訊息不完備下擔保債權憑證之評價

Abstract

本文的主要目的是希望在台灣目前受到信用風險資料庫與債券次級市場過小的限制下，僅透過市場上公開的資訊，建立一套精確擔保債權憑證之評價模型。又因國內會計制度上並不健全，所以在模型建構上，我們是以資訊不完備的違約機率模型估算出各別債務人的違約機率，並利用動態 copula 函數來建構擔保債權群組間的違約相關。最後，再根據風險中立的假設，建構出一套完整的架構以對擔保債權憑證之批次證券進行公平溢酬之定價。研究結果顯示，利用訊息不完備之違約機率模型，能較利用 Merton 違約模型捕捉到更多違約情形，風險溢酬也較高。同時動態 copula 函數也較靜態 copula 函數更能細部描述出違約損失的架構。本文可提供實務界一個可行且精確的擔保債權憑證之評價方法。

The purpose of this study is to employ publicly available data to estimate the default probability and build credit spreads for various tranches of CDO considering default correlation. Because of limited data for credit ranking, small secondary market for bond transaction and insufficient accounting information, we use incomplete information default model to estimate default probability and use dynamic copula functions to construct default correlation. Under the risk neutral measure, we can obtain the fair premium for different tranches of collateralized debt obligations. The empirical results show that the spread of each tranche under incomplete information default model is larger than that under Merton’s default model. Moreover, dynamic copula functions can describe loss structure in more details and affect the spread for each tranche. We think this article can provide a more precise and workable CDO pricing process for an emerging market like in Taiwan.