Conference Agenda

Biodiversity finance is central to achieving the Kunming-Montreal Global Biodiversity Framework (GBF). Yet, donor reporting remains fragmented and inconsistent. We introduce a three-dimensional framework that distinguishes the goal, instrument, and ecosystem of biodiversity development finance. Using large language models (LLMs) to apply this framework, we classify 5.4 million official development finance (ODF) projects from 2000–2023. Despite an increase from USD 2.8 to 15.1 billion over this period, we estimate that biodiversity finance will remain 16-43% below the GBF Target of USD 30 billion annually by 2030. Biodiversity finance is concentrated in activities (e.g., 31% in pollution-control infrastructure) and spatially (e.g., 24% in China, India, Colombia, Brazil, and Vietnam). Overall, we find that the allocation of funding aligns well with countries’ ecosystem needs, as measured by species richness and ecological restoration potential. The presented framework can help monitor and evaluate biodiversity finance across sources globally.

Despite the growing attention to biodiversity losses across the world, there

are considerable concerns of whether this leads to actual behavior for protection. Using a sample of Chinese listed firms from 2008 to 2023, this paper

firstly demonstrates a clear biodiversity attention–behavior gap. This gap has

been narrowing in recent years, which is arguably due to the changing external environment. Using the adoption of the Kunming Declaration in 2021

as a quasi-natural experiment, we show that the biodiversity-sensitive firms

exhibit stronger responses and engage more actively in environmental protection investment. Further analyses explore the role of external governance,

at the government, public, and investor levels, in the attention-behavior relationship. Results confirm that the enhancement of external governance

against biodiversity challenges can help bridging the gap. Overall, our results

highlight the importance of government enforcements and public awareness in

driving corporate eco-friendly actions and avoiding potential greenwashing.

Biodiversity underpins ecosystem services essential to human well-being, yet it is declining at unprecedented rates worldwide. Agri-environmental policies are a central tool for halting these losses, but systematic evidence on which measures effectively conserve biodiversity remains scarce, particularly at the global scale. This paper addresses this evaluation gap by providing a large-scale, cross-country assessment of the effectiveness of agri-environmental policy. We develop a conceptual framework linking policy instruments and targets to biodiversity outcomes and apply it to classify over 200,000 policy documents using large language models. This results in a novel, comprehensive, global database of over 14,000 agri-environmental policy measures implemented between 1988 and 2022. The policy data reveal how policymakers design biodiversity-related interventions and which conservation targets they prioritize. To identify effective policies, we use a machine learning-based extension of difference-in-differences approach and measure biodiversity outcomes using the Biodiversity Intactness Index for more than 170 countries. Our preliminary results identify 38 cases of significant biodiversity improvements across diverse biomes, associated with policy mixes aiming to improve species richness and ecosystem integrity and protect genetic diversity. These findings provide actionable insights for policymakers seeking to design agri-environmental policies that deliver measurable biodiversity gains.

As China is a major contributor to global greenhouse gas emissions, its 2060 carbon-neutrality commitment is pivotal to international climate mitigation. Achieving this goal requires credible evidence on which provincial climate policy measures deliver scalable emissions reductions, evidence that is often obscured by overlapping regulations and heterogeneous enforcement. We conduct a systematic ex-post evaluation of roughly 2,400 provincial low-carbon policy measures implemented between 2007 and 2022. Moving beyond traditional econometric designs, we develop a machine learning-based reverse causal inference framework that first detects structural breaks in sectoral carbon dioxide emissions — power, industry, buildings, and transport — and then attributes these shifts to specific local policy measures.

We identify 53 policy interventions that successfully reduced emissions, collectively delivering a 6.5 billion metric tonnes of carbon dioxide abatement, equivalent to approximately 4.3\% of total scope 1 emissions across all regions and sectors between 2007 and 2022. The distribution of realized mitigation is highly uneven: industry contributes the largest share of observed reductions, while buildings and transport display stronger average responsiveness once policy packages become sectorally aligned.

The results reveal a clear disconnect between policy volume and mitigation effectiveness. Policy-rich provinces such as Shanghai do not necessarily exhibit detectable structural breaks, whereas repeated breaks in provinces such as Yunnan, Henan, Inner Mongolia, and Shandong emerge under more selective policy configurations. Across sectors, the industrial sector is governed almost exclusively through integrated policy mixes, whereas single market-based instruments are frequently sufficient in the power and transport sectors. We further find that policy stringency does not systematically rise at the onset of structural breaks; instead, breaks are associated with concentrated pressure within specific instrument–objective combinations rather than with broad-based policy escalation. These findings support a shift toward targeted, portfolio-focused policy design to better translate climate ambitions into realized mitigation outcomes.

Biomass carbon losses from land use and land cover change have increased in recent decades, driven largely by agricultural land expansion. In response, governments worldwide have adopted a wide range of land conservation policies. Here, we assess how these public policies affect land conservation patterns and associated biomass carbon changes, at the global scale, across different policy schemes and climate zones. We further identify specific policy interventions that stand out as particularly effective. Using global geospatial datasets and a combination of econometric and machine learning methods, we estimate that land conservation policies have globally reduced losses of intact and non-intact forest by 30% and by 36%, respectively. In contrast, their effect on grassland conservation is less clear. The observed forest protection has been achieved through both environmental payment schemes and habitat conservation regulations. We further find that large-scale frameworks with financial and technical support from international organizations are critical for mitigating land-use emissions in the tropical regions, whereas integrated national policies that combine regulatory measures with payments are most effective in the temperate regions.

Clean cooking transition has been a major change in rural household living conditions, yet its broader population-level health effect remains insufficiently quantified. We study this question in rural China, where solid-fuel reliance fell sharply between 2000 and 2020 but varied substantially across counties. Using a county-level panel of 1,740 rural counties observed in 2000, 2010, and 2020, we assess whether larger reductions in the share of households primarily using solid fuels for cooking were associated with better improvements in age-adjusted death rates (ADR) and life expectancy at birth (LEB). The results indicate that a 10-percentage-point decline in county-level solid cooking fuel share is associated with 9.3 fewer deaths per 100,000 population in the ADR and a 0.11-year increase in LEB. Scaled by the mean observed decline, these estimates imply that cooking fuel transition accounts for approximately 9.7% of the observed ADR reduction and 8.1% of the observed LEB increase. The associations are larger in poverty counties and smaller where educational attainment and separate-kitchen prevalence are higher, suggesting that the health returns to cooking fuel transition depend on pre-existing socioeconomic disadvantage and exposure-mitigation capacity. In a counterfactual in which counties retained their 2000 solid cooking fuel share levels, the observed transition corresponds to approximately 362,000 avoided deaths in 2020, valued at about 880 billion yuan. These findings suggest that rural China’s movement away from solid cooking fuels as a socially consequential change in rural living conditions, with implications for survival, longevity, and the economic valuation of health gains.

In the past decades, illegal gold mining has sharply increased in the Tropics. Despite its importance in sustaining livelihoods, it creates negative externalities like increases in malaria due to increasing the habitat suitable for mosquitos. However, the impact of illegal gold mining on malaria remains largely unexplored. Using a 6-year panel dataset at the municipality level and heterogeneity-robust difference-in-differences estimators, we provide evidence of the impact of illegal gold mining on malaria incidence in the Brazilian Amazon, an understudied location with increasing illegal gold mining and high burden of malaria. We find that a 10% increase in illegal gold mining leads to, on average, an increase of 515 malaria cases per 100,000 inhabitants annually. This increase does not seem to be explained by increases in population density or decreases in health expenditures, or deforestation. This increase in malaria transmission translates into social costs ranging from US$39,434 to US$1,725,971 annually due to increases in mortality, cost of treatment, and foregone wages. Our analysis highlights the social cost of unregulated gold mining and the need for the adoption and enforcement of monitoring and mitigation practices concomitant of mining.