Agency MBS Spreads Collapse, Durations Shorten

Fannie Mae and other agency MBS spread to treasuries declined sharply in the last couple of days - to the lowest levels in years in fact - all in reaction to the news from the Fed (see post). With spreads this low, the Fed will be buying paper at some of the richest valuations in recent history. The chart below shows the spread between current coupon FNMA 30y yield and the 10y treasury.

Markets are still digesting this massive move.

Bloomberg: A measure of relative yields on mortgage securities dropped to the lowest on record after the Federal Reserve said it will expand its purchases...

“A typical fundamental-value framework really isn’t applicable here” because the Fed’s goals differ from those of normal investors, said Todd Abraham, co-head of the government and mortgage-backed fixed-income group at Federated Investors Inc. “It makes it pretty challenging to determine at what point you need to change your allocations. It’s almost more of a case of needing to try to anticipate what others are going to do.” Adjusted for the prepayment option, the spread on these bonds (OAS) is close to zero - basically pricing no credit risk associated with agency debt (see discussion).

With mortgage rates expected to decline further, the market is anticipating accelerating prepayments, shortening the effective duration of many MBS bonds. Part of the unprecedented decline in spread against the 10y treasuries (above) was driven by these falling durations, as the "on-the-run" agency paper started trading to shorter dated treasuries, which have lower yield.

Those who hedged MBS with 10-year rate swaps (a traditional instrument to hedge fixed rate mortgage paper) were caught off guard. Many were forced to unwind these hedges because the durations of the MBS and the hedge became mismatched. That unwind sent the 10-year swap spread crashing to recent lows.

It may take some time for fixed income markets to adjust to this new regime. And as Todd Abraham correctly pointed out (above), it's no longer about "fundamental-value framework."