A Multiple of the VaR
By multiplying the VaR output, the aim is to provide a cushion against potential weaknesses in the model itself as well as some cover over and above the confidence level (e.g. 99%) used. Furthermore, the VaR output is the estimated maximum loss at a certain confidence level and in a given holding period. In reality, this level of loss might occur more than once in a short period of time.
- The multiplier covers potential weaknesses of the modelling approach such as fat tail issues.
- Multipliers are used by some firms to manage their market risk internally.
- Emphasises the relative riskiness of products (i.e. it loads significantly more capital onto riskier products).
- A multiplier might encourage firms that would otherwise take a conservative approach to calculating VaR to be less conservative in order to reduce the impact of the multiplier. The multiplier might therefore be a perverse incentive to design a model to minimise regulatory effects rather than optimise its use as a risk management tool.
- A low initial number from a flawed model will not be corrected by a multiplier.
- The particular multiplier chosen is open to the charge that it is arbitrary.
(ii) Add-on Based on Stress testing
Another approach would be to use VaR plus an add-on that reflects, in some fashion, the simulation of extreme market movements, including the breakdown of correlations and other assumptions.
- Risk based.
- Can capture liquidity and fat tail risks.
- Can provide a flexible tool to re-inforce supervisors' qualitative approach.
- It gives firms an incentive to tailor their stress testing to meet supervisory requirements.
- It is difficult to establish a consistent approach to the calculation of the add-on.