Journal of Risk Model Validation

Steve Satchell

University of Cambridge

We live in interesting times. Risk models have been failing throughout the English summer, whether through lack of validation, or because circumstances were far removed from anything hitherto experienced, remains to be seen. The preliminary comment that has emerged has shown the dangerous macroeconomic consequences of prime brokers using VaR models, when the majority of hedge funds have highly correlated active positions and buy/sell lists. This was probably not one of the scenarios used in assessing portfolio risk, although it may well be so in the future. Another observation is the strong role of demand shifts for subprime; again a characteristic not focused on in many models.

This can be seen as an increase in the risk premium, and the risk-neutral philosophy behind many derivative-based models is incapable of dealing with this. I predict an increase in demand by the risk industry for economists and a reduction in demand for mathematicians.

The journal continues to flourish, with subscriptions increasing and a steady stream of quality submissions. Due to the failings experienced over the summer we would welcome submissions on the current sub-prime crisis issues and, if we receive sufficient material, we may go to a special issue.

This issue contains a paper by Andersson and Lindell, titled "Risk capital stress-testing framework and the new capital adequacy rules". This paper deals with the use of expert opinion to generate long-term macroeconomic behavior, and couples this with simulation methods to capture the short and medium term. The application of this method is based on a Basel accord calculation.

The second paper, by Kevin Dowd, titled "Backtesting the RPIX inflation fan charts", also has a macroeconomic flavor. This tests the forecasting performance of the Bank of England's fan charts for inflation. The results of this paper cast doubt on the Banks ability to forecast inflation.

The third paper, by Kalyvas and Sfetsos, titled "Impact analysis of VaR methodologies on regulatory capital", addresses some deep issues that arise when trying to assess the performance of non-linear processes that feed into VaR prediction.

Finally, de la Pena and his co-authors present research on equality testing between different transition matrices. Whilst this paper is of a technical nature, it is likely to find many applications in practice.