Cross-Sector Competition and Corporate Default Risk:Based on Machine Learning and Complex Network Approach
The report to the 20th National Congress of the Communist Party of China pointed out that it is necessary to strengthen and improve modern financial supervision and ensure that no systemic risks arise.However,in recent years,as a significant source of systemic financial risks,firm default on debt is one of the most destructive events in firm development.It's not only posing threats to the firms'healthy operations but also potentially triggering economic catastrophes that could evolve into systemic debt crises through economic interconnections,thereby endangering the stability and security of the whole financial system and socio-economic environment.Therefore,exploring what and how the risk of debt default is caused has become a critical issue in the new era.Furthermore,case studies show that corporate debt defaults are often closely related to firm's aggressive investment behavior across industries,exemplified by firms like Macrolink Holding,Guogou Investment,HNA Group,Founder Group,China Evergrande Group,LeEco,and CITIC Guoan Group.These firms engage in aggressive expansion and cross-sector competition,incurring substantial debts and persistent liquidity pressures,while short-term profitability fails to materialize,eroding their core competencies and innovative capabilities.This confluence of factors finally leads to a default crisis.Thus,the question of this paper focuses on how cross-sector competition affects the risk of debt default.However,this cannot be addressed solely through case studies but also requires rigorous empirical analyses.Our empirical analysis centers on data from China's A-share listed companies over the 2007-2020 period.We collect textual data from the Management's Discussion and Analysis section of firms'annual reports and obtain financial data from the China Stock Market and Accounting Research(CSM AR)Database and Chinese Research Data Services Platform(CNRDS)Database.A simple research approach utilizes traditional diversification metrics as core explanatory variables in empirical design,such as calculating the Herfindahl index or entropy index of sales income from the number of industries an enterprise operates in,as reported in annual financial statements.Nevertheless,measurement bias in this method warrants caution as traditional diversification metrics have several shortcomings.For example,industry breakdowns information in financial reports might not reflect the true extent of an enterprise's industrial diversification.Additionally,firms often change their reported sectors without real operational changes.Thereby,by employing machine learning and complex network methods to analyze unstructured textual data,this paper constructs a novel and firm-specific proxy to measure the behavior of firm's cross-sector competition,validates its informativeness and investigates the effect of firm's cross-sector competition on firm's likelihood of financial distress.Specifically,the results of our empirical analyses demonstrate that higher level of cross-sector competition results in higher risk of debt default.Our findings remained robust after a series of robustness checks.For example,we use a two-stage least squares model with historical instrumental variables to mitigate the endogeneity issues.The results confirm our main findings.In addition,the results of the mechanism analyses provide three possible channels through which cross-sector competition results in higher risk of debt default:increased pressure of debt repayment,inefficient allocation of resources,and erosion of innovative capacities.The main effect is concentrated in subsamples for which supply of credit resources is sufficient,the pressure of market attention is high,and the managers are overconfident and incompetent.Lastly,the rising debt default risk,induced by intense cross-sector competition,impairs the firms'productivity and value-creation capability.This study documents a previously under-identified adverse consequence of competition:its exacerbation of firm's default risk.This paper makes the following three contributions.First,by exploiting machine learning techniques to process unstructured text data and complex network methods,this study creates a more accurate firm-level variable than traditional methods to measure firm's cross-sector competition and validates its informational value.In contrast to extant literature,this measure provides a more direct,objective,and effective approach to quantify firm's competition in non-core industries.It features high accuracy and variability,significantly mitigating problems in existing literature.Also,it provides a robust data foundation for subsequent research and paves the way for future studies to integrate text analysis and network methods.Second,based on some case studies,this paper examines firm diversification strategies from the novel perspective of cross-sector competition.This paper systematically and deeply explores the impact of this competition on the risk of debt default for the first time.Furthermore,the paper discusses how firms'cross-sector competition exacerbates their debt default risk from the channels of debt pressure,allocation of resources,and innovative capacities.Thus,it not only complements the literature on product market competition but also on the factors affecting the risk of debt default.Thereby,this paper is of significant practical relevance and provides guidance,offering valuable insights for regulating healthy corporate development and maintaining the stability of the financial system.Finally,the literature enriches the heterogeneous analysis of the impact of firm's cross-sector competition on the risk of debt default from internal and external aspects,including credit provision,market pressures,and managerial characteristics.Thus,this study has clear policy implications on how to improve the efficiency of credit resource allocation in China,foster the healthy and stable operation of capital markets,and refine corporate governance and business management practices in the new era.
NETL
NSTL
万方数据
