International Finance and Accounting Handbook

of a 40, 45, 50, or 75% LGD. Evidence suggests that historical LGD for bank loans
is significantly lower than 50%^45 and therefore, the shift to the advanced approach is
expected to reduce bank capital requirements by 2 to 3%. However, the quid pro quo
for permission to use actual LGD is compliance with an additional set of minimum
requirements attesting to the efficacy of the bank’s information systems in maintain-
ing data on LGD.
Another adjustment to the Foundation Approach’s BRW is the incorporation of a
maturity adjustment reflecting the transaction’s effective maturity, defined as the
greater of either one year or nominal maturity, which is the weighted average life (=
∑ttPt/∑tPtwherePtis the minimum amount of principal contractually payable at
timet) for all instruments with a predetermined, minimum amortization schedule.
The maturity is capped at seven years in order to avoid overstating the impact of ma-
turity on credit risk exposure.
The Advanced IRB Approach allows the bank to use its own credit risk mitigation
estimates to adjust PD, LGD, and EAD for collateral, credit derivatives, guarantees,
and on-balance sheet netting. The risk weights for the mark-to-market Advanced IRB
Approach are calculated as follows:

(7)

(8)

andBRWis as defined in the Foundation IRB Approach.

The effect of the term in equation (7) is to adjust the risk
of loans for its maturity.^46 For longer maturity instruments, the maturity adjustments
increase for low PD rated borrowers (i.e., higher quality borrowers). The intuition is
that maturity matters most for low PD borrowers since they can move only in one di-
rection (downward) and the longer the maturity of the loan, the more likely this is to
occur. For high PD (low quality) borrowers who are near default, the maturity ad-
justment will not matter as much since they may be close to default regardless of the
length of the maturity of the loan.^47
The Advanced IRB Approach entails the estimation of parameters requiring long
histories of data that are unavailable to most banks.^48 Given the costs of developing
these models and databases, there is the possibility of dichotomizing the banking in-

31 b 1 PD 2  1 M 324

where b 1 PD 2 
 3 .0235 11 PD24>3PD0.44 .0470 11 PD 24

RW
 1 LGD> 502  BRW 1 PD 2  31 b 1 PD 2  1 M 324

3.4 INTERNAL RATINGS-BASED MODELS FOR CREDIT RISK 3 • 15

(^45) Carty (1998) find the mean LGD for senior unsecured (secured) bank loans is 21% (13%). Carey
(1998) finds mean LGD of 36% for a portfolio of private placements. Asarnow and Edwards (1995) find
a 35% LGD for commercial loans. Gupton (2000) find a 30.5% (47.9%) LGD for senior secured (unse-
cured) syndicated bank loans. Gupton et al. (2000) obtain similar estimates for expected LGD, but find
substantial variance around the mean.
(^46) This may incorporate a mark to market adjustment. However, the mark to market adjustment in BIS
II does not incorporate the transition risk (deterioration in credit quality) and spread risk (change in the
market price of credit risk) components of a fully mark to market model. There is also an alternative spec-
ification of the b(PD) adjustment based on the default mode assumption.
(^47) That is, for loans with maturities longer than three years, the increase in the capital requirement rel-
ative to the BRW decreases as the loan quality deteriorates. This could increase the relative cost of long
term bank credit for low risk borrowers. See Allen (2002a).
(^48) See the Basel Committee on Banking Supervision (1999a) for a survey of current credit risk mod-
eling practices at 20 large international banks located in ten countries.