Evidence of long memory and asymmetry in the EUR/PLN exchange rate volatility

Abstract

This paper focuses on capturing the conditional volatility in the foreign ex-change Value-at-Risk forecasts. By implementing a variety of GARCH models under differ-ent return distributions, we model the volatility of daily returns of EUR/PLN exchange rates. Statistically significant long memory and asymmetry effects in volatility are observed. These characteristics implicate some challenges in volatility forecasting. Therefore, we combine these two effects in the Fractionally Integrated Asymmetric Power ARCH model-ling framework which yields the best goodness-of-fit. Furthermore, it outperforms other models in regard to the applied loss functions and is found to provide the best Value-at-Risk estimation results. Our findings contribute to research on volatility of Polish exchange rate and expand the findings related to dynamic volatility in the existing literature and raises awareness of combined volatility effects to practitioners