Bonds started the new year off on a strong note yesterday, rallying to the best levels since ____.  That blank would be filled in with a distinctly different month depending on who you choose as your bond market spokesman.  If you choose "10yr Treasuries," we had the lowest rates since January.  As we move down the line to shorter maturities, the date would get more recent.  For example, we'd only have to go back to early June to see lower 2yr yields.

If we choose MBS and/or mortgage rates as our representative for "bonds," the blank is filled somewhere in between--roughly April for mortgage rates and late May for 2018.  Unlike 10yr Treasury yields, neither mortgage rates nor MBS are anywhere close to the levels seen back in January.  


Several things, really...  We talked about a few of them back in October, which will be an important time frame for the following discussion.  Here's a link to that piece.  In it, I discuss the issue of duration (how long an average MBS coupon lasts) as it relates to broader bond market duration preferences.  Simply put, consider the 10yr vs 2yr performance mentioned in the first paragraph above.  10s have been doing way better as the yield curve flattens.  MBS duration is more than 2 years and less than 10, so it would make sense for the performance to be somewhere in between purely from a duration standpoint.

But there is other stuff going on as well.  "Convexity selling" is a problem for MBS any time rates are spiking aggressively, and especially when the yield curve is flattening (shorter term yields rising faster than longer-term yields).  Convexity sounds complicated, but it's pretty simple.  I'll use bullet points to break it down:

  • If rates fall, MBS are paid off quicker as people refinance.
  • Quicker payoffs mean "shorter duration" as far as investors are concerned.  Treasuries outperform here because investors will be able to lock in today's rates of return for a longer, guaranteed time frame without having to worry about lower rates causing their MBS portfolio to pay off more quickly.
  • If rates rise (especially if they rise while shorter-dated yields are rising to meet longer-dated yields, like they are in 2018), MBS are paid off more slowly due to an absence of refi demand.  
  • Slower payoffs mean "longer duration" as far as investors are concerned.  Treasuries outperform here because investors worry they'll be stuck with below-market rates of return in a rising rate environment.
  • A flatter yield curve (higher 2yr yields vs 10yr for instance) only makes things worse because there's not much benefit to sign up for longer duration if short-term debt is yielding nearly as much. 
  • It's only when investors see the writing on the wall with respect to short-term debt being on the verge of plummeting in yield (like at the beginning of a Fed rate cut cycle or the beginning of a recession) that there's a noticeable rise in demand for longer-term debt (the stuff that will preserve higher rates of return for longer)
  • At these times, MBS can appear to be "on sale" to investors who think rates will continue to fall

The charts below point out a few instances of convexity selling.  The blue line is the spread between an MBS yield benchmark (calculated by Thomson Reuters) and 10yr yields.  In addition to convexity, we also have Fed bond buying considerations.  This was absolutely huge in 2012 with QE3 (the one where they just focused on MBS) caused a massive "re-set" in the performance of MBS vs Treasuries.  The spread has really been trying to find its range since then, and has mostly done a good job of staying calm.  Late 2018, of course, is a special time because Fed MBS purchases dried up as of early October.  The resultant spike in spreads is highlighted, but note that spreads had been weakening all year.

2018-1-3 open2

Here's an alternate view with average 30yr fixed mortgage rates vs 10yr yields (higher = bigger gap between the two):

2018-1-3 open3

An even easier way to understand the mess of information in the charts above would be to view spreads in the context of 10yr yields.  To do that, we'll need to invert the axis such that MBS are weakening when the blue line is falling (in the chart above, the opposite is true).  As you can see, the general trend is for MBS performance to correlate with 10yr yields as long as rates are fairly stable.  Big spikes in rates cause convexity selling.  

2018-1-3 open4

The bottom line is that 2018 suffered from a combination of convexity selling due to the highest rates in years as well as the gradual decline in Fed bond buying.  The additional drop in October was partly due to the Fed being completely out of the MBS market and partly due to Treasuries finding more favor amid the extreme stock market volatility.  Any way you slice it, it was going to be bad for the mortgage market, relatively.

The great news is that spreads, for now, have found support at the same place they did in 2013 and 2016.  If they can build on that defensive baseline, it will do one of two things.  If Treasury yields rise, mortgage rates would rise more slowly.  If Treasury yields hold sideways (or slightly stronger) mortgage rates should be able to fall at a slightly better pace.