Arthur M. Berd
Capital Fund Management, Paris
The ongoing credit crisis has dramatically impacted on many aspects of our profession, even calling into question the very existence of institutions that have long been among the pillars of the capital markets. Quantitative finance in general, and credit analysis in particular, have suffered in reputation as many widely used models have failed or were misinterpreted and misused. It is against this background that The Journal of Credit Risk has decided to dedicate its 2009 summer issue to “Lessons from the Credit Crisis”, the main objective being to uncover the roots of the credit crisis and to highlight the new realities that have emerged and the insights that were gained from it. The special issue features a great lineup of papers covering topics ranging from the tracking of the history of the current crisis and its impact on the structure and functioning of credit markets, to the analysis of the shortcomings of various credit risk models, to offering new methodologies designed to withstand the test of the crisis.
Among the contributing authors are some of the best-known researchers from both academia and the industry, including John Hull, Ashish Das and Roger Stein, Edward Altman and Brenda Karlin, Omar Masood, Jeffrey Rosenberg, Stuart Turnbull, Vineer Bhansali and Alexander Lipton and Artur Sepp.
Before presenting the papers in this issue, allow me to give a few personal observations. Many researchers have noted that the current crisis is bigger than just a single burst bubble in credit markets: it involves many interconnected markets, from real estate to emerging markets, from commodities to equities and almost everything in between. Why is that so? Why have so many markets frozen up or broken down in unison? This is not the first time this has happened, but it is certainly the deepest and longest crisis in many generations, globally. Is it even relevant to call the current situation just a “credit crisis”, or should we instead be talking of an “economic crisis”? I believe the answer to the last question is that the crisis is indeed primarily about credit, and that there are fundamental reasons why it spilled over to other markets once its magnitude grew too large.
The root of the problem is that “credit” is not just an asset class, or a market encompassing specific securities (commercial paper, bonds, loans) or derivatives (credit default swaps (CDSs), collateralized debt obligations (CDOs)). Credit risk in general is also a universal metric for all risks, because all risks are eventually transformed into a risk of holding the obligations of an entity that is carrying these other risks. To use a software analogy, the credit risk is the “interface” through which market participants interact with each other. If the “base risk” is geopolitical, it gets transformed into sovereign credit risk. If it is market risk, it gets transformed into the counterparty risk of market participants. If it is economic risk, it gets transformed into the credit risk of corporations and other “real economy” participants. Therefore, when this interface breaks down, the holders of any other form of risk have no choice but to reduce their exposures because all of a sudden the formerly netted low-risk transactions become open-ended high-risk transactions.
This brings to the forefront the role of intermediaries in the credit markets, because it is the intermediaries that aggregate the vast majority of the financial transactions and rely the most on the netting and diversification of them to manage the risks on their books. Traditionally, only banks were able to play the role of such intermediaries. Moreover, the transactions that accumulated on their books were, until the 1970s, mostly straightforward loan-type exposures to their clients, which were funded in an even simpler way: via deposits.
Gradually, however, the landscape of the intermediation of credit risk has changed. The investment banks and other non-commercial bank entities have entered the intermediation business and have gradually come to dominate it, while the traditional banks got left behind for quite a few years, at least in the US, partly due to Glass–Steagal act constraints. This discrepancy, and the corresponding gap in profits between non-bank entities and commercial banks, eventually drove the pressure to deregulate and repeal the Glass–Steagal act, as a result of which formerly staid banks like Citibank were given a chance to compete with the likes of Lehman Brothers, Merrill Lynch and Bear Stearns for a slice of the securitization and derivatives market. The results of this competition turned out not to be so good for most of the parties involved.
Perhaps the biggest watershed event of the modern finance era was the invention of the swap contract at Salomon Brothers in 1981. In this first transaction, the US dollar/deutschemark/Swiss franc currency swap between IBM and World Bank (IBRD), the express objective was the mitigation of the counterparty risk compared with a simple cash loan. Indeed, the bilateral netting agreement typical in all swap contracts exposes the counterparties to only a small fraction of the notional loss in the case of default. This feature, together with the attraction of the no-upfront-cash requirements of most swap transactions, has led to explosive growth in the size of this market.As the Bank for International Settlements recently reported, the total notional value of the interest rate swaps, currency swaps and, more recently, credit default swaps and equity and commodity linked contracts has grown to US$591 trillion as of December 2008 (a drop of 13.4% from their peak in June 2008). A better measure of the over-the-counter (OTC) market size, its gross market value (defined as the aggregate replacement value by the Bank for International Settlements), is estimated to be US$57 trillion (an increase of 66.5% in six months because of rising risk levels). This is roughly equal to aggregate world GDP (US$54 trillion). And the vast majority of these transactions are concentrated on the books of relatively few financial intermediaries!
So, the swap contracts (and the whole OTC derivatives market that grew out of them) that were hailed as a solution to the problem of counterparty risk have become so successful that they have grown to pose an even greater counterparty risk problem than the initial “plain-vanilla” transactions. In essence, the OTC market, instead of alleviating the counterparty risk of market participants, has concentrated it in the hands of intermediaries and thus transformed it into systemic risk. This was not necessarily a bad thing, as it produced several decades of accelerated growth and prosperity by freeing up other market participants from the clutches of credit risk. Unfortunately, instead of recognizing this metamorphosis, the regulators thought that the risks were being self-contained and thus failed to foresee or prepare for the current crisis.
Note that I intentionally do not make much of a distinction between interest rate or currency swaps and CDSs. Even though CDSs have gotten a very bad rap recently, I do not believe that they per se pose any greater or smaller a risk to the system than all the other types of swap transactions. The problem is not in what type of risk is swapped, the problem is in the off-balance-sheet nature of the swap transaction and in the fact that a large notional amount is reduced to a small cashflow requirement and a small risk weight under most typical risk management methodologies. In other words, the swap contracts have allowed the financial system to vastly expand its leverage because of the belief that the risks were mitigated. While this treatment is justified for any given particular swap transaction in isolation, when the aggregate amount of them is so large that it cannot be unwound, and it is concentrated in a few intermediaries, the result is a vastly magnified systemic risk.
To put things in perspective, the counterparty risk reduction of a typical 10-year interest rate swap in comparison with outright cash lending of the same notional is roughly 5–10 times, depending on volatility projections for interest rates. However, the notional of interest rate swaps is more than 100 times greater than the aggregate notional of bonds. So the net amount of counterparty risk due to swaps is more than 10 times greater than that due to cash lending, even under very tame underlying volatility projections. Things are much worse for cross-currency swaps, where the counterparty risk reduction is substantially less, and credit default swaps, where it is virtually non-existent. Of course, the silver bullet that allows this whole system to function is that the financial intermediaries maintain frequently rebalanced margin accounts for their counterparties and charge progressively larger margins for greater perceived risks, with the aim of reducing the potential shortfall if the client is unable to perform.
Some critics of CDSs point to their extreme payout asymmetry as the cause of the trouble. They argue that such asymmetry makes margin maintenance impossible because whatever you charge is still not going to be enough to cover the shortfall in case of default. This argument is not without merit but I believe it overstates the case, because in the vast majority of situations the creditworthiness of the underlying entity deteriorates sufficiently gradually for counterparties to be able to adjust the margins. Two important exceptions to this are:
1) fraud, where a sudden revelation of hidden loss can render the company bankrupt without much prior warning, such as was the case for Enron, World- Com and Parmalat; and
2) financial companies, where the opacity of the books, convoluted accounting rules and often uncertain and high leverage are such that no one effectively knows what they have until things become distressed enough that they are visible to the naked eye.
In their drive for market share, intermediaries have gradually reduced their margin requirements and thus allowed greater and greater levels of leverage to accumulate for their counterparties. In some cases, such asAIG with its “iron-clad”AAAratings, there were no margins required whatsoever, making even a transaction earning a few basis points a winning trade, given virtually infinite leverage. Of course, the abuse of this special status led to an accumulation of unjustified risks on AIG’s books and to its eventual fall from AAA status. The resulting margin calls appeared to be so severe precisely because any reasonable margin requirement when compared with zero is infinitely large, and therefore the borrower is unprepared to meet it (unless of course they have the internal discipline to set aside sufficient cash reserves and self-insure, asWarren Buffett’s Berkshire Hathaway is used to doing).
The irresistible attraction of high leverage stimulated in counterparties by low margin requirements is not really that different from the irresistible urge to buy a home one cannot really afford stimulated by the proliferation of subprime mortgages. Ironically, both the least sophisticated and the most sophisticated market participants seem to have fallen ill with the same symptoms: too much desire of immediate satisfaction overshadowing prudent assessment of future risks and costs. Perhaps it is just human nature, and financial sophistication only matters in the form of its expression rather than in its substance.
So in such a context it is hardly surprising that valuation and risk management decisions were gradually driven toward the lowest common denominator: the most forgiving models and the most lax assumptions. Yes, the models used by many market participants have been found wanting. But this is not because all quantitative models were flawed, as some pundits would have you believe, but rather because of a peculiar adverse selection favoring models that would justify the most leverage. The Journal of Credit Risk, among other journals, has published many realistic models of credit that, if used, would have provided investors with a much better sense of the potential risks and pitfalls. No model is ever perfect, but it is not an excuse for going back to the Stone Age and refusing to listen to quantitative analysts. The selection of articles in this special issue will hopefully serve to guide investors in their assessment of credit risks and in opportunities in the next phase of the evolution of the credit market.
The opening article of this issue if written by Hull, a leading authority in the quantitative modeling of derivatives in general and of credit risk in particular. He gives a broad overviewof the ongoing credit crisis, starting with the developments in the US housing market and showing how securitization exacerbated the lax lending practices, particularly for subprime borrowers. He then goes on to demonstrate how the more complex transactions, asset-backed security collateralized debt obligations (ABS CDOs), have in fact served as conduits for capital to support subprime securitization and lending, thanks to risk estimates for the senior tranches of such CDOs that were too optimistic. Hull correctly highlights the fact that the biggest problem with ABS CDOs was that they were constructed from the tranches of ABSs, which themselves represent large pools of mortgages. Thus, most idiosyncratic risk had already been diversified at the ABS level, and therefore when those were combined into CDOs it was wrong to assume that there would be any additional diversification benefit.
As a side comment I would like to add that this point highlights a big difference between ABS CDOs and more conventional corporate CDOs or collateralized loan obligations (CLOs). In the latter case, there is still a lot of idiosyncratic risk in the underlying pool components and therefore the securitization makes sense. I would venture to predict that traditional fully funded CDO and CLO markets, unlike those forABS CDOs, will eventually revive and will actually provide investors with valuable return opportunities in the post-crisis recovery.
The next article is presented by Das and Stein. They thoroughly investigate the roles of underwriting standards versus the changing macroeconomic factors in the subsequent subprime losses, and discern substantial differences between the earlier and more recent vintages of subprime mortgages in terms of their performance.While the economic conditions, including the changes in real estate prices, are shown to be the main factor driving subprime losses, Das and Stein are able to attribute a portion of these losses to the quality of underwriting at the time of origination.
In the third paper in the issue, Altman and Karlin highlight the re-emergence of distressed exchanges in corporate restructurings. As a widely recognized pioneer of credit risk analysis,Altman can drawon his vast personal experience, as well as on his thoroughly compiled database of corporate credit events, to pinpoint the important structural changes in the (high-yield) distressed market. Altman and Karlin show that 2008 sawa dramatic increase in distressed exchanges as companies tried to stave off formal bankruptcy filings. However, they warn investors that even though the recovery rates in such exchanges are much higher than in bankruptcy, the probability of eventual bankruptcy and additional losses is very high.
Reflecting similar sentiments to those in this introduction and in Hull’s paper, Masood follows the paper trail of the balance sheet exposures of leading investment banks leading to the credit crisis. In particular, he highlights the effects of leverage and liquidity factors on the balance sheet, and pinpoints salient differences between the cases of Goldman Sachs and Lehman Brothers that help explain the diverging fate of these two companies in the midst of the credit crunch.
Rosenberg, who heads the credit strategy group at Bank of America and has a very good vantage point on the ebbs and flows of the credit market, continues the ever-important theme of examining the causes of systemic risks of large financial institutions. Focusing on the cases of Lehman Brothers and AIG, whose failures arguably marked theworst moment of the credit crisis in September 2008, Rosenberg argues that the earlier bailout of Bear Stearns had the unintended consequence of exacerbating the systemic risks for Lehman Brothers. I wholeheartedly agree with his argument and also point out that only a few days before its collapse, Lehman Brothers CDSs were trading at the same level (450 bps) as they were right after the bailout of Bear Stearns. This is despite the situation being otherwise visibly worse and the stock trading asymptotically close to zero. The only explanation for this is that market participants were pricing in a very high probability of bailout, ie, they had become too complacent about the nature of this systemic risk.
Rosenberg also argues that the downfall of AIG was precipitated by mark-tomarket losses in its arbitrage CDO portfolio, while the necessity of the bailout stemmed from an even larger amount of regulatory capital relief transactions on its books. What is remarkable in light of this analysis is that AIG was bailed out by the US Federal Reserve alone while the systemic risks in these transactions threatened mostly European banks, supervised by the European Central Bank and the Bank of England. Perhaps this is an example of “regulatory mismatch risk”, when large cross-border transactions fall under the jurisdiction of one agency, while the risks are exposing the constituents of another. It is easy to imagine a situation in the future when such risks might be left “orphaned” as either agency is reluctant to act alone, while there is little time for thorough coordination in a crisis situation. This underscores the need for better collaboration on an ongoing basis between national agencies in regulating global capital markets.
Turnbull’s detailed article on measuring and managing risks in innovative financial instruments completes the logical circle of the preceding papers by examining the difficult challenges facing both internal risk managers and external regulators when it comes to modern financial companies. Turnbull, who is one of the pioneers of quantitative credit risk and credit derivative pricing models, gives us a road map to understanding the different dimensions of risk and model uncertainty, and ties it all up into specific instrument design and market characteristics. He goes on to give specific prescriptions for best practice in using and critically testing the models and estimating the risks of complex products.
Taking a complementary point of view, Bhansali shares the buy-side perspective on tail risk management. His company, PIMCO, has been able to successfully navigate the credit crisis despite its size and its inevitable exposure to every type of fixed-income asset partly due to its adherence to consistent and proactive tail risk management. Bhansali explains how the company recognized the presence of the systemic risk in their investment portfolios, how they estimated the need for hedges and their added value and how an investor can pragmatically go about building such umbrella hedges over time. What is remarkable in this article is that none of this is extremely complicated. The ability to see the forest for the trees is what distinguished almost every winning investor in the past couple of years.
In the final paper of this issue, Lipton and Sepp give us a new model to estimate the credit value adjustment (CVA) for CDSs within the structural default framework. The importance of their results cannot be overstated in the present state of the world, where the counterparty risk dominates the client relationships of financial intermediaries. Credit value adjustments will inevitably be a relevant part of the functioning of the credit derivatives markets as long as the risks of bank failures do not dissipate. Whether it is when a CDS contract is novated to face a new counterparty, or whether it is when one tries to put a fair market value on the portfolio of such contracts with large counterparty exposures (which is normal for any buy-side firm facing one or a few prime brokers), CVA is going to be a crucial input to this process. Readers should not be surprised by the analytical complexity of the model as it is indeed a complex problem to solve if one wishes to capture it realistically. Risk managers would be particularly well advised to build intuition about the important features of this model and incorporate it in their practices.
I am confident that this special issue of The Journal of Credit Risk will become a must-read reference for anyone who is trying to comprehend the ongoing credit crisis and to draw the correct conclusions from it. I am also sure that it will give a new impetus to many other researchers to contribute their insights toward the better understanding of the complex issues facing the credit markets. The editors of The Journal of Credit Risk are looking forward to hearing your opinions and receiving your papers on these and other topics.