Sponsor's article > Counterparty risk – the next generation

Discussions continue on allowing the use of simulation-based estimation methods for counterparty credit exposure under Basel II. In the meantime, however, industry practice has continued to advance. David Rowe argues we should at least move regulatory capital calculations into the late twentieth century before the industry is entrenched in the twenty-first

As has been highlighted previously in Risk1, a proposal has been put forward jointly by the International Swaps and Derivatives Association, the British Bankers’ Association and the Bond Market Association to bring simulation-based counterparty credit exposure estimates into the regulatory capital framework. The idea is to estimate expected positive exposure for each counterparty based on a Monte Carlo simulation of future market conditions. The proposal is to base the risk-adjusted assets to be included in the Basel capital calculation on an agreed multiple (alpha) of this expected positive exposure.

Issues continue to be raised by supervisors about the basis for the multiple to be used. One such issue is the impact of new business booked during the standard one-year time horizon that offsets the normal pattern of ageing and deal expiry that tends to reduce simulated exposure over time. It is hard to see the justification for this when a similar treatment is not included for banking book exposures. Furthermore, the actual capital available to satisfy the minimum requirement under the Accord is not allowed to reflect expected future earnings other than in exceptionally stable and predictable businesses such as consumer credit cards. This makes it doubly hard to see the justification for inclusion of an allowance for projected new business when there is no allowance for expected earnings over the same period.

A second rationale for raising the multiplier is potential wrong-way exposure. Insofar as this refers to idiosyncratic wrong-way exposure, the argument is weak. Clearly such wrong-way exposure does exist, but idiosyncratic exposure must be predominantly right-way in nature unless the majority of derivatives market participants are speculating and not hedging. In fact, however, the argument is more subtle and relates to weaker but more systematic forms of wrong-way exposure.

The idea takes two forms. One is that in times of economic contraction interest rates tend to decline substantially. Such a decline will induce broad increases in credit exposure to counterparties who are paying fixed to their dealers and receiving floating interest payments in return at just the time when cyclical factors are eroding credit quality.2 While this argument does have merit, I am sure an even stronger case can be made with respect to the wrong-way characteristics of committed credit lines. Exposure under such lines does increase when companies are in financial distress and their credit quality is deteriorating. It is unclear why market-driven credit exposure should be singled out for special treatment when systemic wrong-way exposure is almost certainly weaker in this area than in other banking book exposures.

A somewhat more convincing case can be made when considering that the simulated exposure includes in-the-money credit default swaps. Certainly these exposures will rise in the face of deteriorating general credit conditions that simultaneously tend to weaken the counterparties to such trades. If such considerations are not reflected in the full credit value-at-risk comparisons used to calibrate the appropriate value of alpha, then this is a basis for increasing it beyond such estimates.

Yet another argument put forward by supervisors is the impact of model risk. There will be a variety of assumptions used to drive simulations of long-term market behaviour, but this can be addressed by a limited set of guidelines for how volatilities, correlations and mean reversion parameters are to be determined in such simulations.

Improvement

As all this discussion has been in progress, leading market-makers have continued to improve their counterparty credit risk management techniques. As indicated by recent conference presentations by Evan Picoult of Citigroup and Charles Monet of Morgan Stanley, some dealers are calculating a daily market-based estimate of the credit valuation adjustment to their counterparty exposure.

Having initiated calculation of this adjustment on a daily basis, the next logical step is to begin hedging it to the extent possible. At this point, such hedging appears to be confined to macro credit instruments such as the DJCDX/DJiTraxx indexes. True micro hedging at the single-name level is, and is likely to remain, unattractive due to the cost of regular adjustments to such hedges. A possible next phase, however, would be to estimate concentrations of credit exposure in the book and to hedge these with regional or sectoral credit indexes as they emerge or with total return swaps on a corresponding notional portfolio of corporate bonds.

One of the major achievements of banking supervision in the past decade has been the emergence of a risk-based approach. This has been accompanied by establishing the implementation of best-practice methods in risk management, a gradually moving target, as a primary objective of regulatory oversight. In the area of counterparty credit risk management, however, the supervisory community has been remarkably hesitant. It is well beyond time to bring this aspect of the Basel Accord into the late twentieth century before derivatives market-makers are well established in the twenty-first.

1 See Risk September 2003, Picoult, Canabarro & Wilde, Analysing counterparty risk, pages 117–122, and Rowe, Reason for Hope, page 109 2 This is a point made over six years ago in a 1998 Federal Reserve Board paper. See Credit risk models at major US banking institutions: current state of the art and implications for assessments of capital adequacy, prepared by the Federal Reserve System Task Force on Internal Credit Risk Models, page 39, available at
http://www.federalreserve.gov/boarddocs/press/general/1998/19980529/study.pdf

Risk