International regulators are asking for recommendations on how to improve credit risk modeling so that it can aid in setting capital requirements.
The Basel Committee for Banking Supervision said Monday that it concluded, from a study of how 20 internationally active banks modeled credit risk, that "significant hurdles" must be overcome before regulators can rely on these techniques.
"There has been a lot of very good work done in measuring credit risk," said Miguel D. Browne, deputy assistant director of capital markets at the Federal Deposit Insurance Corp. and a member of the Basel task force that wrote the report.
"But the state of the art in computer modeling is not now at the point where we can base a capital charge on the results of a credit risk model."
Susan F. Krause, senior deputy comptroller for international affairs at the Office of the Comptroller of the Currency, said bankers should not expect the Basel Committee to reverse course anytime soon. "People generally think it will take at least five years," Ms. Krause said.
The Basel Committee had planned to unveil the report last month as part of its proposed overhaul of the 10-year-old risk-based capital accord. That broader plan would have increased capital charges on loans to hedge funds and to governments in developing countries, but was delayed indefinitely after German regulators raised last-minute objections over real estate lending.
A regulator said the Basel Committee saw no need to further postpone release of the credit risk paper, which was unveiled in Washington by the Federal Reserve Board.
The Basel study found that few banks use the same method to measure credit risk. For instance, some do not record a credit loss until the borrower defaults. Others might record the loan at its market price, which means the loss is recorded gradually rather than all at once.
Similarly, some banks rely on the judgment of credit officers to determine how much of a loan is likely to be recovered, while others use empirical formulas.
Banks also vary in how they aggregate credit risk. Some measure it at the individual loan level, while others measure the risk of an entire portfolio, the report said.
A second major set of problems involves the lack of historical data on credit quality. For instance, few banks have enough data on the performance of loans to make accurate predictions about credit quality. This forces banks to make many assumptions when calculating credit risk, ahd these assumptions lessen the model's accuracy, the report said.
It concluded that there is not a single, industry-accepted approach to verifying the accuracy of credit-risk models. Also, the accuracy of these models often depends upon the amount of management oversight and the quality of internal controls, two factors that vary greatly among banks.
Comments on the study are due Oct. 1. Mr. Browne said regulators want banks to recommend approaches the entire industry could accept for removing the impediments to using credit risk models.
Such a consensus permitted the Basel Committee in 1996 to adopt a rule letting banks use models to set market risk capital requirements for their trading portfolios. "The idea is to say: How we can overcome these obstacles?" Mr. Browne said.
Several observers criticized the report.
"It is disappointing that they would not allow any reliance on credit risk models that have been developed and used by banks for years and years," said Richard M. Whiting, acting executive director of the Financial Services Roundtable, which plans to release a report in two weeks addressing many of the Basel Committee's concerns.
"This just illustrates why the whole Basel process is a failure," said Charles W. Calomiris, a professor at Columbia University and co-director of the project of financial deregulation at the American Enterprise Institute. "The Basel Committee wants to come up with the model, but of course there is no single market."
Instead, Basel should study how investors decide bid prices on bank debt, he said. "The market already models credit risk every day," Mr. Calomiris said. "Anytime the market rates a bank's debt, it is making an implicit measurement of the its risk position. Otherwise it could not have come up with the price for the debt."
Ms. Krause said Mr. Calomiris' proposal would not work, because the available information on a bank varies by country. Only a uniform formula can ensure that banks compete on a level playing field, she said
"The Basel Committee would like the best of both worlds," Ms. Krause. "We would like to improve transparency to get more market discipline, but there is also a desire to have a quantifiable capital requirement."