In a report released yesterday, FTN Financial, analysts discussed their expectation regarding the GNMA market and GNMA/FNMA relationship in the 1H10.
Analysts discussed the reasons why GNMA/FNMA swaps have been cheap on a historical basis in 2009. These include the increase in GNMA supply that resulted from the collapse of the subprime and Alt-A sectors; the uncertainty regarding prepayments as a result of buyout risk, especially in the 2006-2008 vintages; and the lack of Federal Reserve support in the GNMA market compared to the conventional side.
In the report, analysts cited three reasons for liking GNM/FNMA and GNMA/FHLMC Gold swaps going into the first half of the new year. One is the removal of the Fed after the first quarter, which will have an adverse effect on conventionals, analysts said.
Meanwhile, analysts believe that domestic bank and overseas support of GNMA paper will remain strong. The formeris because of cheap funding costs, weak loan demand and a 0% RBC weighting, while the latter could even be stronger as a result of GNMA's full faith and credit guarantee, as well as very attractive yields relative to Treasurys.
FTN analysts provided two examples to get a sense of what value is currently there with the prospect of a "vastly different environment" that lies ahead.
The first example compared buying G2 4559, a new G2 30yr 5, versus FNCL 5 N, a new TBA FN 5. Based on Yieldbook calculations, the trade picks yield (nine basis points) and LOAS (10.9 basis points). Convexity is also higher (0.49), while duration extends 0.69.
While buyout risk is uncertain, analysts think it is less likely on lower-coupon, new-vintage paper than it has been.
They added that, "the main point of the trade is to insulate the portfolio from the relative spread widening in conventionals that will likely occur at some point in 1H10."
Another trade analysts highlighted was buying GNMAs versus Treasurys. Their example considered the same GNMA as above versus a 3.5-year Treasury (3.625% due 5/15/13). In this case, the trade picks over 200bps in yield.
In addition, they pointed out FNMA LOAS levels can widen out by 14 basis points or over 1/2 point over a six-month horizon in the base case before the GNMA would underperform the Treasury. They pointed out as well that based on the historical average spread between CC GNMA and a 5/10 UST blend, GNMAs are fair to Treasurys currently.
FTN analysts concluded that the GNMA market is set "for strong performance despite supply and prepayment uncertainty."