Econometric Modelling of Exchange Rate Volatility

This research is devoted to modelling the yield volatility of exchange rates of currency pairs USD/RUB, EUR/RUB, GBP/RUB, CNY/RUB, CHF/RUB from 2009 to January 2019, as well as determining the presence of co-integration relations between these pairs using statistical and econometric methods. Searching through the various parameters of autoregressive models with conditional heteroskedasticity, the best volatility models with minimal values of Akaike information criterion and with the residuals without autocorrelation and conditional heteroskedasticity were chosen. The best models were symmetric GARCH models for EUR/RUB, CNY/RUB and asymmetric APARCH models for USD/RUB, CHF/RUB and GJR-GARCH for GBP/RUB. Analytical forecasting demonstrated high predictive quality of the models at the step T+1: the mean absolute percentage error for all currency pairs did not exceed 12%. The study of co-integration relationships showed that the time dependence of exchange rates with the price of gold and Brent crude oil was not found but the relationships between the rates themselves were determined: USD/RUB and EUR/RUB, USD/RUB and CHF/RUB, CHF/RUB and CNY/RUB. The presented study can be very useful for companies and private investors who monitor exchange rate fluctuations because these models can be built, for example, on the last days market second-by-second data of exchange rates, then the forecast quality can be checked and after that the results can be applied to make decisions about the currency transactions.