Saturday, April 25, 2009

Stress Test Constraint

From the New York Times:

Officials have said that at least a handful of the 19 banks undergoing the exam will need to raise sizable amounts of fresh capital, a move that will severely dilute existing shareholders. For the tests to be credible, regulators recognize that not all of the banks can pass the exam with flying colors.

So, this is not like a test with objective criteria. There will be 'several' losers. There has to be.

Hat tip to Matt Stichnoth


Anonymous said...

The Bank Exam is a proportionate test. It wants to identify the inadequate, & may punish the lower portion of the adequate, to drive up standards whilst ensuring the capital injections required are not too excessive to be financed & not all major banks are declared capital deficient.

Don't be deceived into thinking absolute standards are good & proportionate standards bad. After all, who sets the absolute pass or fail standard? Would you rather be driven by a driver who qualified to an absolute standard set in a dubious jurisdiction or be in the top 30% of all drivers you meet on the roads?

J said...

Proportionate tests have no real standards. A perfectly fit testee could be punished only because it is different in some parameters from the group's average. On the other hand, the whole group tested could be unfit and liable to fail. Applying a proportionate test to banks reveals that they dont know what they are testing nor why.

Anonymous said...

This the whole fallacy. A proportionate test has the same standard as an absolute test. In an absolute test all above x pass, in a proportionate test all above y pass. The only difference is that x is a fixed value under all conditions whilst y is a variable dependent on conditions.

Absolute standards are important when there is a choice - only drive if you are good enough - but there is no choice with banks. As the government rescues show, we MUST deal with them, so don't all pass, or all fail, identify the "best" & force the others to raise their standards.

J said...

In an absolute test, you have to have to have an idea of what you are doing and why. Since in this case you are unable to calculate x, you assume it is the group's average and force the "weaker" testees to reach it.

Isnt better to let "life" take its course instead of intervening on such a problematic logical foundation?

Anonymous said...

WSJ: "The regulators are asking "a million questions" and it's "very unclear what they're aiming at," one senior executive said earlier this month. "We can't discern a pattern."

"Some bank executives have said that even after meeting with Fed examiners on Friday, they still don't understand details of the government's methodology for conducting the tests."

J said...

The publication of the results of the stress tests has been postponed. I have a feeling that the testers and the testing methodology have failed.

Anonymous, I am expecting your refutation.

J said...

Miller, a former bank examiner, said his estimate assumes regulators will require banks to maintain tangible common equity, one of the most conservative measures of capital, equal to 4 percent of their risk-weighted assets over the next two years, to withstand losses in case the recession worsens.

Well, it appears that this was NOT a proportionate test. 4% is quite reasonable, I daresay.

Anonymous said...

J, I'm not sure what you are expecting me to refute. That the results are delayed? - that's a fact not an argument. That the test has failed? Who knows? But you are categoric that it was NOT a proportionate test. So are you saying it failed because it was absolute? You then say 4% capital is reasonable - proportionate to what? 4% is nothing more than the Basel standard for core Tier 1 capital - overall capital minimum of 8% of RWAs with no more than 50% in Tier 2. There's nothing absolute about 4%, it's the number some banks exceed, others can readily reach & some will have to struggle, & maybe fail to reach. That way we get a proportion of passes & fails. Nobody wants all banks to fail, which an absolute test might produce. If the delay is a result of "failure" it may be because the absolute pass mark was too high.