Short-run fluctuations in foreign exchange rates: Evidence from the data 1973-1979

Abstract The paper examines the statistical properties of daily changes in foreign exchange (FX) rates for nine currencies. It finds that these changes are leptokurtotic, i.e. have long-tailed and sharp peaked histograms. The evidence suggests that this leptokurtosis does not arise from an underlying Paretian stable distribution, nor from a stationary mixture of normal distributions, but rather from normal processes with time-varying parameters. This finding casts doubt on the validity of many standard techniques (t-stats; ARIMA methods). Some moving statistics for estimating the time varying parameters are also presented