This paper proposes the use of synthetic training data generated by large language models
to improve machine learning SDG classifiers. It shows that supplementing existing training data with
synthetic data produced by the ChatGPT tool improves the performance of the SDGClassy classifier.
This addition of synthetic data is especially useful in building SDG classifiers given the limited availability
of properly labeled data and the complex, interconnected nature of the SDGs. Synthetic data thus enables
more effective machine-learning applications in this context.

Lack of fiscal space and the risk of sovereign debt distress remain key stumbling blocks to achieving Sustainable Development Goals (SDGs) in developing countries. Because the allocation of concessional funds and debt relief is essentially reserved to Low Income countries (LICs), official financing strategies and mechanisms to support developing countries provide insufficient support to non-LICs that may need and deserve special consideration concerning official financing. This paper discusses how official financial support allocation could consider countries’ vulnerabilities in critical dimensions, with special reference to Small Island Developing States (SIDS). It explores how a…

As developing countries pursue infrastructure projects, they should aim to address a combination of the pandemic, climate, inequality, and other crises with the right mix of economic and social infrastructure. To do this, governments must invest in a national infrastructure planning process, align planning with the SDGs, and prioritize sustainable infrastructure over infrastructure that does not put people and the planet first. There is no silver bullet for all the challenges; however, incremental changes based on innovative precedents can potentially make a difference on the ground. This paper proposes an analytical framework to consider these challenges and concludes with possible…

In light of the emphasis on “inclusion” in the Sustainable Development Goals (SDGs), this paper contends that social exclusion and inclusion are context-dependent concepts in at least three senses. First, the ideal of an inclusive society varies by country and by region. Second, different places have different histories, cultures, institutions and social structures. These influence the economic, social and political dimensions of social exclusion and the interplay among them. Third, context – where one lives – shapes access to resources and opportunities. Social inclusion is thus spatially uneven. The paper also shows how context matters, identifying some of the mechanisms by which…

This paper synthesizes the evidence of a negative correlation between income inequality and environmental quality. It shows that inequality exerts adverse impact on environmental outcomes through several channels, including the household, community, national, and international channels. These channels however overlap with one another and can thus reinforce the impact of inequality. Other dimensions of inequality, in particular gender inequality, also impact environmental quality negatively. The concept of the Environmental Kuznets Curve (EKC) is not that helpful in explaining the negative correlation between inequality and environmental quality. The findings of the paper suggest that…