Date: 2025-01-13

Degree: Doctoral Thesis

Programme: Doctor of Business Administration

Authors: Manuel Chau

Supervisors: Professor Florence Lei, University of Saint Joseph

Abstract:

This study aims to establish a robust framework for short-term forecasting of Macau’s Gross Domestic Product (GDP) and Gross Gaming Revenue (GGR). Recognizing Macau as a “monotown” with a significant contribution to the gaming industry in Macau’s economy, forecasting GGR and GDP becomes crucial for local businesses’ strategic planning and policymakers’ informed decision-making.

The research investigates Macau’s internal and external economic variables. Using nowcasting techniques, it seeks to locate the leading economic indicators for Macau from statistical departments in China, Macau, Hong Kong, and stock exchanges. Employing machine learning (ML) and econometric algorithms, including Random Forest, Gradient Boost, AdaBoost, KNN (K-Nearest Neighbors), Elastic Net regression, and Dynamic Factor Model (DFM), the study performs short-term forecasts for Macau’s GDP and GGR. These approaches undergo rigorous evaluation. Furthermore, an ensemble approach is created to combine forecasts by trimmed average. This approach showed accuracy and robustness in this task. This study seeks accurate and robust short-term forecasting and provides insights into Macau’s economic dynamics. However, acknowledging the limitations of pre-COVID modeling assumptions, the study highlights the need for continued research to refine these frameworks in the post-pandemic era.

Keywords: Dynamic Factor Model, Machine Learning, Trimmed Average, Gross Gaming Revenue, Gross Domestic Product, Nowcasting, Short-Term Forecasting, Data-Driven Predictions, Sequential Feature Selection