摘要：In this paper, autoregressive fractionally integrated moving average (ARFIMA) model was proposed and was used for modeling and forecasting of liquidity ratio of commercial banks in Nigeria. Augmented Dickey Fuller (ADF) test was used for testing stationarity of the series. The long lasting autocorrelation function of the data showed the presence of long memory structure, and the Hurst exponent test was used to test for presence of long memory structure. The Geweke and Porter-Hudak (GPH) method of estimation was used to obtain the long memory parameter d of the ARFIMA model. Alternatively, a suitable ARIMA model was fitted for the liquidity ratio data. On the basis of minimum AIC values, the best model was identified for each of ARFIMA and ARIMA models respectively. The models were specified as ARFIMA(5,0.12,3) and ARIMA(1,1,1). To this end, forecast evaluation for the two models were carried out using root mean square error (RMSE). Having compared the forecasting result of the two models, we concluded that the ARFIMA model was a much better model in this regard.
关键词：ARFIMA model; ARIMA model; Long-memory; Autocorrelation function; Liquidity ratio
