Modelling the implied tail risk of foreign exchange
22 November 2012
Main findings
Standard risk models based on historical data have deficiencies: the results of these models depend heavily on the period used for calibration. In addition, the models make unrealistic ex-ante assumptions (such as on the distribution that price changes follow) that may lead to underestimation of tail risks.
In this note, we walk through a risk-modelling approach that, as far as possible, uses forward-looking, market-implied information from the currency option market to shed light on the risk of the Fund’s currency position. This model is used at NBIM as a complement to standard risk models.
Analysing the Fund’s new benchmark for fixed-income investments, we find that increasing the allocation to emerging-market currencies adds to short-term currency tail risk, when measured in a common reference currency. The market-implied tail-risk model predicts expected losses in adverse years to be 30-60 percent higher than the standard models.