In the bond market, as a whole, we've never seen yields any lower for any longer than they've been in 2020.  The runners up aren't even close.  Even though bonds have become increasingly commoditzed (think "buy to sell" as opposed to "buy and hold"), the buy and hold crowd still exists, and it still has an impact on trading levels.  The relationship between Treasuries and MBS makes that clear.

Simply put, MBS offer higher yields compared to Treasuries, but with effectively zero default risk (federally back-stopped Fannie/Freddie protect the investor from borrower default).  Investors only need to worry about risks relating to how long any given MBS will last.  If rates fall too quickly, MBS coupons can pay off too quickly as the underlying mortgages are refinanced. If rates rise too quickly, MBS coupons can last too long, thus locking up investor cash that could now be put to work earning higher returns.

Right now, the dominant coupon is the UMBS 2.0.  It contains note rates of 2.25-3.125%.  With rates rising in the broader bond market, there's suddenly much less concern about 2.0 coupons being paid off an an ever-accelerating pace.  The new adverse market fee adds even more insulation by artificially bumping all refi rates 0.125-.25 higher than they otherwise would be.  At the same time, investors don't see rates rising so much or so quickly that 2.0 coupons will go out of fashion any time soon.  In short, it's a no-brainer for the buy and hold crowd... or at least it has been.

Why "has been?"

Well, we don't know if the frenzy is over yet, but we have to imagine we're getting closer to the limit all the time.  Here's what 10yr yields have been doing in October relative to the past 6 months:


Now let's look at Treasuries and MBS together.  To make the comparison easier, the 10yr yield chart is now just a yellow line.  MBS are in blue, and prices are inverted such that stronger levels are lower (thus allowing us to see correlation since MBS trade in price and 10yr Treasuries trade in yield).

20201023 open2.png

In other words, while 10yr yields are quickly moving toward the weaker levels seen in June, MBS aren't even halfway back yet.  They haven't even broken the weakest levels from Aug/Sept--something Treasuries did 3 weeks ago. 

The net effect on actual MBS yields is staggering.  In and of themselves, MBS yields don't mean much, but comparing them to a benchmark like 10yr yields allows us to see how attractive MBS might be by comparison (among other things).  The blue line in the chart below is the MBS yield as calculated by Refinitiv (MBS yields are somewhat subjective and require complex math to determine) minus the 10yr Treasury yield.  The lower it is, the more MBS are outperforming. 

20201023 open.png

Long story short, the only time MBS were doing remotely as well was the brief period following the roll-out of QE3 in which the Fed unveiled MBS-specific bond purchases.  Being back at those levels raises risks that MBS will have a harder time outperforming Treasuries in the event of additional weakness.  The implications aren't quite as bad for mortgage rates as those spreads are setting records in the other direction (i.e. rates are so high compared to MBS that MBS could lose ground without rates having to rise very fast).

It's a light day for financial markets in terms of the calendar.  Bonds are starting out unchanged to slightly stronger with little--if any--reaction to last night's presidential debate.  Stimulus headlines--should we see any significant examples by the end of the day--remain a hot button for Treasuries.  And again, such hot buttons are increasingly likely to matter for MBS as well.

MBS Pricing Snapshot
Pricing shown below is delayed, please note the timestamp at the bottom. Real time pricing is available via MBS Live.
UMBS 2.0
102-29 : +0-00
10 YR
0.8514 : +0.0034
Pricing as of 10/23/20 10:09AMEST

Tomorrow's Economic Calendar
Time Event Period Forecast Prior
Friday, Oct 23
9:45 PMI-Composite (source:Markit) * Oct 54.3
9:45 PMI-Manufacturing (Markit) * Oct 53.4 53.2
9:45 PMI-Services (Markit) * Oct 54.6 54.6