管理评论

›› 2015, Vol. 27 ›› Issue (12): 27-38.

• 经济与金融管理 • 上一篇    下一篇

不良资产零回收和非零回收影响因素研究

陈暮紫1,3, 刘小芳3,4, 杨晓光2   

  1. 1. 中央财经大学管理科学与工程学院, 北京 100081;
    2. 中国科学院数学与系统科学研究院, 北京 100190;
    3. 金诚中科信用风险与数据分析实验室, 北京 100190;
    4. 东方金诚国际信用评估有限公司, 北京 100062
  • 收稿日期:2014-03-07 出版日期:2015-12-30 发布日期:2015-12-25
  • 作者简介:陈暮紫,中央财经大学管理科学与工程学院副教授,博士;刘小芳,东方金诚国际信用评估有限公司分析师;杨晓光,中国科学院数学与系统科学研究院研究员,博士生导师,博士。
  • 基金资助:

    国家自然科学基金项目(71203247;71271223);北京市自然科学基金项目(9152017);教育部人文社科青年基金项目(11YJC790015)。

The Influencing Factors behind Zero Recovery and Non-zero Recovery of Non-performing Loans

Chen Muzi1,3, Liu Xiaofang3,4, Yang Xiaoguang2   

  1. 1. School of Management Engineering, Central University of Finance and Economics, Beijing 100081;
    2. Academy of Mathematics and Systems Science, CAS, Beijing 100190;
    3. Doho-CAS Credit Risk and Data Mining Lab, Beijing 100190;
    4. GoldenCredit Rating International Co.， Ltd.， Beijing 100062
  • Received:2014-03-07 Online:2015-12-30 Published:2015-12-25

PDF (PC)

933

摘要：

不良资产处置不仅是金融机构一项重要的日常工作,而且是政府和金融机构走出金融困境,走向正常的必由之路。本文基于目前国内最大的违约损失率数据库——LossMetrics数据库,针对我国不良资产违约回收率非对称U型Beta分布的特征,研究不良资产的零回收及非零回收的回收率高低的影响因素。本文构建了违约回收率的零回收判别模型和非零回收计量模型,辨识影响因素和估算其贡献度大小。实证结果表明,零回收和非零回收的影响因素和贡献度方面有诸多不同,判别零回收的重要因素是债务五级分类、债务人是否陷入经营停顿、以及债务人所在地区是否存在区域性困难;而非零的回收率大小的影响因素主要是所在区域、经营状况、债务规模、债权转让和担保方式。研究结果不仅揭示了不良资产回收背后的规律,而且有助于高效率地处置不良资产。

关键词: 不良资产, 零回收, 非零回收, 影响因素, 快速处置

Abstract:

Dealing with non-performing assets is not only an important daily work for financial institutions, but also the only way for the government and financial institutions to survive from financial distresses. Based on Loss Metrics database with respect to the non-performing loans which have U shaped asymmetric Beta distribution for loss given default, we study the influencing factors of zero recovery rate and non-zero recovery rate. We construct a discrimination model for zero-recovery rate and the regression model for non-zero recovery rate, in order to identify the influencing factors and estimate their contributions to models. The results indicate that there are a lot of differences in factors and contributions between the two models. The important factors of discrimination model include five-level classification of loans, whether the debtors are in a business standstill or not and whether the region of debtors is economically backward or not. The important factors of regression model include region, operating conditions, the size of loans and methods for loan transferring. The results not only reveal the laws behind the recovery of non-performing loans, but also improve the efficiency of dealing with non-performing loans.

Key words: non-performing loans, influencing factors, zero-recovery rate, non-zero recovery rate, rapid disposal