Value at Risk, or VAR, is a measure of market risk. It expresses the maximum potential loss by an investor on the value of an asset or a portfolio of assets and liabilities, as a function of a ownership timeframe and confidence interval. It is based on a sample of historical data and is deduced from normal statistical laws.
For example, a portfolio worth €100m, with a VAR of -€2.5m at 95% (calculated on a monthly basis) has just a 5% chance to depreciate in value by more than €2.5m in one month.
VAR, which is used intensely by financial establishments as a management tool, is closely correlated to duration.
VAR is beginning to be used by major industrial groups, such as TeleDanmark, which publish it in their annual reports. It nonetheless has two drawbacks:

• It is based on the assumption that markets follow normal distribution laws, which understates the frequency of extreme values;
• It tells us nothing about the potential loss that occurs beyond the confidence interval. Taking the previous example, how much can be lost in less than 5% of the cases: €2.6m? €10m? €100m? Not exactly all in the same ball park, are they? VAR tells us nothing here.