Saturday, April 25, 2009

The Stress Test Cliff Notes

Compliments of reader Tim.

For those too lazy to click on the Federal Reserve link and read, below are some of the key points contained in the SCAP:

(1) “more than 150 senior supervisors, on-site examiners, analysts and economists” spent a month reviewing the 19 BHC’s that hold two thirds of the country’s bank assets and account for one half of the loans

More than 150 means at least 151. Is the US Iceland or something? Ten trillion dollars in assets and five hundred trillion dollars in derivatives in one month? A typical single bank examination utilizes hundreds of examiners and takes several months. Clearly the next release of public sector productivity numbers is going to astonish.

(2) ”the firms were asked to project…..the firms were asked to provide…etc.”

In other words, the banks tested themselves and the 150 examiners took their word for it. Any wonder they passed?

(3) “traditional role of capital, especially common equity, is to absorb unexpected losses and thus to protect depositors and other creditors. ”

Well, it used to be. Now that job has been outsourced to the US Taxpayer.

(4) “As a result of the loss recognition framework for assets in the accrual loan book, the results of this exercise are not comparable with those that would evaluate such assets on a mark-to-market basis”.

Absolutely. What does the market know anyway? The banks’ models got us into this calamity so damn if they can’t get us out!

(5) “The SCAP analysis is forward looking, but over a limited time horizon. Losses and resources are projected over a two-year period (2009 to 2010) …”

Apparently Treasury is absolutely certain that the crisis they never saw coming will be over soon and in no way is today like 1930. Great news!

(6) “Each participating firm was instructed to project potential losses on its loan, investment, and trading securities portfolios, including off-balance sheet commitments and contingent liabilities and exposures over the two-year horizon beginning with year-end 2008 financial statement data. “

Again, Treasury outsourced the testing to the banks themselves. So what was the job of the Treasury staff other than to photocopy , collate and file? Was this a Temp Staff? Kelly Girls?

(7) “Firms were allowed to diverge from the indicative loss rates where they could provide evidence that their estimated loss rates were appropriate. ”

I know it looks bad, but believe me, it’s getting better!

(8) Chart Page 9

Under the baseline case the economy stops its downward acceleration in Q3 2009. In the so-called adverse case this occurs in…Q3 2009.

(9) House Price Index Projections Chart, Page 9

Best case is an additional 17% fall from Dec 2008, worst is 27%. The Treasury tripped over itself here, as WFC and BAC have massive mortgage portfolios in states like CA where prices are already off 45% and the banks have only provisioned for, at most, 30%. Guess we can go back to Point 7 to whitewash this inconvenience.

(10) “Under the baseline scenario, BHCs were instructed to assume no further losses beyond current marks”

It’s over! Great news!

(11) “For other consumer loans and for commercial lending (including various types of commercial real estate lending), the agencies estimated loss rates using techniques such as regressions of historical charge-off or default data against macroeconomic variables such as home price appreciation “

I know there are some modelers out there. How does one reasonably construct a regression analysis on past data when nothing similar has ever happened?

(12) “Supervisors evaluated firm loss estimates using a Monte Carlo simulation that projected a distribution of losses by examining potential dispersion around central probabilities of default.”

Ah…smells like Gaussian distributions. The old standard. We have seen how well that assumption works in these unusual times. An example of the dependability of using Gauss, taken from stock market movements in October, and calculated by Nassim Nicholas Taleb of Black Swan fame, showed that the price movements seen in October 2008 could be expected to occur---using estimates based on Gaussian distributions---once every 73,000,000,000,000,000,000,000 years. For those of you not tied to Biblical strict constructionism, the Universe is around 18,500,000,000 years old. Looks like it will be a few quintillion years before we see October again.

(13) Appendix

Take a good look at just how complex and thorough this “test” was. An online dating questionnaire is more probing. Sphere: Related Content
Print this post
blog comments powered by Disqus