APPENDIX I Key Risk Factors in CRT Markets
Market risk largely affects sellers of single tranche CDOs (buyers of protection). Hedging the short credit risk positions requires selling protection periodically on the names referred in the CDO in credit markets. The originator/seller of the CDO is exposed to risk associated with changes in the credit spreads of names in the underlying portfolio of the tranche. However, any dynamic hedging techniques entered into by the seller would only be effective if the instruments used (CDSs, corporate bonds, CDS indices) are sufficiently liquid. Thus, the seller is often exposed to liquidity risk as well, which could exacerbate losses in the event of market stress.
Liquidity risk arises from the difficulty of selling an asset quickly in an insufficiently liquid secondary market. This risk tends to be higher for new or rapidly growing markets, compared to well-established, mature ones. The secondary market for customized CDOs is non-existent. One potential concern is that some participants in credit derivatives markets may overestimate the liquidity of these products in constructing and hedging their correlation-based portfolios. The issue is not whether occasional losses are sustained by well-diversified large investors, but rather that one of the key player’s portfolios may be too highly concentrated in these instruments. While CDS indices have a certain level of liquidity due to their standard nature, they are not perfect hedges, and thus give rise to basis risk.
Basis risk arises when the hedging instruments available are not perfectly matched to the risks that are to be transferred. Indeed, most hedges are imperfect. For example, standardized tranches on credit derivatives indices are increasingly being used by arrangers to hedge their exposure to single- (or “bespoke”) tranches.65 Given that the attachment points of the tranches of a substantial number of CDO structures, which determine the potential losses faced by CDO investors and arrangers, are different from the attachment points of standardized tranches, these market participants would be exposed to basis risk. This risk is particularly high at the more subordinated end of the capital structure, namely, equity and lower-rated mezzanine tranches, where idiosyncratic factors tend to be important.66Furthermore, arrangers may not be able to hedge unanticipated event risk arising from a sudden default as experienced in the Delphi and Collins & Aikman bankruptcies.
Counterparty risk arises from the possibility that the risk buyer (seller of protection) may default in settling a claim. A fully-funded CDO structure incurs higher costs but is less-exposed to counterparty risk.67 The concentration of counterparty risk may be unavoidable during periods of rising credit risk; in some large CDO transactions, it may become prohibitively expensive to fund the credit risk transfer beyond a certain point (Rule, 2001). Thus, buyers of unfunded protection must assess whether the protection seller would be able to pay up during an extreme credit event, that is, determine the correlation between the realization of credit events and the creditworthiness of the protection seller.
Market imperfections also work against fully realizing the benefits from credit risk transfer. In particular, the lack of comprehensive legal and supervisory frameworks, the still-developing trading infrastructure, and rating arbitrage could contribute to financial instability in the event of an adverse shock in the credit markets.
Legal risk in the relatively new CRT market has largely arisen from imperfect documentation, although the maturing of the market has resulted in the setting of precedents over time. Legal risk is fundamentally important since legal and documentary issues are key considerations in defining the roles of the different parties involved in a CDO structure, and ensuring the efficiency and validity of the risk transfer itself (Cousseran and Rahmouni, 2005). Unsound market practices such as incomplete documentation of trades in the credit derivatives market are key examples of legal risk. In the United Kingdom, the FSA recently asked banks for an update on delays and errors that are occurring in processing credit derivatives trades. The authorities have expressed concern about the apparent high levels of unsigned trade confirmations between credit derivatives counterparties in transactions conducted outside of exchanges.68
The inadequacy of the trading infrastructure is evident by the substantial backlog for processing trades. However, a number of private sector initiatives have been advanced to deal with this problem. Recently, 14 leading investment banks pledged to U.S. and European regulators to address the issue backlogs in trade processing, and to improve operational practices. These initiatives include commitments to (i) reduce backlogs, (ii) provide regulators with information on the progress, (iii) use a new procedure for or transferring trades according to the ISDA protocol produced in September, which requires consent before trading. Additionally, there are plans to increase automation of trade processing via the Depository Trust and Clearing Corporation, and a proposal that cash settlement become the standard and that offsetting trades between the same parties are cancelled.
Ratings risk arises from the fact that the structured nature of CDOs limits the usefulness of their ratings, since ratings only reflect certain aspects of their credit risk. Ratings reflect the average risk of a security, and merely represent an opinion on the probability of default and expected loss. They do not factor in the dispersion of risk around its mean (Cousseran and Rahmouni, 2005), nor can they convey the complexity of a structure or its sensitivity to embedded assumptions, for example, default correlations and recoveries post-default (Miles, 2005). In cases where investors rely on ratings for their CDO investments, model risk is also related to the specific model the rating agency uses to size the credit enhancement for a given tranche and rating (Fender and Kiff, 2004).69
Different methodological approaches used by to rate CDOs could result in “ratings shopping” activity. For instance, Moody’s ratings are based on the concept of expected loss, while Standard and Poor’s and Fitch base their ratings on probabilities of default, which could result in clear differences in the ratings assigned by the agencies to certain tranche structures (Peretyatkin and Perraudin, 2002). Thus, CDO issuers may be tempted to choose rating agencies based on which one assigns the highest rating to their particular issue or tranche, in order to minimize funding costs.70 One possible solution is to encourage investors to require more than one rating as part of their internal control procedures.
Model risk arises in the valuation of CDOs using myriad complex models that continue to evolve. A major shortcoming of existing models is that they do not adequately capture the co-movements of credit spreads and default correlations, leading to some simplifications in modeling the correlation structure. These simplifications, however, may be incorrect (see Box 1). Essentially, model-based prices cannot replicate observed market prices and the associated correlation skew, and hence, may be misleading for assessing risks. Moreover, since market prices are only available for tranches of CDS indices, these cannot be used for valuing synthetic CDOs with non-standard underlying portfolios.
APPENDIX II Estimating the Exposure of U.K. Financial Institutions to the Credit Derivatives Market
The choice of factors in our model is guided by the requirement that the econometric model captures both systemic risk in the financial system and the specific risk associated with credit derivatives products. Market prices are used in our analysis, since they are readily available on a daily basis (as opposed to accounting data), and quickly transmit financial information about individual companies.71 Equity prices for the major financial groups are obtained from Bloomberg L.L.P. The slope of the yield curve, measured as the difference between the yields on the 10-year and 2-year U.K. government bonds, is included as a measure of contemporaneous economic conditions that would lead to simultaneous movements in equity returns and structured credit product prices. The yield data are obtained from the generic 2-year and 10-year yield series constructed by Bloomberg L.L.P.
While there has been a rapid proliferation of CRT products, credit derivatives are among the most widely used products. The introduction and rapid acceptance of benchmark credit derivatives indices, specifically iTRAXX in Europe, has helped develop a two-way market for standardized CDOs. Given that the tranche seniority of a CDO affects its riskiness, we include as factors the prices of the equity tranche and a number of mezzanine tranches with varying degrees of seniority.72 The super-senior tranche is not included in the analysis since its time-series just started in mid-2004. Price data for the different tranches are obtained from JPMorgan.
Given the vector of n endogenous variables, Yt=(y1t, y2t,…, ynt)’, the corresponding unrestricted VAR system of order p is given by:
where c is a n-vector of constant terms, Φi (i=1,…,p) are n-by-n coefficient matrices and εt is a vector of uncorrelated, independent and identically distributed error terms. The error terms are also serially uncorrelated. Under certain technical conditions, described in detail in econometrics texts like Hamilton (1994), the vector autoregression system in equation (1) admits the following vector moving average representation (VMA):
Equation (3) suggests that variance decomposition can be used to quantify the overall importance of innovations to variable yj for explaining subsequent realizations of variable yi vis-à-vis the other endogenous variables. Specifically, the overall importance of variable yj is captured by the relative share of the variance of variable yi it explains:
In interpreting the results, we do not make any assumption as to whether a particular institution is long or short the credit exposure. We assume that losses fall within the attachment points in case of defaults. Therefore, the higher the fraction of equity return volatiliy explained by a senior tranche, the lower the credit exposure of the firm and the potential impact on financial stability.
Carhill, Michael, 2005, “Supervising Securitizations,” presented at the at the High-Level Seminar on Asset Securitization and Structured Finance: Benefits, Risks and Regulatory Implications, International Monetary Fund (Washington, April).
Clementi, David, 2001, “Recent Developments in Securities Markets and the Implicaitons for Financial Stability,” presented at the Euromoney International Bond Congress, London.
Cousseran, Olivier and Imene Rahmouni, 2005, “The CDO Market: Functioning and Implicationss in Terms of Financial Stability,” in Financial Stability Review, (Paris: Banque de France).
Duffie, D., and N. Garleanu, 2001, “Risk and Valuation of Collateralized Debt Obligations,” Financial Analysts Journal 57, No. 1, pp. 41-59.
European Cental Bank, 2004, “Features of CRT Markets,” in Credit Risk Transfer by EU Banks: Activities, Risks and Risk Management (Frankfurt, May).
Fender, Ian and John Kiff, 2004, “CDO Rating Methodology: Some Thoughts on Model Risk and Its Implications,” BIS Working Papers No. 163 (Basel: Bank for International Settlements).
Fitch Ratings, 2005, “Global Credit Derivatives Survey: Risk Dispersion Accelerates,” Financial Institutions Special Report (London and New York, November).
Gibson, Michael S., 2004, “Understanding the Risk of Synthetic CDOs,” Working Paper (Washington: Board of Governors of the Federal Reserve).
Hasbrouck, Joel,, 1991b, “The Summary Informativeness of Stock Trades: An Econometric Analysis,” Review of Financial Studies, Vol. 4, pp. 571–95.
International Association of Insurance Supervisors, 2003, “IAIS Paper on Credit Risk Transfer between Insurance, Banking and Other Financial Sector,” presented at the Financial Stability Forum, Bank for International Settlements (Berlin, March).
International Monetary Fund, 2006, Global Financial Stability Report, World Economic and Financial Surveys (Washington, forthcoming)
Memani, Krishna, 2005, “Asset Securitization and Structured Finance,” presented at the High-Level Seminar on Asset Securitization and Structured Finance: Benefits, Risks and Regulatory Implications, International Monetary Fund (Washington, April).
Miles, Colin, 2005, “Structured Finance,” presented at the High-Level Seminar on Asset Securitization and Structured Finance: Benefits, Risks and Regulatory Implications, International Monetary Fund (Washington, April).
Monks, Allan and Marco Stringa, 2005, “Inter-Industry Linkages between U.K. Life Insurers and U.K. Banks: An Event Study,” in Financial Stability Review (London: Bank of England).
Peretyatkin, Vlad and William Perraudin, 2002, “EL and DP Approaches to Rating CDOs and the Scope for ‘Ratings Shopping’,” in M. K. Ong (ed.), Credit Ratings — Methodologies, Rationale and Default Risk (London: Risk Books).
Rule, David, 2001, “Risk Transfer between Banks, Insurance Companies and Capital Markets: An Overview,” in Financial Stability Review, (London: Bank of England).
Rule, David, Adrian Garratt and Ole Rummel, 2004, “Structured Note Markets: Products, Participants and Links to Wholesale Derivatives Markets,” in Financial Stability Review, June 2004 (London: Bank of England).
Wagner, Wolf and Ian W. Marsh, 2004, “Credit Risk Transfer and Financial Sector Performance,” Discussion Paper No. 4265 (London: Center for Economic Policy Research).
Prepared by Jorge A. Chan-Lau and Li Lian Ong (both MFD).
London is the main center of the global credit derivatives market, ahead of even New York. The size of the London market is estimated to have reached $2.2 trillion in 2004, of the total global size of about $5 trillion—about 44 percent, compared to the New York market at 40 percent.
Examples of credit derivatives include credit default swaps (CDSs), credit-linked notes (CLNs), credit spread options (CSOs) and total return swaps (TRSs).
Rule, Garratt and Rummel (2004) define a structured credit product as “a bond combined with one or more options or forwards linked to market prices or indices” which can “take a variety of contractual forms depending largely on the nature of the target investors.”
A tranche is defined as a certain loss range.
See Box 1.
CDOs issued in Europe targeted at European investors may also include U.S. underlying assets.
These techniques, to a large extent, have been borrowed from mortgage backed securities (MBS). Lessons learned in this market have been transferred to the credit derivatives market.
As an example, credit risk in the banking sector spiked up in early May 2005 on rumors that hedge funds active in credit derivative markets might have incurred large losses following the ratings downgrade of automobile companies General Motors (GM) and Ford. The influence of hedge funds on the banking industry is largely due to the substantial contribution of hedge funds to investment banks’ fee income. The banks generate these fees by providing trading ideas, financing positions, and executing trades on behalf of hedge funds. Hence, factors affecting hedge funds’ performance affect banks’ profitability. For instance, idiosyncratic shocks that reduce hedge funds creditworthiness increase counterparty risk to banks involved in financing these funds’ positions. It was not entirely surprising, then, that banks’ equity prices fell and credit spreads widened when hedge funds investment strategies underperformed following the automobile companies’ downgrades.
See Appendix I for a detailed discussion on the individual risk factors. It should be noted that these risks are not unique to CRT instruments.
It should be noted that rating agencies are constantly refining the rating criteria applied to structured products.
In this model, the shocks are not orthogonalized. This means that the variance decomposition ranks the importance of every shock, but does not represent the actual percentage that each shock contributes to a particular share price, since the shocks may be correlated. In other words, the sum of the individual variances would not be equal to the total variance because of the covariance terms, but the rankings hold since it is equivalent to a renormalization.
The risk levels of a synthetic CDO are determined by the total accumulated loss of the reference pool of assets. In a CDO, the default losses borne by a tranche range between the attachment point and detachment point. The lower bound of the range is called an attachment point and the upper bound a detachment point. For example, a 3-6 percent tranche has an attachment point of 3 percent and a detachment point of 6 percent. When the accumulated loss of the reference pool is no more than 3 percent of the total initial notional of the pool, the tranche will not be affected. However, when the loss has exceeded 3 percent, any further loss will be deducted from the tranche’s notional until the detachment point 6 percent is reached. At this point, the tranche is wiped out.
The data are subsequently divided into two sub-periods—August 28, 2003 to September 6, 2004, September 7, 2004 to September 15, 2005—and the VAR approach is applied to each sub-sample. The results suggest that the major U.K. financial institutions have become more conservative in their involvement in credit derivatives over time. These institutions became more exposed to volatility in the more senior mezzanine tranches over the two sub-periods, with the exception of RBS. The holdings across institutions remained diverse over time.
In the United States, the regulation and supervision of banks are undertaken by the Federal Reserve and the Office of the Comptroller of Currency, as well as by individual state regulators and supervisors.
This refers to the issuing bank’s equity piece of a securitized instrument.
The approach to supervision and regulation taken by the authorities in the United States and the FSA in the United Kingdom is different, reflecting the differences in the structure and history of the banking system in each country.
International initiatives on surveillance of the credit derivatives market include the Joint Forum and the Financial Stability Forum (see Joint Forum, 2004)
The United Kingdom adopted IFRS on 1 January 2005. The European Commission brought in fair value accounting for derivatives on 16 November 2005. The Commission initially accepted most of the proposals presented by the International Accounting Standards Board in 2004, but opted out of its recommendations on fair value accounting and on hedge accounting. It then drew up a restricted fair value option, which was approved in July 2005, and came into force on 16 November 2005, retroactive to 1 January 2005.
The financial institution is required to fully disclose the criteria for which the instrument is used, and to detail how the instrument supports its investment or hedging strategy. Disclosures of details on credit derivatives is covered by IFRS 7, Financial Instruments Disclosures.
As an example, the Dow Jones iTraxx index was created from the merger between two existing CDS indices (TRAC-X and iBoxx) in June 2004. Standardized tranches of CDOs based on the iTraxx—which replicate the behavior of unfunded synthetic CDO tranches whose reference portfolio comprises names in the basket of the CDS index—were issued. These tranche prices are continuously quoted, and have contributed to improving the transparency and liquidity, and lowering the trading costs of the CDO market. Box 1 briefly discusses single-tranche CDO structures.
Counterparty exposure is usually greater for synthetic CDOs, where the risk transfer is usually unfunded. Further, these exposures could potentially increase very sharply if the creditworthiness of the counterparty deteriorates quickly.
When assignments of trades are effected without the correct notifications and/or consents, this could result in institutions having out-of-date or inaccurate information on their exposures to individual counterparties. Tracking down these assignments has contributed to the existing confirmations backlog.
The asset pool for a CDO may sometimes need to be enhanced by one or more types of credit in order to attain the desired credit risk profile for the security being issued. Such enhancements are usually derived from internal sources, that is, they may be generated from the assets themselves or supplied by a third party.
The equity tranche has attachment points of 0-3 percent, and the mezzanine tranches has the following attachment points: 3-6 percent, 6-9 percent, 9-12 percent, and 12-22 percent.