基于稀疏主成分的我国上市公司信用风险评价
首发时间:2020-04-01
摘要:信用是经济正常运转的基石,也是经济不断发展的前提与基础。加强对我国上市公司信用风险进行评价和管理,有利于上市公司、投资者、金融机构等都具有重要的意义。现有研究存在指标体系不完善、输入变量信息重复冗余、样本配对比例不合理等问题,有一定改进空间。本文选取我国2018年沪深A股上市公司中51家ST公司作为信用违约样本,并采用1:3的样本配对比例选取了153家非ST上市公司作为非违约样本进行实证研究。选取包括营运能力、偿债能力、盈利能力、成长能力等多方面财务指标,结合统计检验方法筛选出25个财务指标,结合稀疏主成分分析法提取主成分因子,并加入公司规模、股权结构和股权质押三个非财务指标。构建基于稀疏主成分的Logistic模型进行信用风险评价和预测,并验证Logistic回归模型在预测准确性、稳定性方面的表现。
关键词: 信用风险 稀疏主成分 Logistic回归模型
For information in English, please click here
Credit Risk Evaluation of Chinese Listed Companies Based on Sparse Principal Components
Abstract:Credit is the cornerstone of the normal operation of the economy, and it is also the prerequisite and foundation for the continuous development of the economy. Strengthening the evaluation and management of credit risks of listed companies in China is of great significance to listed companies, investors, and financial institutions. Existing studies have problems such as imperfect index systems, repeated redundant input variable information, and unreasonable sample pairing ratios. There is room for improvement. In this paper, 51 ST companies in the Shanghai and Shenzhen A-share listed companies in China in 2018 are selected as credit default samples, and 153 non-ST listed companies are selected as non-default samples for empirical research using a 1: 3 sample matching ratio. Select financial indicators including operating ability, debt paying ability, profitability, and growth ability, and select 25 financial indicators based on statistical inspection methods. Combine the sparse principal component analysis method to extract the principal component factors, and add the company size, equity structure, and Three non-financial indicators of equity pledge. A sparse principal component-based Logistic model was constructed for credit risk evaluation and prediction, and the performance of the Logistic regression model in prediction accuracy and stability was verified.
Keywords: credit risk sparse principal component logistic regression model
基金:
引用
No.****
同行评议
勘误表
基于稀疏主成分的我国上市公司信用风险评价
评论
全部评论0/1000