Wossen Tesfamicheal, Alene Arega, Abdoulaye Tahirou, Feleke Shiferaw, Manyong Victor
International Institute of Tropical Agriculture (IITA), Nairobi, Kenya.
IITA, Lilongwe, Malawi.
Food Policy. 2019 Aug;87:101742. doi: 10.1016/j.foodpol.2019.101742.
Previous studies on the adoption and impacts of improved crop varieties have relied on self-reported adoption status of the surveyed households. However, in the presence of weak variety maintenance and poorly functioning seed certification system, measurement errors in self-reported adoption status can be considerable. This paper investigates how such measurement errors can lead to biased welfare estimates. Using DNA-fingerprinting based varietal identification as a benchmark, we find that misclassification in self-reported adoption status is considerable, with significant false negative and positive response rates. We empirically show that such measurement errors lead to welfare estimates that are biased towards zero and substantially understate the poverty reduction effects of adoption. While the empirical evidence suggests attenuation bias, our theoretical exposition and simulations demonstrate that upward bias and sign reversal effects are also possible. The results point to the need for improved monitoring of the diffusion process of improved varieties through innovative adoption data collection approaches to generate robust evidence for prioritizing and justifying investments in agricultural research and extension.
以往关于改良作物品种的采用情况及其影响的研究，依赖于被调查农户自我报告的采用状况。然而，在品种维护不力和种子认证体系运作不佳的情况下，自我报告的采用状况中的测量误差可能相当大。本文研究了此类测量误差如何导致有偏差的福利估计。以基于DNA指纹识别的品种鉴定作为基准，我们发现自我报告的采用状况中的错误分类相当严重，存在显著的假阴性和假阳性回应率。我们通过实证表明，此类测量误差会导致福利估计偏向于零，并大幅低估采用改良品种的减贫效果。虽然实证证据表明存在衰减偏差，但我们的理论阐述和模拟表明，向上偏差和符号反转效应也是可能的。研究结果表明，需要通过创新的采用数据收集方法，加强对改良品种扩散过程的监测，以便为优先安排和论证农业研究与推广投资提供有力证据。
