Applications & Limitations of VaR
Value at Risk (VaR) is the most popular risk metric in the world, but it is not perfect. It is like the speedometer of a car—essential, but it won't tell you if the bridge ahead has collapsed.
Applications of VaR
1. Regulatory Capital (Basel Accords)
- Banks: Regulators (RBI, Basel Committee) mandate that banks hold capital based on their VaR.
- Rule: If your 10-day 99% VaR is $100 Million, you must hold a multiple of that (e.g., $300 Million) in cash reserves.
2. Risk Limits & Control
- Trading Desks: Every trader has a "VaR Limit".
- "You can trade whatever you want, as long as your daily VaR stays below $5 Million."
- This allows decentralized risk management.
3. Performance Evaluation (RAROC)
- Risk Adjusted Return on Capital:
- Trader A makes $10M profit taking huge risks (High VaR).
- Trader B makes $8M profit taking tiny risks (Low VaR).
- VaR helps adjust profits to see who is actually better. Trader B likely has a higher RAROC.
4. Comparison Across Assets
- VaR is a universal number. You can compare the risk of a Bond Portfolio vs a Crypto Portfolio vs a Gold Portfolio using a single currency figure (Rs.).
Limitations of VaR
1. "Silent on the Tails" (The 1% Problem)
VaR tells you the minimum loss in the worst 1% of cases, but not the maximum.
- Statement: "99% VaR is 1 Crore."
- Meaning: "You will lose at least 1 Crore."
- Reality: You could lose 1 Crore, or 50 Crores, or 100 Crores. VaR doesn't distinguish between a bad day and doomsday.
- Fix: Use Expected Shortfall (ES).
2. Assumes History Repeats Future
Historical VaR assumes the past 500 days are a good predictor of tomorrow.
- Problem: Structural breaks (Covid, 2008 Crisis) are never in the data until after they happen.
3. Illusion of Safety
VaR gives a precise number ("Risk is exactly 1,234,567 Rs"). This false precision makes managers overconfident.
4. Non-Subadditivity (Not Coherent)
- Mathematically, VaR(Portfolio) can sometimes be greater than VaR(A) + VaR(B).
- This implies that "Diversification increases risk", which is insane and theoretically wrong. (Expected Shortfall fixes this).
Note
Taleb's Critique: Nassim Taleb calls VaR a "fraud" because it ignores "Black Swan" events. It measures the risk of the dog biting you, but ignores the risk of the building collapsing on the dog.
Loading quiz…