Regional Integration Clusters and Optimum Customs Unions: A Machine-Learning Approach

Abstract

A growing body of literature measures and compares the regional economic integration outcomes across regions and subregions of the world. 1) A common problem in comparing the results presented in different studies is the wide variation of the underlying classifications (i.e., groupings) of countries into regions and subregions across studies. 2) This is particularly problematic because the quantitative measures of regional integration used in these studies tend

On the Optimal Size and Composition of Customs Unions: An Evolutionary Approach

“…Consider a set of countries and their set of borders , with being the border for any two bordering countries and in . We follow the definition of a region adopted in De Lombaerde et al ( 2021 ).…”

“…In recent years, there has also been an increasing interest in empirical measures of regional integration outcomes by policymakers and international institutions such as the United Nations, World Bank, and regional development banks (AfDB, 2016 ; Naeher, 2015 ; Naeher & Narayanan, 2020 ; UNESCAP, 2020 ). The paper most closely linked to our analysis is perhaps the study (De Lombaerde et al, 2021 ) which uses a machine learning approach based on a network clustering algorithm to evaluate the composition of real-world CUs. However, this study only considers a single outcome (regional integration shares) and does not attempt to solve for an optimal solution of the composition of CUs as we do below.…”

Intelligent computational methods for economics

“…However, abstraction supports powerful predictive analysis which allows preferences to be estimated from observed actions and helps build understanding that can be useful in informing policy (De De Lombaerde et al, 2023). Furthermore, abstraction can also help to unearth hidden variables that influence the decision‐making process (Athey, 2018 and De De Lombaerde et al, 2021).…”