Climate change is increasingly challenging agricultural livelihoods, especially in rainfed regions like Sub-Saharan Africa, where climate adaptation is crucial for protecting the environment and reducing poverty. To incentivize adaptation, many agro-environmental policies often employ two types of subsidies: cash transfers that directly target environmentally friendly activities (input subsidies); and price premiums for certified green products, indirectly subsidizing adaptation (output subsidies).
Which of the subsidies are more effective in encouraging sustainable adaptation practice?
The relative effectiveness of these subsidies in increasing adaptation depends on individual beliefs, particularly regarding climate risks and the impact of adaptation on production. For instance, people whose beliefs heighten their sensitivity to climate risks may respond favorably to input subsidies but not as much to output subsidies, as the prospect of low income and correspondingly low subsidies during a climate-crisis state may outweigh the subsidy income they would receive in a normal state. Moreover, the effectiveness of output subsidies relies more heavily on accurate beliefs, compared to that of input subsidies. This differential sensitivity to beliefs across the two subsidy types suggests that information campaigns can lead to different responses, even when farmers similarly update beliefs under both. This highlights the critical role of beliefs in shaping subsidy effectiveness.
My job market paper 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 impacts. 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 Program (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.
Why Focus on Ghana's Cocoa Sector?
Cocoa production is highly vulnerable to climate-related shocks, such as prolonged droughts and extreme heat, and saw a global decline of over 10% in the 2023-2024 season. Ghana, the world’s second largest cocoa bean producer, relies heavily on cocoa, which employs 17% of the labor force and contributes 3.2% to the GDP. A key strategy for adapting to these shocks is shade management – planting trees around farms for shade. This stabilizes yields and enhances ecosystem services. Cocoa production in Ghana is thus an ideal context to compare input subsidies for forest trees and output subsidies for sustainable cocoa beans produced from shaded farms.
The Experiment: Two Types of Subsidies with and without Information
Conducting a lab-in-the-field experiment with 1,905 cocoa farmers in Ghana, we investigate the impacts of an information intervention—highlighting both increasing climate risks and the benefits of shade—across two subsidy schemes. The input subsidy provides cash payments based on the number of shade trees planted; and the output subsidy offers price premiums, increasing in shade level, for cocoa produced from shaded farms. We assess how these subsidy types, with and without information, influence farmers' decisions to adopt shade trees.
Farmers in each of the five treatment arms independently participate in games involving shade tree planting and subsidy program participation on a one-acre cocoa plot, with land characteristics matched to their own. They are incentivized through cash earnings tied to game outcomes, determined by their planting decisions, as well as individualized non-tree inputs and simulated climate shocks reflecting real-world weather conditions.
Greater Impact of Information under the Output Subsidy
The lab results reveal a key insight: farmers receiving the information in both subsidy groups plant more forest trees than their subsidy-only counterparts, however, the increase is more pronounced under the output subsidy. Information nudging increases tree planting by 1.4 trees per acre (a 13.8% increase from the control mean) with the input subsidy and 2.4 trees per acre (a 23.6% increase) with the output subsidy, suggesting that the effectiveness of the output subsidy is more sensitive to belief accuracy.
Figure 1: Treatment effects of subsidies and information on forest trees in the lab
Note: The figure presents the treatment effects on the number of forest trees decided to plant on a one-acre cocoa plot in the lab game. Each bar represents the treatment effect magnitude relative to the control mean: input-based treatments are in green, output-based treatments in brown, with darker shades indicating information treatments. Magnitudes and p-values for group differences between each pair are reported.
Moreover, when validating the lab outcomes with farmers’ requests for forest tree seedlings under the Green Ghana Program outside the lab, we observe consistent effects three months post-lab. Information nudging, particularly when coupled with the output subsidy, leads to a larger increase in the number of tree seedlings requested and obtained, compared the input subsidy. These patterns not only corroborate the lab findings but suggest that incentivized subsidies and information interventions in the lab can partially influence farmers’ real-world actions, despite the intervening three months and the fully voluntary nature of the sign-up process.
Why Does Information Work Differently for Output vs Input Subsidies?
Farmers make ex-ante optimization decisions on shade trees to maximize expected profit under two subsidy schemes, shaped by their beliefs about rainfall uncertainty and shade benefits. Our theoretical model provides an intuitive explanation for the different roles of information in each scheme. With the input subsidy, farmers receive a fixed reward for planting shade trees, making information’s role positive but constant, independent of subsidy level. In contrast, under the output subsidy, the subsidy income depends on cocoa production and price premium, both of which vary by the number of shade trees. The impact of information under output is related to beliefs and can complement the effect of an increase in the subsidy level. When information nudging causes farmers to update their beliefs about either rainfall uncertainty or shade benefits, the resulting adjustment in expected yield influences the rate at which marginal returns to planting shade trees evolve as the subsidy level increases, but only under the output scheme.
Empirical Evidence of Information Updating Beliefs
Our findings indicate that information interventions lead to significant belief updates among farmers about shade benefits, with limited impact on perceptions of climate risks. Farmers receiving information are 10% more likely (7 percentage points) to recognize the protective value of shade. Yet, beliefs about rainfall variability remain stable, suggesting that the information primarily enhances farmers' understanding of shade benefits without altering their general perceptions of climate risks. We also gather data on each farmer's perceived optimal shade level, inferred from self-reported expected cocoa yields under varying shade levels. Farmers exposed to the information intervention tend to increase their perceived optimal shade, resulting in a higher posterior optimal shade level under adverse weather conditions compared to their subsidy-only counterparts.
Farmers who anticipate severe rainfall risks or perceive lower shade benefits tend to plant fewer trees under output subsidies when information is lacking, likely due to concerns about lower expected yield-based subsidy income. This highlights the importance of addressing information frictions alongside subsidy design to promote consistent adoption of climate-resilient practices.
Policy Implications: Toward an integrated approach
These findings underscore the importance of incorporating individual beliefs and local contexts when designing and evaluating subsidies for climate adaptation. Output subsidies demonstrate greater potential to drive adaptation, especially when the beliefs are reasonably correct. For policymakers, this suggests an integrated approach that combines financial incentives with information campaigns to enhance belief accuracy and encourage climate-resilient adaptation. As global climate risks and uncertainties intensify, such an approach becomes increasingly important, particularly for vulnerable farmers who often face limited access to reliable information. My results also inform broader discussions on input versus output subsidies across sectors, such as education and energy adoption, highlighting belief accuracy as a crucial factor influencing expected outcomes of subsidies in diverse contexts.
Yunyu Shu is a PhD student at Brown University. Her main research interests are development and environmental economics, with a focus on climate adaptation and green energy adoption.
Join the Conversation