How can machine learning identify the potential of digital trade facilitation in bridging inequality?

Abstract

This paper is the result of an exploratory application of machine learning (ML) methods for identifying the potential that digital trade facilitation can have in bridging inequality gaps between and within countries. Most well-known coun-try classifications guiding country-level technical assistance are based on income level, socio-economic development di-mensions, and geographical location of countries. While policymakers and researchers have approached these indicators as a valid criterion to characterize structural patterns that differentiate countries among themselves, more objective cri-teria could play a potential role in bridging effectiveness gaps in allocating technical assistance efforts among countries. ML methods, such as clustering methods, could enable adopting a more objective approach for classifying countries ac-cording to desired strategic objectives, such as leveraging digital trade facilitation for reducing between- and within-country inequality. This paper then has two objectives. First, it aims to contribute to existent country classification criteria by identifying a set of variables to cluster countries according to their levels of digital trade facilitation, inequality, and other institutional, social, and economic factors. Second, it intends to explore some possible policy implications for coun-tries from Asia and the Pacific region. Section 1 introduces the ML analysis used and states the underlying motivation. Section 2 explains the data sources and variables used, as well as the process followed to clean and explore the data. Section 3 and 4 present two ML clustering methods applied for this paper's analysis and their corresponding results. Section 5 concludes and discusses areas for future improvements. Section 6 includes an appendix with data visualizations that resulted from the analysis presented in this paper.