It's been plain to see in recent months that MBS have had a very tough time keeping pace with the drop in Treasury yields.  That's been especially true of the past 2 months.  Why is it happening?  How bad is this from a historical standpoint?  And will it change?

The "whys" are fairly straightforward.  Without any concern for the changes in the Fed's bond buying strategies, MBS are always going to suffer when there's a big, unexpected move in the broader bond market.  The August rally definitely qualifies.  Prepayment speeds are at the heart of the matter.  If an investor can't be sure that a mortgage is going to stick around long enough for them to earn their interest, that mortgage becomes a less valuable investment relative to alternatives.  

Changes in the Fed's bond buying strategies complicate the matter.  The Fed had long maintained that it would eventually like to return to a Treasury-only portfolio.  As of October 2018, Fed reinvestments into MBS were effectively over, and it showed as far as spreads were concerned.  Notice the upward movement in MBS vs Treasury spreads gradually throughout 2018 (as the Fed reduced its MBS reinvestments) followed by the spike in October to long-term highs/wides.

Then, of course, spreads went on to correct in early 2019 only to blow out to even wider levels over the last 2 months.  Even without the benefit of being able to chart MBS spreads, the damage has been easy enough to see simply by looking at price movement compared to Treasury yield movement.

But how bad are things really?  I won't tell you that spreads aren't the widest they've been in a long time.  They definitely are, but we would also expect them to be given the volatility in the bond market, and the size of the average trend over the past 2 months.  The better question is "how bad is this in a historical context?"  The first chart above suggests it's not too bad yet.

We can take the caveats a step further by asking ourselves how long the average mortgage is lasting these days.  While it's most often the 10yr yield that we are comparing against, 5yr Treasuries are arguably just as relevant a benchmark.  In fact, MBS traders have been using a 5s/10s blended rate for years due to the average mortgage loan lifespan tending to fall around the 7 year mark  on average over time.  

A 5yr benchmark is much kinder to MBS in the bigger picture.  Spreads versus 5s are still wider than they have been for a long time, but not anywhere close to returning to 2012 levels (like spreads vs 10yr yields).

And now altogether on the same chart:

What's the takeaway here?

Yes, MBS have been underperforming, but logically so, and not any more than we should expect, given the changes in Fed's strategy, the volatility in the market, or the shifts in the Treasury yield curve.  It continues to be the case that spreads can move back down/narrower if the bond market manages to stabilize in a narrower trading range.