Echanges is a financial journal published monthly by DFCG, the French association of financial officers and auditors. I was asked by Jean-François Casanova who will be the editor of a special issue devoted to securitization to contribute a piece on securitization in the light of the subprime crisis. Here it is, in an English translation.

Has securitization played a role in the subprime crisis and in the
credit dislocation that followed? If not, did securitization emerge unscathed from the turmoil where the role it played has been rightly or wrongly questioned? These are two valuable questions that need to find an answer.

Securitization’s primary aim is that of creating for vast collections of debts of a relatively low dollar amount, a new debt instrument behaving like a classical bond tradable on a fluid secondary market. Liquidity is supposed to follow although in the absence of a true organized market with a clearing house and market makers this may amount to wishful thinking. Lacking securitization, an individual mortgage will be parked within a “held for investment” portfolio until maturity. Securitization seemed to have changed once and for all the rules of the game until the recent crisis emphasized that the existence of a reliable secondary market remained problematic for all Residential Mortgage-Backed Securities (RMBS) among which Mortgage-Backed Securities backed by “prime” mortgages granted to the
most creditworthy borrowers and Asset-Backed Securities backed by “subprime” granted to borrowers with a spotted credit history.

An RMBS is exposed to risk in a double manner: firstly, because of the potential default of each of the individual borrowers of one of the thousands of individual mortgages that are packaged into an RMBS and, secondly, because of possible prepayment of the loan. Although the latter doesn’t materialize in a true loss, it still impairs the security’s profitability because of the opportunity cost of those interest cash flows that will fail to materialize. In theory securitization allows dispersion of risk between a large number of investors. Alan Greenspan used to underline an additional factor: that risk is now optimally distributed among those best able to bear it. This year’s events have undermined such certainties. It turned out that, first, if securitization does indeed allow in principle to redistribute risk it is far from clear if such dispersion has truly occurred or whether, conversely, concentration has taken place within a restricted number of portfolios; second, while risk has no doubt migrated to those investors who intentionally chose to bear it, there is no guarantee that any of them valuated that risk correctly. Belong no doubt to that second category, all those investors who acquired Asset-Backed Securities backed by subprime mortgages for the only reason of their high yield, with little worry of how permanent the credit arbitrage of that high yield would remain, as it was clearly not sheltered from switching all of a sudden from a profitable situation of risk over-valuation to a hurting one of under-valuation. The real estate bubble bursting led as we know to such a reversal of circumstances.

To assume, as did Alan Greenspan, that the capacity securitization holds for dispersing risk would automatically materialize in its optimal distribution entails some degree of naivety. It amounts indeed to imagining that those who purchase these high yield securities intend to manage them conservatively “for the long haul.” One would have expected that a keen insider such as Chairman Greenspan would not have subscribed to the layman’s view of speculation as a pathological condition rarely encountered among market participants but rather as the markets’ essential component that it is in truth. It was of course the crucially speculative nature of these securities – due to a circumstantial credit arbitrage opportunity – that would lead Asset-Backed Securities to end up concentrated within the hands of a selected number of entities(insurance companies; pension funds; hedge funds; foreign banks) betting that such an arbitrage would persist, instead of getting “statistically” scattered among their optimal stakeholders.

Add another dimension, this time internal to the security’s machinery: the confidence held by the structurers of ABS in their ability of fully concentrating risk in the junior certificates of these securities so to exclude it entirely from the senior ones. From that trust derives its corollary that loans exposed to serious default risk could nonetheless be used as innocuous “stuffing” within CDOs (Collateralized Debt Obligations) assigned top rating. Such trust rests on the multipliers used in the stress tests of these securities’ credit enhancement structure, i.e. the waterfall reallocating cash flows from the subordinate certificates to the senior ones. The value of these multipliers is supposedly supported by statistical argument; e.g. Standard & Poor’s common “stress test” assumption that the probability of a loss 4.5 times higher than the highest historical loss is close to zero. There is in truth nothing to back that assumption which betrays a fatally flawed understanding of statistical inference, i.e. that the frequencies observed in a time-series of arbitrary length can be extrapolated as probabilities when the phenomenon observed is undoubtedly by nature cyclical (or at least unstable) as it is the case with Housing Price Appreciation.

Some have also debated the role played by securitization in severing the link between a lender’s financial responsibility and the individual loans finding their way into an RMBS. Fingers have been pointed at FASB 140 in particular as it allows recording an immediate gain on sale at securitization, encouraging therefore issuers to ignore how well-equipped a borrower is when paying interest and refunding principal over a number of years. Lenders relaxed underwriting standards as Wall Street’s capital markets showed infinite willingness to accept loans whatever the quality of both collateral and borrower. What lenders lost sight of in such a forgiving climate was that on the long term, quality only would
guarantee the survival of an active market for the product. Greenspan and others had envisaged that the man-in-the-street would represent the typical RMBS investor. What the crisis revealed was that RMBS investors would instead be savvy experts in their field, well-aware of risks incurred, who would briskly withdraw from the market as soon as the credit arbitrage had vanished. This is indeed what happened at the end of 2006 when foreclosures shot up as the real estate bubble burst
and the price of housing started to dive. The new context was one where the default risk premium in an ABS’ coupon would this time be under-evaluated.

Answering thus the first question: did securitization play a role in
the subprime crisis and in the credit crunch that followed?

The answer to that question is far from clear-cut. Let’s recap: MBS securitize mortgages where a portion of the coupon is risk-based and represents a premium reflecting – supposedly adequately – the default risk embodied by the individual borrower. If that premium is calculated in such way that a full business cycle has been encapsulated, there will necessarily be an alternation of times when there exists for the investor a potential credit arbitrage – as the risk premium is then over-valuated – and other times where the arbitrage will be absent – as the risk premium is then under-valuated. During periods of the first type, securities will be attractive due to the presence of such an arbitrage while in periods of the second type they will not. During the latter, the product’s appeal lies in the effectiveness of its credit enhancement: i.e. in the robustness of the waterfall redirecting cash flows from the junior certificates to the senior ones. Should that structure then be deficient, the product would lose its appeal and its market would dry up in no time – as the events of August 2007 convincingly confirmed.

As such, securitization is therefore not to blame in the subprime crisis. What is at stake however is an RMBS’ capacity at being at all times in the real estate’s business cycle a valuable product for both issuer and investor. To meet these ends, an RMBS needs to be correctly structured.

Second question, this time in reverse: what is the future of Residential Mortgage-Backed Securities and more particularly of Asset-Backed Securities backed by subprime mortgages; are they likely to survive the crisis they are undergoing this year?

The teaching is straightforward: RMBS’ future is linked to the product getting immunized against the hazards of the real estate’s business cycle. Credit scoring reflects correctly the idiosyncratic differences between individual borrowers (and to that extent it ranks them satisfactorily) but it has no power for expressing the impact on their credit behavior of a full business cycle. This turns out to be a major defect of any credit scoring of individual consumers like with the all-pervasive FICO score. The scoring system turns out to be unable of assessing objectively the risk that the borrower imposes on the lender as the score conflates within a single number a consumer’s credit ability, the good health of the real estate sector as well as that of the global economy. To that extent, embedding in a mortgage’s note rate a risk premium reflecting mechanically a credit score will fail to ensure that a credit arbitrage is maintained for the security at all times in the business cycle. Should this difficulty reveal itself as being unsolvable, a different credit enhancement structure would be called for, possibly fully external as with insurance or a Letter of Credit. Residential Mortgage-Backed Securities‘ viability would then be determined by the capacity of insurers in that specialized field to absorb the shocks resulting from alternating phases of the business cycle. Recent news about the financial health of the companies occupying that niche is not particularly encouraging in that respect.