Abstract: This paper uses labor-saving patents identified by an AI large language model as a proxy for labor-saving innovation. Findings show that such innovation in listed firms does not reduce total employment, wages, or labor income share; instead, it improves workforce skill structure by substituting low-skilled workers while boosting high-skilled employment. Prior research has found that importing robots, a typical labor-saving technology, negatively impacts China's employment and income. The difference between import and domestic R&D may lie in the “buy instead of build” model of imports, which hinders high-skilled employment growth in R&D and manufacturing, resulting in only a substitution effect. Results remain robust after controlling for robot imports and using foreign firms' labor-saving patents as an instrument. This study provides key insights on tech import vs. R&D, leveraging labor-saving innovation for aging populations, and demographic dividend transformation.

Key words: Innovation, Skill Upgrading, Labor Income, Imported Robot

摘要： 本文以AI大语言模型识别的劳动节约型专利为劳动节约型技术创新指标,研究发现上市企业的劳动节约型技术创新并未降低员工总数、工资水平和劳动收入份额,但优化了员工技能结构,具体表现为劳动节约型技术创新在替代低技能员工的同时刺激高技能就业岗位的增加。机器人进口这一典型的劳动节约型技术给我国就业和劳动收入带来负面影响,造成外部引进与自主研发二者差异的原因可能是机器人进口以“购买替代研发”,没有促进甚至还抑制企业研发和高端制造等方面的高技能员工数量,仅对员工产生替代效应。在控制机器人进口后,以行业中外资企业的劳动节约型专利作为工具变量进行回归检验,结果均保持稳健。上述结论对技术研发和引进的利弊、以劳动节约型技术创新应对人口老龄化和加快人口红利转型等问题具有一定的启示价值。

关键词: 技术创新, 技能结构升级, 劳动收入, 进口机器人