Economic and Ecological Balance in IORA Countries: Machine Learning Reveals New Pathways for Sustainable Development in the Global South

From Academic Models to Regional Realities: The Growth-Environment Equation in IORA Countries

The Indian Ocean Rim Association (IORA) consists of 23 member states spanning Africa, the Middle East, South Asia, and Southeast Asia, serving as a key hub for global trade and resource flows. However, most countries in the region face multiple pressures from fragile coastlines, climate risks, and rapid urbanization. A recent study published in *Scientific Reports* used Malaysia, Mauritius, Sri Lanka, and Madagascar as samples, employing a machine learning framework XOS-ELM-GA to predict the relationship between CO₂ emissions, GDP, and agricultural land, providing a quantitative analysis tool for sustainable development in the Global South.

Why Were These Four Countries Selected?

• The four countries selected in the study represent different economic structures and vulnerability types within IORA:
• Malaysia: Highly industrialized, with both services and manufacturing, but agriculture still accounts for a certain proportion;
• Mauritius: A small island nation, economically dependent on tourism and financial services, with extremely high climate vulnerability;
• Sri Lanka: Combines agriculture and light industry, recently facing debt crises and environmental pressures;
• Madagascar: A least developed country, predominantly agricultural with severe deforestation.

These four countries collectively embody the typical contradiction in the development paths of Global South countries: they need to increase income through industrial expansion while avoiding runaway carbon emissions and loss of farmland. The study used historical data from 1960 to 2020, employing CO₂ emissions as a key explanatory variable to predict GDP and agricultural area, rather than making simple causal inferences. This method is more suitable for the characteristics of emerging markets with high data volatility and frequent structural changes.

Machine Learning Empowerment: Prediction Accuracy and Policy Implications

The XOS-ELM-GA model proposed in the study outperforms traditional ELM and OS-ELM methods in prediction accuracy. For example, in predicting Malaysia's annual GDP, the model achieved an average SMAPE of 10.13%; in predicting Sri Lanka's agricultural land, the SMAPE was as low as 3.77%. This means that even without introducing complex mechanistic models, data-driven AI methods can provide reliable near-future economic and environmental indicator predictions for IORA countries.

For policymakers, such predictions help identify conflict zones between growth, carbon emissions, and land in advance. For example, if the model shows an accelerating trend of agricultural land reduction accompanied by high emissions, then priority should be given to sustainable agriculture and low-carbon technology investments. For international investors, this correlation analysis can be embedded into ESG (Environmental, Social, and Governance) assessments to avoid the risk of asset stranding due to environmental degradation.

Global South Perspective: From Data to ActionThe cases of IORA countries highlight the collective challenges faced by the Global South in achieving the Sustainable Development Goals (particularly SDG 8, SDG 13, and SDG 15). Unlike developed nations, emerging markets often lack high-precision satellite monitoring and census data, and machine learning methods precisely fill the analytical gap in "data-poor" regions. Moreover, the study emphasizes correlation rather than causality, reminding us that the coupling between CO₂ emissions and economic growth is not absolute, and emission reduction policies do not necessarily sacrifice development—the key lies in the efficiency of land and energy use.

Additionally, the four countries are all part of the Indian Ocean rim economies, and regional cooperation mechanisms (such as the IORA framework) can promote joint environmental monitoring and low-carbon technology transfer. In the future, if the model can be expanded to more members (such as India, Somalia, and Australia), it will form a dynamic early warning network covering the Global South.

Long-term Trends: Food Security and Climate Resilience

Changes in agricultural land are another focus of the study. The high prediction accuracy for Sri Lanka and Madagascar indicates that the model can capture the sensitive response of arable land to external shocks (such as climate anomalies and international price fluctuations). For Malaysia, which is experiencing rapid urbanization, the coexistence of agricultural land reduction and carbon emission growth suggests the dual pressure brought by land-use transitions, such as the expansion of palm oil cultivation. When investing in agriculture, forestry, or infrastructure projects, such dynamics should be incorporated into long-cycle scenario analysis.

In the long run, IORA countries have young and growing populations, and future demand for food and energy will rise. How to strike a balance between maintaining agricultural land and absorbing CO₂ is the underlying logic of regional economic growth. The data-driven baseline provided by this study can help countries set more practical pathways when formulating their Nationally Determined Contributions (NDCs).

Conclusion

This study not only demonstrates the potential of machine learning in environmental-economic modeling but also reveals the complexity of the Global South's pursuit of development in both "quantity" and "quality." For macro research institutions and international investors, understanding the nonlinear patterns of regions such as IORA is key to grasping the future shift of global growth centers. When AI can accurately predict agricultural contraction or emission peaks, policy responses and capital allocation can be advanced by half a step—this is precisely the indispensable "long-termism" capability in emerging market analysis.