The Volokh Conspiracy - Outstanding Securitization Article by Kenneth E. Scott and John Taylor

in today's Wall Street Journal ("Why Toxic Assets Are So Hard to Clean Up," Opinion, July 20, 2009, probably behind wall), showing with examples how fantastically complex, and effectively impossible to value, securitization derivatives had(ve) become.

Scott and Taylor are Stanford professors jointly affiliated with the Hoover Institution there (as I'm pleased to say I am as well; Go Hoover!) and this is must reading for those trying to contemplate the form of future financial industry regulation. (I won't say much more, I'm on lite-blogging status and am not supposed to be doing this computer work, but this article is must-read.)

As I've occasionally noted here, credit default swaps (including the perverse incentives they can create, such as the "empty creditor" problem) tend to receive most of the attention in the vexed question of the contribution of derivatives to the crisis. Many of the difficulties of CDS are real, of course. Still, in many respects the leverage combined with valuation-stymying complexity created by the credit derivatives layered on-top of the original asset securitization, rather than CDSs, is the larger regulatory problem:

Why are these toxic assets so difficult to deal with? We believe their sheer complexity is the core problem and that only increased transparency will unleash the market mechanisms needed to clean them up.

The bulk of toxic assets are based on residential mortgage-backed securities (RMBS), in which thousands of mortgages were gathered into mortgage pools. The returns on these pools were then sliced into a hierarchy of "tranches" that were sold to investors as separate classes of securities. The most senior tranches, rated AAA, received the lowest returns, and then they went down the line to lower ratings and finally to the unrated "equity" tranches at the bottom.

But the process didn't stop there. Some of the tranches from one mortgage pool were combined with tranches from other mortgage pools, resulting in Collateralized Mortgage Obligations (CMO). Other tranches were combined with tranches from completely different types of pools, based on commercial mortgages, auto loans, student loans, credit card receivables, small business loans, and even corporate loans that had been combined into Collateralized Loan Obligations (CLO). The result was a highly heterogeneous mixture of debt securities called Collateralized Debt Obligations (CDO). The tranches of the CDOs could then be combined with other CDOs, resulting in CDO2.

Each time these tranches were mixed together with other tranches in a new pool, the securities became more complex. Assume a hypothetical CDO2 held 100 CLOs, each holding 250 corporate loans -- then we would need information on 25,000 underlying loans to determine the value of the security. But assume the CDO2 held 100 CDOs each holding 100 RMBS comprising a mere 2,000 mortgages -- the number now rises to 20 million!

The valuation is essentially impossible to contemplate, and it exceedingly doubtful that in fact anyone made a serious effort at fundamental valuation, as distinguished from taking some (bluntly: circle-jerk) market "proxy" in the hope that someone else had done it. But the sheer amount of leverage involved is also a huge issue - the original securitization might rest on some at least contemplable risk level in terms of default, but the top level derivatives might lose close to 100% of their value on tiny increases in the default rates. Consider this specific example:

To better understand the magnitude of the problem and to find solutions, we examined the details of several CDOs using data obtained from SecondMarket, a firm specializing in illiquid assets. One example is a $1 billion CDO2 created by a large bank in 2005. It had 173 investments in tranches issued by other pools: 130 CDOs, and also 43 CLOs each composed of hundreds of corporate loans. It issued $975 million of four AAA tranches, and three subordinate tranches of $55 million. The AAA tranches were bought by banks and the subordinate tranches mostly by hedge funds.

Two of the 173 investments held by this CDO2 were in tranches from another billion-dollar CDO -- created by another bank earlier in 2005 -- which was composed mainly of 155 MBS tranches and 40 CDOs. Two of these 155 MBS tranches were from a $1 billion RMBS pool created in 2004 by a large investment bank, composed of almost 7,000 mortgage loans (90% subprime). That RMBS issued $865 million of AAA notes, about half of which were purchased by Fannie Mae and Freddie Mac and the rest by a variety of banks, insurance companies, pension funds and money managers. About 1,800 of the 7,000 mortgages still remain in the pool, with a current delinquency rate of about 20%.

With so much complexity, and uncertainty about future performance, it is not surprising that the securities are difficult to price and that trading dried up. Without market prices, valuation on the books of banks is suspect and counterparties are reluctant to deal with each other.

The policy response to this problem has been circuitous. The Federal Reserve originally saw the problem as a lack of liquidity in the banking system, and beginning in late 2007 flooded the market with liquidity through new lending facilities. It had very limited success, as banks were still disinclined to buy or trade such securities or take them as collateral. Credit spreads remained higher than normal. In September 2008 credit spreads skyrocketed and credit markets froze. By then it was clear that the problem was not liquidity, but rather the insolvency risks of counterparties with large holdings of toxic assets on their books.

If you saw the essential crisis as one of liquidity, this meant that Fed-injected funds into the markets would allow the necessary breathing space for market participants to discover and incorporate new information that, in a true liquidity crisis, would show that things were not as bad as feared and the panic could stop. A liquidity crisis, in other words, is a crisis of information. Full information will either show investors (or depositors) that the institution is not in trouble, or that a guarantor stands behind it. Full information stops the panic; the injection of funds is to provide a space for full information to develop.

A solvency crisis, by contrast, is what happens when you have full information - and it turns out that, alas, the panic and investor rush for the exits were justified, because the information indicates that the assets were not properly valued and they really are not worth what the prices indicated in the absence of full information.

As Taylor has cogently argued in his short book on the crisis (published by Hoover Press in a brilliant marketing move to get a short, pithy, blunt, highly informed book out fast), the errors of the US government in crisis management - beyond the original sin of loose money - have largely been an insistence on seeing the crisis as a liquidity crisis rather than a solvency crisis. (Of course, the analytic distinction is not, in real life, fast and hard: a true liquidity crisis can turn into a solvency crisis and vice-versa.)

(It is noteworthy that the new Treasury Department White Paper on financial regulation reform is studiously silent and agnostic on the issue of liquidity versus solvency - apparently on the view that although the distinction mattered perhaps in the management of the crisis, in the distinct matter of future financial reform, it doesn't, presumably because if you enact the White Paper's reforms, you won't have crises requiring you to decide between them. Or at least, regulatory policy ex ante does not need to choose between them before there is a crisis and a panic or market freezeup. I'm not sure it is such a good idea to avoid expressing a view on something so fundamental to crisis response, however, even if regulatory policy ex ante would be approximately the same.)

Scott and Taylor call for mandatory transparency in order to address the valuation and information problems. They call for transparency via a mandatory data base that would contain the basic information necessary for third party valuation:

While the original MBS pools were often Securities and Exchange Commission (SEC) registered public offerings with considerable detail, CDOs were sold in private placements with confidentiality agreements. Moreover, the nature of the securitization process has made it extremely difficult to determine and follow losses and increasing risk from one tranche and pool to another, and to reach the information about the original borrowers that is needed to estimate future cash flows and price.

This account makes it clear why transparency is so important. To deal with the problem, issuers of asset-backed securities should provide extensive detail in a uniform format about the composition of the original pools and their subsequent structure and performance, whether they were sold as SEC-registered offerings or private placements. By creating a centralized database with this information, the pricing process for the toxic assets becomes possible. Making such a database a reality will restart private securitization markets and will do more for the recovery of the economy than yet another redesign of administrative agency structures. If issuers are not forthcoming, then they should be required to file the information publicly with the SEC.

My own tentative view is that this is important, but probably not enough in terms of either addressing complexity or leverage.

Complexity is a problem that is only partly addressed by transparency. Just as many have called for CDS to go onto public exchanges, I would say that there is a pretty good argument - not perhaps dispositive, but one I incline to currently, absent some strong counterarguments - for requiring the same of the leveraged credit derivatives, to provide for mandatory disclosure of counterparty relationships, standardization of contracts and terms, mandatory exchange trading, and related measures that would address not just disclosure and transparency, but complexity as such, all by itself. I say this as someone who believes firmly, by the way, that a huge risk in new future regulation of financial services is over-regulation and the stifling of innovation through lack of credit.

(Actually, my biggest fear today has evolved a step further, to a fear not so much that credit, and the rewards it can bring, will be unnecessarily stifled - but instead that our rapidly developing system of crony capitalism will cause credit to flow to politically favored parties, always in the name of high-minded things, naturally, but cronyism just the same and bleeding away vital investment funds from progress and innovation, but that's a post for another day.)

There is then a further question of whether the degrees of leverage in these credit derivatives should be directly constrained by regulation. That question raises very different issues, mostly ones of moral hazard and the ability of regulators to decide better than the market how much credit is optimal, and whether we have embraced the too-big-to-fail view enshrined in the Treasury White Paper. If we have truly accepted too-big-to-fail for institutions outside of insured depository institutions, then regulation of leverage directly is hard to avoid, because the effect is dangerously to skew the market mechanisms for deciding how much credit is too much credit. If we have not, and are willing to contemplate the failure of large financial institutions - as I certainly think we should - then the market ought to be able, in the presence of full, transparent, and uniformly presented information, be able to police its own leverage as between greed and fear.

My fundamental point here is that complexity is something not completely remedied by transparency. It retains characteristics and risks and dangers even where transparency is full and information is complete - but still really, really complicated in ways that in theory permit apples to apples comparisons, but in practical fact do not, and do not require them of a market still shielded in the short term from the consequences of its long term

(a) mistakes in valuation; or

(b) rational decisions not to bother with attempting accurate but complicated and expensive valuations and due diligence given the long term uncertainties about whether anything bad will really happen.

	[Kenneth Anderson, July 20, 2009 at 12:30pm] Trackbacks Outstanding Securitization Article by Kenneth E. Scott and John Taylor in today's Wall Street Journal ("Why Toxic Assets Are So Hard to Clean Up," Opinion, July 20, 2009, probably behind wall), showing with examples how fantastically complex, and effectively impossible to value, securitization derivatives had(ve) become. Scott and Taylor are Stanford professors jointly affiliated with the Hoover Institution there (as I'm pleased to say I am as well; Go Hoover!) and this is must reading for those trying to contemplate the form of future financial industry regulation. (I won't say much more, I'm on lite-blogging status and am not supposed to be doing this computer work, but this article is must-read.) As I've occasionally noted here, credit default swaps (including the perverse incentives they can create, such as the "empty creditor" problem) tend to receive most of the attention in the vexed question of the contribution of derivatives to the crisis. Many of the difficulties of CDS are real, of course. Still, in many respects the leverage combined with valuation-stymying complexity created by the credit derivatives layered on-top of the original asset securitization, rather than CDSs, is the larger regulatory problem: Why are these toxic assets so difficult to deal with? We believe their sheer complexity is the core problem and that only increased transparency will unleash the market mechanisms needed to clean them up. The bulk of toxic assets are based on residential mortgage-backed securities (RMBS), in which thousands of mortgages were gathered into mortgage pools. The returns on these pools were then sliced into a hierarchy of "tranches" that were sold to investors as separate classes of securities. The most senior tranches, rated AAA, received the lowest returns, and then they went down the line to lower ratings and finally to the unrated "equity" tranches at the bottom. But the process didn't stop there. Some of the tranches from one mortgage pool were combined with tranches from other mortgage pools, resulting in Collateralized Mortgage Obligations (CMO). Other tranches were combined with tranches from completely different types of pools, based on commercial mortgages, auto loans, student loans, credit card receivables, small business loans, and even corporate loans that had been combined into Collateralized Loan Obligations (CLO). The result was a highly heterogeneous mixture of debt securities called Collateralized Debt Obligations (CDO). The tranches of the CDOs could then be combined with other CDOs, resulting in CDO2. Each time these tranches were mixed together with other tranches in a new pool, the securities became more complex. Assume a hypothetical CDO2 held 100 CLOs, each holding 250 corporate loans -- then we would need information on 25,000 underlying loans to determine the value of the security. But assume the CDO2 held 100 CDOs each holding 100 RMBS comprising a mere 2,000 mortgages -- the number now rises to 20 million! The valuation is essentially impossible to contemplate, and it exceedingly doubtful that in fact anyone made a serious effort at fundamental valuation, as distinguished from taking some (bluntly: circle-jerk) market "proxy" in the hope that someone else had done it. But the sheer amount of leverage involved is also a huge issue - the original securitization might rest on some at least contemplable risk level in terms of default, but the top level derivatives might lose close to 100% of their value on tiny increases in the default rates. Consider this specific example: To better understand the magnitude of the problem and to find solutions, we examined the details of several CDOs using data obtained from SecondMarket, a firm specializing in illiquid assets. One example is a $1 billion CDO2 created by a large bank in 2005. It had 173 investments in tranches issued by other pools: 130 CDOs, and also 43 CLOs each composed of hundreds of corporate loans. It issued $975 million of four AAA tranches, and three subordinate tranches of $55 million. The AAA tranches were bought by banks and the subordinate tranches mostly by hedge funds. Two of the 173 investments held by this CDO2 were in tranches from another billion-dollar CDO -- created by another bank earlier in 2005 -- which was composed mainly of 155 MBS tranches and 40 CDOs. Two of these 155 MBS tranches were from a $1 billion RMBS pool created in 2004 by a large investment bank, composed of almost 7,000 mortgage loans (90% subprime). That RMBS issued $865 million of AAA notes, about half of which were purchased by Fannie Mae and Freddie Mac and the rest by a variety of banks, insurance companies, pension funds and money managers. About 1,800 of the 7,000 mortgages still remain in the pool, with a current delinquency rate of about 20%. With so much complexity, and uncertainty about future performance, it is not surprising that the securities are difficult to price and that trading dried up. Without market prices, valuation on the books of banks is suspect and counterparties are reluctant to deal with each other. The policy response to this problem has been circuitous. The Federal Reserve originally saw the problem as a lack of liquidity in the banking system, and beginning in late 2007 flooded the market with liquidity through new lending facilities. It had very limited success, as banks were still disinclined to buy or trade such securities or take them as collateral. Credit spreads remained higher than normal. In September 2008 credit spreads skyrocketed and credit markets froze. By then it was clear that the problem was not liquidity, but rather the insolvency risks of counterparties with large holdings of toxic assets on their books. If you saw the essential crisis as one of liquidity, this meant that Fed-injected funds into the markets would allow the necessary breathing space for market participants to discover and incorporate new information that, in a true liquidity crisis, would show that things were not as bad as feared and the panic could stop. A liquidity crisis, in other words, is a crisis of information. Full information will either show investors (or depositors) that the institution is not in trouble, or that a guarantor stands behind it. Full information stops the panic; the injection of funds is to provide a space for full information to develop. A solvency crisis, by contrast, is what happens when you have full information - and it turns out that, alas, the panic and investor rush for the exits were justified, because the information indicates that the assets were not properly valued and they really are not worth what the prices indicated in the absence of full information. As Taylor has cogently argued in his short book on the crisis (published by Hoover Press in a brilliant marketing move to get a short, pithy, blunt, highly informed book out fast), the errors of the US government in crisis management - beyond the original sin of loose money - have largely been an insistence on seeing the crisis as a liquidity crisis rather than a solvency crisis. (Of course, the analytic distinction is not, in real life, fast and hard: a true liquidity crisis can turn into a solvency crisis and vice-versa.) (It is noteworthy that the new Treasury Department White Paper on financial regulation reform is studiously silent and agnostic on the issue of liquidity versus solvency - apparently on the view that although the distinction mattered perhaps in the management of the crisis, in the distinct matter of future financial reform, it doesn't, presumably because if you enact the White Paper's reforms, you won't have crises requiring you to decide between them. Or at least, regulatory policy ex ante does not need to choose between them before there is a crisis and a panic or market freezeup. I'm not sure it is such a good idea to avoid expressing a view on something so fundamental to crisis response, however, even if regulatory policy ex ante would be approximately the same.) Scott and Taylor call for mandatory transparency in order to address the valuation and information problems. They call for transparency via a mandatory data base that would contain the basic information necessary for third party valuation: While the original MBS pools were often Securities and Exchange Commission (SEC) registered public offerings with considerable detail, CDOs were sold in private placements with confidentiality agreements. Moreover, the nature of the securitization process has made it extremely difficult to determine and follow losses and increasing risk from one tranche and pool to another, and to reach the information about the original borrowers that is needed to estimate future cash flows and price. This account makes it clear why transparency is so important. To deal with the problem, issuers of asset-backed securities should provide extensive detail in a uniform format about the composition of the original pools and their subsequent structure and performance, whether they were sold as SEC-registered offerings or private placements. By creating a centralized database with this information, the pricing process for the toxic assets becomes possible. Making such a database a reality will restart private securitization markets and will do more for the recovery of the economy than yet another redesign of administrative agency structures. If issuers are not forthcoming, then they should be required to file the information publicly with the SEC. My own tentative view is that this is important, but probably not enough in terms of either addressing complexity or leverage. Complexity is a problem that is only partly addressed by transparency. Just as many have called for CDS to go onto public exchanges, I would say that there is a pretty good argument - not perhaps dispositive, but one I incline to currently, absent some strong counterarguments - for requiring the same of the leveraged credit derivatives, to provide for mandatory disclosure of counterparty relationships, standardization of contracts and terms, mandatory exchange trading, and related measures that would address not just disclosure and transparency, but complexity as such, all by itself. I say this as someone who believes firmly, by the way, that a huge risk in new future regulation of financial services is over-regulation and the stifling of innovation through lack of credit. (Actually, my biggest fear today has evolved a step further, to a fear not so much that credit, and the rewards it can bring, will be unnecessarily stifled - but instead that our rapidly developing system of crony capitalism will cause credit to flow to politically favored parties, always in the name of high-minded things, naturally, but cronyism just the same and bleeding away vital investment funds from progress and innovation, but that's a post for another day.) There is then a further question of whether the degrees of leverage in these credit derivatives should be directly constrained by regulation. That question raises very different issues, mostly ones of moral hazard and the ability of regulators to decide better than the market how much credit is optimal, and whether we have embraced the too-big-to-fail view enshrined in the Treasury White Paper. If we have truly accepted too-big-to-fail for institutions outside of insured depository institutions, then regulation of leverage directly is hard to avoid, because the effect is dangerously to skew the market mechanisms for deciding how much credit is too much credit. If we have not, and are willing to contemplate the failure of large financial institutions - as I certainly think we should - then the market ought to be able, in the presence of full, transparent, and uniformly presented information, be able to police its own leverage as between greed and fear. My fundamental point here is that complexity is something not completely remedied by transparency. It retains characteristics and risks and dangers even where transparency is full and information is complete - but still really, really complicated in ways that in theory permit apples to apples comparisons, but in practical fact do not, and do not require them of a market still shielded in the short term from the consequences of its long term (a) mistakes in valuation; or (b) rational decisions not to bother with attempting accurate but complicated and expensive valuations and due diligence given the long term uncertainties about whether anything bad will really happen.