Journal of Risk Model Validation

This issue of The Journal of Risk Model Validation commences with a paper by Pilar Abad, Sonia Benito Muela and Carmen López Martín entitled "The role of the loss function in value-at-risk comparisons", which examines in some detail the loss functions that might capture the "losses" experienced by regulators/supervisors and by risk managers/firms. These different loss functions can lead to different value-atrisk models in terms of total loss minimizers. These interesting results bring to mind the actuarial debate in the early 1990s in the United Kingdom when it was realized that different stakeholders (eg, shareholders versus pensioners) would have different views concerning the ways in which pensions should be financed. This in turn led to investigations into contracts and structures that might harmonize these conflicting positions. Perhaps we will see papers that move in this direction in the future.

Our second paper, "Backtesting general spectral risk measures with application to expected shortfall" by Nick Costanzino and Mike Curran, presents (if Imayquote from the abstract) "a simple, practical and easily implementable coverage test to backtest any spectral risk measure". The authors continue: "Our test gives a single decision at a specified confidence level and is perfectly consistent with the binomial test for value-at-risk. Particular attention is given to the special case of expected shortfall." I found this approach helpful, both in theoretical terms and because practitioners are able to actually implement the approach with their own models.

The issue's third paper, "Country risk index and sovereign ratings: do they foresee financial crises?" by Nerea San-Martín-Albizuri and Arturo Rodríguez-Castellanos, presents an interesting critique of existing country risk models (or at least a subset of them). The authors question whether the models considered had any ability to forecast the global financial crisis and they find that they did not. They then propose a new class of models that will hopefully do better. While the editor is not an advocate of the black swan theory of risk, the findings in this paper could be interpreted as positive support for such theories. A view of history as an evolutionary process that bleaches swans in a never-ending sequence is one (rather negative) interpretation of how risk models change.

The final paper in the issue is by Jie Zhang and Lyn C. Thomas and is titled "The effect of introducing economic variables into credit scorecards: an example from invoice discounting". The paper looks at the construction of scorecards, an important topic in retail risk modeling and one that is rarely addressed in this journal. The authors show how the introduction of economic variables into a credit scorecard improves the predictive power of the scorecard. Exactly where the benefits come from is explained in the paper, and the authors present a convincing argument that the proposed approach is an improvement on existing procedures. Finally, an application to invoice discounting is investigated.

In general terms, I can report that the journal is in good health. The administrative side runs smoothly and we have a much-improved supply of papers that are relevant to our readers while possessing enough heterogeneity to provide wide interest.

Steve Satchell
Trinity College, University of Cambridge