银行竞争、企业债务与“脱虚向实”——基于双重机器学习方法

BANK COMPETITION,CORPORATE DEBT,AND BRINGING FIRMS BACK TO THE REAL ECONOMY——Based on Double Machine Learning

本文以2009年中国银监会降低中小商业银行设立分支机构的准入门槛作为外生事件,实证检验了银行业竞争加剧对企业债务和投资带来的影响。在实证方法上,本文采用双重机器学习模型进行因果识别,该方法能克服传统线性回归的模型误设和机器学习模型的正则偏误问题。本文研究发现:银行业竞争加剧能显著降低企业负债成本,提高企业长期借款的融资能力,并促进企业“脱虚向实”。在异质性分析中,本文发现银行竞争所带来的正面影响在小规模企业中表现得更为显著。本文研究对于解决中小企业“融资难融资贵”,改善企业金融化现象,以及优化金融资源配置具有重要启示。

This paper takes the deregulation of bank branch entry barriers in 2009 as an external shock and explores how bank competition influences firms'debt and investment.This paper uses double machine learning method to identify causal relation,which can solve the risk of misspecification in traditional OLS regressions as well as the regularization bias in machine learning.The results indicate that intensifying competition in banking sectors not only significantly reduces the corporate financing costs,increases firms'financing ability of long-term debt,but also brings firms back to the real economy.In addition,SMEs enjoy more benefits brought by this reform.This paper is important for solving the financing problem of SMEs,improving the financialization of firms and optimizing financial resource allocation.