摘要
在外部环境复杂多变的背景下,研究关税变化对中国创新发展的影响机制具有重要意义。本文利用2018年美国对华加征关税这一外部冲击,使用双重差分法探究外部关税冲击对中国企业创新产出和效率的影响,并通过文本分析和Word2vec机器学习技术,测度了企业家注意力配置,借此进一步分析了外部负向关税冲击影响企业创新的传导机制。结果发现:外部负向关税冲击显著降低了中国企业的创新产出和创新效率;企业家注意力内容配置和注意力时间配置是产生以上影响的两个渠道;政府增加R&D补贴、媒体和分析师减少关注、具有抗压能力强特质的企业家,均可以缓解外部关税冲击对企业创新的负向影响。本文结论意味着要为企业家确立良好的政策导向和社会环境,缓解外部关税压力,进而使其在自主创新中更好地发挥企业家精神。
The increasingly complex international environment has put forward more urgent requirements for China to accelerate technological innovation.This paper exploits the external shock of U.S.tariff increase on China in 2018 to explore the impact of external tariff shocks on Chinese firms'innovation output and efficiency using difference in differences method,and measures entrepreneurs'attention allocation through text analysis and Word2vec machine learning techniques,through which we further analyze the transmission mechanism of external tariff shocks affecting firms innovation.The results find that:(1)External tariff shocks can significantly reduce Chinese firms'innovation output and efficiency;(2)Entrepreneurs'attention content allocation and attention time allocation are two important channels to produce the above effects;(3)Government increasing R&D subsidies,the media and analysts reducing excessive attention,and entrepreneurs with high stress resistance can mitigate the negative effects of external tariff shocks on firms'innovation.The conclusion of this paper implies that government need to create a favorable policy orientation and public opinion environment for entrepreneurs to alleviate external tariff pressure and thus enables them to better exercise their entrepreneurial spirit in independent innovation.
作者
余振
李元琨
李汛
Yu Zhen;Li Yuankun;Li Xun
出处
《世界经济》
CSSCI
北大核心
2024年第6期65-94,共30页
The Journal of World Economy
基金
国家社科基金国家应急管理体系建设研究专项“全球重大突发事件中的国际合作:中国参与战略及对策研究”(20VYJ035)
国家社科基金项目“中美贸易摩擦对我全国统一大市场建设的影响和对策研究”(23STA013)
中国博士后科学基金面上项目“美国对华经贸政策与中国企业创新困境及突破路径研究”(2023M732020)的资助。
关键词
关税冲击
企业家注意力配置
自主创新
文本分析
机器学习
tariff shock
entrepreneurial attention allocation
firm innovation
textual analysis
machinelearning