Date of Award

2025

Degree Type

Thesis

Degree Name

Doctor of Business Administration (DBA)

Abstract

上市公司信用风险预测是金融风险管理的关键环节。传统企业信用风险评估方法存在数据滞后、维度单一等问题，难以捕捉微观区域动态变化，随着遥感技术发展，夜间灯光为构建多维度信用风险评估框架带来新思路。

本研究以沪深两市A股制造业上市公司为对象，探索夜间灯光数据在企业信用评估中的价值，构建地理空间风险预警新机制。基于既有研究，选取2019-2023年1964家公司的9820组财务与夜间灯光数据为样本。以制造业上市公司工厂区域夜间灯光亮度（ANTLT）为核心解释变量，违约距离（DD）、违约风险概率（EDF）、借款逾期（Default）、信用评级（Rating）为被解释变量，纳入代表性企业财务指标作为控制变量，运用多元回归模型实证检验夜间灯光数据对企业信用风险的影响。

研究结果显示，制造业上市公司工厂区域夜间灯光强度与企业信用风险呈显著负相关，验证了夜间灯光数据可作为信用风险评估的补充指标。稳健性检验表明，制造业上市公司行业层面信用风险的变化方向与单个公司信用风险的变化方向呈现一致性。在控制企业规模、行业特征及内生性因素的前提下，制造业上市公司工厂区域的夜间灯光强度与企业盈利能力在统计意义上存在显著正相关关系。中介效应检验确证，研发投入费用率在夜间灯光强度影响企业信用风险过程中发挥显著中介作用，但中介效应强度与路径存在指标异质性。异质性分析发现，制造业上市公司夜间灯光强度对不同产权性质企业违约概率的影响存在差异，国有企业的风险变动幅度和显著性均高于非国有企业。

本研究在理论上拓展了非财务数据在信用风险评估中的应用，在实践中为金融机构、投资者、监管机构及企业自身提供决策参考。未来研究可围绕开发动态宏观环境模型、融合多源卫星数据、构建行业差异化评估基准、设立风险预警机制等方向，推动该领域研究向纵深发展。

Predicting listed companies' credit risk is crucial in financial risk management. Traditional corporate credit risk assessment methods face data lag and one-dimensionality, hindering micro-regional dynamic change capture. Advancements in remote sensing technology enable Nighttime Light (NTL) data to offer a novel approach for constructing a multi-dimensional credit risk assessment framework.

This study targets A-share manufacturing listed companies on the Shanghai and Shenzhen Stock Exchanges, exploring the value of NTL data in corporate credit assessment and constructing a novel geospatial early-warning mechanism. Based on prior research, a sample comprising 9,820 sets of financial and NTL data from 1,964 companies over the period 2019-2023 was compiled. The core explanatory variable is the ANTLT within the factory regions of these manufacturing listed companies. Dependent variables include Distance-to-Default (DD), Expected Default Frequency (EDF), Loan Default (Default), and Credit Rating (Rating). Representative corporate financial indicators are incorporated as control variables. Multivariate regression models are employed to empirically test the impact of NTL data on corporate credit risk.

The results reveal a significant negative correlation between the intensity of nighttime lights within manufacturing listed companies' factory regions and their credit risk, verifying that NTL data can serve as a supplementary indicator for credit risk assessment. Robustness checks demonstrate that the direction of credit risk changes at the manufacturing industry level aligns with changes at the individual company level. Controlling for firm size, industry characteristics, and endogeneity concerns, a statistically significant positive correlation exists between factory region NTL intensity and corporate profitability. Mediating effect tests confirm that the R&D expenditure ratio plays a significant mediating role in the process through which NTL intensity influences corporate credit risk; however, the strength and pathway of this mediation exhibit heterogeneity across different indicators. Heterogeneity analysis found that the impact of NTL intensity on the default probability differs based on the ownership nature of manufacturing listed companies, with risk changes for SOEs being both more significant and of greater magnitude than those for non-SOEs.

This research theoretically expands the application of non-financial data in credit risk assessment. In practice, it provides a decision-making reference for financial institutions, investors, regulatory bodies, and companies themselves. Future research could advance the field through directions such as: developing dynamic models incorporating the macro-environment, integrating multi-source satellite data, establishing industry-differentiated assessment benchmarks, and setting up operational risk early-warning mechanisms.

Keywords

夜间灯光, 信用风险评估, 制造业上市公司, 实证分析, Nighttime Light, Credit Risk Assessment, Manufacturing Listed Companies, Empirical Analysis

Language

Chinese (Simplified)

Comments

An empirical study on assessing the credit risk of listed manufacturing companies based on nighttime light data

Recommended Citation

王土生 (2025)。基于夜间灯光数据评估制造业上市公司信用风险的实证研究 (博士論文，香港嶺南大學)。檢自 https://commons.ln.edu.hk/otd_tpg/67/

Download

Share

COinS