Data-Driven Macro-economic Model Analysis Using Non-standard Trimean Algorithm

Abstract

AbstractUnderstanding the relationship of the macro-economic variables is crucial to the economic-related policymaker. In addition, high accuracy in predicting the simulation behavior will also assist the policymaker in planning strategy for the country’s well-being. This study analyzes the type of relationship between two macro-economic variables for the Malaysian data set and predicts the value of both macro-economic variables. The selected macro-economic variables are gross domestic product (GDP) and foreign direct investment (FDI). We construct two data-driven macro-economic models using Lotka-Volterra as the based model. The model relates both GDP and FDI as the prey and predator. We also propose two non-standard trimean approaches, namely Non-standard Numerical Algorithm with trimean applying Nelder-Mead (NSNAT(NM)) and Non-standard Numerical Algorithm with trimean applying Least-Square (NSNAT(LS)) methods to analyze the interaction between GDP and FDI. The simulation results show that the dynamic interaction relationship between GDP and FDI using Nelder-Mead is the predator and prey type of relationship where FDI negatively affects GDP. While the dynamic interaction relationship between GDP and FDI using Least-Square is the mutualism type of relationship where both FDI and GDP positively affect each other. To analyze the prediction accuracy of the proposed methods, the comparison of the simulation results obtained with the existing method that is the fourth-order Adam-Bashforth-Moultan (ABM) method shows that the NSNAT(NM) method can predict GDP and FDI with high accuracy (MAPE value 0.0003% for GDP and 7.3763% for FDI).KeywordsNon-standardTrimeanLotka-VolterraGross domestic productForeign direct investmentPrediction