Fixed or mixed? Farmer‐level heterogeneity in response to changes in salinity
利用加州三角洲农场面板数据，比较固定系数与混合Logit模型，发现农民对灌溉水盐度的反应存在显著异质性，混合模型预测的盐敏感作物面积弹性比固定模型大一个数量级。
Abstract We study supply response to irrigation water salinity—an important and ubiquitous environmental problem facing agriculture around the world. The geographical setting is the Sacramento–San Joaquin River Delta, a main water hub in California, where salinity is a significant part of recent policy debates over water management and infrastructure. We use highly granular farm‐level panel data on Delta farming activity to estimate two sets of response parameters using (i) standard (fixed coefficients) logit model, and (ii) a mixed (random coefficient) logit model. The mixed logit results provide evidence for heterogeneity in supply response across farmers. To put these findings in a meaningful economic context, we use the results from both models to simulate estimates for aggregate acreage responses under two alternative salinity scenarios. We implement a method to simulate individual specific responses together with their confidence intervals, which provides additional insight into the composition of farmer heterogeneity. Based on these post‐regression simulation results, we find the mixed logit model predicts an elasticity of aggregate acreage for salt‐sensitive (hence higher value) crops that is an order of magnitude larger compared to that of the standard fixed coefficients logit method. This result sheds light on category of costs likely to increase in response to rising salinity in the Delta region.