Forgetting the Three Cs of Lending

At its core, lending is a simple business. Its goal is to provide money to people and then to be repaid with a profit. From times immemorial, prudent banking granted loans according to three main principles:

1) Credit and character – The borrower had to have a track record of having met his obligations and be considered trustworthy.

2) Capacity – The borrower had to be able to pay the debt from existing income.

3) Collateral – The value of the collateral had to exceed the loan balance.

The financial crisis of 2008 is a crisis of structured loans that violated, to varying degrees, all of the above principles; particularly the second. As a result, U.S. and foreign financial institutions now hold large portfolios of promises that can no longer be credibly valued. How were these portfolios valued in the first place? We describe a very abstract statistical model that the rating agencies did not, as it turned out, consistently apply. The following discussion is for our readers who are curious about the nature of the investments at the heart of this financial crisis.

In 1996, Moody’s published a report called, “The Binomial Expansion Method Applied to (Collateralized Bond Obligations and Collateralized Loan Obligations)”. The paper utilized the binomial statistical model to estimate the expected losses and therefore the credit ratings of various structured mortgage tranches. The binomial statistical model is a way of handling a series of discrete observations, such as a given number of loan defaults within a mortgage pool. As the number of loans becomes large, the probability of having a given number of defaults converges to the probability given by the bell-shaped Gaussian distribution. The Moody’s paper promised that “(the) Binomial Expansion Technique...combines the best of two worlds: a high degree of accuracy coupled with computational friendliness…” All quantitative models, however, are only as good as their assumptions. Our comments on this report are in italics.

The BET (binomial expansion technique) method is based on the diversity score concept. (This merely says that as the number of cases increases, the probability of defaults decreases due to some diversification.) The idea is to use the diversity score to build a hypothetical pool (as it turned out very hypothetical, that had not much connection with reality) of uncorrelated (subprime mortgages were very correlated with the trend of housing prices) and homogeneous assets (the assets were homogeneous because they weren’t structured to the situations of the borrowers) …that will mimic (they hoped) the default behavior of the original pool…

Finally, as far as defaults are concerned, the behavior of this homogeneous pool of D assets (that’s all the diversity score was) can be fully described in terms of D possible scenarios: one default, two defaults…up to D defaults. The probability P_j that scenario j (j defaults) could happen can be computed using the so-called binomial formula:

(we summarize) Pj = f(D, j, p)

(D= the number of cases, j is the number of defaults, p as defined below)

Where p represents the weighted average probability of default of (any particular mortgage in) the pool…(This assumption was based on historical experience. The concept of an average default probability of any particular mortgage was likely a meaningless concept, since there was an increasing trend of bad mortgage underwriting in 2006 and 2007).

Let Ej be the loss for the note to be rated under scenario j. (The loss, expressed as a percentage, can be easily computed by taking the present value of the cash flows received by the note holder, assuming there are j defaults, and using the note coupon as the discount factor).

Finally, the total expected loss, considering all possible default scenarios, is calculated as follows:

D

Expected Loss = Sum P_jE_j

j = 1

(Banks have loan loss reserves. But loans should not be approved without the justified expectation

of repayment, because otherwise reality is guaranteed to exceed. We also note that financial institutions

are highly leveraged.)

The structured finance ratings methodology contained unrealistic statistical reasoning that was inappropriate to the billions of dollars squandered. According to this methodology, the probability of a large number of simultaneous mortgage defaults in a large portfolio was very small, essentially following a Gaussian distribution. But the underlying loans were not made according to the three simple lending principles above. With the collapse of the Gaussian convention, there is now no accepted way to value large portfolios of problem subprime and Alt-A loans except by analyzing each loan individually, renegotiating if necessary, then simply summing their totals - when house prices stabilize. Call this belated due diligence that now has to occur at the loan servicing level.

We were curious about how these portfolios were valued. The analysis above allows us to trace present financial problems to faulty loan underwriting and deal structure. To say that large portfolios of subprime and Alt-A mortgages cannot be valued is not to say that the individual mortgages are valueless. Some of the mortgages will pay as agreed, others can be renegotiated and restructured.* However, a blind faith in financial risk models that assumed a fairly placid Gaussian market is at the root of the financial crisis of 2008 - when the real markets proved otherwise.

“It is hard for us, without being flippant, to even see a scenario within any kind of realm of reason that would see us losing one dollar in any of those transactions.”

A Former Structured Finance Executive

August, 2007

“We foolishly believed that the firms had a strong culture of self-preservation and responsibility and would have the discipline not to be excessively borrowing…Letting the firms police themselves made sense to me because I didn’t think the S.E.C. had the staff and wherewithal to impose its own standards and I foolishly thought the market would impose its own self-discipline. We’ve all learned a terrible lesson…”

Professor James D. Cox

Duke School of Law

The banks that structured the securities and investors both failed to do “fundamental credit analysis,” said Janet Tavakoli… “They were using correlation models, they were using spread models, but they weren’t doing analysis on the underlying corporations (borrowers).”

Bloomberg

October, 2008

* We do not comment upon the CBO/CLO derivatives that also bedevil the financial system.