Abstract
A network analysis approach, complemented by machine learning (ML) techniques, is applied to analyse the factors influencing Bilateral Investment Treaties (BITs) at the country level. Using the Electronic Database of Investment Treaties, BITs with health safeguards from 167 countries were charted, resulting in 534 connections with countries as nodes and their BITs as edges. Network analysis found that, on average, a country established BITs with six other nations. Additionally, we used node embedding techniques to generate features from the network, such as the Jaccard coefficient, resource allocation, and Adamic Adar for downstream link prediction. This study employed five tree-based ML models to predict future BIT formations with health inclusion. The eXtreme Gradient Boosting model proved to be superior, achieving a 64.02% accuracy rate. Notably, the Common Neighbor centrality feature and the Capital Account Balance Ratio emerged as influential factors in creating new BITs with health inclusions. Beyond economic considerations, our study highlighted a vital intersection: the nexus between BITs, economic growth, and public health policies. In essence, this research underscores the importance of safeguarding public health in BITs and showcases the potential of ML in understanding the intricacies of international treaties.
Introduction
Bilateral Investment Treaties (BITs) are fundamental instruments in international law designed to protect foreign investors. However, they frequently induce a "regulatory chill," where governments hesitate to enact public health regulations—such as tobacco control or access to essential medicines—due to the risk of corporate litigation in international arbitration. While health experts advocate for the inclusion of "health safeguards" to protect a state's right to regulate, the adoption of these clauses has been inconsistent. This study moves beyond traditional regression models to explore the underlying drivers of health-inclusive BITs, asking: what characteristics make a country likely to include robust health safeguards, and can we predict these formations using a data-driven approach?
Methodology
The researchers framed the problem as a link prediction task within a social network. The study utilized two primary data sources: the Electronic Database of Investment Treaties (EDIT) to identify 584 BITs with health provisions across 167 countries, and the World Bank Databank for country-level socio-economic indicators.
Network Construction and Feature Engineering
A graph was constructed where nodes represent countries and edges represent existing BITs containing health-related provisions.
Features were engineered from two perspectives:
- Structural/Network Features: Node embedding and similarity metrics (e.g., Jaccard Coefficient, Adamic Adar, Common Neighbors) captured the topology of international relations.
- Node-Specific Features: Ratios of socio-economic indicators (GDP, FDI, unemployment, Capital Account Balance) between country pairs.
Machine Learning Approach
The study compared five tree-based models: Decision Tree, Random Forest, AdaBoost, CatBoost, and eXtreme Gradient Boosting (XGBoost). Tree-based models were prioritized over deep learning to maintain interpretability for policymakers.
Results and Key Insights
The XGBoost model emerged as the most effective, achieving an accuracy and F1-score of 64.02%. The BIT network itself is relatively sparse (density of 0.139) but highly interconnected, with an average path length of 2.58, suggesting that most countries are only two to three "treaty-steps" away from any other in the health-safeguard network.
Drivers of Treaty Formation
- The "Common Neighbor" Effect: The most influential predictor was Common Neighbor centrality. Countries are significantly more likely to sign a health-inclusive treaty if they share existing treaty partners. This indicates that treaty norms diffuse through established network clusters.
- Economic Stability: The Capital Account Balance Ratio and Unemployment Ratio were major predictors. Countries with similar economic stability and development trajectories are more likely to align on health protections.
- Temporal Evolution: Analysis across three periods (1959–1985, 1986–2005, and 2006–2021) showed a shift. Early treaty formation was driven almost entirely by network connectivity, whereas recent years have seen traditional economic indicators like population size and unemployment become increasingly influential.
Conclusion
This research highlights the critical intersection of international investment law, economic growth, and public health policy. While structural network factors remain dominant, economic stability is playing an increasing role in the adoption of health safeguards. Future work will aim to integrate political indicators and the history of Investor-State Dispute Settlement (ISDS) cases to provide a more holistic view of why countries choose to protect public health in the global marketplace.
Summary updated