Skip navigation

Monthly Archives: January 2010

QuantumRisk is starting a new CMBS service – providing econometric loss forecast for CMBS deals.

Why econometric CMBS loss forecasting? Our studies show that

1. The Triangular  Matrix Method is incorrrect (Excel 2007 example).

2. DSCR are not good predictors of defaults (see brochure).

3. Stress testing usually not done correctly (see post).

4. Black Swans can be used to differentiate between two deals with similar cashflows (see brochure).

Contact  Ben Solomon ( for more details. Company brochure here.

Several news items (WSJ’s Tishman Venture Gives Up Stuyvesant Project, Control of Stuyvesant Takes Center Stage and the NYT’s Fallout Is Wide in Failed Deal for Stuyvesant Town, As Doubts Grow, U.S. Will Judge Banks’ Stability) got me thinking about stress testing.

Stress testing is an approach to evaluating how a deal or a balance sheet will behave when selected factors are changed to induce a loss. It is done to answer the question what is the downside financial risk should a bad situation occur?

Around February 2009, the Feds and the Treasury impleted stress testing of the largest banks. This illustrates that stress testing is used in both the private and public sectors. And for those who are not familiar with this methodology, it is not a Republican or Democrat issue, as we can see that Ben Bernake is a Republican and Timothy Geithner is an independent having been a Republican.

Figure 1 illustrates a single factor that is used to stress test the deal or the balance sheet. I purchased some structured finance models a few weeks ago and was reviewing them. My observations about 3 important failings of the stress tests are based partly on these models:

Figure 1: Factors Skewed Right

1. Right/Wrong Distribution: I realized that the inherent assumption behind one of the stress parameters, the loss statistic, is that it is Normally distributed, because it is based on average values. In this example the mean is the 5.0% dashed green vertical line.  Therefore, this factor should behave like the red-dashed Normal curve in Figure 1. However, losses are skewed right and fat tailed per the blue curve. The mean is still 5.0%, however, the mode is 4.6% (solid purple vertical) with a long tail extending out beyond 17.9%.

2.  Insufficient: Stress testing multiplies the loss number by 2x to 10.0% or 3x to 15.0%. In this example 15.0% is insufficient to cover 17.9% tail loss.

Figure 2: Factors Skewed Left

3. Irrelevant: In Figure 2, we see that the skew is to the left. As an example some returns distributions can be skewed to the left. Therefore, the stress test has no bearings with reality. The stress test stresses the loss statistic in a manner that can never happen because it is outside both Normal and Left Skewed distributions, and therefore, the results are irrelevant to the real world.

So take care with your stress testing.


Disclosure: I’m a capitalist too, and my musings & opinions on this blog are for informational/educational purposes and part of my efforts to learn from the mistakes of other people. Hope you do, too. These musings are not to be taken as financial advise, and are based on data that is assumed to be correct. Therefore, my opinions are subject to change without notice. This blog is not intended to either negate or advocate any persons, entity, product, services or political position neither is it to be construed as investment advice.

Contact: Ben Solomon, Managing Principal, QuantumRisk