What is Conditional VaR (CVaR / Expected Shortfall), and why did Basel III replace VaR with ES for market risk capital?
I keep seeing Expected Shortfall (ES) described as 'better than VaR' for capturing tail risk. Can someone explain the calculation with a simple example and articulate why regulators moved from VaR to ES in the Fundamental Review of the Trading Book?
Conditional VaR (CVaR), also called Expected Shortfall (ES), measures the average loss in the worst alpha-percent of scenarios. While VaR tells you the threshold loss at a given confidence level, ES tells you how bad it gets beyond that threshold.
Formula (Continuous)
ES_alpha = E[Loss | Loss > VaR_alpha]
For a discrete sample of n sorted losses (L_1 >= L_2 >= ... >= L_n) at alpha = 1%:
ES_{1%} = (1 / floor(n x 0.01)) x SUM of the worst floor(n x 0.01) losses
Worked Example
Dunbar Fixed Income Fund has 1,000 daily P&L observations. At the 99% confidence level:
- VaR_{99%} = $4.2 million (the 10th worst loss out of 1,000)
- The 10 worst daily losses (in $M): 9.1, 7.8, 6.5, 5.9, 5.3, 5.0, 4.8, 4.6, 4.4, 4.2
ES_{99%} = (9.1 + 7.8 + 6.5 + 5.9 + 5.3 + 5.0 + 4.8 + 4.6 + 4.4 + 4.2) / 10 = $5.76M
VaR says: "On 99% of days, losses won't exceed $4.2M."
ES says: "On the 1% of worst days, losses average $5.76M."
Why ES Replaced VaR in Basel III (FRTB)
| VaR Weakness | How ES Addresses It |
|---|---|
| Not subadditive — diversification can appear to increase risk | ES is subadditive — always rewards diversification |
| Ignores severity beyond the threshold | ES averages the entire tail |
| Can be gamed by shifting risk just beyond the VaR cutoff | ES penalizes all tail scenarios |
| No information about tail shape | ES is sensitive to tail thickness |
Subadditivity Example
Consider two portfolios X and Y:
- VaR(X) = $5M, VaR(Y) = $5M
- VaR(X + Y) could be $11M (superadditive!) — VaR says combining them increases risk
- ES(X) = $7M, ES(Y) = $7M
- ES(X + Y) will always be <= $14M — ES correctly reflects diversification benefit
FRTB Specifics:
- Basel moved from 99% VaR (10-day) to 97.5% ES for internal models
- The confidence level was lowered because ES already captures tail severity
- Stressed ES is calibrated to the worst 12-month period in the bank's history
For more on market risk capital, visit our FRM Part I question bank.
Master Part I with our FRM Course
64 lessons · 120+ hours· Expert instruction
Related Questions
How exactly do futures margin calls work, and what happens if I can't meet one?
How do you calculate the settlement amount on a Forward Rate Agreement (FRA)?
When should I use Monte Carlo simulation instead of parametric VaR, and how does it actually work?
Parametric VaR vs. Historical Simulation VaR — when does each method fail?
What are the core components of an Enterprise Risk Management (ERM) framework, and how does it differ from siloed risk management?
Join the Discussion
Ask questions and get expert answers.