Econ Papers Daily / 期刊论文
对“加拿大北方营养补贴传导率”一文的评论
Comment on “On the pass‐through rate of the Nutrition North Canada subsidy”
摘要
Naylor, Deaton and Ker provide estimates of Nutrition North Canada subsidy pass‐through using data we collected and an identification strategy we devised. They propose an alternative econometric specification based on estimating passthrough for separate monopoly and non‐monopoly sub‐samples and aggregating the coefficients using subsidy share weights. They argue that this specification is more appropriate for estimating overall subsidy pass‐through and conclude that its application to our data provides evidence for full subsidy pass‐through. We show that their conclusions are not warranted by making two distinct but related points. First, we show that conclusions based on their specification are not robust to two changes in the estimating sample. Correcting two subsidy assignment and one monopoly classification error in the original data set, or using a time period more relevant for comparing heterogeneous effects across monopoly and non‐monopoly communities, leads to estimates that either reinforce our original finding of incomplete passthrough or that are too imprecise to be informative. Second, we provide a graphical illustration and power calculation to make a more general point about the appropriateness and limitations of the type of sub‐sample analysis they propose in settings with limited sample size and statistical power.
中文摘要
Naylor、Deaton和Ker利用我们收集的数据及我们设计的识别策略，估算了“营养北方加拿大”补贴的传导效应。他们提出了一种替代性计量模型，基于对垄断与非垄断子样本分别估算传导效应，并通过补贴份额权重对系数进行加总。他们认为该模型更适合估算整体补贴传导效应，并据此得出结论：将该模型应用于我们的数据可证明补贴完全传导。我们通过两个独立但相关的论点表明，其结论缺乏依据。第一，我们证明其模型结论对估算样本的两项调整缺乏稳健性。修正原始数据集中的两项补贴分配错误和一项垄断分类错误，或采用更适用于比较垄断与非垄断社区异质性效应的时段，所得估算结果要么强化了我们最初关于不完全传导的发现，要么因精度过低而无法提供有效信息。第二，我们通过图形说明和统计功效计算，就其在样本量有限且统计功效不足的背景下所提出的子样本分析方法的适用性与局限性，提出更一般性的观点。