Conference Agenda

"Can development be decoupled from carbon emissions where investments actually land? To address this question, I use development-finance portfolios as a laboratory: projects arrive at different times, intensities, and sectors across locations, generating variation in local exposure. I compile a global ADM2-year panel (2000–2023) linking geocoded development projects to local CO2 emissions and night-time lights. To measure climate finance consistently across donors, I classify project description using natural language processing techniques into mitigation, adaptation/environment, and non-climate flows. Exploiting staggered project arrival with not-yet-treated difference-in-differences and a dose–response design, I find that climate portfolios generate modest average emissions reductions, with effects becoming meaningfully more negative only at higher cumulative exposure. In contrast, non-climate development finance raises local emissions in the medium run. Night-time lights indicate visible expansion in mid-brightness regions, while lights are muted in already bright places even as emissions rise, consistent with compositional shifts and lighting/land-use changes. Disaggregating climate finance highlights solar and environment protection as “win–win” investments where lights increase while emissions fall, implying that decarbonization requires both scaling effective climate finance and greening mainstream development spending."