Alexander Levin

Prepayment Risk- and Option-Adjusted Valuation of MBS

This exploration of risk-neutrality in MBS prepayments introduces option-adjusted valuation of MBS with consideration of prepayment risk, i.e., the risk that future prepay speeds systematically drift away from a model. A new measure, prepayment risk- and option-adjusted spread (prOAS), is suggested to replace the traditional option-adjusted spread, which varies widely across instruments. PrOAS levels the playing field by assuming that all liquid MBS and their derivatives are traded flat to the same debenture curve, on a risk- and option-adjusted basis. There are two alternative and theoretically equivalent methods of pricing with prOAS: 1) solving pricing PDEs with risk terms explicitly for pass-throughs using the active-passive decomposed prepay model, and 2) using a risk-neutral prepayment model, which is more universal. Although the prOAS method generally works well in case studies, there are some exaggerated dynamics of prices of risk and market dislocation in times of panic that offer arbitrage opportunities.

Interest Rate Model Selection

Analysis of the most popular single-factor mortgage valuation models unambiguously rejects the lognormality of interest rates. Recent trends in the swaption market indicate normalization, supporting use of the Hull-White model, which can be quickly and accurately calibrated to at-the-money swaptions. Durations for TBA instruments come up shorter by the Hull-White model and are generally more in line with the empirical measures. The issue of negative rates is not found to be detrimental to the standard OAS pricing practice.

The Concept of Credit OAS in Valuation of MBS

Neither the traditional option-adjusted spread (OAS) method nor a static loss forecasting is capable of providing a rational explanation of the pricing of traded asset-backed securities (ABS), credit default swaps, and loan protections. In this article, the authors introduce the concept of credit OAS, which revolves around the stochasticity of both interest rates and home prices. The credit OAS approach requires a risk-neutral stochastic model of home prices, a model of defaults and losses (theoretical or empirical), and a rigorous and efficient valuation scheme. The authors discuss how the risk-neutral conditions can be formed from concurrently observed prices of ABS tranches up and down the credit structure. An alternative is to analyze the cost of a GSE guarantee inclusive of capital requirements to support the guarantee. The authors illustrate their method with case studies in both liquid and illiquid markets. Even distressed market prices can be explained by a combination of modeled losses and a properly selected credit OAS level (i.e., liquidity spread). The greeks (risk measures), however, depend strongly on the modeling details (such as the link between interest rates and home prices), vary widely from model to model, and tend to be far away from traditional measures.