Index-Based Yield Protection for Smallholder Farmers: Insights and Data-Driven Policies

Abstract

Smallholder farmers in emerging economies are vital to the global food supply but remain highly vulnerable to yield losses from risks like extreme weather. Unlike in developed countries, where government subsidies are often used to compensate farmers in low-yield scenarios, implementing such yield protection subsidies in emerging economies is often impractical due to the high costs of yield assessment for small farms. To address these challenges, an innovative index-based approach to yield protection has been gaining popularity in emerging economies, under which payments to farmers are triggered when a pre-determined index, such as rainfall, predicts a low yield, thereby avoiding costly yield assessment. This dissertation examines the government's optimal design of index-based yield protection subsidies.In Chapter 1, we study the optimal design of index-based yield protection from an analytical perspective. We develop a game-theoretic model that consists of a local government and a unit mass of farmers who make planting decisions for a certain crop. The model captures three sets of salient features of our problem: (1) yield uncertainty and the impact of aggregate supply on the market price; (2) farmer risk aversion; and (3) an index-based subsidy payment, where the index is imperfectly correlated to the actual crop yields. Our analysis generates several intriguing findings. First, we show that unlike subsidies that are based on actual yield, an increase in index-based subsidy may increase farmer income variance due to potentially inaccurate yield prediction of the index. Based on this result, we uncover a non-monotonic relationship between the optimal subsidy amount and the accuracy of the index. Second, although price and yield protection are often viewed as strategic substitutes since both can incentivize more planting, we show that they act as strategic complements when the index accuracy is low. Finally, when the government can exert a costly effort to improve index accuracy, contrary to expectations, we find that a tighter budget can lead to a higher optimal investment in index accuracy. Collectively, these insights contribute to a more nuanced understanding of index-based yield protection policies, aiding in developing effective agricultural subsidies.In Chapter 2, we study the optimal design of index-based yield protection from a data-driven perspective. In practice, when determining the index-based subsidy payment, governments typically use predicted yields as indicated by the index as direct substitutes of the unknown actual yields. However, such a commonly used approach has two key limitations: first, while generating predicted yields, this approach usually overlooks an effect of farmers' risk aversion which influences how prediction errors impact farmers across different actual yield levels; second, by treating predicted yields as actual yields, this approach does not explicitly consider the inaccuracies in these predictions which critically determines the effectiveness of a subsidy payment. In this study, using historical yield data and yield-related features, we propose a new data-driven framework to determine the optimal payment amounts for index-based yield protection policies. First, we establish high-probability performance guarantees for our approach compared to the oracle approach, assuming full knowledge of the underlying true yield model. Next, we analytically demonstrate that our approach addresses the two key limitations of commonly used approaches, thereby achieving greater effectiveness in benefiting farmers and enhancing robustness to model misspecification. We further show that, compared to the commonly used approach, our method can create a win-win scenario where farmer welfare is improved and government expenditure is reduced. Finally, we highlight that our approach requires minimal adjustments to the commonly used approaches, preserving their advantages of being easy to implement, interpretable, and logically intuitive for farmers.