Center for Stochastic Dynamics Seminar with Zhihui Yang

Speaker:

Zhihui Yang, Controls and Optimization at GE Global Research

 

Title: 

A Credit Value at Risk (CVaR) Model

 

Abstract: 

Credit risk management is an important responsibility for financial institutions with a portfolio of loans and leases. Risk managers need access to an array of tools to actively manage and monitor credit risk, such as understand risk dynamics, monitor risk concentrations, compute Economic Capital (E-Cap), compute allocated E-Cap value for each transaction or sub-portfolio and perform stress testing.  A Credit Value at Risk (CVaR) software was developed by GE Capital and GE research team and is being used internally to estimate quarterly E-Cap and manage capital allocations. A CVaR model includes four major sub-components:  correlation model to capture the default correlation of the obligors within the portfolio,  valuation model to price each loan or derivative in the portfolio, aggregation model to sum up all the loan value and generate a portfolio value/loss distribution, allocation model to allocate the portfolio level VaR value to obligor level or sub portfolio level. In this talk, I will focus on construction of a constraint multi factor correlation model where model parameters are calibrated with default data using maximum likelihood method. I will also talk about an analytic approximation of the Ecap model using the central limit theorem. 

 

Zoom link