加载中...
The Demand of New Quality Human Capital and Great Power Strategic Competition: Evidence from Large Language Models Data Mining
人才是发展新质生产力的第一要素。本文基于互联网招聘大数据,创新性地使用大语言模型测度企业新质人才需求。根据工作任务分析框架构建理论模型,将美国对中国企业实施的“清单式”制裁作为外生冲击,考察大国战略竞争背景下,美国制裁冲击如何影响中国企业的新质人才需求。分析结果显示,制裁冲击整体上显著提高了企业的新质人才需求,任务配置效应和要素替代效应是产生上述影响的主要路径,然而,由制裁所引致的生产规模收缩对企业新质人才需求产生负向影响。此外,制裁冲击还具有显著的供应链溢出效应和数字任务偏向性特征,并对中国整体劳动力市场的新质人才需求产生促进作用。
Abstract
Human capital serves as the primary driver in developing new quality productive forces. Drawing on micro-level data from Chinese A-share listed enterprises and online job vacancy datasets spanning 2014-2022, this study employs two large language models (LLMs) – GPT and SBERT – to classify standardized occupational skill categories in online job postings and quantify enterprise demand for new quality human capital. Building on this foundation, the paper constructs a theoretical model within a task analysis framework, treating U.S. "list-based" sanctions against Chinese enterprises as a quasi-natural experiment. Using a difference-in-differences (DID) approach, it examines how U.S. sanction shocks influence Chinese enterprises' demand for new quality human capital, as well as the underlying mechanisms, within the context of great power strategic competition. Theoretically, U.S. sanction shocks generate competing effects on Chinese enterprises' demand for new quality human capital: the task allocation effect and factor substitution effect increase demand, while the production scale effect reduces it due to contracted output of final products. Empirically, U.S. sanctions significantly increase overall demand for new quality human capital among affected enterprises. Mechanistically, these shocks operate primarily through positive task allocation and factor substitution channels, though their net impact is partially offset by reduced demand stemming from production scale contraction. Heterogeneity analysis reveals that U.S. sanction shocks induce Chinese enterprises to significantly increase demand for highly skilled human capital while boosting employment of high-efficiency and high-quality talent – thereby improving talent chain allocation through new quality human capital demand effects. Moreover, the policy effects vary substantially across different U.S. sanction lists: export control and investment restriction measures positively influence Chinese enterprises' human capital demand, whereas sanctions imposing financing constraints exert negative effects. Furthermore, the positive effect of sanction shocks on human capital demand proves particularly pronounced in technology-intensive enterprises and those with higher productivity levels. Beyond enterprise-level effects, extended analysis demonstrates that U.S. sanction shocks generate significant backward supply chain spillover effects, elevating demand for new quality human capital among upstream suppliers to sanctioned Chinese enterprises. These shocks also substantially reshape labour skill demand and task composition patterns – particularly through increased employment of digitally-skilled talent. Preliminary estimates derived from the empirical results suggest that these sanction shocks additionally produce measurable economic effects on new quality job creation across China's broader labour market. This study provides theoretical and empirical evidence on structural shifts in Chinese enterprises' new quality human capital demand amid geopolitical upheaval, offering important policy insights for cultivating new quality productive forces.
美国制裁冲击显著提高企业新质人才需求
基于多期DID模型,核心解释变量Sanction的估计系数为0.0562,在1%水平下显著为正,表明美国制裁冲击平均促使受制裁企业的新质人才需求份额提高约18.34%,具有显著的经济意义。
任务配置效应和要素替代效应是主要作用路径
机制检验发现,制裁冲击通过任务配置效应(企业将任务由开放模式转向自主模式,扩大新质人才雇佣)和要素替代效应(用本国新质人才替代价格上升的进口技术投入)正向促进企业新质人才需求。
生产规模收缩对新质人才需求产生负向影响
机制检验表明,制裁冲击显著降低了受制裁企业的主营业务收入(生产规模),而生产规模收缩进一步降低了企业的新质人才需求,验证了生产规模效应的存在。
制裁冲击存在显著的供应链后向溢出效应
进一步分析发现,中心企业受制裁后,其上游供应商的新质人才需求显著提升(估计系数0.1408),表明存在供应链'绑定效应',但对下游客户的新质人才需求无显著影响。
制裁冲击具有数字任务偏向性,并促进整体新质就业
制裁冲击显著提升了企业对数字技术创新型、融合型和技能改进型人才的需求。利用包络反推法估算,2018-2022年美国制裁冲击使中国劳动力市场新质就业岗位整体增加约328,315个。
核心解释变量
核心解释变量为美国制裁冲击(Sanction),即企业是否被列入美国各类制裁清单的虚拟变量。企业被列入清单的当年及以后赋值为1,否则为0。制裁清单包括未经验证清单、实体清单、最终军事用户清单、中国涉军企业制裁清单、非SDN中国军工复合体清单和特别指定国民清单。
被解释变量
被解释变量为企业新质人才需求份额(NT),即企业新质人才招聘数量除以企业招聘总量。新质人才涵盖高技术、高效能和高质量三类人才,基于《中华人民共和国职业分类大典(2022年版)》标准职业分类体系识别。
样本与数据
样本为2014-2022年中国A股上市公司。数据来源包括:企业招聘大数据来源于各知名互联网招聘平台和CnOpenData数据库;制裁信息来源于美国各政府机构官方网站;企业财务数据来源于CSMAR、Wind和CNRDS;城市数据来源于EPS数据平台和城市统计公报。剔除金融行业、ST、财务数据缺失及负债率异常的样本,连续变量进行1%缩尾处理。
识别方法 / 模型设定
采用多期双重差分模型(DID),利用美国'清单式'制裁分批实施的特点,将企业首次被列入制裁清单视为外生冲击。通过比较受制裁企业与未受制裁企业在制裁前后新质人才需求的变化来识别因果效应。
内生性及稳健性检验
稳健性检验包括:事前趋势检验、异质性处理效应分析、安慰剂检验、排除同期外生冲击、更改模型设定形式、替换核心变量、调整样本范围、控制选择偏误、数据代表性讨论、制裁外生性检验、工具变量法以及排除其他潜在干扰因素等。