A top U.S. banker once said that the only time he gets worried is when he doesn't have a problem. A healthy bank, this gentleman believes, is a diversified one, and a diversified bank, by definition, always has a problem somewhere, one that is being offset by good to outstanding performance elsewhere in the organization.

The banker in question is a devotee of diversification. Counterparts in other banks aren't always as enthusiastic, however. Some have taken to minimizing the importance of diversification, arguing, as do many academics, that the shareholder doesn't value internal bank diversification because he or she can replicate its benefits simply by owning a basket of different bank stocks.

Not true. The academic maxim applies only in a perfect world where the cost of bankruptcy is minimal. In the real world the cost of bankruptcy is high, equal to its probability times an estimated 30% of company assets. So improved diversification should lift stock prices.

But there is such a thing as too much diversification. It occurs when the bank stretches so far that the costs of monitoring, coordinating, and planning for a group of disparate businesses (or loans) exceed the attendant risk-spreading benefits.

Examples of imprudent diversification include merging institutions solely to join two loan portfolios with different credit-loss profiles and, recalling the regrettable experiences of a few decades ago, expanding via loan production offices just to obtain geographic credit diversity.

Many banks recognize the threat of imprudent diversification. However, this threat should not blind institutions to the fact that most are as yet insufficiently diversified. This is almost axiomatic if, as is usually the case, the standard deviation of loan losses (the extent to which actual losses diverge from expected magnitudes) at individual banks substantially exceeds that for the industry as a whole.

Short of merging or interstate branching, diversification can be improved through loan sales and participations, loan securitization, and, importantly, through credit derivatives. More extensive use of such vehicles should have profound implications for the reallocation of bank capital.

The issues involved in capital allocation are too complex to be discussed exhaustively in this brief comment. Suffice it to note that the capital assigned to each loan should reflect its contribution to the bank's overall risk of unexpected loss. This, in turn, depends on the linkages among three major risk types: credit, interest rate (market), and operating.

Since these risks are not perfectly correlated, the bank benefits from having multiple sources of exposure rather than a unitary one. Otherwise stated, the fact that variations in the returns derived from taking credit, interest rate, and operating risk do not occur synchronously means that an institution exposed to all three risks can be safer than one exposed to only one.

It follows that the typical bank needs less capital to support its credit portfolio than it would if credit risk were its only source of earnings variation.

Unfortunately, interactions among the three major risk types are difficult if not impossible to measure. More happily, however, precise measurement may not be essential.

The reason is that the loss correlations in a well-diversified loan portfolio are already low enough to allow for substantial capital economies. That is, extensive diversification within the lending operation itself may reduce the unexpected loss potential of the bank so significantly that it isn't terribly vital to measure the extra fillip of protection resulting from, say, the imperfect loss correlation between the bank's loan business and its mismatch activities.

The more important question is one of allocating the benefits of growing diversification of the loan portfolio. A bank's credit portfolio consists of a group of loans with different degrees of loss linkage or correlation. Those loans with the weakest linkages to the portfolio to which they are added make the greatest contribution to reducing bank risk.

But this contribution is obviously a function of two influences-that of the loans themselves and that of the portfolio of which they became a part. How, then, does one allocate the capital benefit accruing from reduced loss exposure to each?

Stated somewhat differently, the marginal contribution of any given loan to risk reduction depends on when it is added to the portfolio. If it is the first loan to enter (say, when the bank opens its doors), its contribution to diversification and thus risk reduction is nonexistent.

If it is, say, the fifth to enter, its contribution can prove quite large. The fact that the benefits of diversification are "orderdependent" makes it singularly difficult to assign equitable capital rebates.

Yet, some resolution of the issue of how to parcel out diversification benefits seems needed. One of the reasons banks have been losing share to the public, nonintermediated markets is price. And one of the reasons for relatively high bank prices is underdiversification, which in the past has produced large loss volatilities and the concomitant need for hefty and expensive economic capital cushions.

As banks continue to improve their diversification they will be able to reduce their economic capital needs and thus relative loan prices. To turn this circumstance into an opportunity for enhanced market share, banks will have to come up with a reasonable plan for sharing the diversification dividend with their expectant customers.

If the interaction of a particular loan with the portfolio reduced the capital requirement for that loan by 200 basis points, that could yield a 30-basis-point potential price reduction, assuming a 15% bank hurdle rate. If the capital benefit were split fifty-fifty between the loan and the portfolio, the rebate would be 15 basis points, equivalent to the discount a bank could offer if it reduced its fully loaded credit operating costs by some 10%.

In the meantime, the gains from improved diversification are being pocketed by bank shareholders, which isn't such a bad thing, though it may be only temporary. To ensure that these gains are appropriate-i.e., individual loans and the loan portfolio as a whole are being assigned the right amount of economic capital-management must be extremely careful.

For example, the amount of capital to be assigned each loan is a function not only of its contribution to the unexpected loss risk of the bank (as we have noted), but also of the amount of overall unexpected loss risk the bank is willing to tolerate. Obviously, if the bank wants to be protected against bankruptcy 99.5% of the time, it needs a much plusher capital cushion than if it is content with being safeguarded 95% of the time.

The questions of how much protection a bank needs and should aspire to and, just as important, how precisely to achieve that level of protection are by no means easy ones to answer. In fact, answering these questions satisfactorily is one of management's toughest assignments, since it presupposes both sound judgment and considerable statistical and management science know how.

Subscribe Now

Access to authoritative analysis and perspective and our data-driven report series.

14-Day Free Trial

No credit card required. Complete access to articles, breaking news and industry data.