Financial Engineering Seminar
October, 10, 2005(Mon)
Time: 11:45 am,
Place: Weil 279
We will be discussing some practical issues and methodologies relating to optimization of a large bank credit portfolio, including about $200 billion in corporate and retail loans, many thousands of corporate obligors and many more thousands of retail obligors, and still larger numbers of individual credit facilities and accounts. This is a live project, very much a work in progress, at a major Asian bank. Accordingly, critiques and suggestions from seminar participants will be especially welcome
The first set of topics to be discussed is the approach for measuring RAROC performance of the current bank portfolio. We will describe the use of Credit Manager (Credit Metrics) to simulate portfolio behavior, in order to estimate risk and economic capital for the overall portfolio, and then to attribute these to the constituent credit exposures. A key decision is the level of aggregation with which to input credit exposure data, e.g. individual accounts at one extreme vs. loan pools at the other extreme. We also discuss other data-related issues, such as development of ratings, transition probabilities, transition correlations, and instrument valuations.
The second set of topics to be discussed is the approach for using current RAROC performance measures and other simulation results in order to optimize future portfolio performance. A key decision here is the definition of “Asset Clusters”, to map the very large number of portfolio credit exposures used for input to Credit Manager, onto a relatively smaller number of building blocks that will be practical to use for portfolio planning and optimization. We also discuss formulations of the optimization problems, potential optimization methodologies, and practical implementation of the results.
Presentation (PPT)
Credit Risk (PPT)