加载中...
Can Efficiency and Equity be Balanced? Analysis Based on Wage Rigidity
实现企业与员工利益协同增长是促进共同富裕的必要途径。工资刚性历来被视为阻碍劳动力市场出清的摩擦,本文则将其视作降低员工收入风险的有效机制,认为工资刚性能够提高企业生产效率。本文采用分位数回归森林模型测度企业层面的工资刚性,计量分析表明,工资刚性通过节约劳动力成本、减少员工流失和降低融资成本显著提高生产效率。异质性分析表明,工资刚性究竟提高还是降低生产效率取决于企业面临的风险环境,在非极端负面冲击下保障员工收入稳定能够为企业带来积极回报,在极端负面冲击下企业需要通过及时削减工资转嫁经营风险。此外,劳动力流动性越强、信息不对称程度越强,工资刚性的积极影响越显著。
Abstract
How to achieve the coordinated growth of corporate and employee interests is a core issue that must be addressed in advancing common prosperity. This paper approaches the problem from the perspective of downward wage rigidity, arguing that it is not merely a labor market friction but also an implicit insurance. Based on a quantile regression forest model to measure firm-level downward wage rigidity, the empirical results show that it significantly improves production efficiency by reducing labor costs, lowering employee turnover, and decreasing financing costs. Heterogeneity analysis reveals that whether downward wage rigidity leads to efficiency loss or efficiency gain depends on the firm's risk environment: under non-extreme negative shocks, income stability generates incentive effects and positive returns, whereas under extreme negative shocks, firms must cut wages to transfer risk. Moreover, the positive impact of downward wage rigidity is more pronounced in contexts characterized by high labor mobility and severe information asymmetry. Compared with existing literature, this study makes two major contributions. First, it reinterprets downward wage rigidity from the perspective of implicit insurance, showing that it is not simply a market friction but, under certain conditions, a mechanism that enhances production efficiency. It also clarifies the boundary conditions under which efficiency loss and efficiency gain perspectives apply. Second, this paper introduces machine learning into the traditional nonparametric measurement framework, estimating the firm-level downward wage rigidity. This approach overcomes the limitations of prior studies that focused only on industry or regional dimensions and provides fine-grained data support for subsequent research. The empirical analysis of firm-level downward wage rigidity not only fills a gap in studies of China's labor market but also offers robust evidence for academic debates on labor protection policies. The policy implications of this study provide an actionable framework for achieving the dual goals of efficiency and equity, thereby advancing common prosperity. Policymakers should strengthen implicit insurance mechanisms represented by downward wage rigidity and build a tripartite framework involving firms, employees, and government. Specifically, labor protection laws should be improved, incentive-compatible fiscal and tax policies should be designed, and wage insurance mechanisms should be established to ensure income stability while enhancing productivity. At the same time, in the face of extreme shocks, the government should play a stabilizing role to prevent firms from falling into bankruptcy due to downward wage rigidity. In this way, the study offers practical policy recommendations for balancing efficiency and equity in labor market governance.
工资刚性显著提升企业生产效率
基于双重机器学习方法,使用分位数回归森林测算企业层面工资刚性,回归结果显示工资刚性系数在1%水平上显著为正(以Lasso为例系数为0.036),表明工资刚性提升企业生产效率,验证了隐性保险视角的理论假说。
工资刚性降低企业劳动力成本
机制检验发现,工资刚性作为隐性保险机制,员工愿意接受工资折价以换取收入稳定,导致企业劳动力成本(人均工资对数)显著下降,各机器学习方法下系数均在1%水平上显著为负。
工资刚性减少员工流失
工资刚性通过保障员工收入稳定,增强员工满意度和信任感,显著降低员工流失率(以从业人数年增长率度量),所有机器学习方法下系数均在1%水平上显著为正。
工资刚性降低企业债务融资成本
工资刚性向债权人传递积极信号,降低信息不对称程度,从而降低企业债务融资成本,系数在5%或1%水平上显著为负。
核心解释变量
企业层面工资刚性(dwr),采用分位数回归森林方法测度,基于工资增速分布测算,条件变量包括企业规模、财务风险、流动性、经营效率、盈利水平、成长性、登记注册类型、行业、地区和年份
被解释变量
企业生产效率(tfp),采用LP方法测算全要素生产率,使用企业营业收入加增值税销项税额度量产出,固定资产度量资本存量,营业成本减折旧和劳动者报酬度量中间品投入,员工数量度量劳动投入
样本与数据
2007-2020年全国税收调查数据,非金融行业样本403,934个,剔除财务指标缺失和明显异常样本,开业年份缺失用工商注册数据补充,连续变量进行5‰截尾处理
识别方法 / 模型设定
采用去偏差/双重机器学习(DML)方法,分别使用Lasso、Ridge、CatBoost和LightGBM四种机器学习算法控制混淆因素;通过更换被解释变量测算方法(OP法)、引入相关随机效应控制企业固定效应、考虑企业进入退出、动态先行-滞后模型检验反向因果、工具变量估计等多种方法缓解内生性
内生性及稳健性检验
更换生产效率测算方法(OP法);引入相关随机效应(CRE)控制企业固定效应;选择至少存续6期企业样本;分布先行-滞后模型动态关系检验;使用最低工资标准、社会保险政策缴费率和同类企业工资刚性水平作为工具变量;排除固定资产加速折旧政策和工业机器人应用等替代性假说
更多相关数据正在补充