演講公告
新聞標題: ( 2009-04-27 )
演講主題:Efficient Importance Sampling for Credit Risk
主講人:韓傳祥教授(清華大學計量財務金融學系)
演講日期:98年5月19日(星期二)<br>下午2:30-3:30
演講地點:(光復校區)科學一館223室
茶會時間:當天下午2:00科學一館205室
摘要內容:
I will first introduce the CDS (Credit Default Swap) as a mean to reduce the default risk of asset, for example a corporate bond. Its generalizations such as BDS (Basket Default Swap) and CDO (Collateralized Debt Obligation) are considered to be very complex due to the embedded characteristics of high dimensionality and nonlinearity. These financial products cause a huge difficulty to manage the risk of credit portfolios and so on.
Second, I address a fundamental problem of estimating joint default probabilities (JDP) among many “names,” including firms, bonds, loans, etc. Structural-form and reduced-form models are used to trigger the time to default. Under these models, a stochastic method called Monte Carlo simulation is used to estimate JDP. An importance sampling technique is developed to improve the accuracy, and shown to be “efficient” by means of large deviation theory. When random environment such as stochastic correlation/volatility happens, asymptotics of JDP under suitable scaling in time and space appears to be crucial for the design of importance sampling.
