What Would a More Stressful Test Tell Us?

The outline (PDF) of the methodology used to stress test the banks is out, and I already see some people arguing that it's not nearly stressful enough. And they have a point; the adverse scenario only has unemployment heading a little north of 10%.

Is that a problem? I don't actually think so. For one thing, it's likely to be damning enough as it is. It has to be, or markets will roll their eyes and move on. Tim Geithner can't help but give us all some useful information at this point, which is the point, after all.

And secondly, I actually don't know that there's much point in pushing for a more adverse adverse scenario. Things don't have to get much worse than the one they chose before every bank is once again in serious trouble, and before things get unpredictable in any case. What would the banking system look like if the United States hits 13% unemployment? Maybe there are models to figure this out, but I suspect no one can say much more than "awful, and with a lot more government intervention, of immediate necessity."

I'll feel silly for having written this if everyone walks out of the Treasury building with gold stars on their test papers, but I really don't think this is indicative of the administration trying to give banks a free pass.

Comments

[Moon Kil Woong:]

First all, banks are allowed to make their own assumptions about the effects of all their assets themselves based upon such a nebulous assumption. What if I assume there is 10% unemployment but well, I think my default rate is still about the same, especially if I have derivatives insuring them. Well why would a 10% defect ratio make them less solvent (probably they wouldn't since they are already not solvent).

What would be infinitely more helpful in determining bank risk is if they banned the off balance sheet accounting and required all banks to disclose their derivatives positions.

Oops that wouldn't work would it? Considering to do that would probably get you a 75% failure rate among the top 20 largest banks the next day. Well, maybe a bell curve is in order. A grade of D can be an A. An F can equal a C. If an F- equals a D then it works out the same. No one can fail. Then we can all go on our merry way totally oblivious again.

After all isn't that what Geithner wants us to do?

Apr 26 06:11 AM

[mrktcap:]

THEY ARE CLOSING MANY BANKS ON THE WEEKENDS: EVERY WEEKEND THERE HAS BEEN BANKS TAKEN OVER BY THE FDIC. BANKS WERE "FAIRLY PRICED" AT MARCH LOWS.

Idaho bank is fourth closed in a single day

By John Letzing
Last update: 8:16 p.m. EDT April 24, 2009

SAN FRANCISCO (MarketWatch) -- Ketchum, Idaho-based First Bank of Idaho became the fourth bank closed by regulators Friday, as the credit crisis continued claiming victims. The Federal Deposit Insurance Corp. said Minneapolis-based U.S. Bank (USB:US Bancorp
News , chart , profile , more
Last: 18.97+0.60+3.27%

4:01pm 04/24/2009

Delayed quote dataAdd to portfolio
Analyst
Create alert Insider
Discuss
Financials
Sponsored by:
USB 18.97, +0.60, +3.3%) has assumed the failed bank's deposits. First Bank of Idaho had $374 million in deposits as of Dec. 31, the FDIC said. The bank's closure follows that of other banks in Georgia, Michigan and California on Friday.

Apr 26 09:13 AM

[CautiousInv...:]

If this exercise can be conducted with baseline and adverse assumptions, why cannot it be done with yet more adverse assumptions?

The purpose of the stress test exercise is to give us some idea as to how bank capital will hold up in the face of growing losses and what must be done to raise more capital should it be required.

The more adverse scenario would simply be an extension of this process and reasoning and would shed light on a likely more severe capital deficiency and provide a ballpark estimate on what would be required to redress the problem.

Some believe knowing this type of information, even though subject to assumption and error, might allow better advance contingency planning. So if it happens, we will know how to respond unlike last fall when we were developing one-off solutions as the problems cropped up.

Apr 26 09:19 AM

[Fabien Hug:]

I read that the unemployment rate used for the stress test was 8.5%.
Difficult to get clear and accurate information in this context.
Anyway this stress test is not really stressful it seems.

Apr 26 10:09 AM

[Duke Wong:]

We don't know enough detail of the stress-test. wait and see.
If stress-test can evaluate the asses value transparently then it's good, the investors more comfortable and then finance will start lending and it will be good for economic.
Mark to market rule was eased that was make many investors uncomfortable.
I believed, last market rally was just a technical rally because many investors still stay on the side line. Hopefully, more investors will get in after the stress-test.
As Obama mentioned, the problem have to fix right, it can not change quickly. He believed, we are going in the right path.

Apr 26 11:58 AM

[henarl:]

The government does not want a more stressful test that might prove some very large banks to be terminally insolvent. As a previous poster noted, there have already been many small and medium banks fail and the FDIC could not stand to have a big one go down. This "stress test" is nothing more than a govt. PR scheme to pacify an angry population and prevent bank runs.

Apr 26 01:35 PM

[Rumpole:]

When looking at the stress test model, a SA contributer noted that the test focused on the most recent types of problem areas like commercial and residential real estate securities. But much in this most recent crisis has focused on derivatives like CDSs. Does the stress test adequatly deal with the derivatives issue? It doesn't seem like it.

Apr 26 02:35 PM

[JC09058:]

Having read the outline from the stress test and understanding the basic statistical methods being employed, I have a concern about the data collection date that was employed.

Credit utilization was one of the require items used in evaluating Credit Cards. While some institutions decreased credit lines prior to the February 2009 Data acquisition point, a larger number of credit line reductions have occurred afterward. This difference will skew the credit card data in a false positive effect than what currently exists today. In truth, credit card utilization currently has a more negative impact now.

How much more of an impact I can not say because I don't have the actual numbers. But this must be taken into consideration for it's impact on the whole.

Secondly, FICO made a major change, in 2008, in its scoring engine that is used by banks today. Knowing how the Scoring engine works at the detail level and knowing that many banks have not switch to the new scoring engine as this time, my concern here is that they are mixing apples and oranges if they do not correct for the differences between the engine types.

The key factor that I do not know is what scorecards are used by what banks and there in lies the problem. How an individual is score depends on the quality of the cardholders at each institution. So, each scorecard a bank uses is different than any other scorecard at all the rest of the institutions. Plus there are four different types of scorecards used within the scoring engine based on whether a cardholder is an excellent one ranging to the bottom of the poorest scoring cardholder.

So, FICO either had to supply the Fed another engine to create a correction factor for all of the credit card score information or apples are being compared to oranges and tangerines and pomegranates and etc.

Plus, FICO Loan scoring engine is different from the Credit card scoring engine as well because of the type of loan notes and how each individual pays those loans. It is a know fact the people in delinquent condition will favor certain credit instruments over other when paying loans or credit cards. Basically, loans will receive a more favorable payoff and reduced delinquency rate because people like sleeping in house over cars.

Lastly, no mention was made in the document stating whether FICO provided any kind of score adjustment engine or not to bring all of the scores into a baseline common alignment.

Those are my thoughts and concerns.

Apr 26 03:56 PM