Title: FINAL RULE--Federal Agricultural Mortgage Corporation Funding and Fiscal Affairs; Risk-Based Capital Requirements--12 CFR Part 652
Issue Date: 06/05/2008
Agency: FCA
Federal Register Cite: 73 FR 31937
___________________________________________________________________________
[6705-01-P]

FARM CREDIT ADMINISTRATION

12 CFR Part 652

RIN 3052-AC36

Federal Agricultural Mortgage Corporation Funding and Fiscal Affairs; Risk-Based Capital Requirements

AGENCY: Farm Credit Administration.

ACTION: Final rule.

SUMMARY: The Farm Credit Administration (FCA, Agency, or we) adopts a final rule that amends capital regulations governing the Federal Agricultural Mortgage Corporation (Farmer Mac or the Corporation). The final rule updates the Risk-Based Capital Stress Test (RBCST, RBC model, model) in response to recent changes in Farmer Mac's operations that are not addressed in the current version (Version 2.0). The final rule also amends the current model's assumption regarding the carrying costs of nonperforming loans to better reflect Farmer Mac's actual business practices. In addition, the final rule adds a new component to the model to recognize counterparty risk on nonprogram investments through application of discounts or "haircuts" to the yields of those investments and makes technical amendments to the layout of the model's Credit Loss Module. The effect of the rule is to update the model so that it continues to appropriately reflect risk in a manner consistent with statutory requirements for calculating Farmer Mac's regulatory minimum capital level under a risk-based capital stress test.

EFFECTIVE DATE: This regulation will be effective the later of 30 days after publication in the Federal Register during which time either or both Houses of Congress are in session, or June 30, 2008. We will publish a notice of the effective date in the Federal Register.

FOR FURTHER INFORMATION CONTACT:

Joseph T. Connor, Associate Director for Policy and Analysis, Office of Secondary Market Oversight, Farm Credit Administration, McLean, VA 22102-5090, (703) 883-4280, TTY (703) 883-4434;

or

Rebecca S. Orlich, Senior Counsel, Office of the General Counsel, Farm Credit Administration, McLean, VA 22102-5090, (703) 883-4420, TTY (703) 883-4020.

SUPPLEMENTARY INFORMATION:

I. Purpose

Under section 8.32 of the Farm Credit Act of 1971, as amended,1 the FCA established the RBCST for Farmer Mac in 2001. It is the Agency's objective that the RBCST continues to determine regulatory capital requirements in a manner consistent with statutory requirements and constraints. The purpose of this final rule is to revise the risk-based capital regulations that apply to Farmer Mac to more accurately reflect changes in Farmer Mac's operations and business practices. The substantive issues addressed in this final rule include the treatment of program loan volume with certain credit enhancement features (e.g., Off-Balance Sheet AgVantage volume, subordinated interests, and program loan collateral pledged in excess of Farmer Mac's guarantee obligation (hereafter, "overcollateral")), counterparty risk on nonprogram investments, and the carrying costs associated with the funding of nonperforming loans. We also describe minor formatting changes to the structure of the Credit Loss Module and the RBC model that are in the nature of technical changes. The preamble to the proposed rule, which was published in the Federal Register on September 13, 2007, contains a full description of the proposed changes. The proposed rule provided for a 45-day comment period that ended on October 29, 2007.2 Below we discuss only those provisions on which we received comments.

The final rule (Version 3.0 of the RBC model) is adopted with one revision from the proposed rule. The revision permits the Director of the Office of Secondary Market Oversight to reduce the haircut level applied to unrated investments.

II. Background

Our analysis of the RBCST has identified a need to update the model in response to changing financial markets, new business practices and the evolution of the loan portfolio at Farmer Mac, as well as continuing development of industry best practices among leading financial institutions. Our goal is to ensure that the RBCST reflects changes in the Corporation's business structure and loan portfolio that have occurred since the model was originally developed by FCA, while complying with the statutory requirements and constraints on the model's design.

III. Comments

We received one comment letter on the proposed rule from Farmer Mac. In general, Farmer Mac agreed with FCA's objective to revise the RBCST to reflect Farmer Mac's actual business risks more accurately but offered specific comments on three aspects of the proposed rule--the method of calculating the loan loss resolution time factor (LLRT), funding rate assumptions applied to nonperforming loan volume, and the treatment of unrated Government-sponsored enterprises (GSE) for purposes of applying discounts (or "haircuts") to nonprogram investments.

IV. Description of Comments on the Proposed Rule and FCA's Response

Below is a description of the three specific comments on the proposed rule and FCA's responses to the comments.

A. Treatment of Unresolved Nonperforming Loans in the LLRT Calculation

The proposed rule’s method for calculating the LLRT called for first calculating the average LLRT of nonperforming loans for all such loans that have resolved by the calculation date.3 This average is then adjusted to incorporate the LLRT to date of unresolved nonperforming loans currently on Farmer Mac’s books where the individual unresolved loan’s LLRT to date is greater than the average LLRT of resolved loans. The average is calculated on an Unpaid Principal Balance (UPB)-weighted basis. Farmer Mac did not object to the proposed UPB weighting or generally to the method for measuring time in nonperforming loan status. Farmer Mac disagreed with the specific method for incorporating the influence of censored data.4 Farmer Mac asserted that excluding data from the portion of the data set made up of unresolved nonperforming loans with individual LLRTs lower than the average of resolved loans would bias the overall LLRT calculation. To correct this perceived bias, Farmer Mac suggested either using only loans that have resolved or employing statistical tests that formally accommodate censored observations in order to accommodate the influence of the unresolved defaults in the data set. Farmer Mac suggested that such an approach would improve the LLRT accuracy by providing an unbiased estimate of "life expectancy" of a nonperforming loan (i.e., LLRT).

In developing the proposed approach, we considered several issues related to the application of duration or survival models, including the uniformity of the "arrivals" into default, the possible impact of UPB at time of default on remaining resolution experience, and general sample characteristics including length of observation window, fraction censored, and average life relative to observation window. The proposed approach was intended to balance the demands of a more complex modeling approach with the limits of the data set over the relatively short window (roughly 11 years), the relatively small set of loans in default and the observed high relative rate of default in a period centered near 2002 that substantially departs from a uniform arrival pattern. Farmer Mac correctly implies that excluding loans with relatively short durations in default as of the calculation date avoids a downward influence on the calculated LLRT. However, the treatment of unresolved nonperforming loans that have individual LLRTs greater than the average of those that have resolved as of the calculation date carries the opposite effect (i.e., avoids an upward influence) relative to their eventual resolution experience, because the current life at the calculation date is used in the weighted average calculation rather than its yet-to-be-determined actual life. The current life of this subset of loans at the calculation date necessarily understates their eventual LLRT and, thus, exerts an offsetting influence on the excluded subset. While there is not a formal statistical test for the relative impact of these two effects (treatment of both longer-than- and shorter-than-average LLRT), the adopted approach is intended to balance the two offsetting influences.

Farmer Mac suggested consideration of a more formal method to accommodate censored data in a duration or life-survival type model, and we conducted several related analyses. Importantly, the bulk of the defaults occurred in a period of time relatively early in the observation window. While the rate of arrival into default is non-uniform, the censored distribution displays the statistically useful property of increasing smoothly toward the censoring date. We calculated several measures of mean time in default on both UPB-weighted and unweighted bases, with alternative treatments of the unresolved data. Under all subsets of data examined, the UPB-weighted LLRT values are consistently 15 to 20 percent larger than the unweighted LLRT estimates.

We also estimated alternative specifications of the related hazard and survival functions using data supplied by Farmer Mac on all loans that had entered default status as of October 1, 2007, under (i) standard direct life tables with censored data, (ii) Kaplan-Meier methods, and (iii) Cox censored regression methods. The Kaplan-Meier method provides a direct method for recovery of the mean survival time accommodating the influence of the censored data at 1.79 years on an unweighted UPB basis. This value can be contrasted with a value of 1.60 on an unweighted basis using the method in the proposed rule for the same data set. Including the influence of UPB-weighting results in the proposed rule’s method increasing from 1.6 to 1.88, a value below that which we expect to find from any form of a censored regression or Lifetest model after weighting by UPB. Importantly, the survival function models we estimated generally confirm the significance of UPB on time-in-default and further argue for the use of UPB-weighted LLRT. Our testing of the suggested general approaches has shown that the joint treatment of excluded loans with lower than average current LLRTs and the conservative treatment of loans with longer than average but currently unresolved LLRTs results in a similar but slightly lower LLRT value compared with the censored regression methods suggested by Farmer Mac.

We conclude that the simplicity of the proposed approach is warranted because of the similarity in estimated values and the fact that Farmer Mac would have to re-run this test every quarter to update the LLRT. We note that, as the observation window continues to lengthen and the influence of censored loan data continues to decline, the specific treatment employed becomes less important because we expect the censored data effects to become more diluted.

B. Carrying Costs of Nonperforming Loans

Farmer Mac commented that the proposed funding rates applied to nonperforming loan volume do not reflect its actual operations and reiterated the comments in its letter of April 17, 2006, which related to the proposed rule for Version 2.0 of the RBC model.5 That letter encouraged FCA to treat on- and off-balance sheet nonperforming loans in the model as being funded at the less than 1 year (short-term) rate or in keeping with Farmer Mac’s actual practice of using the lowest funding rate available at the time a loan became nonaccrual given yield curve conditions existing at that time. Given the consolidated reporting of funding in only two categories--less than 1 year and greater than 1 year–-we determined that tying the incremental carrying costs to the short-term rate was acceptable.

The Agency acknowledged in the proposed rule that, under unusual conditions, the short-term rate may not be the minimum rate, and Farmer Mac could potentially reallocate to some degree debt on its books in order to fund nonperforming loans at a point on its corporate yield curve that might be more advantageous than the short-term rate. Such a reallocation could necessitate a corresponding reallocation of funding to a different asset to offset the debt associated with the now-optimally funded nonperforming loan position. We did not attempt to reflect forward discretionary management behavior or develop an "optimal" funding practice that would result in effective funding durations changing throughout the modeled 10-year period of the RBCST. In the proposed rule, we discussed this possibility and rejected a more complex LLRT funding assumption in favor of the proposed approach, particularly in light of the fact that the model is cast with only two maturity groupings ("buckets") of debt securities. To do otherwise would require adding substantial complexity to the components of the model reflecting funding costs – components which we believe are reasonably well calibrated to actual operations of Farmer Mac in their current aggregated form (i.e., two duration buckets).

We believe the proposed approach reflects Farmer Mac’s typical practices under normal conditions, and Farmer Mac has confirmed this is true in the preponderance of cases. To attempt to build an "optimal" or "discretionary" future duration-of-funding model that depends on the projected forward balance sheet composition in the model is beyond the scope of the model.

C. Treatment of Unrated GSE Securities

Farmer Mac commented that the proposed method of applying haircuts to unrated GSE securities should be changed. Specifically, Farmer Mac believes the model should treat such securities as AAA-rated, rather than limiting such treatment only to GSE securities that are fully guaranteed by a GSE. Farmer Mac asserts that this approach would both reflect the low risk of default on all GSE securities and be consistent with FCA’s approach to risk-weighting similar assets on the balance sheets of other Farm Credit System (System) institutions.6 FCA regulations of other System institutions permit a 20-percent risk weighting to "all securities" of GSEs without regard to credit rating. Farmer Mac asserts that FCA has recognized the low risk associated with GSE securities in the context of Agency regulations governing nonprogram investments and liquidity because they permit much higher obligor limits for eligible GSE investments than other types of nonprogram investments.7 Lastly, Farmer Mac asserts that the Agency would be justified in applying an automatic AAA-rating equivalent treatment to both unrated and GSE securities rated lower than AAA because the GSEs are closely regulated by Federal regulatory agencies that have access to more comprehensive and current information concerning the financial condition of the regulated GSE. The comment effectively encourages FCA to supersede the ratings of nationally recognized statistical rating organizations (NRSRO). This would be contrary to our stated goal for the regulation to avoid such a de facto re-rating process by the Agency in applying investment haircuts. However, we acknowledge there could be circumstances under which a reduction in the haircuts applicable to unrated investments that are not guaranteed by a GSE might be appropriate based on the risk characteristics of the investment. We believe that such circumstances could exist for non-GSE instruments as well as for GSE instruments. Therefore, in the final rule, while the default haircut on unrated instruments will remain as proposed, we have made a change in response to this comment that gives the Director of the Office of Secondary Market Oversight the discretion to apply a lower haircut on unrated investments on a case-by-case basis in accordance with the risk characteristics of the instrument.

We disagree with Farmer Mac’s assertion that the risk-based capital framework for other System institutions provides support for a policy that would apply AAA haircuts to all GSE securities regardless of their rating. The risk-based capital framework for other System institutions is fundamentally different from the RBCST applied to Farmer Mac as required by section 8.32 of the Farm Credit Act. The purpose of the regulations governing System capital requirements is to protect a System institution against unexpected losses arising from all types of risk, unlike this component of the RBC model, the purpose of which is to estimate counterparty risk. Comparing the proposed haircuts with capital requirements is not a relevant comparison because equity requirements to cover all types of unexpected losses applied as a percentage of volume are not comparable to haircuts to reflect counterparty risk that are applied by reducing estimated future cashflows over the RBC model’s 10-year time horizon on a gradually increasing basis. Accordingly, GSE investments with ratings will be haircut in accordance with the schedule in this rule.

V. Technical Changes to the RBCST in the Final Rule

In Version 3.0, we have revised the loan seasoning codes previously used in the Credit Loss Module to make off-balance sheet loan seasoning codes the same as those used for on-balance sheet loans and made other conforming data entry changes in the RBCST module. We have also incorporated a specification for senior subordinated loans in the Credit Loss Module to reduce the loss impact by the degree of subordination as referenced in the proposed rule.

VI. Impact of Changes on Required Capital

Our tests indicate that changes related to the LLRT would have the most significant impact on risk-based capital calculated by the model. The table below provides an indication of the relative impact of each revision for the quarter ended December 31, 2007, using preliminary model submission information for the fourth quarter 2007. The lines labeled "Impact of Carrying Costs of Nonperforming Loans within Ver. 3.0 (estimated)," "Impact of Investment Haircuts within Ver. 3.0 (estimated)," and "Impact of Treatment of Off-Balance Sheet AgVantage Program Volume and Other Credit-Enhanced Program Volume (e.g., Subordinated Interests) within Ver. 3.0 (estimated)" present the minimum risk-based capital level calculated if that revision were excluded from the final rule, Version 3.0 of the RBCST. The scenario used to estimate the impact of AgVantage Program Volume and Other Credit-Enhanced Program Volume excluded those two portfolios completely. As the table shows, the individual estimated impacts do not have an additive relationship to the total impact on the model relative to Version 2.0. This is due to the interrelationship of the changes with one another when they are combined in Version 3.0.

Calculated Regulatory Capital
($ in thousands)
12/31/2007
RBCST Version 2.0
42,754
RBCST Version 3.0 (estimated)
59,965
Impact of Carrying Costs of Nonperforming Loans within Version 3.0 (estimated)
20,623
Impact of Investment Haircuts within Version 3.0 (estimated)
707
Impact of the Treatment of Off-Balance Sheet AgVantage Program Volume and Other Credit-Enhanced Program Volume (e.g., Subordinated Interests) within Version 3.0 (estimated)
(2,620)

VII. Regulatory Flexibility Act

Pursuant to section 605(b) of the Regulatory Flexibility Act (5 U.S.C. 601 et seq.), FCA hereby certifies the rule will not have a significant economic impact on a substantial number of small entities. Farmer Mac has assets and annual income over the amounts that would qualify it as a small entity. Therefore, Farmer Mac is not considered a "small entity" as defined in the Regulatory Flexibility Act.


________________________________
112 U.S.C. 2279bb-1.

272 FR 52301 (Sept. 13, 2007).

3By "resolved," we mean loans that were in default for some period but were later paid current, paid off, liquidated, or transferred to real estate-owned, and are therefore no longer in nonperforming loan status.

4Censored data are loans that have entered nonperforming loan status but have not resolved as of the calculation date.

570 FR 69692 (Nov. 17, 2005). We discussed this comment in the preamble to the 2007 proposed rule (72 FR at 52305, Sept. 13, 2007).

6The FCA's capital rules for System banks and associations are set forth at 12 CFR part 615, subparts H and K. The risk weightings are in 12 CFR 615.5210-615.5212.

7See 12 CFR 615.5140.


List of Subjects in 12 CFR Part 652

Agriculture, Banks, banking, Capital, Investments, Rural areas.

For the reasons stated in the preamble, part 652 of chapter VI, title 12 of the Code of Federal Regulations is amended to read as follows:

PART 652--FEDERAL AGRICULTURAL MORTGAGE CORPORATION FUNDING AND FISCAL AFFAIRS

1. The authority citation for part 652 continues to read as follows:

Authority: Secs. 4.12, 5.9, 5.17, 8.11, 8.31, 8.32, 8.33, 8.34, 8.35, 8.36, 8.37, 8.41 of the Farm Credit Act (12 U.S.C. 2183, 2243, 2252, 2279aa-11, 2279bb, 2279bb-1, 2279bb-2, 2279bb-3, 2279bb-4, 2279bb-5, 2279bb-6, 2279cc); sec. 514 of Pub. L. 102-552, 106 Stat. 4102; sec. 118 of Pub. L. 104-105, 110 Stat. 168.

Subpart B--Risk-Based Capital Requirements

2. Amend § 652.65 by redesignating paragraph (b)(5) as new paragraph (b)(6) and adding a new paragraph (b)(5) to read as follows:

§ 652.65 Risk-based capital stress test.
* * * * *
(b) * * *
(5) You will further adjust losses for loans that collateralize the general obligation of Off-Balance Sheet AgVantage volume, and for loans where the program loan counterparty retains a subordinated interest in accordance with Appendix A to this subpart.
* * * * *

3. Amend § 652.85 by revising paragraph (d) to read as follows:

§ 652.85 When to report the risk-based capital level.
* * * * *
(d) You must submit your quarterly risk-based capital report for the last day of the preceding quarter by the earlier of the reporting deadlines for Securities and Exchange Commission Forms 10-K and 10-Q, or the 40th day after each of the quarters ending March 31st, June 30th, and September 30th, and the 75th day after the quarter ending on December 31st.

Appendix A--Subpart B of Part 652--Risk-Based Capital Stress Test
1.0 Introduction.
2.0 Credit Risk.
2.1 Loss-Frequency and Loss-Severity Models.
2.2 Loan-Seasoning Adjustment.
2.3 Example Calculation of Dollar Loss on One Loan.
2.4 Treatment of Loans Backed by an Obligation of the Counterparty and Loans for which Pledged Loan Collateral Volume Exceeds Farmer Mac-Guaranteed Volume.
2.5 Calculation of Loss Rates for Use in the Stress Test.
3.0 Interest Rate Risk.
3.1 Process for Calculating the Interest Rate Movement.
4.0 Elements Used in Generating Cashflows.
4.1 Data Inputs.
4.2 Assumptions and Relationships.
4.3 Risk Measures.
4.4 Loan and Cashflow Accounts.
4.5 Income Statements.
4.6 Balance Sheets.
4.7 Capital.
5.0 Capital Calculations.
5.1 Method of Calculation.

* * * * *

2.0 Credit Risk
Loan loss rates are determined by applying the loss- frequency equation and the loss-severity factor to Farmer Mac loan-level data. Using this equation and severity factor, you must calculate loan losses under stressful economic conditions assuming Farmer Mac's portfolio remains at a "steady state." * * *
* * * * *

2.4 Treatment of Loans Backed by an Obligation of the Counterparty, and Loans for which Pledged Loan Collateral Volume Exceeds Farmer Mac-Guaranteed Volume
You must calculate the age-adjusted loss rates for these loans that include adjustments to scale losses according to the proportion of total submitted collateral to the guaranteed amount as provided for in the "Dollar Losses" column of the transformed worksheets in the Credit Loss Module based on new data inputs required in the "Coefficients" worksheet of the Credit Loss Module. Then, you must adjust the calculated loss rates as follows:
a. For loans in which the seller retains a subordinated interest, subtract from the total estimated age-adjusted dollar losses on the pool the amount equal to current unpaid principal times the subordinated interest percentage.
b. Some pools of loans underlying specific transactions could include loan collateral volume pledged to Farmer Mac in excess of Farmer Mac's guarantee amount ("overcollateral"). Overcollateral can be either: (i) Contractually required according to the terms of the transaction, or (ii) not contractually required, but pledged in addition to the contractually required amount at the discretion of the counterparty, often for purposes of administrative convenience regarding the collateral substitution process, or (iii) both (i) and (ii).
1. If a pool of loans includes collateral pledged in excess of the guaranteed amount, you must adjust the age-adjusted, loan-level dollar losses by a factor equal to the ratio of the guarantee amount to total submitted collateral. For example, consider a pool of two loans serving as security for a Farmer Mac guarantee on a note with a total issuance face value of $2 million and on which the counterparty has submitted 10-percent overcollateral. The two loans in the example have the following characteristics and adjustments.

LoanOrigination BalanceAge-Adjusted Loss Rate (Percent)Estimated Age-Adjusted LossesGuarantee Amount Scaling Adjustment   (2/2.2) (Percent)Losses Adjusted for Overcollateral
1
$1,080,000
7.0
$75,600
90.91
   $68,727
2
$1,120,000
5.0
$56,000
90.91
   $50,909

2. If a pool of loans includes collateral pledged in excess of the guaranteed amount that is required under the terms of the transaction, you must further adjust the dollar losses as follows. Calculate the total losses on the subject portfolio of loans after age adjustments and any adjustments related to total submitted overcollateral as described in "1." above. Calculate the total dollar amount of contractually required overcollateral in the subject pool. Subtract the total dollars of contractually required overcollateral from the adjusted total losses on the subject pool. If the result is less than or equal to zero, input a loss rate of zero for this transaction pool in the Data Inputs worksheet of the RBCST. A new category must be created for each such transaction in the RBCST. If the loss rate after subtracting contractually required overcollateral is greater than zero, proceed to additional adjustment for the risk-reducing effects of the counterparty's general obligation described in "3." below.
3. Loans with a positive loss estimate remaining after adjustments in "1." and "2." above are further adjusted for the security provided by the general obligation of the counterparty. To make this adjustment, multiply the estimated dollar losses remaining after adjustments in "1." and "2." above by the appropriate general obligation adjustment factor based on the counterparty's whole-letter issuer credit rating by a nationally recognized statistical rating organization (NRSRO).
A. The following table sets forth the general obligation adjustment factors and their components by whole-letter credit rating (Adjustment Factor = Default Rate x Severity Rate).15
Whole-Letter Rating
Default Rate
(Percent)
Severity Rate
(Percent)
General Obligation Adjustment Factor
(Percent)
AAA
0.897
54
0.48
AA
2.294
54
1.24
A
2.901
54
1.57
BBB
7.061
54
3.82
Below BBB and Unrated26.827
54
14.50

B. The adjustment factors will be updated annually as Moody's annual report on Default and Recovery Rates of Corporate Bond Issuers becomes available, normally in January or February of each year. In the event that there is an interruption of Moody's publication of this annual report, or FCA determines that the format of the report has changed enough to prevent or call into question the identification of updated factors, the prior year's factors will remain in effect until FCA revises the process through rulemaking.
4. Continuing the previous example, the pool contains two loans on which Farmer Mac is guaranteeing a total of $2 million and with total submitted collateral of 110 percent of the guaranteed amount. Of the 10-percent total overcollateral, 5 percent is contractually required under the terms of the transaction. The pool consists of two loans of slightly over $1 million. Total overcollateral is $200,000, of which $100,000 is contractually required. The counterparty has a single "A" credit rating, and after adjusting for contractually required overcollateral, estimated losses are greater than zero. The net loss rate is calculated as described in the steps in the table below.

Loan A
Loan B
1
Guaranteed Volume
$2,000,000
2
Origination Balance of 2-Loan Portfolio $ 1,080,000 $ 1,120,000
3
Age-adjusted Loss Rate
7%
5%
4
Estimated Age-adjusted Losses
$75,600
$56,000
5
Guarantee Volume Scaling Factor
90.91%
90.91%
6
Losses Adjusted for Total Overcollateral
$68,727
$50,909
7
Contractually required Overcollateral on Pool (5%)
$100,000
8
Net Losses on Pool Adjusted for Contractually Required Overcollateral
$19,636
9
General Obligation Adjustment Factor for "A" Issuer
1.57%
10
Losses Adjusted for "A" General Obligation
$ 308
11
Loss Rate Input in the RBCST for this Pool
0.02%

A. The net, fully adjusted losses are distributed over time on a straight-line basis. When a transaction reaches maturity within the 10-year modeling horizon, the losses are distributed on a straightline over a timepath that ends in the year of the transaction's maturity.
* * * * *

4.1 Data Inputs
* * * * *
e. Weighted Haircuts for Non-Program Investments. For non-program investments, the stress test adjusts the weighted average yield data referenced in section 4.1 b. to reflect counterparty risk. Non-program investments are defined in § 652.5. The Corporation must calculate the haircut to be applied to each investment based on the lowest whole-letter credit rating the investment received from a NRSRO using the haircut levels in effect at the time. Haircut levels shall be the same amounts calculated for the general obligation adjustment factor in section 2.4 b.3.A. above. The first table provides the mappings of NRSRO ratings to whole-letter ratings for purposes of applying haircuts. Any "+" or "-" signs appended to NRSRO ratings that are not shown in the table should be ignored for purposes of mapping NRSRO ratings to FCA whole-letter ratings. The second table provides the haircut levels by whole-letter rating category.

FCA WHOLE-LETTER CREDIT RATINGS MAPPED TO RATING AGENCY CREDIT RATINGS
FCA Ratings CategoryAAAAAABBBBelow BBB and Unrated
Standard & Poor's Long-TermAAAAAABBBBelow BBB and Unrated
Fitch Long-TermAAAAAABBBBelow BBB and Unrated
Moody's Long-TermAaaAaABaaBelow Baa and Unrated
Standard & Poor's Short-TermA-1+A-1A-2A-3SP-3, B, or Below and Unrated
SP-1+SP-1SP-2
Fitch Short-TermF-1+F-1F-2F-3Below F-3 and Unrated
Moody'sPrime-1 MIG1 VMIG1Prime-2 MIG2 VMIG2Prime-3 MIG3 VMIG3Not Prime, SG and Unrated
Fitch Bank RatingsABCDE
A/BB/CC/DD/E
Moody's Bank Financial Strength RatingABCDE
FARMER MAC RBCST MAXIMUM HAIRCUT BY RATINGS CLASSIFICATION
Ratings ClassificationNon-Program Investment Counterparties (excluding Derivatives)
(Percent)
Cash
0.00
AAA
0.48
AA
1.24
A
1.57
BBB
3.82
Below BBB and Unrated
      14.50
1. Certain special cases will receive the following treatment. For an investment structured as a collateralized obligation backed by the issuer's general obligation and, in turn, a pool of collateral, reference the Issuer Rating or Financial Strength Rating of that issuer as the credit rating applicable to the security. Unrated securities that are fully guaranteed by Government-sponsored enterprises (GSE) such as the Federal National Mortgage Corporation (Fannie Mae) will receive the same treatment as AAA securities. Unrated securities backed by the full faith and credit of the U.S. Government will not receive a haircut. Unrated securities that are not fully guaranteed by a GSE will receive the haircut level in place at that time for "Below BBB and Unrated" investments unless the Director,at the Director’s discretion, determines to apply a lesser haircut. In making this determination, the Director will consider the risk characteristics associated with the structure of individual instruments.
2. If portions of investments are later sold by Farmer Mac according to their specific risk characteristics, the Director will take reasonable measures to adjust the haircut level applied to the investment to recognize the change in the risk characteristics of the retained portion. The Director will consider relevant similar methods for dealing with capital requirements adopted by other Federal financial institution regulators in similar situations.
3. Individual investment haircuts must then be aggregated into weighted-average haircuts by investment category and submitted in the "Data Inputs" worksheet. The spreadsheet uses these inputs to reduce the weighted-average yield on the investment category to account for counterparty insolvency according to a 10-year linear phase-in of the haircuts. Each asset account category identified in this data requirement is discussed in section 4.2, "Assumptions and Relationships."
* * * * *

4.2 Assumptions and Relationships
* * * * *
b. * * *
(3) Elements related to income and expense assumptions. * * * These parameters are the gain on agricultural mortgage-backed securities (AMBS) sales, miscellaneous income, operating expenses, reserve requirement, guarantee fees and loan loss resolution timing.
* * * * *
(C) The stress test assumes that short-term cost of funds is incurred in relation to the amount of defaulting loans purchased from off-balance sheet pools. The remaining unpaid principal balance on this loan volume is the origination amount reduced by the proportion of the total portfolio that has amortized as of the end of the most recent quarter. This volume is assumed to be funded at the short-term cost of funds and this expense continues for a period equal to the loan loss resolution timing period (LLRT) period minus 1. We will calculate the LLRT period from Farmer Mac data. In addition, during the LLRT period, all guarantee income associated with the loan volume ceases.
(D) The stress test generates no interest income on the estimated volume of defaulted on-balance sheet loan volume required to be carried during the LLRT period, but continues to accrue funding costs during the remainder of the LLRT period.
(E) You must update the LLRT period in response to changes in the Corporation's actual experience with each quarterly submission.
* * * * *

4.4 Loan and Cashflow Accounts
The worksheet labeled "Loan and Cashflow Data" contains the categorized loan data and cashflow accounting relationships that are used in the stress test to generate projections of Farmer Mac's performance and condition. As can be seen in the worksheet, the steady-state formulation results in account balances that remain constant except for the effects of discontinued programs, maturing Off-Balance Sheet AgVantage positions, and the LLRT adjustment. For assets with maturities under 1 year, the results are reported for convenience as though they matured only one time per year with the additional convention that the earnings/cost rates are annualized. For the pre-1996 Act assets, maturing balances are added back to post-1996 Act account balances. The liability accounts are used to satisfy the accounting identity, which requires assets to equal liabilities plus owner equity. In addition to the replacement of maturities under a steady state, liabilities are increased to reflect net losses or decreased to reflect resulting net gains. Adjustments must be made to the long- and short-term debt accounts to maintain the same relative proportions as existed at the beginning period from which the stress test is run with the exception of changes associated with the funding of defaulted loans during the LLRT period. The primary receivable and payable accounts are also maintained on this worksheet, as is a summary balance of the volume of loans subject to credit losses.

4.5 Income Statements
a. Information related to income performance through time is contained on the worksheet named "Income Statements." Information from the first period balance sheet is used in conjunction with the earnings and cost-spread relationships from Farmer Mac supplied data to generate the first period's income statement. The same set of accounts is maintained in this worksheet as "Loan and Cashflow Accounts" for consistency in reporting each annual period of the 10-year stress period of the test with the exception of the line item labeled "Interest reversals to carry loan losses" which incorporates the LLRT adjustment to earnings from the "Risk Measures" worksheet. Loans that defaulted do not earn interest or guarantee and commitment fees during LLRT period. The income from each interest-bearing account is calculated, as are costs of interest-bearing liabilities. In each case, these entries are the associated interest rate for that period multiplied by the account balances.

* * * * *

Dated: May 28, 2008

Roland E. Smith,
Secretary,
Farm Credit Administration Board.










_____________________________
15Emery, K., Ou S., Tennant, J., Kim F., Cantor R., "Corporate Default and Recovery Rates, 1920 – 2007," published by Moody’s Investors Service, February 2008 – the most recent edition as of March 2008; Default Rates, page 24, Recovery Rates (Severity Rate = 1 minus Senior Unsecured Average Recovery Rate) page 20.