Image
May 19, 2014

Moving from VaR to ES | What Difficulties are Involved?

According to a Risk Magazine article (subscription required), medium-sized financial institutions are going to have a tough time moving from Value at Risk (VaR) to Expected Shortfall (ES) for the purposes of calculating trading-book capital requirements.

According to a Risk Magazine article (subscription required), medium-sized financial institutions are going to have a tough time moving from Value at Risk (VaR) to Expected Shortfall (ES) for the purposes of calculating trading-book capital requirements.

The difficulty lies not in the theory behind the ES risk model (VaR and ES are calculated from the "heads" and "tails" respectively of the same distribution of returns), but in the practical complexities of implementing ES, given the operational burden needed to update existing risk platforms and trading systems.

The Risk article explains the different challenges facing firms looking to move from VaR to ES. Below is a comprehensive summary of the main challenges: For illustrative purposes, I will assume a confidence level of 99% for VaR.

  • Limited Scenarios: Whereas VaR is calculated based upon 99% of all market scenarios in a dataset, ES is calculated based upon the remaining 1% only. This makes it difficult to back-test.
  • Issues with Back-Testing: VaR can be easily back-tested by seeing whether historical portfolio losses exceed VaR 1% of the time. However, given the subjective nature of the tail-distribution in ES, it is more difficult to judge the results of the back-tests. In addition, back-testing ES requires significantly more historical data sets than back-testing VaR. 
  • Extrapolation Issues: With VaR, one doesn't have to worry about the size and shape of the distribution tail, as it is "cut off" from the risk-measure. However, for ES (presumably Monte-Carlo based ES) one has to make a judgment call on how to extrapolate the size of the tail.
  • Data Availability: To make historical ES statistically meaningful, one has to use large historical data sets. The article doesn't explain exactly why this is an issue, but this is probably because:
    • Reliable historical data for securities traded in opaque markets may be difficult to source
    • Purchasing large historical data sets from 3rd party data vendors may be expensive
  • Additional Calculation Requirements: The raw ES for a trading-book must be multiplied by a ratio to determine the book capital. This ratio is calculated using a further set of stressed ES calculations, which require subjective judgments on how to determine the stressed risk factors. Institutions may not have the necessary expertise to do this.

Popular
Most Viewed

Image

Related Articles


June 30, 2022

SIMM Falls Short says PRA Letter to Banks




2 MIN



Risk Management


June 28, 2022

FMSB Statement of Good Practice on Trading Platform Disclosures




2 MIN



Regulation


June 20, 2022

Regulatory change and data fragmentation are key challenges for 85% of firms




2 MIN



Regulation