An eigenvalue approach to risk regimes in currency markets

Over the past decades, both academics and market practitioners have produced much research on forecasting financial risk. In view of the levels of risk that the markets have operated over in recent times, this research is very topical. One growing area of investigation is the construction and use of risk indicators that aim to detect changes in market risk regimes. For investors, these have tangible applications as risk affects the accuracy of asset valuations, portfolio construction and risk management. In this article, in contrast to other studies that deal with cross-asset risk indicators, we propose a risk indicator that derives solely from implied volatilities of currency exchange rates. Our logic is that derivative markets have been shown to exhibit a degree of forward-looking element into their pricing. In addition, foreign exchange is the largest financial market, suggesting that a large amount of information is therefore reflected in the time series of exchange rate volatilities. In our article, we first propose an eigenvalue-based approach to construct a risk indicator that focuses on the reduction of the number of significant factors during periods of market stress. We then go on to study its informational content, and conclude on possible avenues of research and applications.