Testing normality in the time series of EMP indices: an application and power-comparison of alternative tests

The Exchange Market Pressure Index (EMPI) is an indicator of pressure on a currency. Because of the presence of serial correlation, financial time series may not be normally distributed even for large sample sizes. They may have undefined parameters and hence parametric tests of normality may give misleading results. In this paper, we look at the time series of EMPI of eleven countries of the world, put the data to normality check using tests suggested by various scholars. We also apply a test used exclusively for serially correlated data. No one has used this test earlier. In this context, we also compare the power of these statistical tests, which is another novel contribution of this paper. On the basis of these tests the EMPI time series is found to be non-normal. Two tests are found to be the most powerful. The test which is designed exclusively for time series data is found to be powerful only for China and South Korea, the countries which had the lowest EMPI- standard- deviation in the group of all the eleven countries studied in this paper.