Nowcasting Philippine Economic Growth using MIDAS Regression Modeling

• Main Author: Rufino, Cesar C.
• Format: text
• Published: Animo Repository 2017
• Subjects: Nowcasting; MIDAS Regression; Mixed Frequency Problem; Temporal Aggregation; Ragged Edge Problem; Bridge Equations; Collection Development and Management; Growth and Development
• Online Access: https://animorepository.dlsu.edu.ph/res_aki/44; https://animorepository.dlsu.edu.ph/context/res_aki/article/1043/viewcontent/2017_12_044.pdf
• Institution: De La Salle University

Among the most anticipated data releases of the Philippine statistical system is the quarterly real gross domestic product. This all-important variable provides the basis for deriving the economic growth performance of the country on a year-on-year basis. Official publication of this statistics, however, comes at a significant delay of up to two months, upsetting the planning function of various economic stakeholders. Under this backdrop, data scientists coined the term “nowcasting,” which refers to the prediction of the present, the very near future, and the very recent past, based on information provided by available data that are sampled at higher frequencies (monthly, weekly, daily, etc.). Nowcasting, however, opens up the “mixed frequency” problem in forecasting, which is the data frequency asymmetry between the dependent and independent variables of regression models that will be used in forecasting. The central objective of this study is to demonstrate the viability of using a state-of-the-art technique called MIDAS (Mixed Data Sampling) Regression to solve the mixed frequency problem in implementing the “nowcasting” of the country’s economic growth. Different variants of the MIDAS model are estimated using quarterly Real GDP data and monthly data Inflation, Industrial Production, and Philippine Stock Exchange Index. These models are empirically compared against each other and against the models traditionally used by forecasters in the context of mixed frequency. The results indicate the feasibility of adopting the MIDAS framework in accurately predicting future growth of the economy using information from high-frequency economic indicators. Certain MIDAS models considered in the study performed better than traditional forecasting models in both in-sample and out-of-sample forecasting performance.