FRANÇOIS-SERGE LHABITAN T 193Table 10.1A few examples of model risk and its consequences
Period Institution Problem Loss (M)
1970s Merrill Lynch Pricing of bond issues with an inadequate US$ 70
US interest rate curve.
1970 Merrill Lynch Use of an incorrect yield curve to price US$ 70
stripped Government bonds
1987 Merrill Lynch Incorrect pricing model for stripped US $350
mortgage-backed securities
1992 J.P. Morgan Inadequate models of prepayments US$ 200
for mortgage-backed securities.
1997 Natwest Markets Pricing of options based on a naïve GBP 90
volatility feed.
1997 Bank of Tokyo- Inadequate calibration of a model for US$ 83
Mitsubishi swaptions.
1997 UBS AG Inadequate pricing model for structured CHF 120
equity derivatives
1998 LTCM Inadequate models for arbitrage Unknown
between US and European interest
rates, and excessive leverage
2001 Lipper Convertible Inadequate mark to model for US$ 600
Hedge Fund convertible bond arbitrage.
2003 Fourth District Modelling error in its auto-leasing US$ 67
Institution Provident business identified 7 years later
Bank
2005 Various hedge funds Market reaction to the GM and Ford Unknown
credit downgrades too improbable an
event for credit risk models to capture it
Far from remaining a theoretical concern, the percentage of losses
attributed to model risk has been consistently increasing over recent years,
and the deficiencies in current quantitative models continue to be brought to
light by market events (see Table 10.1). In 1999, a report by Capital Markets
Risk Advisors (CMRA) and Meridien Research estimated the annual model
risk losses reach $5.5 billion. Since, it is almostde rigueurto have a major
model risk-related disaster every year. The loss may be absorbable, but the
acute embarrassment is not. It is, therefore, essential for model users and
builders to be aware of the existence of model risk, to understand its poten-
tial sources and to implement business practices and technology solutions
to mitigate it.
In this chapter, we discuss the potential impact on model risk in the
particular case of derivatives pricing and hedging. Section 10.2 recalls the
evolution of pricing models for derivatives. Section 10.3 illustrates the exam-
ple of implied volatility – a necessary input in most pricing models that is