Title: State space and Box-Jenkins approaches: a comparison of models prediction performance in finance

Authors: Obinna Damain Adubisi; Ikwuoche John David; Ogbaji Eka; Awa Erinma Uduma

Addresses: Department of Mathematics and Statistics, Federal University Wukari, Taraba State, 23401, Nigeria ' Department of Mathematics and Statistics, Federal University Wukari, Taraba State, 23401, Nigeria ' Department of Mathematics and Statistics, Federal University Wukari, Taraba State, 23401, Nigeria ' Department of Statistics, University of Port-Harcourt, River State, Nigeria

Abstract: This paper describes a study that used data collected from the Central Bank statistical web database system in Nigeria to evaluate and compare the forecasting performance of the nonstationary linear state space model and Box-Jenkins (ARIMA) model at different historic time periods. The comparison uses data series on inflation rates (core and non-core) in Nigeria for a specified period. The performances were evaluated based on three metrics: mean absolute error (MAE), mean absolute percentage error (MAPE) and root mean square percentage error (RMSPE). The one-year forecast evaluation results indicated that predictions from the nonstationary linear state space model outperformed the seasonal ARIMA model at different time periods. Furthermore, the proposed nonstationary linear state space model captured the dynamic structure of the inflationary series reasonably and requires no new cycle of identification and model estimation given the availability of new data.

Keywords: ARIMA; autoregressive integrated moving average; filtering; inflation rate; smoothing; state space model.

DOI: 10.1504/IJDS.2019.102789

International Journal of Data Science, 2019 Vol.4 No.3, pp.181 - 195

Received: 14 Apr 2018
Accepted: 10 May 2019

Published online: 07 Oct 2019 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article