广西民营上市公司运营效率测度、驱动因素与治理效应研究

Study on Measurement, Driving Factors and Governance Effects of Operational Efficiency of Private Listed Companies in Guangxi

摘要: 本文选取2018~2025年广西7家产权始终保持民营属性的上市公司为研究样本，采用超效率SBM-Malmquist模型测算企业运营效率，从纯技术效率、规模效率、技术进步三个维度解析全要素生产率的内在驱动来源，并利用OLS分层回归(HC1稳健标准误)检验公司治理因素对效率的影响。研究发现：样本企业运营效率分化突出，桂林三金是唯一持续处于生产前沿之上的企业(效率几何均值1.413且稳步上升)，皇氏集团效率深陷低位(几何均值0.198)，莱茵生物效率由前沿之上回落跌破前沿；样本整体全要素生产率明显改善(TFPCH几何均值1.116)，改善以效率追赶为主导(EFFCH = 1.070)，技术进步同步贡献(TECH = 1.043)，莱茵生物是唯一TFPCH未达1的企业；分层回归显示，第一大股东持股在仅含治理变量时显著为正，但加入财务控制变量后不再显著，说明H1的正向效应受到财务特征混淆；两职兼任在基准模型中显著为正，但其结论对投入产出指标集合较为敏感；董事会规模在基准模型及多数稳健性设定下保持显著为负。受广西本地民营上市企业数量限制，本研究样本规模有限，研究结论适用范围需谨慎界定。

Abstract: Taking seven consistently private listed companies in Guangxi from 2018 to 2025 as research samples, this paper adopts the super-efficiency SBM-Malmquist model to measure corporate operational efficiency, and analyzes the internal drivers of total factor productivity from pure technical efficiency, scale efficiency and technological progress. OLS hierarchical regression with HC1 robust standard errors is used to examine the impact of corporate governance on efficiency. The study finds that the operational efficiency of the sample enterprises is highly differentiated: Guilin Sanjin is the only firm persistently above the production frontier (geometric mean 1.413 with steady improvement), Huangshi Group remains deeply inefficient (geometric mean 0.198), and Laiyin Biological falls from above the frontier to below it. Overall TFP improved markedly (TFPCH geometric mean 1.116), driven mainly by efficiency catch-up (EFFCH = 1.070) with technological progress contributing in tandem (TECH = 1.043), and Laiyin Biological is the only firm with TFPCH below unity. Hierarchical regression reveals that largest shareholder ownership is significantly positive in the governance-only model but loses significance after financial controls are added, indicating that the positive effect hypothesized in H1 is confounded by financial characteristics. CEO duality is significantly positive in the baseline model but its conclusion is sensitive to the choice of input-output indicators. Board size is negatively significant in the baseline model and most robustness specifications. The study sample size is limited due to the number of local privately owned listed companies in Guangxi, and the applicable scope of the research conclusions needs to be carefully defined.