Efektivitas ECM - MIDAS berbasis Principal Component Analysis (PCA) dalam Memprediksi PDB di Indonesia

Abstract

GDP is closely related to monetary policy, because changes in GDP often affect decisions taken by the central bank in formulating policies to maintain economic stability. This study aims to predict the value of Gross Domestic Product (GDP) by developing a more accurate and efficient model. The variables analyzed include primary money, net domestic assets, net foreign position, and foreign exchange reserves as independent variables, and gross domestic product (GDP) as the dependent variable. The method used combines the Error Correction Model (ECM) into the Mixed Data Sampling (MIDAS) and Principal Component Analysis (PCA) models, this approach provides a more comprehensive analytical framework to capture complex interactions between variables with different frequencies, while taking into account long-term and short-term dynamics that influence each other. The results of the study indicate that the combination approach of PCA and MIDAS with the Almon distribution is more effective in capturing data patterns than other approaches that only use PCA with the average or median of economic indicators. The ECM-MIDAS-PCA model with the Almon weight function showed the best results, marked by an Adjusted R-Square value of 22.33% and low prediction error. The Error Correction Term (ECT) coefficient of -0.1579 indicates a correction towards long-term equilibrium of 15.79% per quarter, so that the process towards equilibrium can be achieved in 6.33 quarters.