Mike Landrigan (Rimrock Capital, San Juan Capistrano, CA).

Portfolio Optimization for Valuing Collateralized Mortgage Obligations

Collateralized Mortgage Obligations (CMO) can have a high degree of variability in cash flows and a complex embedded optionality structure. Because of this, it is generally recognized that a yield to maturity or static spread calculation is not a suitable valuation methodology. To assess the value of a CMO it is common to rely on benchmark-calibrated Monte Carlo simulation of interest rates and projection of cash flows along the various interest rate paths. The often quoted measure of value derived from such simulation is the Option-Adjusted Spread (OAS). We find that OAS has serious flaws and introduce a variance reducing portfolio optimization technique to value CMO. The optimization is applied to liquid, benchmark interest-rate derivatives. In our talk we will introduce CMOs and OAS, present the general framework of our analysis, and discuss results from numerical experiments.