Abstract:Energy security and energy trade are the cornerstones of global economic and social development. The structural robustness of the international oil trade network (iOTN) plays an important role in the global economy. We integrate the machine learning optimization algorithm, game theory, and utility theory for learning an oil trade decision-making model which contains the benefit endowment and cost endowment of economies in international oil trades. We have reconstructed the network degree, clustering coefficient, and closeness of the iOTN well to verify the effectiveness of the model. In the end, policy simulations based on game theory and agent-based model are carried out in a more realistic environment. We find that the export-oriented economies are more vulnerable to be affected than import-oriented economies after receiving external shocks. Moreover, the impact of the increase and decrease of trade friction costs on the international oil trade is asymmetrical and there are significant differences between international organizations.
What problem does this paper attempt to address?
### Problems the Paper Attempts to Solve
This paper aims to address the issue of structural robustness in the international oil trade network (iOTN). Specifically, the authors seek to construct a decision model that reflects the characteristics of gains and costs for economies in international oil trade by integrating machine learning optimization algorithms, game theory, and utility theory. This model not only effectively reconstructs the network degree, clustering coefficient, and closeness centrality of the iOTN but also allows for policy simulation in more realistic environments, thereby providing practical strategic support for governments and organizations.
### Main Contributions
1. **Construction of a Heterogeneous Economies Decision Model**: Based on utility theory and game theory, the authors consider the endowment characteristics of heterogeneous economies in oil trade and construct an interpretable decision model.
2. **Application of Machine Learning Optimization Algorithms**: The authors use machine learning optimization algorithms to learn the endowment characteristic vectors of heterogeneous economies and test the interpretability of the model parameters.
3. **Policy Simulation and Evaluation**: The authors explore the evolution of economic entities under changing trade friction "scenarios," conduct policy simulations and evaluations, and consider the impacts of trade wars and the COVID-19 pandemic.
### Research Background
Energy security and energy cooperation are important factors affecting global economic stability and development. In the face of challenges such as natural disasters, local wars, and climate change, ensuring energy security and stable energy cooperation is a problem that needs to be addressed. The uneven distribution of oil resources, large-scale and long-distance transportation leading to tight transport capacity and increased costs, affects the coordinated development of the energy industry. Additionally, the vulnerability of the global energy supply-demand balance and frequent oil market fluctuations, as well as various non-economic factors such as local wars, sanctions, and international oil price shocks, also impact international energy cooperation. Solving these problems requires interdisciplinary cooperation, and the development of artificial intelligence technology provides new tools and methods for addressing unreasonable energy structures and expanding international energy cooperation.
### Methods and Data
1. **Construction of iOTN**: Using global oil trade data from 1990 to 2019, the authors obtain data from the United Nations Comtrade database to construct a weighted oil trade network.
2. **Decision Model**: Based on the oil trade decision model of heterogeneous economies, the authors consider various trade factors such as oil prices, import and export routes, energy production, and economic culture. They measure the gains and costs of trade relationships through utility functions.
3. **Machine Learning Model**: The authors use machine learning optimization algorithms to learn the endowment matrix and weight coefficients to reconstruct oil trade relationships. They optimize model parameters by minimizing the loss function to ensure the interpretability and predictive accuracy of the model.
### Results and Discussion
1. **Spatial Distribution Analysis**: By visualizing the endowment characteristic vectors of different economies, the authors find that major energy demand and supply countries (such as the United States, UAE, Nigeria, Indonesia, China, and Singapore) have unique endowment characteristics.
2. **Interpretability Analysis**: Despite the black-box problem of machine learning models, the authors can further explore the specific economic significance of the endowment vectors through correlation analysis and the combination of existing economic variables.
### Conclusion
This study successfully reconstructs the international oil trade network by constructing an interpretable machine learning model and conducts policy simulations under different trade friction scenarios. The research results not only help to understand the structural robustness of the international oil trade network but also provide practical decision support for governments and organizations.