By Arthur Frank, director of mortgage research at Nomura Securities
In April 1999, both Fannie Mae and Freddie Mac launched their callable benchmark note programs, motivated by the success of their bullet benchmark note programs in 1998. Prior to this program, the callable agency debenture market consisted of a large number of relatively small issues, many of them under $100 million and therefore not included in the major bond indices. These small issues had relatively poor liquidity compared to mortgage passthroughs. There was not an active two-sided market in inter-dealer broker screens, so that dealers could not readily short these issues to facilitate investor trades. As a consequence of this illiquidity, many large money managers who actively traded their portfolios avoided the callable debenture market.
Now that the agencies have issued large 10NC3 and 5NC2 benchmark notes, and ended their prior practice of issuing such securities as Medium-Term Notes, the callable agency debenture market has received greater attention from money managers. All of these securities are in the major bond indices, there is an active two-sided inter-dealer market in them, major dealers are comfortable shorting them in reasonable size to facilitate investor trades, and they can be financed on slightly more favorable terms than mortgage passthroughs (currently 7 basis points behind general Treasury collateral, compared to 10 basis points for fixed rate passthroughs). Some mortgage investors who previously shunned callable agency debentures now consider them viable alternatives to passthrough securities.
Passthroughs Yield More Than Callable Agencies
For the sake of comparison, we examine the yields and option-adjusted spreads on Fannie Mae's 10NC3 benchmark, the 6.5 of 4/09, compared to 30-year FNMA 6.5s, and Fannie's 5NC2 benchmark, the 5 7/8 of 4/04, compared to Dwarf 6s. The comparison could have been done using Freddie Mac callable benchmarks, which also are large, liquid issues, with virtually identical results. The dollar prices and effective durations for the securities being compared are fairly close, while the effective convexities are slightly more negative for the callable benchmarks, since the agencies exercise the embedded call option more efficiently than homeowners exercise their refinancing options. The last two months comparative yield history is given below:
Despite their slightly higher option costs, callable agencies consistently yield less than comparable mortgage passthroughs. This varying difference can be considered compensation for the MBS investor's assumption of idiosyncratic prepayment risk (i.e., the risk of prepayments coming in different from investor expectations given the path of interest rates), as well as compensation for the administrative inconvenience of holding large numbers of mortgage pools with uncertain monthly cash flows. The systematic prepayment risk (the dependence of prepayment speeds on interest rates, causing negative convexity) is already accounted for in the option-adjusted spread calculation. The mortgage agencies can profitably issue callable debentures and buy passthroughs with the proceeds because of this difference, but that does not mean that the bond portfolio manager is always better off purchasing passthroughs instead of callable agencies. When the compensation for idiosyncratic prepayment risk is relatively meager, investors may reasonably prefer callable debentures over comparable passthroughs.
Option-Adjusted Spread (OAS) Comparisons
The callable agency debenture market has established a 14% constant volatility standard for measuring the option cost and option-adjusted spread of securities, so given agreement on a price at a given point in time, all market participants use the same option-adjusted spread. This is convenient for trading purposes, but the resulting OAS is not very useful for comparing callable debenture OASs to those of comparable MBS. This OAS calculation ignores the impact of changes in implied volatilities in interest rate derivative markets, which has a major impact on the option cost of callable securities (e.g., mid-market implied volatilities on 5x10 swaptions have varied between 11.5% and 16% over the past year).
In contrast, in the MBS market, agreement on price is compatible with sharp disagreement on OAS, since the option cost is dependent upon the particulars of a prepayment model. Nomura's MBS option valuation model, like many competing models, uses market-based implied volatilities as inputs to the OAS calculations. On a daily basis, Nomura's derivatives trading desk provides mid-market volatility levels for caps, floors, and swaptions, and the resulting data is used to calibrate a term structure of implied volatility that drives the interest rate process used for MBS valuation. We use the identical term structure of volatility and interest rate process to calculate OASs for benchmark callable agencies, providing a consistent framework for OAS comparisons. Of course, different prepayment models will yield significantly different OASs for passthroughs, but the comparison is worthwhile as a supplement to yield comparisons. We provide the comparative OAS history below:
With the introduction of callable agency benchmark notes by Fannie Mae and Freddie Mac, the callable agency market became a viable, liquid alternative to the passthrough market for large money managers. To date, spread relationships (both yield and OAS) have been in a tight range for the two products, but with enough variability to create attractive swap opportunities for total return investors. As of this writing, the 10NC3 benchmark is at fair value compared to 30-year passthroughs, while the 5NC2 benchmark currently appears inexpensive compared to discount 15-year passthroughs. The comparison requires that quoted callable benchmark OASs be recalculated on a basis consistent with mortgage option valuation methods.