Market risk is the risk that the value of a financial instrument will fluctuate as a result of changes in market prices. These changes are caused by factors specific to the individual security or its issuer, or by factors affecting all securities traded in the market.
Depending on the specific asset class or market parameter, we talk about Equity Risk, Commodity Risk, Interest Rate Risk, Spread Risk or Currency Risk. In spite of formally matching the above definition, credit risk, i.e. the dramatic decrease in market price due to default of the issuer of a credit instrument, is generally treated separately, since methods differ considerably.
A standard approach to assess market risk is to generate a Profit/Loss distribution of the portfolio under consideration. The instruments of a portfolio are evaluated under a set of scenarios for the risk factors. As often, there is no single "right" method for the generation of these scenarios. Every institution has to choose one which fits best its specific needs. This can be a historical simulation, sensitivity approach or Monte Carlo simulation. Clearly, for each chosen method the quality of the market data is pivotal.
In many situations quantile based measures of the P/L distribution are used as risk measures, such as Value at Risk (VaR) or Expected Shortfall (TailVaR), where only the latter is a coherent risk measure.
An old truth that has been recalled violently by the current financial crisis is that these statistic measures can only predict what history has shown us already. They lose their value for risk that arises from elementary or takeover events (Event Risk), sudden changes of market conditions (Regime Shift Risk) or liquidity (Liquidity Risk Link). Neither can they account for sudden changes in volatilities or correlations, where special techniques or stress test come into play.
If two currencies have been pegged to one another, they will exhibit a historical correlation of basically one. A VaR analysis based exclusively on historical information will not address the currency risk that becomes very real the day one of the currencies may be devalued relative to the other. If this scenario is critical to the business, a simple stress test will offer more insight than a VaR analysis, even if performed with a modified correlation assumption.