高涵

内容提要：当前，新一轮财税体制改革意义重大，重塑地方政府的以地谋发展行为和激发市场主体的高质量发展迫在眉睫。“营改增”是我国财税体制现代化的重大改革，本文以“营改增”政策为准自然实验，基于土地市场网 2010 年至 2016 年的微观土地出让信息和匹配了地块的税收调查数据，实证检验了此改革如何影响产业用地市场的供需双方，以及基于机器学习方法提出政策优化具体方向。结果表明，相比其他城市，试点城市在政策推行后工业用地的出让面积显著减少了 10%，交易规模下降 7%，实际单价上升 5%，且效应在建设用地指标约束较强的城市中更为突出，在与服务业关联更紧密的行业以及质量较好的高等级地块更为明显。进一步地，采用因果路径分析考察了“财政激励效应”和“需求冲击效应”及二者的相对重要性，发现“营改增”使得制造业实际增值税率下降，这对政府而言是负向的财政激励，进而减少了出让工业用地培育增值税税源的动机；而在需求侧，这一税制改革使得制造业企业资源配置更合理，放缓了粗放式购置土地扩张厂房的动机，因此最终工业用地交易规模减少。最后，基于市场均衡处理效应框架的分析表明，工业用地的市场化交易与集约节约利用仍是未来改革的重点。本文的研究结论能为后续深化财税体制现代化改革以及城市土地配置发展目标提供优化方向。

关键词：税制改革；营改增；工业用地；因果路径分析；市场均衡下的处理效应

摘要：随着地理编码（geocoded）微观数据的普及，以“到某一地点的地理距离”界定处理状态、考察其对邻近单元影响的研究设计，在经济学经验研究中日益普遍，广泛应用于污染、交通、城市更新、产业集群与企业区位等议题。然而，如何依据距离合理界定处理组与控制组、距离效应的理论基础何在，相关方法仍缺乏系统梳理。本文以潜在结果框架与“基于设计”（design-based）的研究范式为统一基础，澄清“空间处理”的概念内涵与识别假设，系统梳理距离缓冲区、环形法、距离梯度、空间双重差分、邻里固定效应与处理分配建模等主流方法，并结合蒙特卡洛模拟与图示介绍非参数处理效应曲线估计、反事实候选地点的机器学习构造、基于设计的标准误等前沿进展。在此基础上，本文给出一套可操作的“标准化操作流程”与“可行性检验”清单，回顾权威中英文期刊的应用实践，剖析处理环与控制环设定任意、控制组溢出污染与不可比、对照组构造不当、空间相关推断失当等常见误区，并提出相应的解决方案，以期为相关经验研究提供方法论参考。

关键词：地理距离；空间处理变量；因果推断；空间双重差分；环形法

Abstract: While there is a consensus that widespread competition exists among local governments within multi-layered large economies, the specific mechanisms of how this competition unfolds remain debated. Existing literature has predominantly focused on the instruments of competition rather than the process and its market implications. This paper investigates the behavioral logic and dynamic processes of local governments from a social interaction perspective, focusing on low-price subsidies for industrial land. Using tax survey data matched with micro-level industrial land transaction records, we explore the origins of the “race to the bottom” in land prices. We find that local government decisions regarding industrial land supply are not based solely on local factor endowments; rather, governments exhibit strong peer effects, mimicking and following the decisions of peer cities, often converging toward those offering the highest subsidies. Further analysis reveals that this behavioral response is more sensitive in key industries and high-tech sectors identified in planning documents, indicating significant industrial heterogeneity. Sub-sample analyses based on economic development levels and official turnover show that homogenous competition among local governments is complex, involving both imitative behavior and strategic adjustments driven by economic status and political cycles. Finally, we examine the economic consequences and find that this mimetic behavior distorts land supply without effectively enhancing firm productivity or promoting innovation.