中文摘要
Hsieh 和 Moretti (2019) 发现,放宽美国三个高生产率城市的土地使用管制将使 GDP 提高 3.7%。在本评论中,我重新审视了他们的研究结果。我首先尝试复现其结果,却发现其反事实情形会降低产出。我记录了其代码中的错误,这些错误解释了这一差异。接着,我展示其模型的结果取决于人口单位的任意选择。我对他们的模型提出了一种消除单位依赖性的修改。在修改后的模型中,他们的实验提高了产出,但效应比他们所报告的结果小两个数量级。(JEL E23, J24, J31, R23, R31)
Abstract
Hsieh and Moretti (2019) find that relaxing land use regulations in three productive US cities would increase GDP by 3.7 percent. In this comment, I revisit their findings. I first attempt to replicate their result and find that their counterfactual would lower output. I document errors in their code that explain this discrepancy. I next show that the results of their model depend on the arbitrary choice of population unit. I propose a modification to their model that eliminates unit dependence. Their experiment raises output in the modified model, but the effect is two orders of magnitude smaller than what they report. (JEL E23, J24, J31, R23, R31)