What is CoVaR, and how does it measure the systemic risk contribution of individual financial institutions?

CoVaR (Conditional Value at Risk) measures the Value at Risk of the financial system conditional on a specific institution being in financial distress. It captures spillover effects and interconnectedness that individual VaR completely ignores.\n\nDefinitions:\n\n- CoVaR_{system|i}: the VaR of the financial system given that institution i is at its VaR level (in distress)\n- DeltaCoVaR_i: the difference between CoVaR when institution i is in distress versus when it is at its median state\n\nDeltaCoVaR_i = CoVaR_{system|i in distress} - CoVaR_{system|i at median}\n\nDeltaCoVaR captures the marginal systemic risk contribution of institution i.\n\nWhy VaR Alone Is Insufficient:\n\nConsider two banks with identical $500 million 99% VaR:\n\n| Bank | VaR | Interconnectedness | DeltaCoVaR |\n|---|---|---|---|\n| Greystone Mutual (retail bank) | $500M | Low (local deposits, mortgages) | $200M |\n| Ironbridge Capital (dealer bank) | $500M | High (derivatives counterparty to 50 banks) | $3.2B |\n\nBoth look equally risky in isolation, but Ironbridge's distress would cause 16x more damage to the system. DeltaCoVaR reveals this hidden systemic importance.\n\nEstimation Method (Quantile Regression):\n\nThe standard approach uses quantile regression to estimate CoVaR:\n\n1. Define system returns R_system and institution returns R_i\n2. Run a quantile regression at the 5th percentile:\n R_system = alpha + beta x R_i + gamma x M_t + epsilon\n where M_t includes macro state variables (VIX, credit spreads, yield curve slope, liquidity proxies)\n3. CoVaR is the fitted value when R_i equals its 5th percentile (distress state)\n4. DeltaCoVaR = fitted value at R_i = 5th percentile minus fitted value at R_i = 50th percentile\n\nNumerical Example:\n\nAnalyst Kamala estimates CoVaR for Ashfield Securities using weekly return data:\n\nRegression results (5th percentile):\nR_system = -0.012 + 0.85 x R_Ashfield - 0.003 x VIX_change + 0.02 x CreditSpread_change\n\nAshfield's 5th percentile return: -6.2%\nAshfield's 50th percentile return: +0.3%\n\nCoVaR (distress): -0.012 + 0.85 x (-0.062) - 0.003 x 2.5 + 0.02 x 0.8 = -5.87%\nCoVaR (median): -0.012 + 0.85 x (0.003) - 0.003 x 0.5 + 0.02 x (-0.1) = -1.55%\n\nDeltaCoVaR: -5.87% - (-1.55%) = -4.32%\n\nInterpretation: when Ashfield is in distress, the financial system's 5th percentile loss is 4.32 percentage points worse than when Ashfield is at its median.\n\nDrivers of High DeltaCoVaR:\n\n| Factor | Effect on DeltaCoVaR |\n|---|---|\n| Size (total assets) | Larger institutions contribute more |\n| Leverage | Higher leverage amplifies spillovers |\n| Maturity mismatch | Short-term funding with long-term assets |\n| Interconnectedness | More counterparty relationships |\n| Complexity | Opaque structures harder to resolve |\n\nRegulatory Application:\nCoVaR has influenced the development of systemic risk surcharges (G-SIB buffers). Institutions with higher DeltaCoVaR face additional capital requirements, ranging from 1% to 3.5% of risk-weighted assets depending on their systemic importance score.\n\nStudy systemic risk measurement in our FRM study materials.