The third panel session considered how Basel II capital standards are being applied to retail credit in general and credit card asset-backed securities (ABS) in particular. William Lang, of the Federal Reserve Bank of Philadelphia, moderated the discussion; panelists Marc Saidenberg, Federal Reserve Bank of New York; Randy White, Bank One Corporation; and Hugh Van Deventer, Citicorp, presented alternative perspectives on calibrating ABS capital requirements for Basel II.
Saidenberg began the session by describing the current Basel approach toward risk-based capital allocation to ABS issuers and discussing the recent decision to apply the Basel II internal ratings-based (IRB) framework to calibrating capital only on unexpected loss. Basel II seeks to capitalize ABS exposures that pose risk to banks’ balance sheets. Basel II breaks these risks down into direct exposures to privately placed ABS residuals, direct exposures to the rated investment-grade tranches, and capital risk arising from early amortization.
The most obvious risk that ABS issuance poses to banks’ balance sheets lies in direct exposures to residuals and other lower-level noninvestment grade tranches of the securities that remain on balance sheet. Hence, Basel II currently provides that banks hold 100 percent capital against exposures to ABS pieces that have credit enhancement levels lower than the bank’s own internal risk-based capital requirement.
As mentioned earlier, however, publicly issued asset-backed securities have experienced far fewer downgrades than corporate debt. Thus, Basel II generates risk weights for investments in “thick, granular pools”? that are very different from those required for commercial loan pools. Capitalization of these asset-backed securities is based on the initial ratings of the tranches, and required capitalization increases with lower initial credit ratings.
Last, as described in a previous session, ABS can become riskier if the pool of loans underlying the ABS begins to perform poorly and becomes manifest only when the asset-backed securities (and possibly the issuer) default. Hence, Basel II proposes that banks hold increasing capital against their ABS issues as the deals approach early amortization triggers. If excess spread is less than 450 basis points away from early amortization triggers, the bank begins to hold nonzero capital against the securities. If excess spread on a typical U.S. credit card ABS is less than 112.5 basis points away from its early amortization trigger, the bank will be required to treat the pool underlying the ABS as if it were on the bank’s balance sheet.
Other panelists’ comments (and a great deal of the discussion that ensued) focused on the accuracy of estimated internal ratings-based capital requirements that form the baseline for capitalizing direct exposures to privately placed ABS and residuals. At the heart of this debate is a fundamental difference in risk measurement and management for retail and commercial credit.
Basel II seeks to establish capital for both retail and commercial credits based on the probability of default and the loss given default for the loan type (commercial loans and four types of retail consumer loans). This approach is more commonly used for commercial loans, as most proprietary commercial loan default models work with these parameters.
Randy White, of BankOne, noted that because of the granularity of retail portfolios, banks manage retail loan risk differently from commercial loan risk. Credit card lenders typically estimate expected loss the product of probability of default and loss given default for their retail loan portfolios. Of course, expected loss may be high either because of high probability of default and low loss given default or low probability of default and high loss given default. Hence, White argued for more analysis of the correspondence between retail expected loss calculations and the simulated decomposition of probability of default and loss given default currently proposed under Basel II.
Of course, even if a high correspondence existed between the two estimation processes, panel members noted that certain parameters of relative risk used in Basel II remain to be accurately quantified. Three parameters were discussed at length: the correlation of within-asset credit risk across different quality borrowers; the correlation of credit risk across assets; and the appropriate threshold for capitalizing early amortization exposures.
Some argue that the correlation of within-asset credit risk across borrowers of different quality is too high, particularly in the case of credit card portfolios. White showed results from simulations suggesting that it takes substantially lower correlations than those assumed by Basel II to align regulatory risk-based capital with his bank’s required economic capital estimates. On the other hand, White noted that Basel II understated the required capital for off-balance-sheet credit card exposures and suggested that a more appropriate approach would be a lower correlation applied to on- and off-balance-sheet receivables.
White also argued that the simulations of appropriate within-asset correlations are significantly influenced by the level of correlation of credit risk across assets. White asserted that his simulations equate his own economic capital measurements and Basel II internal risk-based estimates only by assuming dramatically higher cross-asset correlations than those in simulated bank portfolios.
Hugh Van Deventer and some audience members who are issuers of ABS also commented that well-managed banks may have different across-asset correlations than others and should receive capital credit for managing the quality their portfolios. However, most agreed that the whole point of Basel II is to generate a better quantitative benchmark for capital adequacy decisions, and the resulting model would not be a complete substitute for discretionary supervisory judgment.
Furthermore, other conference participants suggested that capitalizing early amortization risk might well be a significant potential source of procyclicality, wherein banks will be required to raise capital in periods of distress. Forcing banks to raise capital during periods of distress is widely believed to choke off lending at precisely the time when the credit is needed to fuel economic recovery.
Overall, participants seem to agree that more work is necessary to better understand retail credit portfolio risk and more accurately parameterize Basel II provisions. The regulatory comment period on Basel II’s third Quantitative Impact Study closed just before the conference date; more than 400 comments were received. An important conclusion of the session, echoed in Wright’s message in his keynote address, was that there are significant benefits from regulators and issuers working together to better address retail credit risk issues in Basel II.