Securitisation: was the tail wagging the dog?

Keynote speech by Mr Malcolm D Knight, General Manager of the BIS, on the occasion of the 33rd Annual Conference of the International Organization of Securities Commissions (IOSCO), Autorité des Marchés Financiers (AMF), Paris, 29 May 2008.

Abstract

Securitisation is essential to the evolving global financial system because it provides a number of important economic benefits. But securitisation has also been a central element of the dynamics of the financial turmoil that began in the summer of 2007, largely caused by inadequate risk management practices and a lack of due diligence on the part of market participants. Complex financial engineering also led to fundamental illusions among investors about credit and liquidity risks and increased the dependence of banks on the originate-to-distribute (OTD) model. There are four important areas where corrective action could be particularly fruitful in enhancing the benefits of securitisation and reducing its risks: fixing the securitisation chain; understanding tranche exposures better; improving the OTD model; and enhancing regulation and supervision.

Full speech

I. Introduction

The topic of today's panel, securitisation, is of course very important: the securitisation process has been at the heart of the ongoing financial market turmoil that began almost a year ago as the "US subprime mortgage crisis".

A key issue is whether securitisation is the proverbial dog that has been wagged by its own tail. As you probably know, in North America when a factor of minor importance - or an event that is regarded as unlikely to occur - comes to dominate a situation, this is often characterised by saying that the "tail is wagging the dog". One question for today is whether the realisation of "tail risks" wagged the "dog" of the securitisation process to the point where it is no longer seen as a useful element of the financial system. Or, on the contrary, did the basic weaknesses lie in the way securitisation was implemented? And can the securitisation process be strengthened and its shortcomings redressed so that it can continue to play a key role in the operation of the financial system?

My short answer to these questions is that securitisation is essential to the evolving global financial system because it provides a number of important economic benefits - through increasing the diversification of risk and reducing the costs of intermediation between savers and borrowers. Thus, we need to strengthen the securitisation process to safeguard its benefits and mitigate its risks.

To draw useful lessons from the current financial turmoil, we must start right where the problems began - with the adverse credit events that originated in a specific segment of the US mortgage markets and quickly reverberated to the heart of global financial system. One view is that market participants were taken by surprise by low-probability events, or "tail risks", that were not predictable on the basis of observed historical loss distributions. This view would suggest a fundamental weakness in securitisation. Another view, however, is that there were a number of adverse developments in the supply of and demand for housing, particularly in the United States, that should have made the possibility of a marked fall in the price of housing and associated mortgage defaults a central element of systematic risk, rather than an unlikely "tail event".

So I will first say a bit about what went wrong, then turn to the things that can be done to address them, and end by listing four broad issues that I think deserve discussion by this panel.

II. How we got where we are

Essentially, the financial turmoil that began in the summer of 2007 stemmed from the fact that participants in the securitisation chain ignored the possibility of a general decline in US house prices leading to unexpected mortgage delinquencies. My emphasis here is on the word "ignored". Frankly, it is hard to believe that this resulted from the materialisation of a true "tail risk". By 2007, many economists and observers had been saying for some time that the US housing market was overvalued and that there was an increasing risk with a significant probability (not just a "tail risk") that US house prices would correct downwards. We now know that these predictions, based on simple economic analysis, proved to be right. Indeed, it was because market participants had not assessed the mean likelihood of this risk properly that, once the risk of a serious housing market downturn materialised, problems quickly spread. The unwinding of leverage and collapsing liquidity, of course, have severely aggravated the situation. Asset prices declined dramatically; and liquidity guarantees to off-balance sheet vehicles and other conditional liabilities (all of which would have been unlikely to be drawn under normal circumstances) were triggered in the wake of the ensuing uncertainties, giving rise to severe disruption in the interbank money markets.

This whole set of events is frequently described by observers as "unprecedented". But it was not solely the result of bad luck. It was also the consequence of a clear analytical perspective on how the housing market operates, inadequate risk management practices, and a lack of due diligence on the part of market participants, who had not taken sufficient consideration of alternative but plausible scenarios in their stress tests.

That said, there is little doubt, in my view, that securitisation has been a central element of the dynamics of this financial crisis. One key reason is that modern securitisation involves rather long chains of production. Putting considerable distance between ultimate investors and cash-generating assets led to adverse incentives along the securitisation chain and contributed to deteriorating credit underwriting and monitoring standards - enabling the build-up of credit event risk, in particular in the US housing market.

A related feature was tranching, which allowed pools of various liabilities to be engineered in ways - in particular, with the benefit of various forms of credit enhancement - that enabled a large proportion of the value of these instruments to obtain a AAA rating from the credit rating agencies. Investors in such tranches had few incentives to bear the costs of undertaking serious due diligence because they thought they could rely on these ratings for their investment decisions. And as lower-rated tranches were re-securitised, the burden of borrower scrutiny was passed along the chain, to ever smaller groups of investors who faced an ever harder task of analysing the credit quality of what they were buying. All these weaknesses and perverse incentives were reinforced by the sheer complexity of the instruments that were being developed.

Complex financial engineering also led to fundamental illusions among investors about the characteristics of the instruments they were holding. One key fallacy was the belief that complex instruments could be created that were both tailored to the needs of individual investors and, at the same time, tradable in liquid markets. Another illusion was that securitisation could offer higher yields without the corresponding elevated risk - ignoring the analytical relationship between risk and return. In particular, investors did not realise that tranching techniques had given rise to rating transitions and valuation adjustments that could be much more pronounced than those for corporate bonds. Many of them also failed to appreciate that while they could expect to obtain slightly higher and stable returns by holding these complex instruments under most circumstances, this was at the risk of incurring exceptionally large losses in adverse environments.

A final set of reasons for the turmoil was the dependence of the banks' originate-to-distribute (OTD) business model on securitisation markets as a means of funding and risk transfer. Being no longer supposed to keep the loans they had originated on their books, banks became increasingly exposed to the risk of illiquid markets: from warehoused exposures that suddenly could not be offloaded; or from liquidity guarantees to off-balance sheet vehicles.

III. Where do we go from here?

Again, it is important not to forget that securitisation has quite a number of desirable features that deserve to be preserved. When it functions properly, securitisation can help to spread risks widely, allocate capital efficiently, diversify the revenue streams of banks, and foster financial innovation in the system. Given the time constraints of this panel session, let me focus on four areas where corrective action could be particularly fruitful in enhancing the benefits of securitisation and reducing its risks.

1. Fixing the securitisation chain

  • Reduced complexity. It seems obvious that the riskiness of re-securitisations (such as structured finance CDOs) was vastly underestimated. In a way, this problem is currently being addressed by the markets themselves in that many of these structures have vanished, perhaps forever. The new CDOs now appear to be of a more "plain vanilla" type: no more than one securitisation layer, better collateral, more subordination. Could this process, perhaps, be pushed further, with some of the more standardised products taken onto exchanges? Fewer re-securitisations and shorter securitisation chains should also help to better align incentives.

    Information transparency. Something that is often forgotten when thinking about "complexity" is that even analysis of "simple" mortgage securitisations is an information-intensive job. Often it is not so much modelling that is the problem, but the dispersion of information and the fact that its quality tends to deteriorate as it is passed along the securitisation chain. Clearly, shorter chains would help. Beyond that, better information is required. The Financial Stability Forum has made proposals in this regard, calling - for example - for more and timelier information to be made available, in a standardised format, on the nature and performance of the assets underlying securitised products.

2. Understanding tranche exposures: less reliance on ratings

  • Did the rating agencies get their expected loss estimates right? Well, certainly not in all cases. Indeed, in the past few weeks additional mistakes have been reported regarding the valuation of structured products. Inadequate incentives between developers of securitised products and those paid for rating them may also have played a role. Perhaps more fundamentally, investors did not pay sufficient attention to the fact that ratings cover only one dimension of risk, and that the variability of the expected returns could be high. They also failed to realise that tranched instruments often have risk properties - especially in terms of illiquidity and rating transition probability - that are different from those of cash products. In any case, a key point was that one should not have generalised across all structured products. This is where information on risk properties other than expected loss comes in: multidimensional ratings, as proposed by various bodies recently, are one possible solution.
  • At the same time, we would probably want less - or less mechanistic - reliance on ratings. One benefit of the proposal to have a different rating system for structured finance instruments would be that the broad investor community would have to review existing investment mandates and guidelines.

3. Improving the OTD model: better risk management

  • The key insight here is not at all new, but it is worth repeating: tail risk exposures - including the risk of illiquidity - are not well measured by simple tools such as value-at-risk. So risk managers must rely on a wide ranger of tools to capture the multidimensionality of risk.
  • One important aspect is better stress testing, as noted by the Basel Committee recently. More emphasis must also be placed on risk interactions, such as those between credit and liquidity risks, in hedging existing exposures. One example would be prime brokers' exposures to hedge funds and how these, through collateral arrangements and margin calls, can interact with liquidity risks.

4. Implications for regulation and supervision

  • An area that I have already mentioned is undue reliance on ratings. To the extent that existing regulation encourages such behaviour, authorities need to make adjustments. Regulatory action may also help strengthen incentives at the beginning of the securitisation process, by ensuring that originators have some "skin in the game". Efforts along these lines are already in train.
  • For the banks, which are often both arrangers of and investors in securitised instruments, the Basel Committee has made proposals to establish higher capital requirements for certain products and to strengthen the capital treatment of liquidity facilities for off-balance sheet vehicles. In cooperation with IOSCO, it has also proposed to adjust existing guidelines for the treatment of event risk in banks' trading books.
  • A key goal of these measures is to better align banks' risk-taking with their capital and liquidity buffers, with specific attention to be given to the evolution of these risks over the full business cycle. At the same time, they would sharpen existing incentives to limit the complexity and, ultimately, improve the liquidity of securitisation markets.
  • And of course the recent turmoil has highlighted the need for better coordination of financial regulation and supervision across institutions, markets and jurisdictions.

IV. Issues for discussion

In my view, all this translates into four broad areas that deserve further discussion in this morning's panel:

  1. Better transparency and disclosure, including issues of product complexity and valuation.
  2. Better risk management, including issues of integration across risks and the role of ratings.
  3. Better ways to align incentives along the securitisation chain.
  4. Enhancements in the role of regulators and supervisors, including the efforts being undertaken by IOSCO.

With these comments, I am glad to open today's panel discussion by handing over to the moderator, Jochen Sanio.