The global financial system’s institutional framework has been evolving over time.Every crisis has helped decipher a gap in the financial structure which is then fixed by the regulating authorities.It hasn’t been very often that the regulators were able to identify the gaps before the market identified it.This does not serve the purpose of existence of regulatory authorities.In future the role of regulatory authorities should be pro active in nature rather than reactive mode of undertaking corrective actions.
The subprime crisis which originated in the united states led to a global melt down which was severe.The mortgage market in the United States saw a tremendous growth in the initial years of the 21st century. Subprime borrowers started obtaining mortgages due to availability of cheap credit, lenient lending practices and appreciation in real estate values. These mortgages were inturn sold by the lenders to investment banks who packaged them into exotic securities and sold them to high risk taking investors seeking high returns.
Investors had faith in these packaged securities primarily because of Credit Rating Agencies’ (CRA) ratings of these securities as ‘investment grade’. In 2007, the tide turned and credit became expensive. Home values dropped. Majority of the subprime buyers started defaulting their loan payments. The CRAs rapidly downgraded all the securities for which they had given favourable ratings.
This dissertation is undertaken to understand the emergence of structured financial products, the rating process followed by the credit rating agencies for rating them and the mistakes done by the rating agencies, a major contributor to the subprime mess in the United States which had ripple effect across financial markets all over the world.
The following research papers and articles have been referred and reviewed in order to gain indepth knowledge about the work done about the dissertation topic under consideration. This would facilitate a clear understanding of different view points to the issue and enable a comprehensive analysis of the topic.
According to V.Gupta, R.K.Mittal & K.Bhalla (2010), low interest rates, abundant liquidity and a chase for yield led to the emergence of sub prime lending which was given undue support by the credit rating agencies. Credit rating agencies gave investment grade ratings to securitization transactions based on subprime mortgage loans. The CRAs combined lower rated mortgage loans with equity to form mezzanine CDO enabling a higher credit rating. Also CRAs used the same risk metric for assessment of all instruments. The CRAs assigned supersafe, triple-A ratings to structured products that later turned out to be extremely risky, and in some cases worthless. This has been illustrated with few examples of downgrades.The paper concludes that The regulatory framework should also facilitate the conduct of stress tests by users on key model parameters, and provide for the disclosure by credit rating agencies of the economic assumptions underlying their rating of structured products.
According to Katz and Salinas (2009), faulty credit ratings and the flawed rating process have been the key drivers to the financial crisis 2007-2008. While the easy availability of (what turned out to be flawed) ratings fueled the growth of thismarket, the subsequent downgrades in ratings accelerated the market’s collapse.The paper suggests that While corporate debt ratings are based on publicly available, audited financial statements, structured debt ratings are based on nonpublic, nonstandard, unaudited information supplied by the originator or nominal issuer. Moreover, rating agencies had no obligation to perform due diligence to assess the accuracy of the information and often relied on representations and warranties from the issuers about the quality of the data, which later proved to be inadequate. The researchers note that the credit rating agencies have always been slow to react to market events and a few examples have been quoted.Few measures suggested by the researchers include managing conflict of interest, better transparency, direct government oversight etc.
According to Fender and Kiff (2004) , rating od collateralised debt obligations involves assumptions such as default probability, recovery rates and correlated defaults of pool assets. The research paper analyses one of the rating methodologies used which is termed as Binomial Expansion Technique.A comparative analysis of this method and Monte corlo Simulation is done. The paper elaborates the implications of usage of different techniques on the rating outcomes. It finally discusses how methodological differences might induce issuers to strategically select rating agencies to get CDOS rated.
According to Barnett- Hart(2009), Collateralized debt obligations (CDOs) have been responsible for $542 billion in write-downs at financial institutions since the beginning of the credit crisis.The poor CDO performance has been attributed to inclusion of low quality collateral with exposure to U.S residential housing market.The role of CDO underwriters and credit rating agencies in the crisis have been discussed. The credit rating agencies failed to rate the performance of CDOS precisely due to over automation in rating methodologies and heavy reliance on input whose accuracy was not verified. The researcher concludes that by understanding the CDO market meltdown story more effective regulatory and economic policies and practices to prevent history from repeating itself in the future.
According to Securities and Exchange commission(2008), few observations about credit rating agencies with respect to CDOS have been made.SEC claims that few credit rating agencies could not deal with the substantial increase in the number and complexity of the CDOS since 2002. Rating agencies failed to document significant steps in rating of CDOS including reasons behind deviation from the models. Also the internal audit procedure of rating agencies varied significantly.The report summarises the remedial actions that the Nationally Recognised Statistical Rating Organisations(NRSRO) would take after the SEC examined them and came up with issues to be looked into. Under the new law and rules, NRSROs are required to make certain public disclosures, make and retain certain records, furnish certain financial reports to the Commission, establish procedures to manage the handling of material non-public information and disclose and manage conflicts of interest. The Commission’s rules additionally prohibit an NRSRO from having certain conflicts of interest and engaging in certain unfair, abusive, or coercive practices.
According to Partnoy (2008), Credit rating agencies have been the primary drivers of second level securitisation.Investors did not examine the underlying assets and depended on parameters set by rating agencies to assess the CDOS. If the Credit rating agencies had used reasonable and accurate models and assumptions , the CDO transactions would not have been problematic. The paper suggests some policy prescriptions which include elimination of explicit reliance on credit ratings and the claims made by rating agencies that the ratings are mere opinions should not be accepted any longer. The researcher suggests that rolling average of market measures is a much better representation of the instrument than the unchanged credit rating .Credit default swap spreads would provide a warning about the CDOs and their true performance in the market.
According to M.K.Datar(2011), the role of CRAs in the crisis has attracted attention basically owing to the severe downgrades during the initial stages of the crisis. The conflict of interest in the payment model has been discussed and the author suggests that investor pay model should be adopted as the issuer pay model creates a bias as rating agencies might be prone to give good ratings because the issuers are paying for it.An alternative platform pay model has been suggested in the paper wherein an issuer approaches a clearing house (platform) with a preset fee to get a rating. The platform would get the ratings done from a pool of recognised CRAs. This process avoids direct contact between the issuer and the rating agency.The paper concludes that better disclosures by CRAs and their subsidiaries in respect of details of earning from rating and non-rating revenues, default and transition statistics would play a key role in improved governance in CRAs.
The dissertation work is undertaken to understand the reasons behind the emergence of the subprime crisis in late 2000s and the role of credit rating agencies in the crisis.The study is divided into two parts studying the pre crisis and post crisis situations and analysing the change in credit ratings of various complex instruments in response to the crisis.
The objectives are briefly stated as below:
- Understand the evolution of structured financial products
- Understand the causes of subprime crisis
- Study the credit rating process for CDOS
- Study the factors that drove the rapid downgrade of CDOs in the initial meltdown stages
- Analyse the flaws in the rating process which led to failure in forecasting true performance.
- Suggestions and corrective action for facilitating accuracy in credit ratings of complex products.
The method adopted for research is causal research wherein the problem in question is understood and the degree of impact of the cause on the effect under study is analysed. The financial crisis that began in 2007 is studied and the contribution of credit ratings to the crisis is analysed. Credit ratings serve as the control group in this research. Finally suggestions for improvement in credit ratings and measures to be taken are proposed.
Methods and Techniques of data collection and analysis:
To achieve the research objectives, secondary data from reliable sources are being used. Thorough study of the existing literature is being done to understand different ideas and view points on the topic which would facilitate a comprehensive analysis of the issue. Methodology adopted for rating complex products by leading credit rating agencies has been studied in detail which includes statistical tools and financial models.
Data is being obtained from various secondary data sources for study and analysis.
The major sources used for research are as follows:
- Credit Rating Agencies websites and reports
- Banking for International Settlements(BIS) working papers and reports
- Securities and Exchange Commission reports
- Journals and papers published on Credit ratings contribution to the Crisis.
DRIVERS TO EMERGENCE OF FINANCIAL CRISIS 2007-2008:
The financial crisis was fuelled right from the early 2000s through various factors , the most important of which is sub-prime lending. This inturn led to construction of CDOs at a later stage in order to transfer the concentrated risk of banks to the investors . Hence it is vital to get a clear idea about the emergence of sub prime lending and evolution of CDOs .
The sub-prime mortgage market caters to customers who are unable to meet normal credit and/or documentation requirements for mortgages. Subprime lending is riskier than normal lending for the banks. Hence banks tend to charge a higher interest rate to compensate for the risk. Over the past decade, this mark-up over prime rates has been about 2%, making lending potentially very lucrative. Only by the mid-1990’s did the subprime mortgage market begin to take off as a number of factors emerged which apparently mitigated the default risk on such loans and hence led to an increasing number of banks lending ever-larger amounts to this sector. Some important factors which contributed to a boom in subprime lending are discussed below.
Introduction to Sub Prime Lending:
Evolution of Structured Financial Products:
Collateralised debt obligations have been one of the complex financial products which have been instrumental in driving the financial system into a crisis. The evolution of CDOs needs to be understood in order to study the emergence of the financial crisis.
The basic principle behind a CDO involves re-packaging of fixed income securities and division of their cash flows according to a specified structure. A CDO is constructed by creating a ‘brain-dead’ company, a special purpose entity (SPE) or structured investment vehicle (SIV), which buys assets and issues bonds backed by the assets’ cash flows. The bonds are divided into a number of tranches with different claims on the principal and interest generated by the CDO’s assets. The mechanics of a typical CDO are illustrated in Diagram A. essay_footnotecitation">
In order to understand the sudden growth in the demand for CDOs which in turn led to the financial crisis , it is vital to list out the reasons behind the growth of CDOs which are as below.
Rationale behind growth of CDOs:
Securitisation has been a way that helped banks to bundle loans and sell it to investors or make it off-balance sheet items .Once these items are removed from the balance sheet the capital adequacy gets more space and hence banks make new loans and the process continues. This basically facilitates banks to free up cash and easily meet BASEL norms for capital adequacy.
The second rationale is re-allocation of risk.CDOs helps banks reduce the concentration of risk and also create securities as per specific requirements and risk profiles of the investors. This facilitated institutional investors to purchase CDOs as they can invest only in highly rated ‘ investment grade securities’.
CDOs allowed these investors to gain exposure to assets that, on their own, had been too risky, while investors looking to take more risk and receive potentially higher returns could buy the most junior or ‘equity’ CDO tranches. essay_footnotecitation">
These are the major reasons behind growth of CDOs . Banks only thought of their own benefits and growth and the aftermath of this action was left to the market to face in reality few years down the lane . The consequences of this act of the highly knowledgeable financial community has been faced by people across the globe.
Credit Ratings and CDOs: An overview
Investors invest in securities based on various criteria one such being reliable ratings given by well known credit rating agencies. Credit rating agencies(CRA) were basically formed to guide investors assess risk of fixed income securities. CRAs have played a major role in the growth of CDOs market as investors relied on the ratings given to these complex structures and based their investments majorly on these credit ratings. They used credit ratings in place of their due diligence for assessment of CDOs.
Credit rating agencies are approved by Nationally recognised Statistical rating organisation(NRSRO) . There are three well known players in the U.S financial market which are
Standard and Poor’s
These three agencies rated CDOs and the fees generated by rating CDOs were quite high which led to record profits . The percentage of CDO deals that were rated by the credit rating agencies has been given in the below diagram.
Source: UBS CDO research
Note: The percentage total exceeds 100 as the same instruments have been rated by more than one agency.
Revenue earned by the rating agencies has grown since 2002 which has been depicted in the diagram below:
According to Mark Adelson, current Chief Credit Officer at S&P:
‘ The advent of CDOs in the mid-1980s was a watershed event for the evolution of rating definitions. Until the first CDOs, rating agencies were only producers of ratings; they were not consumers. With the arrival of CDOs, rating agencies made use of their previous ratings as ingredients for making new ratings â€“ they had to eat their own cooking. For rating CDOs, the agencies used ratings as the primary basis for ascribing mathematical properties (e.g., default probabilities or expected losses) to bonds. essay_footnotecitation"> ‘
Credit rating agencies failed to examine the accuracy of the prior collateral ratings. They also used other rating agencies ratings as base for rating CDOS without verifying accuracy.To adjust for the shortcomings in other agencies’ ratings they used a system called notching where the rating would be decreased by one notch if the rating has been done by another rating agency.
For example , if Moody’s is rating a CDO which has a collateral rated BB+ by Fitch , Moody’s would consider the rating as BB and plug it into their rating model. No analysis of accuracy had been done and it would be assumed that the notching would compensate for any errors in the rating done by Fitch.
Figure below illustrates a comparison between the collateral ratings and the corresponding CDO ratings at the beginning and also the current scenario.
This shows that the CRAs somehow gave huge amounts of AAA rated CDO securities from collateral with much lower ratings, reassuring the fact that that main reason why CDOs were so profitable in 2005-2007 is that it was possible to generate a high proportion of highly rated securities from lower quality assets. That practice backfired, resulting in massive downgrades of the CDO tranches as it became apparent that the rating agencies had been overly optimistic. While in 2005-2007, the initial ratings given to CDO tranches were on average better than the ratings of their underlying collateral assets, current CDO tranche ratings are worse than their associated collateral pool ratings which is an area that needs attention.
The following figure shows the downgrades of CDOs over the years. The numbers on the y-axis correspond to the rating scale with lower numbers equal to higher-quality ratings (1=AAA, 22=D).
An overview about the credit ratings and CDOs has been done. The following section elaborates the rating methodologies adopted by the rating agencies which have different variables considered for the purpose of rating the complex financial instruments and the assumptions behind them.
CDO Rating Methodology:
CDOs are based on portfolios of instruments combined together and not on a single loan. Rating these complex structures requires ascertaining a probability of default (PD) to each instrument in the portfolio and involves assumptions relating to recovery rates and default correlations. Thus it combines credit risk assessments of the individual assets and estimates about default correlations using credit risk modelling.
There are two widely used methodologies for rating CDOs namely:
- Binomial expansion technique
- Monte Carlo Simulation
Each method is discussed initially and then a comparison is done between the techniques and their impact on the ratings.
Steps in the Rating Process:
The reliability of a CDO rating depends on the agency’s ability to assess the credit risk in the underlying asset pool and accurate modelling of the distribution of cash flows from the asset pool to different groups. All rating agencies generally follow a two stage rating process. In the first stage, analytical models are used to assess credit risk. The tools applied for analysing CDO pools differ according to the nature of underlying assets and are also based on the rating agencies.
The second stage of the process comprises of structural analysis. This stage involves detailed modelling of cash flows, legal assessments and evaluations of third parties involved in the deal such as asset managers. The results of the cash flow analysis are used as input in the credit model in the form of adjustments in particular model assumptions. Finally, all information is aggregated and combined into a single, alphanumeric rating which is benchmarked to the historical performance of bonds.
The famous CDO rating methodology is based on Moody’s quantitative approach for determining expected losses for CDO tranches which is called the binomial expansion technique (BET). BET was introduced in the year 1996 and is still used in CDO analysis along with a number of other new methodologies. The method relies on the use of ‘diversity score’ (DS) which is used to map the underlying CDO portfolio with a hypothetical portfolio that consists of homogeneous assets equal to the diversity score. For calculating expected loss distributions a simpler hypothetical portfolio of homogeneous, uncorrelated securities is used in place of the original portfolio. As the number of assets in the hypothetical pool is assumed to equal the diversity score, it will be lower than the number of assets in the actual CDO portfolio to account for uncorrelatedness under the BET.
Given the homogeneous nature of the hypothetical portfolio, the behaviour of the asset pool can be explained by DS+1 default scenarios with default occurring for 0 assets, 1 asset, DS assets, where the probability of each scenario is calculated using the binomial formula. After working out the cash flows and losses under each default scenario, the obtained output from the binomial distribution are converted into estimates of the portfolio and tranche loss distributions.
An alternative method that is used in by three major rating agencies is Monte Carlo simulation technique which estimates the default properties of the underlying CDO asset pool based on repeated trials of random defaults with correlation structure that is assumed. In this process, default events are simulated within a credit risk model, where default occurs when the value of assets fall below that of its liabilities. The model’s main inputs are asset-level probabilities of default and pair-wise correlations of assets, which are converted into an estimate of the entire pool’s loss distribution. This distribution is used with other inputs, to determine the required subordination level (level of credit enhancement) for each CDO tranche, where desired tranche ratings are assumed constant or given.MC approaches give more accurate loss distribution estimates, they are computer intensive and take a long time to provide accurate results. Especially for cash flow CDOs it is very difficult to construct an efficient MC simulation that accounts for all cash flow nuances .Sometimes it takes hours for an MC simulation to determine the subordination level for an AAA tranche and this can be complicated when further assumptions are made. In managed portfolios, the relative value of the simulation approach’s asset-by-asset analysis is less while some of the BET’s implicit simplifying assumptions (like equal position sizes) closely resemble typical covenants in managed deals. The choice of rating methodology basically considers a trade-off between accuracy and efficiency, and the result may differ for certain types of CDO structures. This is one of the reasons for Moody’s to introduce a new Monte Carlo simulation-based method called CDOROM to rate static synthetic CDOs, while it continues to use the BET and its modifications for rating cash CDOs and managed structures.