Value at Risk and Bank Capital Management

194 CHAPTER

♦

7 Value at Risk and Risk Control for Market and Credit Risk

obviously support dynamic management of trading limits by the head of the trading units

or decisions by a market risk management committee about authorizations for temporarily

exceeding limits.

In the case of credit risk, the use of VaR to support risk control is different, but it can

also be very important. First of all, autonomy limits defi ning at which hierarchical level

a credit can fi nally be approved can be defi ned in terms of VaR or (more frequently) of

the expected loss of the loan rather than in terms of its size, as typically happened in the

past. Second, risk policies can defi ne a risk-adjusted pricing target that can be proposed

to the relationship manager as the benchmark to price a new loan given the riskiness of

the borrower. Of course, the application of risk-adjusted pricing guidelines needs to be

tailored, based on the organizational structure of the bank and the characteristics of the

counterparty, as we discussed by comparing two benchmark cases: a loan from an invest-

ment bank to a large borrower and a loan from a smaller, commercial bank to an SME.

7.7 Further Reading

VaR limits have sometimes been blamed for potentially forcing traders to liquidate posi-

tions after increases in market volatility, thereby self-inducing further volatility. This

argument has been critically discussed in Persaud (2000) and Jorion (2002). The issue of

how to harmonize VaR limits for the time horizon of one day, given the maximum accept-

able loss, has been discussed in Saita (1999b) and Härtl and Johanning (2005); both papers

also discuss the linkage with cumulative losses. The concepts of VaRDelta, incremental

VaR, and component VaR are described in Garman (1996, 1997). Litterman (1997a, b)

explains how component VaR can be used to identify natural hedges and to optimize the

portfolio of exposures of the bank. Carroll et al. (2001) discuss how component VaR can

be measured when risk measures are based on scenarios, while Hallerbach (2002) consid-

ers the case of the absence of joint normality. The issue of defi ning VaR limits with ex

ante uncertain correlations and exposure directions has been discussed by Strassberger

(2004). Hallerbach (2004) extends the analysis of portfolio optimization discussed by

Litterman (1997a, b) for market risks to the credit risk portfolio. Finally, risk-adjusted

pricing for loans is discussed in Matten (2000) and Bessis (2002).