The effectiveness of applying machine learning to forecast the Singapore economy

• Main Authors: Foo, Benedict, Koh, Deng Yao, Tan, Juan Pang
• Other Authors: Wang Wenjie
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2021
• Subjects: Social sciences::Economic theory
• Online Access: https://hdl.handle.net/10356/152966
• Institution: Nanyang Technological University

This paper applies novel machine learning methods to high-dimensional macroeconomic data from Singapore. Our data consists of 220 monthly time series which captures the macroeconomy of Singapore from various aspects such as output, retail and trade, price indices and others. Our machine learning methods are combined with two popular econometric models for forecasting macroeconomy: the Vector Autoregression (VAR) model and the Diffusion Index (DI) model. For the VAR model, we use penalized regression techniques such as the least absolute shrinkage and selection operator (LASSO), the elastic net (ENET), and the group LASSO (GLASSO) to achieve dimension reduction. For the DI model, we combine it with cross-validation and LASSO techniques for the selection of the number of factors. We find that our machine learning methods, especially those based on the VAR model, have outstanding forecasting performance for the macroeconomy of Singapore in both short and long horizons.