以修正Merton模型評估之信用風險：以第一銀行為例

本國銀行為因應新巴塞爾資本協定（Basel II）之需要，已於2004年開始鼓勵銀行採取現代信用風險管理，為了精確衡量不同客戶其信用風險，因此必須先了解影響信用風險之三要素，其中以違約機率（probability of default, PD）最為重要，故本文以修正後的Merton模型為基礎，選取台灣地區的第一銀行為研究樣本，估算2004至2007年該樣本銀行之違約機率，本文進一步將總體經濟變數納入考量以評估違約機率，本文將未考量總體因素所估算之違約機率稱為無條件違約機率，相對地，受到總體變數限制下所估算之違約機率稱為有條件之違約機率，並比較與分析無條件違約機率與有條件違約機率。本研究實證結果顯示，無條件違約機率可能具有落後景氣之現象，而有條件違約機率與景氣循環較其相關性，即有條件違約機率較能即時反應信用風險，有鑑於此本文建議銀行在審核貸款案件時除了注意放款公司本身償債能力以外，還必須加以觀察影響違約機率之總體經濟變數其變動情形，以作為制定適切的授信條件之參考所用，進而降低違約事件的發生以維持穩定的風險程度。

關鍵字

信用風險；違約機率；新巴塞爾資本協定

並列摘要

According to new Basel capital accord (Basel II), bank regulators have to develop an effective credit review process to measure loans' credit risk such as probability of default (PD). In this paper, we estimate the unconditional and conditional default probabilities for bank loans of First Commercial Bank in Taiwan by the modified Merton model. We compare differences between unconditional and conditional default probabilities. The unconditional default probability is estimated whereas ignoring the influence of macroeconomic variables. The conditional default probability includes macroeconomic variables. Therefore, the conditional default probability is more flexible than unconditional default probability for measuring credit risk. The empirical results show that that conditional default probability is close to business cycle. However, unconditional default probabilities may lag the business cycle. In other words, the conditional default probability is more accurate in investigating credit risk. Consequently, macroeconomic variables play an important role in gauging the probability of default and financial institutions should be paid attention to these variables.