General and specific statistical properties of foreign exchange markets during a financial crash

We investigate minute-by-minute foreign exchange rate (FX) data of 14 currencies with different exchange-rate regimes during a financial crash, and divide these data into several stages according to their respective tendencies: depreciation stage (stage 1), fluctuating stage (stage 2), and appreciation stage (stage 3). The tail distribution of FX rate returns satisfies a power-law structure for different types of currencies. We find the absolute value of the power-law exponent is smaller in emerging markets than in developed markets, especially during the stage 1, and is greatest in pegged currencies.We also find that the correlation properties of the FX rate return series have quite disparate results among the various types of currencies. Currencies in developed markets respectively have weak persistence and anti-persistence in short and long timescales; whereas the pegged currencies and currencies in emerging markets show different degrees of anti-persistence in various timescales. Further analyses on the data in divided stages indicate that emerging markets and pegged currencies have more prominent dual fractal structures after the depreciation stage, while the developed markets do not. Hurst exponent analyses on the sign series yield similar results to that on the original return series for most currencies. The magnitude series of the returns provide some unique results during a crash. The developed market currencies have strong persistence and exhibit a weaker correlation in the depreciation and appreciation stages. In contrast, the currencies of emerging markets as well as pegged currencies fail to show such a transformation, but rather show a constant-correlation behavior in the corresponding stages of a crash. These results indicate that external shocks exert different degrees of influence during different stages of the crash in various markets.