Access to Credit and Agricultural Productivity: Evidence from Kenyan Maize Farmers

The primary data for this study came from a survey designed to determine the impact of credit on maize productivity in Kenya. According to Yadav & Rao (2024), 400 smallholder maize farmers in five sub-counties of Trans Nzoia, Kenya, were surveyed between 2020 and 2023. The study focused on 200 maize smallholders who received agricultural credit (treatment group) and 200 who did not receive credit (control group). By Wanzala et al., (2023), the 200 quasi-control group was chosen using a propensity matching score (PMS) that runs a logistic regression of the observable characteristics listed in Table 1. PSM was used to prevent selection bias because the treatment variable (access to agricultural credit) was not randomized. The authors also use sensitivity analysis with "Rosenbaum bounds" to assess the impact of hidden bias. All the farmers in this study produced maize for household consumption, demonstrating that they were credit constrained. Farmers who produced improved maize seed but were not credit-constrained were excluded from the study. This is because these smallholder farmers do not face the same technical production constraints and risks. The treatment group was drawn from the Chereng'any, Kwanza, and Saboti sub-counties. The control group was drawn from the Endebess and Kiminini sub-counties. To reduce potential agricultural input spillovers to the control group in their sublocations, this study separated the treatment and control groups with a 6 km buffer zone (Wanzala et al., 2023).

本研究的主要数据来源于一项旨在探究信贷对肯尼亚玉米生产效率影响的研究调查。依据 Yadav & Rao (2024) 的研究，2020年至2023年间，对肯尼亚Trans Nzoia地区的五个次区域共计400名小型玉米种植户进行了调查。研究聚焦于200名获得农业信贷（实验组）和200名未获得信贷（对照组）的玉米种植户。Wanzala等人（2023）通过倾向得分匹配（PMS）选择了200名准对照组，该匹配分数通过运行包含在表1中的可观测特征的逻辑回归得出。为了防止选择偏差，由于实验变量（农业信贷的获取）并非随机分配，因此采用了PSM。作者还运用了带有“Rosenbaum界限”的敏感性分析来评估隐藏偏差的影响。所有参与研究的农民均生产玉米以供家庭消费，这表明他们面临信贷约束。那些生产改良玉米种子但未受信贷约束的农民被排除在研究之外，这是因为这些小型种植户不面临相同的技术生产约束和风险。实验组来自Chereng‘any、Kwanza和Saboti次区域，而对照组则来自Endebess和Kiminini次区域。为了减少潜在农业投入对对照组次区域的溢出效应，本研究在实验组和对照组之间设置了6公里宽的缓冲区（Wanzala等人，2023）。