situation in history. We believe that the most appropriate approach
                  would be for Farmer Mac to use its own data, which provides relevant
                  and comparable loan characteristics, in its loan loss methodology while
                  also applying appropriate “stress testing” approaches to reflect any
                  potential or likely future downturns or economically depressed
                  conditions.

               2. Farmer Mac commented that an outside expert on prepayment
                  modeling has validated the adjustments that Farmer Mac made to its
                  prepayment model, and that since no useable agricultural mortgage
                  database exists, Farmer Mac will revise its prepayment modeling
                  accordingly when its historical database becomes statistically
                  significant. In making that comment, Farmer Mac seems to disagree



               Page 100                                                GAO-04-116 Farmer Mac
Appendix VII

Comments from the Federal Agricultural

Mortgage Corporation





    with our recommendation that it should improve the quality of its
    prepayment model to ensure accurate measurements of interest rate
    risk. However, as stated in our report, Farmer Mac management noted
    that they are currently working with an outside expert to develop an
    agricultural mortgage prepayment model to better model prepayment
    risk.

3.	 Farmer Mac commented that the reduction of capital requirements for
    mortgage loans that bear Farmer Mac credit enhancements is not
    arbitrage but is analogous to the regulatory capital treatment of loans
    enhanced by Fannie Mae and Freddie Mac guarantee or commitment.
    Referring to table 2 in the report, Farmer Mac commented that
    comparing Farmer Mac's statutory capital minimum requirement to the
    capital requirement for primary lenders is irrelevant and stated that
    Farmer Mac is required to maintain the higher of statutory minimum
    and risk-based capital. First, because all of Farmer Mac's current
    participants in standby agreements are FCS institutions (another GSE),
    the report discusses the potential reduction of the sum of capital
    required to be held by the Farm Credit System and Farmer Mac without
    a corresponding reduction in risk. In this regard, a reduction in capital
    requirements for loans bearing Farmer Mac credit enhancements is not
    analogous to the housing GSEs because these GSEs are enhancing loan
    credit from commercial lenders, not another GSE. The intent of table 2
    is not to compare the capital levels of Farmer Mac with primary
    lenders, but rather, to demonstrate the reduction of capital for loans
    enhanced by Farmer Mac guarantee or commitment. We agree and the
    draft report clearly states that Farmer Mac must meet the higher of
    statutory minimum or risk-based capital requirement. As such, we have
    analyzed the risk-based capital model and have identified some
    limitations that are discussed in the report.




Page 101                                                GAO-04-116 Farmer Mac
Appendix VIII

Comments from the Farm Credit
Administration

Note: GAO comments
supplementing those in
the report text appear
at the end of this
appendix.




                         Page 102   GAO-04-116 Farmer Mac
Appendix VIII

Comments from the Farm Credit 

Administration





Page 103                          GAO-04-116 Farmer Mac
                        Appendix VIII

                        Comments from the Farm Credit 

                        Administration





See comments 1 and 2.




See comment 3.




                        Page 104                          GAO-04-116 Farmer Mac
                        Appendix VIII

                        Comments from the Farm Credit 

                        Administration





See comment 4.




See comment 5.




See comment 6.




See comments 7 and 8.




                        Page 105                          GAO-04-116 Farmer Mac
                  Appendix VIII

                  Comments from the Farm Credit 

                  Administration





See comment 9.




See comment 10.




                  Page 106                          GAO-04-116 Farmer Mac
               Appendix VIII

               Comments from the Farm Credit 

               Administration





               The following are GAO’s comments on the Farm Credit Administration's
               letter dated August 21, 2003.



GAO Comments   Steady-state Approach

               1.	 FCA commented that it had the authority to use whatever approach is
                   reasonable to produce a stressful model that is most suitable for
                   Farmer Mac and agricultural loans. It also stated that the 1987 Act is
                   best read to treat Farmer Mac as a going concern. We agree and believe
                   that our report clearly indicates that FCA had the authority to build a
                   risk-based capital test using either a steady-state or a run-off approach.
                   However, we do not agree with FCA's view that the statute's
                   requirement for positive capital throughout a stress scenario implies a
                   preference for a steady-state approach. The Federal Housing
                   Enterprises Financial Safety and Soundness Act of 1992 (FHEFSA),
                   which sets the requirements for OFHEO's risk-based capital test, also
                   requires positive capital throughout a stress scenario, but requires an
                   initial run-off approach, followed by mandated studies of the steady-
                   state approach.

               2.	 FCA commented that using a steady-state approach resulted in their
                   having to make fewer assumptions. We believe that the assumptions
                   required for a steady-state approach are difficult to support. The key
                   assumption of a steady-state approach is that the volume of new
                   business will exactly match the run-off of old business, even during a
                   stressed period. Additionally, assumptions concerning the level of
                   profitability, or unprofitability, of new business during a stressed period
                   must be made in order to implement a steady-state approach. Both of
                   these assumptions are difficult to base on data for financial institutions
                   in stressed time periods.

               Data Limitations

               3.	 FCA commented that GAO staff were unable to provide suggestions for
                   a more suitable data set. In our report, we recognized that FCB Texas
                   data was the most comprehensive data source available and did not
                   suggest that the FCB Texas data be replaced with a more suitable data
                   set. Rather, we recommend that the FCB Texas data be brought current,
                   if possible, and that data from other sources be used to model risks such
                   as payment shocks on adjustable-rate mortgages or amortization terms
                   that cannot be easily modeled with the Texas data. Updating the Texas



               Page 107                                                  GAO-04-116 Farmer Mac
Appendix VIII

Comments from the Farm Credit 

Administration





   Bank data would improve the credit risk estimation model by
   addressing an issue raised in the report, that loans originated in 1992 or
   earlier, some of which would have experienced the land price stresses of
   the mid 1980's, may still result in credit losses after 1992. Only by
   updating the data set with post 1992 foreclosures can the model capture
   the lifetime credit experience of these loans.

   Additionally, the credit risk model uses a regression framework to
   extrapolate losses based on the Texas stress event to a more severe
   stress event such as that which occurred in the Upper Midwest. The
   extrapolation relies on the slope of the land price decline variable
   estimated by the regression. Since 71 percent of the loans in the Texas
   data file used by the contractors, comprising 176 of the 180 credit losses,
   are associated with a land price decline of 17 percent, and another 25
   percent of the Texas data are associated with land price declines of 2 or
   4 percent (these loans have no credit losses), there is very little variation
   in the data used to estimate the slope with respect to minimum land
   price changes. The loans associated with a 17 percent price decline are
   all observed for 7 to 13 years after origination, while the loans
   associated with 2 or 4 percent declines are all observed for only 0 to 5
   years after origination. Augmenting the Texas data to include credit
   losses over less stressful time periods should reduce the bias and
   increase the precision of the estimate of the land price decline - credit
   loss relationship, upon which the extrapolation used by FCA is based.

4.	 FCA commented that the magnitude and location of worst-case
    conditions of its model validation process is consistent with findings in
    studies and data compilations by a number of economists. Therefore, it
    is evident that the FCA model has strong forecasting capability in
    determining risk-based capital requirements. We do not dispute FCA's
    finding that the Upper Midwest in the mid-1980's was a high stress event
    for agricultural real estate. We disagree that FCA has presented
    evidence of the model's forecasting ability. Without post 1992 data on
    loans, there are no data with which out of sample forecasts can be made
    to test the model's forecasting ability. Additionally, we have noted in the
    report that in-sample goodness-of-fit statistics presented with the model
    are likely to be biased, based on the fact that nonlinear transformations
    of certain variables, such as loan-to-value, were fitted prior to the
    estimation of the regression model.

5.	 We have modified the text of the report to indicate that FCA did not
    have post 1992 data available in a ready to use format, and to recognize



Page 108                                                   GAO-04-116 Farmer Mac
Appendix VIII

Comments from the Farm Credit 

Administration





   that FCA is engaged in an effort to incorporate post 1992 FCB Texas
   data.

Servicing Records

6.	 In commenting on a section of the draft report that discusses data
    limitations, FCA stated that it had reviewed the servicing records of FCB
    of Texas, which were not detailed payment records, and concluded that
    they did not contain information that would enhance the quality of the
    loss estimates. We modified the text of the report to delete references to
    servicing records.

Yield Maintenance Provisions

7.	 In commenting on a section of the draft report that discusses the effect
    of yield maintenance provisions and prepayment penalties in the credit
    risk model, FCA stated that falling interest rates, with other factors held
    constant, would tend to increase rather than decrease present market
    values of farmland. We agree that, with other factors held constant,
    declining interest rates will tend to increase the value of agricultural real
    estate. However, other factors are often not constant. For example, a
    decline in inflation will lower both interest rates and anticipated cash
    flows, so that real estate values will not necessarily increase when
    interest rates decline. FCA's stress test is based upon falling Texas land
    prices in 1985 and 1986. From their peak in 1985, Texas agricultural real
    estate values fell by 25 percent over the next 2 years, although the
    interest rate on 10 year Treasury bonds had fallen from 10.6 percent to
    8.4 percent over the same time period. Additionally, yield maintenance
    provisions increase the borrower's obligation even when interest rates
    are unchanged, because the present value of the spread between the
    loan rate and the rate on comparable Treasury securities must be paid
    when a loan is terminated. Farmer Mac's seller-servicer manual gives an
    example in which there is an 8 percent yield maintenance penalty
    despite unchanging interest rates. Because yield maintenance penalties
    and land prices do not always move in equal and opposite proportions,
    we believe that each should be considered as independent variables in a
    credit risk regression.

8.	 FCA stated that default rate studies generally indicate that default
    frequencies are considerably higher earlier in the lives of loans and that
    these time patterns characterize the FCBT data and the risk-based
    capital model. We do not agree that default frequencies are higher in the



Page 109                                                   GAO-04-116 Farmer Mac
Appendix VIII

Comments from the Farm Credit 

Administration





   early years of a loan's life based on performance of the loans in the
   FCBT data. In this data, no credit losses occurred in the year of loan
   origination, and less than 4 percent of the credit losses occurred within
   the subsequent 2 years. Further, about 25 percent of the credit losses
   occurred 9 years or more after origination and the median year of
   foreclosure in the FCBT data is the 7th. Nevertheless, large yield
   maintenance penalties and substantial refinancing incentives can occur
   early in a loan's life. Therefore, we continue to believe that it is
   important to consider the effects of prepayment and yield maintenance
   when estimating a credit risk model.

The Use of Land Value Decline

9.	 FCA referred to a section of the draft of this report that discusses how
    the variable of minimum land price decline affects the accuracy of the
    credit loss regression. FCA commented that GAO's position, which
    suggests the use of this variable would reduce the accuracy of the
    model, is inconsistent with theory and empirical evidence and that the
    direction of the functional relationship in the risk-based capital model is
    valid. We agree that the direction of the effect of land prices on credit
    losses in FCA's credit risk model is consistent with theory and empirical
    evidence. However, FCA's implementation of the risk-based capital test
    relies on the magnitude, as well as the direction, of this relationship. It
    is still the case that using a land price decline variable that is defined, in
    part, by the event that the regression seeks to predict, will produce a
    biased estimate of the magnitude of the effect of land price changes on
    credit risk.

Credit Risk Not Captured

10.In commenting on the draft report discussion of the three types of
   instruments that are not subject to credit risk in the risk-based capital
   model, FCA stated that the current risk exposures on these instruments
   were immaterial. We recognize that Agvantage bonds are backed by
   both mortgage collateral and the general obligation of the issuing
   institutions. Issuing institutions are likely to be stressed at a time of
   falling farmland values as contemplated by the RBC stress test. While
   we agree that multiple layers of backing for these bonds is likely to
   result in a small credit risk, they are still at some risk of loss in a
   stressed time period. The Federal Home Loan Bank System uses a
   similar product (Advances) with even more layers of backing (mortgage
   pools, general obligations of the originating institutions, and the so-



Page 110                                                    GAO-04-116 Farmer Mac
Appendix VIII

Comments from the Farm Credit 

Administration





   called superlien, giving Home Loan Banks first priority on the assets of
   originating depository institutions), yet the Federal Housing Finance
   Board assigns a small, but nonzero credit risk charge to these assets.

   We agree that the credit risk stemming from counterparty risk on swap
   transactions, and the credit risk on many of the assets in Farmer Mac's
   liquidity portfolio, is likely to be small. However, we believe that credit
   risk can be easily accounted for, and the text of our report notes that it
   is accounted for in the risk-based capital models of other regulators,
   such as OFHEO and the FHFB. Doing so would increase the accuracy of
   FCA's risk-based capital calculation for Farmer Mac, and would provide
   an incentive for Farmer Mac to do business with higher rated
   counterparties and to hold lower risk assets, if the risk based capital
   constraint becomes binding.




Page 111                                                 GAO-04-116 Farmer Mac
Appendix IX

GAO Contacts and Staff Acknowledgments




GAO Contacts	      Davi D’Agostino, (202) 512-8678
                   Jeanette Franzel, (202) 512-9471



Acknowledgments	   In addition to those individuals named above, Rachel DeMarcus, Debra
                   Johnson, Austin Kelly, Paul Kinney, Bettye Massenburg, Kimberley
                   McGatlin, Nicholas Satriano, John Treanor, and Karen Tremba made key
                   contributions to this report.




                   Page 112                                           GAO-04-116 Farmer Mac
Glossary of Terms




Amortization	         The process of making regular, periodic decreases in the book or carrying
                      value of an asset.



Basis points	         A basis point is equal to one hundredth of a percent. It is used to measure
                      changes in or differences between yields or interest.



Capital	              For financial purposes, capital is generally defined as the long-term funding
                      for a firm that cushions the firm against unexpected losses.



Credit risk	          The possibility of financial loss resulting from default by borrowers on
                      farming assets that have lost value or other parties’ failing to meet their
                      obligations.



Discount notes	       Discount notes are unsecured general corporate obligations that are issued
                      at a discount but mature at face value. Their maturities range from
                      overnight to one year.



Duration	             A measure of the average timing of cash flows from an asset or a liability. It
                      is computed by summing the present values of all future cash flows after
                      multiplying each by the time until receipt, and then dividing that product by
                      the sum of the present value of the future cash flows without weighting
                      them for the time of receipt.



Interest rate risk	   Interest rate risk is the potential that changes in prevailing interest rates
                      will adversely affect assets, liabilities, capital, income or expenses at
                      different times in different amounts.




                      Page 113                                                  GAO-04-116 Farmer Mac
                              Glossary of Terms




Interest rate swap	           A financial instrument representing a transaction in which two parties
                              agree to swap or exchange net cash flows, on agreed-upon dates, for an
                              agreed-upon period of time, for interest on an agree-upon principal amount.
                              The agreed upon principal amount, called the notional amount, is never
                              exchanged. Only the net interest cash flows are remitted. In the simplest
                              form of interest rate swap, one party agrees to swap fixed-rate loan
                              payments with the floating-rate payments of another party.



Liquidity	                    Both the capacity and the perceived capacity to meet all obligations
                              whenever due, without a material increase in cost, and to take advantage of
                              business opportunities important to the future of the enterprise. The
                              capacity and the perceived ability to meet known near-term and long-term
                              funding commitments whole supporting selective business expansion.



Liquidity contingency risk	   The risk that future events may require a materially larger amount of
                              liquidity than the financial institution currently requires. It is of the three
                              primary components of liquidity risk along with mismatch liquidity risk and
                              market liquidity risk.



Medium term notes (MTN) 	     Medium term notes are debt securities that may be issued with floating or
                              fixed interest rates with maturities ranging from nine months to thirty years
                              or longer. An advantage of MTNs over corporate bonds is that they tend to
                              be more flexible in terms of maturities and interest rates.



Operations risk	              The risk that an entity may be exposed to financial loss from inadequate
                              systems, management failure, faulty controls, or human error.



Prepayment risk 	             The risk that prepayments will speed or slow and therefore change the
                              yield and/or life of the security.



Return on average assets 	    Return on average assets is net income for the year divided by the average
                              total assets of the year.




                              Page 114                                                  GAO-04-116 Farmer Mac
                     Glossary of Terms




Return on equity 	   Return on average common stockholder equity is the net income for the
                     year less preferred stockholder dividends divided by the average common
                     stockholder equity for the year and demonstrates how well the company is
                     performing for its common stock shareholders.



Yield maintenance	   A prepayment premium that allows investors to attain the same yield as if
                     the borrower made all scheduled mortgage payments until maturity.




(250095)             Page 115                                              GAO-04-116 Farmer Mac
GAO’s Mission	            The General Accounting Office, the audit, evaluation and investigative arm of
                          Congress, exists to support Congress in meeting its constitutional responsibilities
                          and to help improve the performance and accountability of the federal government
                          for the American people. GAO examines the use of public funds; evaluates federal
                          programs and policies; and provides analyses, recommendations, and other
                          assistance to help Congress make informed oversight, policy, and funding
                          decisions. GAO’s commitment to good government is reflected in its core values of
                          accountability, integrity, and reliability.


Obtaining Copies of       The fastest and easiest way to obtain copies of GAO documents at no cost is
                          through the Internet. GAO’s Web site (www.gao.gov) contains abstracts and full-
GAO Reports and           text files of current reports and testimony and an expanding archive of older
                          products. The Web site features a search engine to help you locate documents
Testimony                 using key words and phrases. You can print these documents in their entirety,
                          including charts and other graphics.
                          Each day, GAO issues a list of newly released reports, testimony, and
                          correspondence. GAO posts this list, known as “Today’s Reports,” on its Web site
                          daily. The list contains links to the full-text document files. To have GAO e-mail this
                          list to you every afternoon, go to www.gao.gov and select “Subscribe to
                          e-mail alerts” under the “Order GAO Products” heading.


Order by Mail or Phone	   The first copy of each printed report is free. Additional copies are $2 each. A check
                          or money order should be made out to the Superintendent of Documents. GAO
                          also accepts VISA and Mastercard. Orders for 100 or more copies mailed to a single
                          address are discounted 25 percent. Orders should be sent to:
                          U.S. General Accounting Office
                          441 G Street NW, Room LM
                          Washington, D.C. 20548
                          To order by Phone: 	   Voice: (202) 512-6000
                                                 TDD: (202) 512-2537
                                                 Fax: (202) 512-6061


To Report Fraud, 	        Contact:
                          Web site: www.gao.gov/fraudnet/fraudnet.htm
Waste, and Abuse in       E-mail: fraudnet@gao.gov
Federal Programs          Automated answering system: (800) 424-5454 or (202) 512-7470



Public Affairs	           Jeff Nelligan, Managing Director, NelliganJ@gao.gov (202) 512-4800
                          U.S. General Accounting Office, 441 G Street NW, Room 7149
                          Washington, D.C. 20548
United States
                  Presorted Standard

General Accounting Office
      Postage & Fees Paid

Washington, D.C. 20548-0001
           GAO

                                  Permit No. GI00

Official Business

Penalty for Private Use $300

Address Service Requested