What are the four axioms of a coherent risk measure, and how does VaR violate subadditivity?
The FRM curriculum defines coherent risk measures with four properties, and VaR famously fails one of them. I can list the axioms but I don't understand the intuition behind each, especially positive homogeneity and translation invariance. Also, can you show a concrete example where VaR violates subadditivity?
A coherent risk measure satisfies four axioms proposed by Artzner, Delbaen, Eber, and Heath (1999). These axioms ensure the risk measure behaves sensibly for portfolio management and capital allocation.\n\nThe Four Axioms:\n\n1. Monotonicity: If portfolio X always has worse outcomes than Y, then rho(X) >= rho(Y). Intuition: a riskier portfolio should have higher measured risk.\n\n2. Subadditivity: rho(X + Y) <= rho(X) + rho(Y). Intuition: diversification should not increase risk. The risk of a combined portfolio should be at most the sum of individual risks.\n\n3. Positive Homogeneity: rho(lambda x X) = lambda x rho(X) for lambda > 0. Intuition: doubling a position doubles the risk. No economies of scale in risk.\n\n4. Translation Invariance: rho(X + c) = rho(X) - c for constant c. Intuition: adding a risk-free cash amount c reduces the risk measure by exactly c.\n\nVaR Violates Subadditivity — Concrete Example:\n\nConsider two binary credit positions at Greystone Bank:\n\n- Bond A: 4% chance of $10M loss, 96% chance of zero loss\n- Bond B: 4% chance of $10M loss, 96% chance of zero loss\n- Losses are independent\n\nIndividual VaR at 95%:\n- VaR(A) = $0 (the 5th percentile loss is zero, since losses occur only 4% of the time)\n- VaR(B) = $0\n- Sum = $0\n\nCombined portfolio VaR at 95%:\nProbability of zero total loss: 0.96 x 0.96 = 92.16%\nProbability of exactly $10M loss: 2 x 0.04 x 0.96 = 7.68%\nProbability of $20M loss: 0.04 x 0.04 = 0.16%\n\nThe 95th percentile loss falls in the 7.68% region (since 92.16% < 95%), so:\nVaR(A + B) = $10M\n\nSubadditivity violation: VaR(A + B) = $10M > VaR(A) + VaR(B) = $0\n\nDiversification (combining independent risks) appears to INCREASE risk under VaR, which is nonsensical.\n\nExpected Shortfall Passes:\n\nES(A) at 95% = (0.04/0.05) x $10M = $8M\nES(B) at 95% = $8M\nES(A + B) at 95% = (0.0768/0.05) x $10M + (0.0016/0.05) x $20M adjusted = approximately $10.64M\n\nSince $10.64M < $8M + $8M = $16M, subadditivity holds.\n\nExam Tip: The subadditivity violation in VaR typically arises with concentrated, binary-type risks. For normally distributed returns, VaR does satisfy subadditivity. The FRM tests understanding that VaR is not generally coherent.\n\nStudy risk measure theory in our FRM Market Risk course.
Master Part I with our FRM Course
64 lessons · 120+ hours· Expert instruction
Related Questions
Why is DV01 so much smaller than dollar duration if both are supposed to measure rate risk?
When should I stop using modified duration and switch to effective duration?
How should I think about the relationship between Macaulay duration and modified duration instead of memorizing two separate definitions?
Why do hedge calculations often use dollar duration or DV01 instead of just modified duration?
When should I prefer historical simulation VaR over delta-normal VaR?
Join the Discussion
Ask questions and get expert answers.