Standard & Poor's Ratings Services is updating the way it assesses the capital adequacy of financial institutions worldwide. So far, this update has taken the form of proposed new charges for certain asset classes for U.S. financial institutions. In addition, Standard & Poor's is developing new approaches to assessing capital adequacy that use company-specific data in capital charges and allow issuers to discuss with Standard & Poor's the way they manage risks internally. Standard & Poor's is soliciting comments from market participants about this evolving methodology.
Proposal Summary
Standard & Poor's primary objective in developing capital models is to establish a robust analytic process for assessing the risks of the financial institutions we rate. Although a number of critical attributes contribute to the overall credit rating on an institution, a key component of that analytical structure is assessing the financial institution's capital requirements.Standard & Poor's approach to assessing a financial institution's capitalization produces a capital adequacy ratio, which compares adjusted total equity with risk-charge assessments consistent with an 'A' category rating. The main objective of our capital model is to compare capitalization ratios among financial institutions consistently across geographies and types of institution. The proposed model will be risk sensitive, reflecting credit risk, market risk, asset/liability risk, and operational risk.
The capital adequacy ratio is only a starting point for judging capital adequacy. Qualitative and quantitative enhancements are applied as warranted to derive a more complete picture of a financial institution's capital position. The analyst plays a critical role in adjusting the model to best assess risks that are unique to a company while maintaining a standard of comparability among companies.
Ratings Impact
Although considerable attention has been focused on risk-based capital ratios, our assessment of capital adequacy is only one of many factors (see chart) used in arriving at a credit rating. Our rating process will continue to be based on the belief that capital adequacy ratios are not a substitute for a broad-based analysis of a financial institution's credit quality. Strength or weakness in other key areas—such as a company's market position, management and strategy, credit risk, liquidity risk, earnings performance, enterprise risk management, and financial flexibility—can more than offset relative strength or weakness in capital adequacy.
Response Deadline
We request that interested parties submit their comments by Feb. 8, 2007, by E-mailing them to criteriacomments@standardandpoors.com.
Philosophy And Approach
In recognition of the diverse quality of information available across financial institutions—whether because of required public disclosures, regulatory requirements, or the institution's willingness to share confidential information—our assessment and analysis process will use publicly available information and that which is only disclosed confidentially.
Basel I and Basel II regulatory capital adequacy ratios, though important in assessing capital adequacy, have their shortcomings. Basel II in particular—despite its risk-sensitive framework, which is a clear improvement compared with Basel I—will not allow consistent comparisons because of the significant impact of alternative methodological choices, national discretions, and banks' specific risk assessments. For these reasons, as well as our differences in risk assessment, we will continue to rely primarily on our own capital measures. Clearly, qualitative and quantitative enhancements need to be applied to derive a more complete picture of an institution's capital position compared with those of its peers. We believe global consistency and comparability add significant value to our approach.
The tools that we will use to measure financial institutions' capital adequacy will be updated continually. The evolution of these metrics will reflect the improving availability of data (both public and institution-specific), the willingness of the rated institutions to disclose proprietary information on an ongoing basis, the nature and extent of the risks undertaken by these institutions, and the evolution of the techniques and methodologies used to quantify and evaluate these risks.
Standard & Poor's will employ the following hierarchy of approaches to assessing and evaluating capital adequacy:
Base case assessment (BCA). This approach (described more fully later) will assign flat charges to each asset class and region and will apply to all financial institutions.
Specific case assessment (SCA). This approach is intended to customize the capital charges based on institution-specific data.
Economic capital assessment (ECA). We will develop this approach over the next two years. It is intended to provide institutions an opportunity to discuss with Standard & Poor's the evaluation of their internal economic capital assessment framework.
For institutions in the U.S. where data is available, this approach will be applied over the coming year. For institutions elsewhere in the world, for data availability reasons, we will probably need to wait for Basel II enhanced Pillar III disclosure to apply industry capital charges that will be calibrated to account for regional specifications. The underlying analytical concepts and methodology (a combination of worst-case, statistical analysis and expert knowledge), however, will be similar to those applied in the U.S. to ensure consistency in our underlying framework.
Standard & Poor's is aware that institutions have allocated extensive time and effort in building and fine-tuning their capital models. We welcome the opportunity to engage in constructive and fruitful dialogue to better understand the quality and framework of their approach to assessing economic capital requirements.
Fundamental Building Blocks And Open Issues
Each of the approaches will build on the previous method and assess a capital charge for market risk, credit risk, and operational risk. In addition, other material risks—such as strategic/business risk, liquidity risk, and country risk—will be considered both qualitatively and quantitatively. The total capital requirements would then be an aggregation of these individual charges.
Different types of financial institutions have different proportionate exposures to these risks, and such characterizations will be incorporated into the analysis. Securities broker/dealers, for example, have relatively high degrees of market and operational risks but smaller amounts of credit risk, whereas banks and finance companies tend to have most of their capital allocated to credit risk. Currently, no diversification benefits will be attributed across risk types for the BCA approach, as this is not trivial to assess from purely publicly disclosed data. As part of the SCA approach, it is important to note that Standard & Poor's is open to discussions with institutions to better understand their approach and internal process for estimating diversification benefits both within and across risk types.
Currently, assessing an operational risk capital charge is still the most difficult, and Standard & Poor's has opted to use an approach that is similar in construct to Basel's Basic Indicator Approach and Standardized Approach. If an institution adopts the AMA (advanced measurement approach) as described by Basel, Standard & Poor's will review and assess the robustness and integrity of that process as part of its overall capital adequacy assessment and would be willing to recognize a risk-mitigation impact of operational risk insurance to the operational risk capital charge. Standard & Poor's will explore alternative approaches to assign a capital charge for this risk type within the coming year.
Some of the broad issues that are still under review are those related to structural interest rate risk and liquidity risk. Standard & Poor's will revise its current capital methodology and will provide commentary to its constituents for feedback and comments.
Standard & Poor's has re-addressed a capital charge for the credit risk component for a set of U.S. asset classes that covers about 70% of the loan portfolio of large U.S. commercial banks. This covers the larger and more important set of asset classes, and for the remaining 30% of the assets, Standard & Poor's existing capital charges will apply. An enhanced approach to assessing the capital charges for these asset classes will also be addressed over the coming year and will be communicated to our constituents for comment prior to being implemented.
BCA Approach
Credit risk capital charge
In assessing a credit risk capital charge based on publicly available information, we started with the underlying premise of estimating a potential worst case or total loss for key asset classes that are described later. In addition to this metric, we analyzed a number of other statistical measures, including distribution fitting to the data. In each case, the view is that capital will be required to be in place to offset these potential scenarios.
We would not be in a position to estimate the true loss distribution. In addition, the net charge off (NCO) can be viewed as probability of default multiplied by loss given default, and in North America, such data is available at the industry level for each of the asset class that we revisited. Therefore, Standard & Poor's performed a statistical analysis of this data and coupled that information with expert judgment of our analysts to arrive at a credit risk capital charge for each of the specified asset class.
As previously mentioned, these flat capital charges per asset class will be applied to all institutions. With this approach, at this stage we are not making a distinction between expected and unexpected losses. However, in this analysis, we do recognize the availability of reserves to offset expected losses. In the more advanced models that we will develop, we will model probability distributions that will result in estimates of expected losses and the associated volatility statistics.
Flat capital charges for each Basel II asset class will be calibrated, based on prior Standard & Poor's loss experience and expert judgment validated in committees. These will be used in a more general framework, which may also factor average industry probability of defaults and loss given defaults reported by banks as part of their Pillar III disclosure requirements. However, Standard & Poor's will need to wait for Pillar III implementation to start using the model, probably sometime between the second half of 2007 and 2008 depending on countries and banks. To ensure global consistency, such an approach will eventually converge with that used in the U.S. once Basel II comes into force.
In the U.S., where NCO data is available for broad asset classes at the industry level, Standard & Poor's has established credit risk capital charges for the following asset classes:
Auto loans.
Other consumer loans.
Commercial and industrial loans.
Commercial real estate loans. These loans are further subdivided into three categories: construction, multifamily, and all other.
Credit cards.
Derivatives counterparty exposure.
Residential mortgages.
To estimate a capital charge, Standard & Poor's used three statistical measures coupled with the analysts' expert judgment to arrive at a base case capital charge. As indicated, the analysis was based off publicly disclosed NCOs for the asset classes that were subject to this analysis.
The first is the worst-case cumulative NCOs for a given time period. In other words, based on the historically available NCOs for a given asset class, the largest cumulative NCOs for a period of one to five years was considered as being the appropriate capital charge. The second method used a quantile-type cut-off based off an empirically fitted non-normal distribution. The third method for estimating the capital charge was based off a six-sigma-type estimate.
Standard & Poor's realizes that each method has its mathematical limitations. As a result, we decided to triangulate across various measures coupled with Standard & Poor's analysts' opinions to arrive at a reasonable estimate for a base-case capital charge. Subjective judgments by our analysts were based on trends in losses, relative capital charges across asset classes, and economic capital models currently in use at financial institutions.
Auto and other consumer loans. In our review of auto loans as well as other consumer loans, we collected FDIC data and reviewed GAAP data from auto lenders. We also have reviewed economic capital models from major lenders.
In reviewing U.S. GAAP data, we used the annual NCOs for each of the prime and sub-prime subsectors of the auto lending industry. Data was collected for 32 lenders considered to be prime lenders and 27 subprime lenders. The data are for individual companies that are active loan securitizers. Using the data, the worst-case scenarios were derived by selecting the actual worst loss year for each of the two subsectors.
We also reviewed data from the FDIC for banks with more than $1 billion in total assets in the Other Consumer Loans category. We obtained the net charge-offs for 20 years of quarterly data and fitted a non-normal distribution to the one-year (rolling four-quarter) losses and to the three-year (rolling 12-quarter) cumulative losses.
Armed with the data from the individual company's GAAP losses, the FDIC data, and what we know from the economic capital models of various consumer finance companies, we decided that the appropriate charge should be 5% for prime auto loans, 15% for sub-prime auto loans, and 6% for the category All Other consumer loans.
As with all asset classes, these base-case capital charges will be scaled upward or downward in cases where a financial institution's loss experience varies materially from industry averages.
Commercial and industrial (C&I) loans. We used the FDIC data to determine quarterly net charge-offs for the past 20 years. Under the assumption that banks do not always recognize losses in their accounts as they occur economically, we determined that a three-year cumulative period was appropriate to look at losses that would arise in a worst-case scenario (i.e., through a severe economic downturn, it could take as long as three years for a bank to fully recognize all the losses in a C&I portfolio). We also fitted a non-normal probability distribution to rolling four-quarter NCOs.
Based on the results from these two measures and considering that C&I loan losses are likely to increase sharply, the committee decided on 8% as the appropriate base-case capital charge.
Commercial real estate. Commercial real estate loans were subdivided into three types: construction loans, multi-family loans, and all other. We reviewed both quarterly and annual charge-off data for commercial real estate portfolios. We also looked at whether there were important differences in the size of the institution. To the extent there was, we chose the more conservative methodology.
One method was to find the worst three-year and five-year periods of NCOs for the 100 largest banks. In addition to the FDIC data, we analyzed publicly disclosed data on Other Real Estate (ORE) write-downs for the top 30 banks in the five years for 1990-1994. The banks did not breakout their ORE write-downs by property type, so we apportioned ORE write-downs to property types. In total, ORE write-downs increased the loss rate by 50% on average. The large amount of ORE write-downs is because of the banks' practice of aggressively classifying loans as foreclosed, even when that was not the case in a legal sense. As real estate prices deteriorated, further loss recognition was recorded as a non-interest expense rather than as a charge against reserves.
Construction, multi-family, and All Other CRE loans. The results obtained from fitting a non-normal distribution to the NCOs in the case of construction loans was meaningless and omitted from the analysis. After triangulation and incorporating analysts' opinions, we decided on a 20% charge for construction and a 13% capital charge for each multi-family loan and all other CRE loans. Credit cards. In addition to FDIC data, we reviewed quarterly NCO data collected for managed portfolios from the first quarter of 1997 through the first quarter of 2006 on a 12-month lagged basis. Lagged NCOs were used instead of coincident NCOs, as rapid growth in the industry can hide credit problems. However, the results were similar to the FDIC data over the same period. Based on these analyses, we decided on a capital charge of 8.5%. Derivatives counterparty exposure. In assessing a capital charge for the derivative asset class, we used a regression technique to estimate the potential future exposure for an institution's derivative portfolio based on quarterly U.S. GAAP disclosure for a suitably long period (at least five years) depending on data availability. In all cases, the numbers represented values after netting agreements and collateral were applied. This estimate of the potential future exposure was then bucketed into four broad credit qualities. An approximate probability of default for each bucket was then applied to the appropriate exposure to arrive at an expected loss estimate. Based on this analysis, a charge of 1.8% was applied. This will be applied on a case-by-case basis. Residential mortgages. The residential mortgage capital charge is based on our review of FDIC data. The worst three-year cumulative loss experience (1992, 1993, and 1994) adds up to a net charge-off rate of 92 basis points, which we rounded up to 1%. However, this does not capture the different risk characteristics of the individual banks' residential loan portfolios (FICO score ranges, loan-to-value, and other underwriting differences). Certain institutions might have mortgage portfolios with substantially different risk profiles than the average for the industry. To adjust for this, we will look at how a financial institution's loss experience deviates from industry norms and adjust the charge upward if there are meaningful differences.
Home equity lines of credit, because they are second mortgages, warrant a higher capital charge. We will use a charge of 3%. This is based on a review of the peak three-year cumulative nonperforming loans from FDIC data (banks with total assets of more than $1 billion).
We are also quite aware that sub-prime mortgage portfolios can generate substantially higher credit losses than prime portfolios. Therefore, we will increase our base-case capital charge for mortgage portfolios that, in our opinion, have credit characteristics that deviate significantly from industry averages.
Market risk capital charge
In the U.S., for example, we look at securities portfolios based on accounting under U.S. GAAP because it sets the stage for the information that is disclosed by companies and often indicates their uses. The three categories are securities held for trading, available for sale, and held to maturity. For trading portfolios, required capital will be a multiple of value-at-risk as reported by the company or a multiple of the standard deviation of principal transactions revenue. Both approaches have their strengths and weaknesses and are not the only methodologies that we might consider.
Value at risk (VaR) is a useful measure because it captures the market risks in long and short positions and in derivatives that might not even appear on the balance sheet. However, given the range of assumptions and the different portfolio re-evaluation techniques that an institution can employ in calculating its VaR, a market risk charge based on VaR might not be the best choice.
One alternative is to use the volatility of historical trading revenue for each firm. The key problem with this approach, however, is that the current risks of the firm could differ from those that were consistent with the trading portfolio's profits and losses over time. In addition, trading revenue reporting is not consistent across firms. Some include prime brokerage interest and fees, and others do not; some include commissions, but not all do.
Nevertheless, in the BCA approach Standard & Poor's has assigned capital for market risk based on the volatility of trading revenue (measured by the standard deviation) for the past several years. Although interest rate risk is a form of market risk, Standard & Poor's has decided to deal with it separately because in banks and finance companies, it exists separately from the trading activities. As Standard & Poor's engages in dialogue with financial institutions, it would propose to consider using alternative market risk capital measures such as, for example, one that is a function of the institution's 99th percentile conditional expected shortfall calculated from the past year's daily trading profit or loss.
Operational risk capital charge
For operational risks, we use a charge that would reflect the size of the financial institutions' operations. Our approach mirrors the basic indicator approach of Basel II. This is one of the most important risks. The large losses that can arise from legal costs, in particular, are Standard & Poor's primary concern. Settlements for lawsuits such as Enron and WorldCom are the largest examples we can think of. The BCA approach would apply a flat charge across all business lines.
A possible indicator that could be used for this purpose is revenues or expenses. However, using revenues or expenses as an indicator has its pros and cons, and to that extent, Standard & Poor's proposes to use a weighted measure that would be a combination of 8% of the average revenues from the prior three years and 12% of the average expenses over the same period, again tempered with analyst judgment. Clearly, this is an evolving measure, and we are open to alternatives to a flat charge and will develop more refined charges based on the operational risks unique to each business segment and the integrity of the institutions' operational risk management framework. In addition, there could be operational risk add-ons unique to certain businesses. For example, we are considering an add-on for securities trading based on the volatility of trading revenue similar to what was described above under "Market Risk Capital Charge."
The proposed capital charges for the BCA approach are based on publicly available data and U.S. GAAP disclosure and have been summarized below for the key asset classes.
Summary Of Proposed Capital Charges For Key Asset Classes For U.S. Financial Institutions BCA Approach
Credit Risk
Risk Category/Asset Type
Proposed Capital Charge (%)
Loans
Home equity*
3.00
First mortgage*
1 (prime), 5.00 (sub-prime)
Other commercial real estate*
13.00
Construction and land development*
20.00
Farmland
12.00
Commercial real estate mortgage*
13.00
U.S. C&I*
8.00
Foreign C&I
12.00
U.S. banks
1.50
Foreign banks
5.00
Acceptances
3.00
Agricultural loans
10.00
Credit card*
8.50
Other consumer*
6.00
Foreign government
15.00
Other loans*
5.00 (prime), 15.00 (sub-prime)
Leases
12.00
Securitized residential real estate
4.00
Securitized consumer loans
8.00
Securitized C&I
3.00
Securitized other
10.00
Commercial paper conduit
0.00
Off balance sheet
Commitments residential real estate
1.00
Commitments commercial real estate
10.00
Commitments credit cards
0.00
Commitments securities underwriting
20.00
Commitments other
1.50
Standby letter of credit
8.00
Commercial and similar letter of credit
8.00
Credit derivatives (net guarantor)*
1.80
Securities lent
0.75
Interest rate derivatives
0.02
Foreign exchange derivatives
0.06
Equity derivatives
0.15
Commodity and other derivatives
0.15
Assets under management—equities and fixed income
0.10 sliding scale
Assets Under Management - Money Market Mutual Funds
0.25 sliding scale
Assets under administration
0.00
Investments
Trading account
U.S. Treasuries
0.75
U.S. government agencies
0.75
FNMA pass-through
0.75
Other FNMA
0.75
Other MBS
0.75
Municipal securities
1.50
U.S. corporate debt
4.00
Other domestic debt
6.00
Foreign trading debt
6.00
Total trading
Held to maturity
U.S. Treasuries
0.75
U.S. government agencies
0.75
Total MBS
0.75
Total ABS
0.75
Municipal securities
1.50
U.S. corporate debt
4.00
Foreign debt
6.00
Available for sale
U.S. Treasuries
0.75
U.S. government agencies
0.75
Municipal securities
1.50
Total MBS
0.75
Total ABS
0.75
U.S. corporate debt
4.00
Foreign debt
6.00
Portfolio equities
0.00
Private equity
50.00
Other assets
Premises
10.00
Mortgage servicing rights
20.00
Accrued receivables
1.00
Deferred acquisition costs
100.00
Insurance risk
100.00
Operational risk*
8% of revenues plus 12% of expenses plus add-ons as needed
Market risk*
7x (function of volatility of trading revenues)
*Revised charge based on the BCA approach described in the article. The total is the required capital. Capital adequacy is ATE divided by required capital.
SCA Approach
The SCA approach is intended to customize the capital charge based on an institution's specific data. However, it is possible to use this approach in conjunction with the BCA approach to estimate and assess a capital charge for an institution. For example, based on data availability, the SCA approach could be used to assess a credit risk capital charge, while the BCA approach is used to estimate a market risk and operational risk capital requirement.
It is conceivable that for some institutions, the credit component of the BCA capital charge for some or all of the asset classes turns out to be unjustly conservative. In other cases, it could be too liberal. The SCA approach is designed to address these concerns. Over time, we believe that this hybrid approach would lead to a stronger convergence and comparability of capital requirements across banks and across regions.
ECA Approach
Over the next two years, we also intend to develop an economic capital model evaluation framework that will complement the two above approaches of our hierarchical framework. This will be the ECA approach, in which Standard & Poor's concentrates on reviewing the principles underlying the risk models, the quality of data, the appropriateness of the assumptions, and the ways in which the results were produced and integrated into the institution's overall ERM process. It is not the intent to either question or replicate the task of the regulators in this regard. The most important consideration will be for Standard & Poor's to understand whether all of a company's major risk exposures have been addressed appropriately. Key elements are whether the main risk drivers have been identified and whether the form and level of interaction among them is understood. Special attention is paid to risk-mitigation activities allowed in the model—in particular, the assessment of management actions to be undertaken in adverse circumstances.
Standard & Poor's will assess how a firm ensures that the data used in the model is representative of a company's portfolio, how deficiencies in the data have been dealt with, and the approaches used to aggregate and validate the data. We will review approaches used to set assumptions, as well as how they are stress-tested, and identify how the institution tackles key challenges in that process—in particular, the availability and quality of data used. Standard & Poor's will also compare an institution's assumptions with those used by its peers.
Analytical Adjustments
Our capital adequacy model creates a reasonably consistent initial approach to measuring financial institutions' capital adequacy. Still, results are primarily guideposts, not absolute benchmarks, by which to gauge capital adequacy. A vital part of the assessment of capital adequacy incorporates adjustments-both qualitative and quantitative-to the model.
These adjustments may consider:
An institution's ability to generate capital internally and self-fund growth through earnings. All else being equal, Standard & Poor's views companies with long track records of consistently good earnings as having a stronger capacity for reliable capital development than companies with more volatile performance. We also consider an institution's prospective growth plans in conjunction with management's commitment to maintaining or enhancing capital adequacy or running a leaner capital structure.
We consider potential calls on capital by affiliates that may look to the rated entity for future capital support or by a parent's potentially increasingly aggressive appetite for dividends. Conversely, a parent's, subsidiaries, or affiliate's ability to provide future capital support could have a positive effect on how we view an institution's capital strength.
