According to the New York Times, “the European Central Bank said its analysis was intended to strengthen banks’ balance sheets, improve transparency in reporting and persuade investors that these institutions are sound.”
This intent would have some credibility if before 2008 regulators had pointed out the banks that were not sound and then if some of them failed during the crisis. But there were no such judgments implying that no one could tell before the crisis which banks were sound and which were unsound.
Bank failures generally come as a surprise in crises. This is because today’s assessment of the soundness of banks is based upon so many subjective assumptions that it fails to prepare banks for the surprises extreme crises bring. The weakest link in the way current stress testing is done is the assigning of values to assets. In relation to the survival of banks in crises, it is a meaningless exercise. This is because no one knows what value these assets will have during crises. What compounds this problem is that such values-in-crisis are assigned during normal times when it is hard to envision a crisis environment.
In a conversation a couple of years ago, a senior risk manager at a large financial institution said that no one had foreseen before 2008 that the value of some the mortgage-related assets would be down by 10-15%. Everyone thought, he said, that a 5% loss would be the maximum deterioration. Actual loss turned out to be more than 20%.
So today’s stress testing is designed with the last crisis in mind. Actual results may turn out to be not as bad as assumed in such testing in some areas considered critical, but some other surprise that no one had thought of may doom institutions.
There is a better way to make stress testing useful to both regulators and institutions. (See “What Stress Testing is Not”)
Also, more on how can stress testing can a very useful indispensable tool in the next blog.