Climate change has been increasingly challenging agricultural livelihoods, especially in Sub-Saharan Africa which relies heavily on rainfed agriculture. Therefore, climate change adaptation is crucial for protecting the environment and reducing poverty. To incentivize adaptation, agro-environmental policies mainly employ two types of subsidies: input-based subsidy where cash transfers directly target environmentally friendly activities; and output-based subsidy where price premiums are added to certified green products, indirectly subsidizing adaptation. The relative effectiveness of these subsidies in increasing adaptation depends on individual beliefs, particularly regarding climate change risks and adaptation benefits (if any) on production. Effectiveness of input- and output-based subsidy policies respond differently to these beliefs, primarily through the production function. However, there remains only limited empirical evidence on the policy effectiveness accounting for the target population’s belief distribution as well as limited theoretical framework on ways to improve and select policies considering potentially shifting beliefs.
Shu and Zhang (2025) investigates how information shapes adaptation under different subsidy schemes among cocoa farmers in Ghana. Conducting a lab-in-the-field experiment, we explore whether a subsidy targeting output from sustainable production is as effective at promoting adaptation as an input subsidy that directly compensates for pro-environment practices, especially when paired with information on climate risks and adaptation benefits. We further examine the impact of these lab interventions on farmers’ real-life decisions by distributing tree seedlings to lab participants through the Green Ghana Project (GGP), a nationwide reforestation initiative by the Ghanaian government which provides free forest tree seedlings. The GGP can be viewed as an input subsidy that increases linearly with the number of trees. Therefore, it provides a practical context to evaluate the real-world implications of agro-environmental subsidies and information nudging. We find that information does shift individual beliefs about adaptation benefits, which in turn increases adaptation more under the output-based policy than the input-based. We rationalize this finding with a model where beliefs affect the inputs to the production function in an uncertain environment, and output-based policy becomes more effective when beliefs are reasonably correct. The effects of information intervention within lab persists in the real-life participation in GGP.
Effectiveness of conservation programs in developing countries
There is a rich literature evaluating the impact and effectiveness of conservation programs, particularly Payment for Ecosystem Services (PES) programs, in promoting conservation without exacerbating poverty in developing countries. In one of the earliest influential studies, Jayachandran et al. (2017) conduct a randomized evaluation of a deforestation avoidance PES program in Uganda and find it significantly reduced forest cover loss, with the value of carbon storage benefits estimated at 2.4 times the program costs. More recent work explores design variations among PES programs to improve cost-effectiveness. For example, Jack et al. (2025) study a PES aimed at reducing agricultural burning in India. By experimentally varying payment levels, conditionality and upfront versus ex-post payments, they show that incorporating a partial up-front payment into the contract increases compliance. Building on this literature, our paper provides novel evidence comparing traditional PES programs with output-based subsidy policies, highlighting differences in how their effectiveness depends on correct beliefs among the target population.
Role of individual beliefs
While the growing literature on climate adaptation in agriculture has examined various frictions to adaptation such as credit constraints and market structure, relatively fewer studies explore the role of environmental beliefs in individual adaptation decisions. Zappala (2024) develops a theoretical model of behavioral frictions to study the role of individual drought beliefs and their accuracy in irrigation decisions, showing that farmers exhibit asymmetric responses to dry shocks depending on the accuracy of prior beliefs. Patel (2024), in turn, leverages an experimental setting with rice farmers in Bangladesh to investigate both the causes and consequences of environmental beliefs. He finds that farmers’ past exposure to salient versus subtle shocks shapes their salinity beliefs differently, and that correcting misperceptions significantly alters farmers’ demand for salinity-tolerant seeds, leading to substantial profit gains. We consider a wider range of beliefs, incorporating not only farmers’ perceptions of climate risks but their beliefs about how adaptation tools can reshape their production function. We then connect their beliefs to the relative effectiveness of two kinds of adaptation subsidies and show how information campaigns that generated updated beliefs could yield different impacts under two types of subsidies.