Market Risk Measurement - VaR and Beyond › Limitations of VaR

Michael BrenndoerferDecember 20, 20252 min read
Quantitative FinanceAll chapters
Part 10 of 15Risk Management and Financial Risk Controls
Chapter 53Market Risk Measurement - VaR and Beyond › Limitations of VaR
Reading section 6 of 13 · Limitations of VaR · 2 minSection 6 of 13 · 2 min

Book contents

Quantitative Finance15 parts · 79 chaptersView the whole book →
Active chapter 53 of 79

Section 6 of 13 · 2 min read

Market Risk Measurement - VaR and Beyond › Limitations of VaROne section · Sections in this chapter6 / 13 ↓
Sections in this chapter13
Reading tools Definitions & inline links

Choose your expertise level to adjust how many terms are explained. Beginners see more tooltips, experts see fewer to maintain reading flow. Hover over underlined terms for instant definitions.

Article links

Make inline references clickable

Despite its widespread adoption, VaR has fundamental limitations that became painfully apparent during the 2008 financial crisis. Understanding these limitations is essential for proper risk management.

VaR Tells You Nothing About Tail Severity

The most critical limitation is that VaR is silent about losses beyond the threshold. A 99% VaR of $10 million says losses will exceed this level 1% of the time, but it provides no information about whether those tail losses average $11 million or $50 million. Two portfolios with identical VaR can have vastly different risk profiles in the tail.

Consider a portfolio that sells deep out-of-the-money options. Most days it earns small premium income, producing a favorable VaR. But on the rare days when the options go in-the-money, losses can be catastrophic. VaR completely misses this asymmetric risk profile.

VaR Is Not Sub-Additive

A mathematically troubling property is that VaR can violate sub-additivity: the VaR of a combined portfolio can exceed the sum of individual VaRs. This means diversification can appear to increase risk under VaR, which contradicts financial intuition and makes risk aggregation problematic.

Coherent Risk Measures

A risk measure ρ\rho is coherent if it satisfies four axioms:

  1. Monotonicity: If portfolio A always loses more than portfolio B, then ρ(A)≥ρ(B)\rho(A) \geq \rho(B)
  2. Sub-additivity: ρ(A+B)≤ρ(A)+ρ(B)\rho(A + B) \leq \rho(A) + \rho(B) (diversification reduces risk)
  3. Positive homogeneity: ρ(λA)=λρ(A)\rho(\lambda A) = \lambda \rho(A) for λ>0\lambda > 0 (doubling position doubles risk)
  4. Translation invariance: Adding cash reduces risk by the cash amount

VaR satisfies all axioms except sub-additivity for general distributions.

Backtesting Challenges

Validating VaR models requires backtesting: comparing predicted VaR to actual realized P&L. For 99% VaR, you expect exceedances about 1% of trading days. With 250 trading days per year, that's only 2-3 expected exceedances, making it statistically difficult to distinguish a good model from a bad one. You need years of data to have statistical power, but by then market dynamics may have changed.

Model Risk and Parameter Uncertainty

All three VaR methods depend on historical data to estimate parameters or scenarios. When markets enter regimes not represented in the lookback period, VaR estimates can be dangerously wrong. The 2008 crisis saw correlations spike toward 1.0 across asset classes, volatilities explode, and return distributions shift dramatically. VaR models calibrated to pre-crisis data severely underestimated risk.