- 學位論文
組合資產之信用風險管理：理論與應用
Portfolio Credit Risk Management: Theory and Application
摘要
「新巴賽爾資本協定」於2007年開始實施之後，已促使各銀行致力於提升風險管理能力，其中新巴賽爾資本協定第一支柱所規範的內部評等法，更成為眾多銀行努力追求的目標。然內部評等法用來計提資本的公式，卻是建立在許多簡化假設上，這些假設包括忽略違約機率及違約損失率之相關性、沒有集中度風險等等，且公式中所使用之參數亦直接引用國外資料。除了內部評等法，常見的商業化信用風險管理模型（如CreditMetrics、CreditRisk+等）亦均假設違約損失率為固定值，或假設為與違約機率獨立的隨機變數，無法處理違約機率及違約損失率相關議題。 有鑑於此，本文嘗試建立一個違約機率及違約損失率聯合模型，其中違約機率是由受總體因子影響的股票報酬率與受信用評等影響的報酬門檻所共同決定，而違約損失率則是由公司資產價值及預期負債所決定，其中資產價值由三種不同擔保力、折扣率資產所共同決定，預期負債比率則為總體因子決定。因此這個聯合模型的基本特色是違約機率及違約損失率將因同時受到總體因子影響而彼此相關。 將由台灣上市櫃公司實際資料所估計而得的參數值納入前述違約機率及違約損失率模型後，我們便可得到一個能夠進行蒙地卡羅模擬的完整架構，並得以此估算對應的經濟資本。模擬結果顯示，我們所估得的經濟資本要遠大於未考慮違約機率及損失率間相關性的經濟資本，也要比內部評等法所需計提之資本為大。此外為充分利用經濟資本作為風險定價的基礎乃至於考核風險調整後的績效，我們更進一步採用蒙地卡羅模擬方式將經濟資本配置至個別交易對手，這個模擬結果顯示個別交易對手相對風險的高低排序，與由內部評等法所得到的排序有相當大差異。 本文的主要結論是，銀行如欲精確衡量及妥善管理其信用風險，必須探究新巴賽爾資本協定所提供內部評等法資本計提公式的合理性，並依據其內部資料建構符合銀行特性的經濟資本模型，如此方有助於銀行有效的配置資本，也才更能符合新巴賽爾資本協定第二支柱的要求。若銀行能達此境界，則不僅銀行可賺得較大的風險調整後利益，其交易對手也可降低借貸成本，整體社會福利將因此獲得提升。
並列摘要
The implementation of Basel II in 2007 has driven banks to enhance their risk management capability, and the IRB approach has become the goal that many banks are aiming for. The IRB capital calculation formula is based on many assumptions including independence between PD and LGD, no concentration risk and identical parameters applying to all countries. Popular credit risk models like CreditMetrics or CreditRisk+ also make assumptions that LGDs are constant or LGDs are independent of PDs and do not explicitly deal with the dependence issue between PD and LGD. For this reason, we try to build a joint model for PD and LGD. The PDs depend on expected stock returns, which are in turn affected by a common factor, as well as return thresholds that are determined by risk ratings. In contrast, LGDs depend on asset values and expected equity/debt ratios, in which asset values are decomposed to three categories with different guarantee powers, while expected equity/debt ratios are influenced by the common factor. The central idea of this joint model is that PD and LGD are both affected by the common factor and hence are correlated. Viewing listed companies in Taiwan as a portfolio, we could estimate all parameters of our joint model and run Monte Carlo simulation to generate portfolio’s credit loss distribution and calculate the corresponding economic capital. The simulated economic capital is larger than the economic capital under the independence assumption and is also larger than the IRB capital. In addition, to employ economic capital as the basis for risk pricing and risk adjusted performance measurement, we must allocate portfolio economic capital to individual counterparty, which requires further Monte Carlo simulations. From the simulations, we find the ranking of allocated economic capital substantially differs from the ranking of IRB capital. Our main conclusion is that banks need to explore the basic idea of IRB capital formula and use their internal data to construct economic capital models if they want to correctly measure and manage credit risks. Banks could then meet the requirement of the Pillar II in Basel II and allocate economic capital much more efficiently. Banks need to access loans according to comparative risk. Only when this is accomplished, will banks earn more risk adjusted profit. Their counterparty would also bear less financing cost. Social welfare could therefore improve.