Credit risk

Portfolio risk measures

Expected loss

Credit losses must be expected as an inherent cost of doing business. But the occurrence of credit losses is erratic in both timing and amount and those that arise usually relate to transactions entered into in previous accounting periods. In order to reflect the fact that future credit losses are implicit in today's portfolio, we use the concept of "expected loss".

Expected loss is a statistically based concept from which we estimate the annual costs that will arise, on average, from positions in the current portfolio that become impaired. It is derived from the three components described above – probability of default, exposure at default and loss given default.

Expected loss is the basic measure for quantifying credit risk in all our credit portfolios. For products carried at amortized cost, it can be used for risk-adjusted pricing, and to assess credit loss for management accounting purposes, which differs from the credit loss expense reported in the financial statements. For fair valued products, it provides a basis for determining credit valuation adjustments, which are required both for financial reporting and for the planning process. It is also the starting point for further portfolio analyses – statistical loss and stress loss.

Statistical and stress loss

Our credit portfolio is heterogeneous, varying significantly in terms of client type, sector, geographical diversity and the size of exposures. For the assessment of both statistical loss and stress loss in material credit portfolios, we start our analysis at the level of sub-portfolios with more homogeneous characteristics.

We aggregate statistical loss across these portfolios using our own proprietary credit Value at Risk (credit VaR) methodology. This provides an indication of the level of risk in the portfolio and the way it changes over time. It is also a component of our Earnings-at-Risk measure described in the "Risk Management and Control" section of this chapter.

Modeling stress losses is complex because they are only partially driven by systematic factors. We apply scenarios which allow us to assess the impact of variations in default rates and asset values, taking into account risk concentrations in each portfolio. We also measure industry and geographical contributions to stress loss results.