Lawler: Mortgage/Treasury Spreads Part II: “Decomposing” the Widening This Year

CR Note: The following is from housing economist Tom Lawler. Here is Mortgage/Treasury Spreads, Part I

From Tom Lawler:

From the end of last year to May 6th of this year 30-year fixed mortgage rates increased by about 237 basis points, the largest increase over such a short period of time since the early 1980’s.  Over that same period other interest rates went up sharply as well, including Treasury rates, but the increases in Treasury rates was significantly lower than the rise in mortgage rates.

During this same period, Fannie Mae’s required net yield on purchases of 30-year fixed-rate mortgages for delivery in 30-days, which is set based on prices of mortgage-backed securities, also increased by 233 basis points.  As such, almost all of the “widening” in the spread between mortgage rates and Treasury rates reflected a widening in MBS yields relative to Treasury yields.

So:  why did MBS yields increase by much more than Treasury yields over this period?

There are a few likely reasons: (1) interest rate volatility, both “actual” and that implied by market prices on options/swaptions, increased substantially over this period.  Since mortgage borrowers have the option to prepay their mortgage without penalty, a rise in interest volatility, all other things equal, causes the value of that mortgage to the investor to decline, and the quoted yield on that mortgage to increase.  (2)  Early this year the Federal Reserve said that it was likely to reduce its bloated balance sheet sometime after its first-rate hike, and more recently it gave guidance on the likely pace of the reduction in its almost inconceivably high holdings of Agency MBS, and that future MBS sales were possible.  Given that the Federal Reserve had not only been the single largest purchaser of MBS for quite a while, and that Fed purchases were relatively price insensitive and not based on relative value measures, these announcements could easily have led other MBS investors either to curtail purchases or sell positions, resulting in a widening of mortgage rates to other interest rates from what had been historically low levels after adjusting for interest rate and prepayment risk.

In terms of relative value, one of the key metrics used by mortgage analysts and investors is the so-called “option-adjusted spread” (OAS).  To compute an OAS one generates a very large number of possible future interest rate paths based on the current term structure of interest rates and assumptions about interest rate volatility, with those assumptions typically calibrated to market implied volatility measures.  One then generates mortgage prepayment rates along each interest rate path, with the prepayment rates typically based on models using historical prepayment rate behavior.  The OAS is the “spread” above the benchmark interest rate (often the Treasury rate) that produces an expected present value of mortgage cash flows equal to the observed mortgage/MBS price.

While there are many shortcomings of OAS as a relative value tool, virtually everyone agrees that it is a vastly superior relative valuation measure than a simple “MBS-10Year Treasury” spread or even a MBS “Z-spread,” which is the spread over the entire Treasury curve assuming no volatility.

As one might guess, there is no “one” mortgage OAS observed an any given time, as the OAS is model dependent – both in the interest rate model used the prepayment model used.  However, OAS generated by various market analysts/investment firms typically move in the same direction.

With that in mind, there is little doubt that MBS OAS to Treasury spreads widened considerably from the end of last year to early May, and a few widely followed OAS measures increased by about 50 basis points from the end of last year to May 6.  However, that widening was from extremely narrow levels from an historical perspective.  Indeed, two widely followed “current coupon” MBS OAS to Treasury measures – one available on Bloomberg and one available on Yield Book – showed that MBS OAS were NEGATIVE during most of last year, including the last day of last year.  Stated another way, after adjusting for interest rate/prepayment risk, MBS yields were LOWER than Treasury yields!!!!  There is little doubt that the Federal Reserve’s ill-advised decision to continue buying massive amounts of MBS even as overall financial market conditions no longer warranted such purchases, and even as the housing market had become historically “frothy.” (I’ll go into more on that later.)

In addition, interest rate volatility – both actual and market-implied – is up significantly from the end of last year.  All other things equal, an increased in implied interest rate volatility will increase stated mortgage to Treasury spreads even if option-adjusted spreads don’t change (assuming the OAS model is calibrated to market-implied volatility).

Finally, the change in Treasury rates from the end of last year to May 6 varied considerably across different maturities, as shown on the first page of this report.  As a result, the “mortgage to 10-year” spread widened somewhat (though just a little) more than the so-called “Z-spread” of mortgages to the whole Treasury curve.

While a total “decomposition” of the 77 basis point widening in mortgage rates relative to Treasury rates from the end of last year to May 6 would require more knowledge about OAS models than I currently have, my “best guess” would be as follows:

Curve:                                   5 bp

OAS:                                   50 bp

Volatility:                            18 bp

Primary/Secondary Spread:  4 bp

Now let’s shift back to the Federal Reserve:  why did the Federal Reserve continue to buy massive amounts of agency MBS and continue to add to its overall MBS holdings during all of last year when (1) it seemed clear that MBS/Treasury spreads adjusted for risk were at the lowest levels ever, and by some accounts negative; and (2) the housing market had become “red hot” with home prices soaring?  There does not appear to be a good answer to this question.  And yes, there were at least some in the Federal Reserve system who know that the Fed was driving down MBS spreads well below normal levels.  Here, e.g., is a chart from an article by officials at the Federal Reserve Banks of Chicago and New York from late August of last year.

And here is an excerpt from the article.

“While OAS is typically positive, the measure shown here—produced by Bloomberg for a hypothetical MBS priced at par—has declined steadily since March 2020 and turned negative for the first time since 2013, the only other time in the series history that this has occurred. Taken together, the data in Chart 1 suggest that the Federal Reserve’s recent large-scale purchases of agency MBS have contributed to historically tight spreads in the secondary market.”

Overall, it appears that 30-year fixed rate mortgages last year were about 45 basis points lower than one normally would have expected given other interest rates and prevailing market conditions because of Federal Reserve actions in the MBS market..

One question is:  did the housing market NEED excessive stimulation last year?  Obviously, based on what home sales and housing prices did last year, the clear answer is no. 

Another question is:  did Federal Reserve policy contribute to last year’s housing boom and the unprecedentedly rapid home price gains last year? The clear answer is yes.

I may have more on this later.

Note: This article was from housing economist Tom Lawler.