⚬ 论文

How Responsive is Investment in Schooling to Changes in Redistributive Policies and in Returns?

收录时间:2025-12-02 摘要:This paper uses an unusual pay reform to test the responsiveness of investment in schooling to changes in redistribution schemes that increase the rate of return to education. We exploit an episode where different Israeli kibbutzim shifted from equal sharing to productivity-based wages in different years and find that students in kibbutzim that reformed earlier invested more in high school education and, in the long run, also in post-secondary schooling. We further show that the effect is mainly driven by students in kibbutzim that reformed to a larger degree. Our findings support the prediction that education is highly responsive to changes in the redistribution policy. Slides

⚬ slides

AI_innovation

收录时间:2025-12-31 摘要:2022年全面启动的“东数西算”工程是中国数字经济与人工智能版图重构的战略性举措,旨在通过构建国家算力枢纽节点,解决算力供需的空间配置问题。本文旨在通过严谨的计量经济学框架,评估该政策在早期的实施效果。研究首先采用双重差分模型,基于54个地级市的宏观面板数据与AI发明专利存量,检验政策对枢纽城市创新产出的因果效应。实证结果显示,DID估计系数在统计上不显著,即政策并未在短期内引发政策实施地区的专利爆发。 针对这一看似“无效”的零结果,本文并未止步于此,而是进一步构建了基于2025年信通院“算力分”的截面回归模型,将2021年专利存量作为控制变量。结果发现,在控制了经济与电力等基础设施变量后,历史专利存量对未来算力评分不具有解释力(系数不显著),而政策实施组虚拟变量显著为正。这一“零相关”结果,结合残差诊断中识别出的“廊坊”离群点(非节点城市但拥有超高算力),揭示了算力基础设施与传统创新积累的“空间脱钩”现象。 结合南水北调等历史工程的经验证据与大模型时代的技术逻辑,本文提出核心论点:实证结果的背离实际上反映了人工智能创新范式正从“专利导向”的离散算法创新(小模型时代),向“模型导向”的工程化基建(大模型时代)发生深刻转型。在这一新范式下,算力基础设施的先行部署构成了创新的必要非充分条件,政策的有效性应由“算力规模”而非“专利数量”来衡量。本文的研究不仅解释了“廊坊离群”,也为理解新基建背景下的区域数字经济与人工智能发展政策提供了新的评估维度。