Pace of model development

The pace of model development over the past several years has accelerated to support
the rapid growth of financial innovations such as caps, floors, swaptions, spread
options and other exotic derivatives. These innovations were made possible by developments
in financial theory that allow one to efficiently capture the many facets of
financial risk. At the same time these models could never have been implemented on
the trading floor had the growth in computing power not accelerated so dramatically.
In March 1995, Alan Greenspan commented, ‘The technology that is available has
increased substantially the potential for creating losses’. Financial innovations,
model development and computing power are engaged in a sort of leapfrog, whereby
financial innovations call for more model development, which in turn requires
more computing power, which in turn results in more complex models. The more
sophisticated the instrument, the larger the profit margin – and the greater the
incentive to innovate.
If the risk management function does not have the authority to approve (vet) new
models, then this dynamic process can create significant operational risk. Models
need to be used with caution. In many instances, too great a faith in models has led
institutions to make unwitting bets on the key model parameters – such as volatilities
or correlations – which are difficult to predict and often prove unstable over time.
The difficulty of controlling model risk is further aggravated by errors in implementing
the theoretical models, and by inexplicable differences between market
prices and theoretical values. For example, we still have no satisfactory explanation
as to why investors in convertible bonds do not exercise their conversion option in a
way that is consistent with the predictions of models.