Assessing hybrid time series models for analyzing currency in circulation volatility in Nigeria

Abstract

The volatility of currency in circulation in Nigeria poses significant challenges to the country’s economic stability. This study contributes to the literature by identifying the most suitable hybrid time series model for capturing volatility in Nigeria’s currency in circulation. The study utilizes secondary data obtained from the documented records of Central Bank of Nigeria, covering the period from 1960 to 2023. Stationarity checking was carried using ADF and KPSS tests. The presence of heteroscedasticity in the residuals was examined using Scatter Plots and Breusch-Pagan test. Two phase methods were applied in fitting the four different models. In the first method ARIMA (1, 1, 1) model was fitted. In the second method the residuals of the ARIMA (1, 1, 1) were extracted to fit GARCH, FIGARCH, EGARCH and TGARCH, which result to the four different hybrid models. ARIMA-GARCH (1, 1, 1) was found to be the best model with least AIC, Bayes, Shibata, and Hannan Quinn. The model was diagnosed using ARC-LM Test and Ljung-and Box test. The results of the forecast revealed that since the conditional standard deviations (0.9476, 0.9476, 0.9475) are close to each other, this indicates stable volatility, low volatility and similar volatility level in the currency in circulation for the next three years. Thus, there is need for continuous monitoring sigma values for potential changes, adjust strategies if volatility increases or decreases significantly and future research can use TAR-FIGARCH model for future forecasts of the volatility.