隱含界限做為破產邊界 --- 演算法與其實證

Abstract

信用風險模型可以分為縮減式模型（reduced-form）與結構式模型（structural form）兩類。本篇論文主要使用結構式模型並加入移動邊界選擇權的概念，從而提出了一種新的預測公司破產的方法。我們採取了非常彈性且有效率的演算法去找到這些隱含邊界並把這些邊界視為公司的破產邊界。在一些美國知名的公司的個案分析上也顯示出我們的隱含邊界對於即將到來的破產事件有非常高的資訊含量。更重要的是，本論文提出的利用隱含邊界預測破產的方法有非常高的精確率跟準確率，這個新方法不只讓破產機率變得可以操作也有非常扎實的財務意涵。

Credit risk models can be classified into reduced-form or structural form. This thesis focuses on the structural model with the idea of moving-barrier options to develop a new default prediction scheme. We adopt a flexible and efficient tree algorithm to find the implied barriers to serve as a company's default boundary. Case studies of some well-known companies in the U.S. suggest the implied barrier is very informative about coming defaults. For listed U.S. companies between 1991 and 2018, the proposed implied barrier-based default prediction performs very well in terms of precision and accuracy. This new prediction methodology not only makes the default probabilities operative but is also founded upon good financial rationale.