交易對手風險,信用傳染與財務危機預測

Abstract

由次貸危機所引起的全球金融風暴，已對全球金融系統產生顯著的衝擊，亦使投資組 合信用風險模型的建模更加受重視。違約叢集(Default clustering，也稱為違約感染 credit contagion)的現象，可能導致更多銀行在發生財務危機時破產，以及擔保債權 憑證的損失超過其原始最保守估計。因此，學者嘗試建立更符合實際狀況的違約相關 模型，以冀望改善投資組合信用風險模型。先前的學者已經檢視許多可能的違約叢集 解釋原因，然而，現有信用風險模型中的因子或產業效用似乎無法產生符合實際的違 約叢集模式。本研究係針對其中一特別不同於其他違約感染的管道，亦即交易對手風 險進行實證分析，由於資料難以取得，以及確認直接公司交易對手的困難性高，目前 在交易對手風險的實證分析上研究非常稀少。 本研究將首先針對違約相關、信用感染以及交易對手風險的理論與實證研究進行文獻 探討，繼而採取Jorion 與Zhang (2008)的研究方式，針對市場對於破產宣告中的最大 的二十家無擔保債權公司進行事件分析，檢視市場反應。接下來，我們將針對事件發 生後債權公司的財務危機進行分析，檢視其與結構信用風險模型隱含的違約機率之關 連，嘗試建立交易對手風險與財務危機的直接關聯。此外，各個不同結構信用風險模 型的違約預測能力亦可再做進一步的探討，我們主要將Merton(1974)、Black and Cox (1976)與Leland (1994)模型進行實證分析。

The global financial turmoil resulted from the subprime mortgage crisis of 2007 has significantly impacted the financial systems around the world, and raises the importance of portfolio credit risk modeling. Unexplained default clustering is a major issue for traditional credit risk models and could lead to more bank failures in periods of stress, or losses on CDOs that exceed the worst estimates. Therefore, researchers are intended to model the default correlation more realistically in order to improve the portfolio credit risk modeling. Prior researches have been examining several possible structural explanations for default clustering, also called “credit contagion.” Nonetheless, current factor or industry effects in credit risk modeling seem to be unable to reproduce the actual pattern of default clustering. Therefore, our study aims at one particular different channel of credit contagion, which is counterparty credit risk. The empirical evidences on counterparty risk are rare due to the availability of data and the difficulties in identifying direct business links between companies. This paper will first review theoretical and empirical studies of default correlation, credit contagion, and counterparty risk. Next, we will follow the pioneered work by Jorion and Zhang (2008) and examine the market reaction of the top unsecured creditors in bankruptcy filings. Thirdly, the financial distress of creditor firms subsequent to the loss of the filing of bankruptcy is only preliminarily analyzed by Jorion and Zhang (2008) in terms of the number and fraction of firms that were delisted or downgraded within one and two years. In our empirical study, we will investigate the relationship between the changes in default probabilities implied by the structural models and the subsequent events of financial distress of creditors. This is the attempt to construct a direct linkage between counterparty risk and financial distress. Furthermore, the default prediction capabilities of various structural models can also be explored further. In addition to the commonly used KMV-Merton default probability (which is commonly regarded as the volatility-adjusted leverage), we will also incorporate additional features of the well-developed structural model literature, including the exogenous default boundary (Black and Cox (1976)) as well as the endogenous default boundary (Leland (1994)) models. These structural models have been shown to be able to improve default prediction capabilities (Chen, Hu, and Pan (2006) and Chen, Lee, and Lee (2008)).