An evaluation of the Basel III market risk proposals

Jon Danielsson

The Basel committee has proposed significant changes in the way market risk is forecasted. These pages provide some quantitative analysis of those proposals and accompany the VoxEU article « An evaluation of the Basel III market risk proposals » on 28 November 2013.

An updated version of the analysis is here.

Mon voisin du dessus
Un certain Blaise Pascal,
M’a gentiment donné
Ce conseil amical :
«Mettez-vous à genoux
Priez et implorez,
Faites semblant de croire
Et bientôt vous croirez.»

Georges Brassens
Advice to Pascal, loosely translated as “Kneel, pray, make it look as if you believe and soon you will believe”.

Summary of the results

The Committee has proposed for significant changes to how market risk is forecasted

The original 1996 version of the market risk regulations, essentially still in effect, are about to see significant changes, in particular:

  • a move to expected shortfall (ES) from Value-at-Risk (VaR);
  • lowering of the confidence level from 99% to 97.5%;
  • the holding period is calculated by overlapping windows;
  • ES is to be adjusted by stressed observations.

The results show that there is little difference between 99% VaR and 97.5% ES. In the conditional Gaussian case they are essentially the same, and since such a model is probably the most common in industry, this seems to be no gain to be had from moving to ES. For fat tailed conditional distributions, ES 97.% Is slightly higher than 99% VaR but by a factor of 10% or less, so the impact seems to be minimal.

However, the ES results are more volatile, so by making the switch, the magnitude of the risk forecast is essentially the same, but the uncertainty it higher.

The move to overlapping holding periods seems to provide biased risk forecasts, and significantly increasing forecast uncertainty, in some cases so large that the actual forecast is statistically insignificantly different from zero.