 |
 |
since August 31, 2007. One thing we will discuss further is that only some of the 2007 MBS cheapening is prepayment-related. As for the hybrid ARMs, even if we lowered the prepay scales further, we would not be able to explain the OAS widening – hybrids would become even wider.
The prOAS Method: Adjustments are Wanted
The prOAS method (aka OAS using the risk-neutral prepay model) has become a part of AD&Co’s analytical regimen since 2003. We attribute the existence of OAS, in part, to the price of prepayment model risk, i.e. market fear that prepayment models are biased or can become biased going forward. We postulated that agency TBAs should be valued flat to agency debentures, once we account for the price of this risk.
It is obvious that the prOAS basis has widened, especially for the hybrid ARMs. It is impossible and unreasonable to attribute the 50 – 70 bps of OAS for the agency hybrids to prepayment concerns. The demand-supply balance of the entire MBS market has been shaken thereby forcing us to add a spread to the agency curve. How large should this add-in be and how should it change from one market sector to another? One can’t pinpoint a unique number because we are solving one equation with two unknowns, prepay risk and liquidity risk.
Despite the seeming ambiguity, alternative methods exist to address the problem:
• One method often used by dealers is calibrating the prepay tunings to Trust IOs and POs by equating their OAS. This method is limited to the fixed-rate MBS and can’t be used for ARMs. In addition, dealers’ publications on Trust IOs employ various methods of “matrix pricing,” i.e., reconstructing and extrapolating quotes from a few daily trades.
• Since prepayment tunings strongly affect the interest rate sensitivity, one can argue that tuning the prOAS model to OAD, rather than to OAS, can help. This approach is not limited to any particular MBS type. AD&Co has tools for carrying out this task too, and I plan to discuss the implications in next-month’s article.
