WASHINGTON - Federal Reserve Board economists have found snags in a model for figuring the capital levels banks should hold to guard against trading losses, which central bankers from large industrialized countries approved just a month ago.
The issue is critical because the model banks use to measure risk from trading activities in foreign-exchange or commodities markets determines the capital levels they must hold to cushion against losses.
Right now, regulators don't impose a capital requirement for the market- risk exposures banks take in trading account activities.
As U.S. bank regulators prepare to propose rules implementing the Basel draft, which came out mid-April, the study points out provocative shortcomings with it.
"The analysis suggests that bank internal models are not capable of measuring risk exposures over the relatively lengthy time interval of regulatory interest," wrote Fed researchers Paul H. Kupiec and James M. O'Brien. This is because statistical complications hinder estimates banks make of risk exposure over the long term, they say.
What's worse, regulators have no way to adequately verify whether bank internal models are reliable, say Mr. Kupiec and Mr. O'Brien, who clearly state that their opinions don't represent the views of the Fed. Gauging a model's reliability grows even trickier when using it to guard against long-shot cataclysmic losses, they say.
The Fed researchers say their results "suggest that an internal models approach for setting market risk capital requirements may be less than ideal if the accuracy of risk exposure estimates and the potential for independent verification are valued aspects of the regulatory capital system."
Officials at the Fed and the other bank regulators favor allowing banks to use their own models to measure these risks.