Event: Recent Advances in Credit Risk Modeling
Date: Monday, May 20, 2013
Time: 5:45pm Registration, 6:00pm Seminar, 7:15pm Reception
300 Madison Aveue
(at 42nd Street)
New York, NY
The International Association of Financial Engineers is pleased to invite you to
Linda Kreitzman, Master of Financial Engineering,
Haas School of Business, UC Berkeley
Terry Benzschawel, Managing Director
Bond Portfolio Analysis -- Quantitative Strategy
Citi Fixed Income Currency & Commodities
This presentation will consider two topics. These are (1) inferring physical default probabilities from credit spreads; and (2) generating credit-cycle dependent ratings transitions matrices for estimating expected changes in credit spreads.
I first describe how to estimate physical probabilities of default (PDs) for risky obligors from credit market spreads. The motivation for this work is that structural (i.e., Merton) models can not generate PDs for obligors without tradable equity (e.g. sovereigns, municipalities, and private firms) and is of questionable use for financial firms. The method is to first estimate the daily risk premium (i.e. amount of credit spread compensation per unit of spread volatility) using risk-neutral valuation of PDs from a structural model. The estimate of the risk premium is then multiplied by obligors' spread volatilities and, along with an assumed recovery value and its current market spread, used to calculate firms' market-implied PDs. I will present performance of market-implied PDs at predicting defaults relative to agency ratings and structural models of default.
I also describe how to generate credit state transition matrices that are credit-cycle dependent. That is, average credit-state transition matrices published by the agencies are rarely realized amid the changing macroeconomic environment. The proposed method incorporates current default rates and credit upgrade-to-downgrade ratios to project probabilities of migrating among credit states over horizons from one-month to ten years. In addition, by superimposing the term structure of credit spreads on these cycle-adjusted transition matrices, one can infer probabilities of expected losses on risky bonds, both from default and spread moves, over arbitrary horizons.