Value at Risk (VaR)
Value at Risk is the most widely used risk measure in finance, quantifying the maximum expected loss over a specified time horizon at a given confidence level. It answers the question: "What is the worst loss I can expect under normal market conditions?"
Overview
Value at Risk (VaR) was popularized in the 1990s by J.P. Morgan's RiskMetrics system and has since become the industry standard for measuring and reporting market risk. Regulatory frameworks such as the Basel Accords require financial institutions to compute VaR for determining capital reserves against potential trading losses.
VaR provides a single, intuitive number that summarizes the downside risk of a portfolio. For example, a one-day 95% VaR of $1 million means that there is a 95% probability that the portfolio will not lose more than $1 million over one trading day. Equivalently, there is a 5% chance of losses exceeding $1 million.
Despite its widespread adoption, VaR has important limitations. It does not describe the magnitude of losses beyond the VaR threshold, it is not subadditive (meaning the VaR of a combined portfolio can exceed the sum of individual VaRs), and it can be misleading for portfolios with non-normal return distributions or significant tail risk.
Mathematical Formulation
Formal Definition
Given a random variable representing portfolio returns and a confidence level , VaR is defined as the negative of the -quantile of the return distribution:
where is the cumulative distribution function of portfolio returns. The negative sign ensures VaR is expressed as a positive number for losses.
Parametric (Variance-Covariance) VaR
Under the assumption that portfolio returns are normally distributed with mean and standard deviation , VaR has a closed-form expression:
where is the -quantile of the standard normal distribution. For commonly used confidence levels:
- At 95% confidence: , so
- At 99% confidence: , so
Confidence Level Interpretation
The choice of confidence level has significant practical implications:
95% VaR
- Exceeded approximately once per month (1 in 20 trading days)
- More suitable for day-to-day risk management
- Easier to backtest due to more frequent exceedances
- Less sensitive to tail behavior assumptions
99% VaR
- Exceeded approximately 2-3 times per year (1 in 100 trading days)
- Required by Basel regulatory framework for capital calculations
- More difficult to backtest accurately
- Highly sensitive to distributional assumptions in the tails
Multi-Period Scaling
Under the assumption of i.i.d. returns, VaR scales with the square root of time:
The Basel Committee uses a 10-day holding period and 99% confidence level, computed by scaling the daily VaR: . This square-root-of-time rule is an approximation that may understate risk when returns exhibit serial correlation or volatility clustering.
Alternative Estimation Methods
Historical Simulation
Historical simulation computes VaR directly from the empirical distribution of past returns without imposing any distributional assumptions. The portfolio is revalued using each historical scenario, and VaR is read off as the appropriate percentile of the resulting P&L distribution. This method captures non-normality, fat tails, and nonlinear instrument payoffs naturally, but assumes that the historical window is representative of future risk. A typical lookback period is 250-500 trading days.
Monte Carlo Simulation
Monte Carlo VaR generates a large number of simulated return scenarios from a specified stochastic model (e.g., geometric Brownian motion, GARCH, or copula models). The portfolio is revalued under each scenario, and VaR is computed as the appropriate percentile of the simulated P&L distribution. This method is the most flexible, accommodating complex instruments and non-linear payoffs, but is computationally expensive and sensitive to model specification. Typically 10,000 to 100,000 scenarios are generated.
Advantages & Limitations
Advantages
- Intuitive: Expresses risk as a single dollar amount or percentage that is easily communicated to non-technical stakeholders.
- Universally adopted: Industry standard supported by regulators, risk managers, and portfolio managers worldwide.
- Flexible time horizon: Can be computed for any holding period, from intraday to multi-year horizons.
- Backtestable: Exceedances can be tracked and statistically tested to validate model accuracy.
Limitations
- No tail information: VaR says nothing about the magnitude of losses beyond the threshold -- a critical blind spot during crises.
- Not subadditive: The VaR of a combined portfolio can exceed the sum of individual VaRs, violating a basic property of coherent risk measures.
- Model dependent: Parametric VaR assumes normality, which systematically underestimates tail risk in real financial data.
- Procyclical: Historical VaR tends to be low during calm markets and spikes during crises, precisely when it is least useful.
- Manipulation risk: Traders can construct portfolios with low VaR but extreme tail risk (e.g., short out-of-the-money options).
References
- Jorion, P. (2006). Value at Risk: The New Benchmark for Managing Financial Risk. 3rd Edition, McGraw-Hill.
- Alexander, C. (2008). Market Risk Analysis, Volume IV: Value-at-Risk Models.John Wiley & Sons.
- Dowd, K. (2002). Measuring Market Risk.John Wiley & Sons.
- Basel Committee on Banking Supervision (1996). "Amendment to the Capital Accord to Incorporate Market Risks." Bank for International Settlements.
- RiskMetrics Group (1996). "RiskMetrics -- Technical Document." J.P. Morgan/Reuters, 4th Edition.