How to Measure The Adequacy Of Loss Reserves
Bankers are now besieged with regulatory demands that they increase loss reserves. Are these demands justified?
Before addressing this question, it must first be pointed out that there is no economic distinction between capital and reserves. Together they represent a buffer between the value of the bank's claims on others (borrowers) and the value of the claims of others (depositors and creditors) on the bank.
If the buffer is in the form of capital, the bank's gross asset value will be larger than if reserves are subtracted, but, absent these reserves, the valuation is more questionable, necessitating a commensurately larger amount of capital backing. If the buffer is in the form of reserves, the value of the assets will be lower but also more reliable, necessitating a commensurately smaller amount of capital support.
From the marketplace's vantage, the division between capital and reserves is a distinction without a difference. The evidence suggests that, in valuing banks, the marketplace considers the size and adequacy of the overall buffer rather than its separation into capital and reserves. As a corollary, banks that penalize current income in order to create large reserves fare no worse in a valuation sense than those of comparable asset quality that report more income and thus have larger retained-earnings accounts.
What banks and regulators should be debating is the adequacy of the total buffer. More specifically, both should be interested in determining whether the buffer is sufficient to cover expected- and unexpected-loss risks in the credit portfolio. It is distressing that the regulators are pressuring many banks to increase their reserves (capital) without having made this determination. It is equally distressing that many banks reflexively oppose regulatory demands without having made a comparable determination.
A Model Solution
Establishing the adequacy of a bank's buffer presupposes the following steps: The expected probability of default for all loans must first be assessed. This is best done through the use of models that have been described in previous articles (see this space, Sept. 12).
The KMV Corp., my firm's joint venture partner, has proven models that measure default probabilities for both publicly traded and privately held firms. That for public companies, for example, calculates default probabilities by analyzing stockprice movements. A company's default likelihood equals the amount of downside stock-price volatility that would lower the value of equity to the point where, together with the associated decline in the value of the over-one-year debt, the value of the company's assets, definitionally equal to the value of its liabilities, is no greater than the face amount of its short-term debt.
The default probabilities churned out by this and the other models must then be converted into expected-loss magnitudes by adjusting for likely recovery rates and for the cost of carrying nonperforming loans and the administrative expenses of loan workout.
Banks should assign a risk charge to each loan that is equal to its expected loss. Whether the bank actually collects this charge is in some senses immaterial. The charge represents a kind of depreciation reserve that should be subtracted from the income derived from the loan. If the bank doesn't subtract it, it is in effect overstating its income.
Hence, if the bank has correctly computed its risk charges, the cumulative total of these charges - that is, those for under-one-year loans and the present value of those for multiyear loans and facilities - should represent the amount of income that is likely to be surrendered in future years because of loan problems. As noted, it doesn't matter where the bank holds this buffer against future losses. It could, in principle, be lodged in the capital accounts in the form of monies unavailable to shareholders.
In practice, it probably makes a lot of sense to segregate the total of risk charges in a reserve account nourished by annual infusions from the provision, since this would enable the marketplace to form a clearer idea of the bank's basic income potential (equal to income less risk charges). And let us remember that the more unequivocal information the bank provides the marketplace, the lower is investor uncertainty and therefore the higher is the market value of the institution.
Measuring Loss Linkages
Although, in the long run, the total of all risk charges, if correctly computed, should be adequate to cover losses, in the short run, actual losses may diverge quite markedly from expected or average ones. Therefore, the bank must have an additional buffer (capital) to cover deviations from the norm. (In practice, of course, when unexpectedly large losses occur, banks do not invade the capital account directly, but rather indirectly through reduced earnings.)
Banks can determine the appropriate size of their secondary capital buffers by knowing, first, their expected losses, and, second, the average coefficient of loan-loss correlation in their portfolios. Such a coefficient, which measures the potential for loss linkages or loss simultaneity, can in turn be calculated from the same models that establish the expected-loss probabilities.
If, based on this type of calculation, which should of course be updated periodically, the bank has an adequate total buffer against losses, then it is in a position to debate any regulatory demand for buffer supplementation - either in the form of additional reserves or more capital. Indeed, in a rational world, the regulators should be doing the same calculation and, if it yielded this kind of result, should not be bothering the bank.
If the bank's total buffer is insufficient, then it should forgo expansion, sell assets, or pare its dividend payout. Given too large a buffer, the institution should be free to entertain opposite courses of action.
In summary, whether the regulators are unreasonable in their demands for augmented reserves is a proposition that can be tested empirically. However, until both the banks and the regulators make use of a currently available methodology that is not now in common use, the debate between them is likely to remain an arid one.
Mr. Rose, formerly senior columnist for this newspaper, is now associated with Oliver, Wyman & Co., a management consulting firm in New York.