Measure of Volatility and Its Forecasting: Evidence from Naira / Dollar Exchange Rate

Abstract

In the last five decades, Box Jenkins methodology has been in existence to model univariate time series data but fails or has limitations on modeling volatility. Most financial time series data do exhibit heavy tail and thick distribution, to this effect various parametric and semi-parametric non –linear time series models have been proposed two or three decades ago to capture volatility. However, this research entails measuring volatility and its forecasting using time series exchange rate annual data over the period from 1981 to 2020 (wide periodicity). The exchange rate was transformed to return, and parametric non –linear time series was modeled on it. It was found out that GARCH (1,2) reveals continuous volatility for short while and was the best model to predict the exchange rate volatility based on the evidence from measurement volatility tool; RMSE, MAE, MAPE among other extensions of GARCH models; EGARCH and TGARCH. EGARCH (1, 4) captures the asymmetry effect revealing that negative shocks will persistently have an effect on the volatility of the naira/dollar exchange rate.