ABSTRACT
Using high-frequency data on US$/AU$ exchange rate futures, we construct realized volatility models. We investigate whether return, standardized return, variance and logarithmic standard deviation series are normally distributed in a thinly traded market. The behaviour of these series is analysed with the help of statistical normality tests. We find that logarithmic realized standard deviations, as opposed to variances, are fairly normally distributed. For returns as well as for standardized returns we cannot reject a normal distribution for roughly half of the series under inspection. This is an unexpected result as it is not in line with earlier results obtained with this methodology. The Asian crisis, falling into the time period covered by our data sample, is not found to have a significant impact on the distributional characteristics of the series.
