开发者:上海品职教育科技有限公司 隐私政策详情

应用版本:4.2.11(IOS)|3.2.5(安卓)APP下载

eee · 2018年06月10日

问一下reduce form 的缺点



Marlin asks Betta if reduced form models of corporate credit risk provide advantages over structural models. Betta responds that reduced form models have less restrictive assumptions than structural models because of the one critical assumption that cannot be relaxed. He notes two weaknesses with regard to reduced form models: (1) The use of past observations to predict the future requires that it be formulated and back tested properly, and (2) credit risk measures are biased by implicit estimation procedures.

Finally, Betta would like to examine the term structure of credit spreads for Bay Corp. The company has a number of coupon bonds outstanding across the maturity spectrum that should facilitate his analysis. He notices the market prices of the bonds are lower than would be implied by results of either a structural or reduced form model. Betta is confident he has correctly estimated the price of the zero-coupon bonds implied by the coupon bond prices, which in this case rank equally in the capital structure.


Q. Which comment made by Betta is least likely correct with regard to reduced form models relative to structural models?

  1. His comment regarding assumptions
  2. His comment regarding the first weakness
  3. His comment regarding the second weakness


这里答案和课程讲解都说到第二个weakness是

structural models的弱点,可是我记得reduce model 参数好像有两个估计方式,一个是历史数据,另一个好像也是implicit estimation,不知我记得对不对,如果是这样,那这个模型不是也有这个缺点吗



1 个答案
已采纳答案

发亮_品职助教 · 2018年06月12日

对于参数:

使用Structural model,只能用implicit estimation

而对于reduced model,可以用两种方法;implicit estimation;Historical Estimation


对于Structural model,他的弱点就是只能用implicit estimation,这是structural model估计参数的唯一方法

再用structural model时,本身model的假设就很严格不靠谱,所以用implicit estimation估算出来的参数肯定是Biased。由于structural model只有这一种方法估算参数,所以用这个model算出来的违约概率、expected loss肯定是biased。


而对于Reduced model,参数估计有两个方法可选,即便用implicit算出来的数据有biased,但是用reduced model算出来的credit risk数据不一定会受到Implicit的biased影响,因为还可以参考historical方法。

  • 1

    回答
  • 2

    关注
  • 504

    浏览
相关问题