Wednesday, May 20, 2009

Whither Stress Testing?

OK, the stress of stress testing has come and gone, now what? If that is the first and last time we do this, it was a meaningless PR exercise.

Stress testing needs to be put on a solid, repeatable, and repeated basis.

First - Publish the stressors. What was the formula used to compute capital adequacy? Can I have that in XBRL, please?

Second - This is fun, let's do it every quarter! How long have people been complaining about financial statements being retrospective instruments? We need to embrace forecasting fully. We need to build up a body of time series of these things. Published in XBRL with no tweaking by the Fed. Transparency rocks!

Third - What's the error bar? Let's say I got my wish, and every quarter every bank in the US reported a matrix of capital adequacy forecasts across some increments of inflation and unemployment. Is that enough? No. We also want the probability assigned by the bank to each cell, and the uncertainty band around their reported figure.

Fourth - Why stop at banks? In this era of large companies sucking at the government teat to survive, we should be able to demand similar forecasts of other recipients. The capital market needs to eventually take up the responsibility of holding public companies up to a requirement of disclosure, not the regulators. In the meantime, regulators should prime the pump, and establish the set of data items to be provided. Standard setters, make some standards for prospective disclosures, please.

No comments: