Joseph Pimbley, a finance professional in New York, explains why ratings agencies could provide a far better barometer for bank capital needs than the current BIS guidelines.

In the 1988 Basel Capital Accord, the Bank for International Settlements promulgated a set of fairly simple rules to determine the adequate capital a bank must hold to earn the right to operate internationally. The greater the capitalization of a bank, of course, the less likely it is that the bank's depositors or senior creditors will suffer any loss. Requiring a bank to hold adequate capital is a reasonable method of ensuring the soundness of the banking system.

Now, because of weaknesses in the capital adequacy rules and in response to new developments in the banking world, the BIS is seeking to amend the rules for computing adequate capital. Its approach ignores a natural, and far more accurate, method for assessing a bank's ability to perform on its senior obligations: reliance on the ratings that nongovernment rating agencies assign to the bank.

The current rules

The 1988 accord requires that banks hold minimum capital equal to 8 percent of the debt obligations they hold as assets. If a bank lends $10 million to a corporate client, for example, it must hold at least $800,000 in capital in support of the loan. This calculation presumes that the bank's risk is that the default of many of its borrowers would prevent the bank from honoring its own obligations to depositors and other senior creditors. A bank must also hold capital against its noncredit (that is, market) risks, but this market risk capital tends to be far less than credit risk capital.

The accord gives disastrously wrong answers that translate into a systemic inability of the bank to make intelligent lending decisions.

Since loans to different clients expose the bank to different levels of credit risk, the accord seeks to reflect disparities in risk by assigning lower risk weightings to some specified borrowers. Under the accord, most borrowers have a 100 percent risk weighting, which means that, as in the example above, the bank needs to hold at least 8 percent of the full $10 million loan size to the corporate client. For some borrowers, however, the bank is able to compute a reduced capital requirement by multiplying the loan size by a risk weighting that is less than 100 percent. Loans to OECD banks, for example, carry a 20 percent risk weighting. Hence, a bank treats a $10 million loan to an OECD bank as if it were only a $2 million loan. The necessary capital is only $160,000 (8 percent of the $2 million "risk asset” size of the loan).

Loans and other forms of credit exposure to OECD governments, meanwhile, carry a 0 percent risk weighting. The requirement to hold absolutely no capital against such loans is consistent with the premise that these loans have no credit risk. Also in this 0 percent category are unfunded commitments to lend to any entity that mature in less than one year. Finally, there is a 50 percent risk weighting category that includes unfunded commitments with a maturity greater than one year, some subsovereign government obligations and sundry other instruments.

The problem

These rules form the basis for BIS capital adequacy, and provide a simple framework for a bank and its regulators to judge whether the bank is sufficiently safe (admittedly a vague term). The rules are useful in that they specify increased capital for increased risk. Even if the resulting capital level is wrong, the BIS model of capital adequacy forces all players to think in terms of capital vs. risk.

But there are two principal failures of the accord. First, it gives disastrously wrong answers. Second, the simplicity and power of the rules infiltrate and permeate the culture of a bank so that the wrong answers translate into a systemic inability of the bank to make intelligent lending decisions.

The accord gives wrong answers in a number of ways. There is no distinction between a triple-A-rated corporate borrower and a single-B borrower. Both carry a 100 percent risk weighting. A bank that lends primarily to speculative-grade entities will intuitively need much more capital than a bank that lends mostly to investment-grade borrowers. But the accord is blind to this obvious difference in loan quality. Furthermore, there are many OECD banks and governments that are significant credit risks and thus do not deserve risk weightings of 20 percent and 0 percent, respectively. Finally, the accord gives no benefit to the diversification of credit risk.

Consequently, the 1988 accord is clearly and deeply flawed in its assignment of capital to bank credit risk. An optimist might expect that bank managers, who are experts in credit risk, would ignore the accord's distorted risk measurements. The optimist would be wrong! In banks below the top tier, the return on BIS capital, or, alternatively, the return on risk assets, is a critical factor in whether management approves a new transaction. An excellent example is the 364-day unfunded lending commitment (which carries a 0 percent risk weighting). Quite often the bank asks for a facility fee that is 10 basis points per annum or less, even when the true default risk of the borrower exceeds 30 basis points per annum. The bank greatly underprices these unfunded facilities due to the favorable risk weighting. It is as if the bank actually believes the BIS prescription that there is zero risk. That is, even if individual bank managers claim to understand that the accord rules do not reflect true risk, their decisions do not similarly reflect that alleged comprehension. The institutional stupidity is completely sensible. The manager has two objectives: maintain 8 percent capital adequacy and make as much net income as possible. As the manager evaluates a new loan or extension of credit, it is quite natural to measure the transaction desirability by the income divided by regulatory capital.

Here's the point. Not only does the 1988 accord fail to distinguish between high-quality and low-quality banks, it also encourages banks to behave in a low-quality fashion. In June 1999, the BIS proposed relatively minor modifications to the accord. These modifications are generally positive and in the right direction. Double-A and triple-A borrowers, for example, would now have a 20 percent risk weighting, as would the unfunded commitments maturing in less than one year. The former reduces the current risk weighting, while the latter increases the current risk weighting. These steps, as well as others, are both sensible and incremental.

Very interestingly, note that the BIS appears to accept in principle the dependence of its capital adequacy regime on public (that is, nongovernment) agency credit ratings. This development is a tremendous change from the 1988 accord. Why else would the BIS have used OECD rather than investment-grade as a qualifier?

The solution

The BIS should simply defer to the credit ratings that nongovernment agencies such as Moody's Investors Service, Standard & Poor's, Fitch/IBCA and others assign. Throw out the rules. Throw out the crude risk weighting categories. If Moody's, S&P, Fitch or another qualified agency assigns a single-A rating or higher to a particular bank, then the BIS will view that bank as suitable for international operations.

Whether the new method is single-A or higher, or investment-grade, would be at the BIS's discretion. The BIS should review the ratings of current banks it views as adequate and inadequate in order to set the threshold it desires. The BIS should also study agency default probabilities for various ratings. The BIS default probability tolerance will impact the rating threshold.

The bis should defer to the credit ratings that nongovernment agencies assign.

The advantages of "rating regulation” are numerous. The primary advantages are simplicity, high quality and flexibility for market innovation. The simplicity advantage is, in truth, simplicity for the regulator. The regulator will no longer need to define and defend a framework that must fit all banks. If the agency credit ratings are correct in the sense that they are equal to or better than any other assessment in terms of the accuracy of predicting default, then the regulators should be greatly relieved to step out of the process.

Simply put, the rating agencies have in place a much more profound and accurate ability to assess bank credit quality than does any regulator. A key market measure of credit quality for a bank, for example, is its funding cost. It's a safe prediction to say that a bank's credit rating is far more predictive of its funding cost than is the bank's BIS regulatory capital adequacy.

Rating agencies are also much faster to adapt to changing environments than are regulators. They do not hesitate to use their judgment to factor in new risks and new risk mitigants as the market produces them. A rating agency can lose its entire business franchise if the market perceives its analyses to be anachronistic (or just plain wrong). Regulators don't lose their jobs for moving too slowly.

There are a number of serious questions one must raise and answer regarding this sort of rating regulation. First and foremost, is it really proper for regulators to defer to nongovernment credit rating agencies? More specifically, should the BIS tell Bank XYZ that it cannot operate internationally without a single-A rating from, say, S&P? This means Bank XYZ's future is strongly contingent on S&P. Wouldn't this give XYZ an incentive to pay S&P under the table for a favorable rating? After all, the security of government regulation is that there's no profit motive.

Yes, these issues are present. But they exist already. The S&P rating is already critical to XYZ for purposes of funding costs. Given the extreme importance of funding cost to the bank, there is already the potential this bank could pay S&P for a favorable rating. That situation, then, will not change. Rather, the BIS simply becomes another customer for agency credit ratings.

Another objection is the question, Which rating agencies are allowed? Is it fair to include only the three or four main agencies? Probably not, but while the BIS need not be fair, it must have a mechanism for determining which firms that call themselves rating agencies will be part of the club for BIS bank review purposes. To avoid sham rating agencies, the BIS should require, among other things, that such firms have adequate market share for bank ratings and that investors use the ratings. In other words, the BIS should use new rating agencies only if real investors use these agencies.

Will banks disclose all relevant information to rating agencies? There are two types of nondisclosure: open refusal to disclose and fraud. If a bank tells a rating agency that it will not describe certain practices or positions, then the agency is free to withhold the rating. In fact, the agency SHOULD withhold the rating, or assign a conservatively low rating, if and only if the agency feels that such information would affect the rating. This open refusal to disclose, then, is not a problem for current ratings or for these future bank ratings for BIS purposes.

Fraud, on the other hand, may be different. A bank that sells debt while misstating financial information to rating agencies and investors is guilty of fraud. The government can prosecute the responsible managers and directors. If the penalties for lying to a government regulator are more severe than those for misleading rating agencies and investors, then critics of rating regulation might argue that it would make fraud more likely.

Healthy debate should ensue! The current regulatory rules have hurt banks more than they have helped by providing a measure of risk that is often wrong. A transition to rating regulation or something better will boost the global financial community tremendously.