预期效用下非极端违约回收率的贝塔分布修正模型  

Beta Distribution Model Based on Expected Utility for Non-Extreme Recovery Rate

作　　者：陈暮紫[1] 陈浩[2] 陈敏[2] 杨晓光[2] 

机构地区：[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

出　　处：《数理统计与管理》2015年第6期1016-1028,共13页Journal of Applied Statistics and Management

基　　金：国家自然科学基金(No.71203247;No.71271223);教育部人文社科基金(11YJC790015;14YJA790048)资助

摘　　要：在预期效用分布理论的框架下,对非极端违约回收率构造了贝塔分布修正模型。该模型与传统分布模型相比,保持了贝塔分布刻画违约回收率的适用性,同时具有一定的经济理论基础,还避免了过拟合问题,进一步采用提出的模型对我国违约贷款数据作了实证分析。结果显示,采用上述模型得到的结果能够直观有效地解释各因素对违约回收率分布的影响,有助于违约回收率分布的影响因素和原理研究;在模型效果上修正的贝塔模型进行违约回收率分布的样本外预测效果也优于传统贝塔拟合和广义贝塔回归模型。Under the framework of expected utility distribution, this paper constructs a modified beta distribution model to study the non-extreme recovery rate. Compared with traditional distribution methods, this model is not only with the applicability of fitting, but also based on economics theory. Meanwhile, this model avoids the over-fitting problem. Further, the empirical analysis is made based on this model. The results show, this model could effectively explain the impact of each economic factor on recovery rate, and contribute to the recovery rate research. In addition, the performance of modified beta distribution model is better than traditional beta distribution and generalized beta regression model.

关 键 词：预期效用 违约回收率 贝塔分布 

分 类 号：F830.33[经济管理—金融学] O212[理学—概率论与数理统计]