考虑错判风险的上市公司违约预测临界点 |
The critical point of default prediction for listed companies considering the misjudgment risk |
摘要点击 16 全文点击 0 投稿时间:2023-06-01 修订日期:2024-02-26 |
查看/发表评论 下载PDF阅读器 |
中文关键词 公司信用; 违约预测; 最优临界点; 逻辑回归; 敏感度 |
英文关键词 corporate credit; default prediction; optimal critical point; logistic regression; sensitivity |
基金项目 国家自然科学基金项目(面上项目,重点项目,重大项目) |
作者 | 单位 | 邮编 | 迟国泰 | 大连理工大学经济管理学院 | 116024 | 李存 | 大连理工大学经济管理学院 | | 张舒茗 | 大连理工大学经济管理学院 | |
|
中文摘要 |
针对中国上市公司违约预测临界点的确定问题, 提出了一种考虑错判风险的临界点确定方法. 以违约概率与临界点的距离定义了样本敏感度, 距离越小, 敏感度越高, 错判风险越大, 根据敏感度最小为目标函数确定了逻辑回归模型的最优违约预测临界点. 研究表明, 中国上市公司不同时间窗口的最优违约预测临界点不同, 均远远偏离现有研究的0.5, 最大偏离程度为53.64%, 最小偏离程度为-21.34%; 未来2年的短期预测的最优临界点均小于0.5, 未来3~5年的中期预测的最优临界点均大于0.5. 本文的临界点确定方法能够有效提高上市公司违约预测的精度. 此外, 还揭示了影响上市企业违约的关键因素, 为信贷决策、股票投资等提供有益参考. |
英文摘要 |
This paper proposes a method for determining the default prediction critical point of Chinese listed companies considering the misjudgment risk. The sample sensitivity is defined by the distance between default probability and critical point, the smaller the distance, the higher the sensitivity and the greater the misjudgment risk. The optimal critical point is determined by minimum sensitivity. The results show that the optimal critical points are different in different time windows, which are far from 0.5, with the maximum deviation degree of 53.64% and the minimum deviation degree of -21.34%. The optimal critical point of short-term forecast in the next 2 years is less than 0.5, and for medium-term forecast in the next 3~5 years is greater than 0.5. The proposed method can effectively improve the prediction performance. It also reveals the key factors influencing the default of listed companies, which provides useful reference for credit decision-making and stock investment. |
关闭 |
|
|
|
|
|