Title: A new combined linear-artificial neural network-based model for accurate inflation forecasting in Tunisia
Authors: Asma Ouerghi; Marwa Hasni; Zied Jaidi; Safa Bhar Layeb
Addresses: UR-OASIS, National Engineering School of Tunis, University of Tunis El Manar, Tunisia ' UR-OASIS, National Engineering School of Tunis, University of Tunis El Manar, Tunisia ' Central Bank of Tunisia, 25, Rue Hédi Nouira – BP 777 – 1080 Tunis, Tunisia ' UR-OASIS, National Engineering School of Tunis, University of Tunis El Manar, Tunisia
Abstract: Accurate forecasts of the inflation rate are essential for tracking the economic streaming. In the early inflation forecasting research, focus was directed toward developing regression-based methods. Despite their performance, several lines of evidence consider that using them is prohibitive due to the variable nature of the inflation series. Henceforth, a number of empirical-data based tools have been put forward. Amongst these, ANN models have been shown to yield satisfactory results. This has motivated several researchers to question the best way for efficient inflation forecasting and accordingly, conduct several comparative studies on the relative performances between allied methods. In our sense, it is more judicious to make use of the advantages provided by each forecasting approach rather than exclusively choose one between them. In this paper, we develop a combined forecasting model which integrates time-series and ANN and test its performance in forecasting inflation by means of an empirical comparative study.
Keywords: inflation; artificial neural network; forecasting models; time series; comparative study; Tunisia.
International Journal of Decision Sciences, Risk and Management, 2019 Vol.8 No.4, pp.220 - 233
Received: 09 Feb 2019
Accepted: 27 Mar 2019
Published online: 22 Apr 2020 *