Volume 3 — Valuation and Risk Metrics © Copyright 2002, CCRO. All rights reserved. 10 3.0 Value at Risk (VaR) 3.1 Discussion of Metric1 The value at risk (VaR) of a position or a portfolio is the loss or change in value that is not expected to be exceeded with a given degree of confidence over a specified time period. VaR is therefore a statistical measure of variability in the value of a portfolio of positions or earnings from economic activity arising from the changes in the market prices of the commodities or other variables underlying the portfolio or activity. For example, a VaR value of $10 million at a confidence level of 99% and a holding period of 5 days using a one-tailed confidence level can be understood to mean the following: given the current structure of the market, one could say with a confidence level of 99% that over this period the loss in the value of the position/portfolio under consideration will not exceed $10 million. Equivalently, one can also say that there is a 1% probability that the actual loss over the liquidation period will exceed $10 million. The VaR measurement is most appropriate for the MTM portion of the portfolio. The following are general qualities of VaR: • Provides a measurement technique to estimate risk of positions, financial assets, and transactions • Intended to predict the potential future range in the MTM value of the forward portfolio, not future realized losses • If tracked and compared over time, has value as a trending tool for the portfolio to analyze changes • Can measure risk across different assets, markets, and tenor while accounting for hedges and rolling them up into a single number • Flexible enough to incorporate both implied volatilities/correlations as well as historical information if necessary • Assumes that the portfolio remains static over the holding period. Measures of risk are useful and necessary, even when imperfect, for all components of a company’s portfolio. The following calculation parameters are most easily verified and validated for liquid (MTM) transactions or portfolios, regardless of accounting method: • VaR methodology – Historic, parametric (variance/covariance), or Monte Carlo methodologies are all acceptable. • Holding period – 1 day to 10 days for trading activity varying time frame for assets • Confidence interval – 99% or 95% one-tailed or two-tailed approach 1 For more discussion of this metric, please refer to the extensive literature on the subject such as P. Jorian’s work on VaR (see References).