An attractive methodology for measuring loss percentiles is Earnings-at-Risk, or "EaR." It is based on historical distributions of earnings. The wider is the dispersion of time series of earnings, the higher is the risk. Earnings at risk have benefits and drawbacks.

Several measures of earnings can be used: accounting earnings, interest margins, commercial margins, cash flows, and market values, notably for the trading portfolio. Of course, the larger the data set, the more relevant the measure will be. The concept applies to any sub-portfolio as well as for the entire bank portfolio. Once earnings distributions are obtained, it is easy to derive loss percentiles by looking for some aggregated level of losses that is not likely to be exceeded in more than a given fraction of all outcomes.

The major benefits of EaR are that they are relatively easy to measure because they are obtained from accounting data. There are some technical difficulties. For example, the volatility calculation raises technical issues, for instance when trends make times series, unadjusted for trends, look highly volatile. In fact, the volatility comes from the trend rather than instability. Hence, relative or percentage variations of earnings are a better measure of their volatility around the trend. The technique requires assumptions, but it remains tractable and easy.

EaR provides a number of outputs. The earnings volatility shows the magnitude of variations. The reduction of earnings volatility, when the perimeter of aggregation increases, measures the diversification effect. The capital is a loss percentile, or the amount not exceeded by adverse deviations of earnings in more than a fraction equal to confidence level. It is difficult to conceive a simpler method of producing a number of outputs without too much effort.

However, the major drawback of EaR relates to risk management. It is not possible to define the sources of the risk making the earnings volatile. Various types of risks materialize simultaneously and create adverse deviations of earnings. The contributions of these risks to the final earning distribution remain unknown. Unlike VaR models, EaR captures risk as an overall outcome of all risks. Without connection to the sources of risk, market, credit or interest rates, EaR does not allow tracing risks back to where they come from. EaR is an additional tool for risk management, but not a substitute.

Found a mistake? Please highlight the word and press Shift + Enter