As anyone knows that has pushed a car to its limits, “S” curves can be dangerous. They are hazardous because you have to go into them more cautiously than a simple curve, even though you can’t see the second curve coming when you need to be taking cautionary measures.
The biggest danger of an “S” curve if you’re driving too fast is the second part of the curve. Even though you manage to get through the first half with a four wheel drift or some seriously frightening G-forces, chance are that you won’t have enough control to negotiate the second curve in the opposite direction.
The curve you get by graphing prepayment rates against the basis points of difference between the current rate and the rate on an existing mortgage traces out an “S.”
Much of our effort in the 1980’s was focused on choosing the right kind of “S” and adding in functions driven by employment, house prices, general inflation, etc. to modify that curve.
At one end (the slow end), some prepayments happen every year because people have to pay off their mortgages, even if they are at very low rates compared to the current rate. Those are families that have deaths or loss of income, job transfers or divorces. There are even a small number of houses that are destroyed every year, and the old mortgage is paid off by fire or flood insurance.
At the other (fast) end of the prepayment “S” curve, `there are 30 to 40 percent of a pool of borrowers in normal markets who don’t prepay when it appears they have a financial advantage (lower rates). Like the borrowers whose life circumstances make them prepay a very low rate loan, the borrowers who stay in a higher rate loan have specific reasons they aren’t refinancing.
Some may have lost their jobs, so they don’t qualify (common today). Others have houses that have lost enough value that a lender won’t create a new loan for the full balance (very common today). Still others are planning to move in a year or two, so they don’t want to incur the costs associated with originating a new loan. I’ve even heard about borrowers left in GNMA 15’s after rates had dropped to 8% who told their bank that they believed in completing the original contract they had signed.
Back when I started this blog, I was expecting the end of the Fed buying in March to present a buying opportunity with wider spreads. For a variety of technical and macro reasons, exactly the opposite happened.
The PIIG sovereign scare in Europe, for example. Some amount of the money that flowed toward US Treasuries for safety ended up in Agency MBS with their currently explicit US Government guarantee and slightly higher yield (today just over 50 basis points).
There were also hundreds of billions of “fails” in the system after the Fed stopped buying. In part, this was because the Fed wasn’t like most buyers, even those “real money” accounts like pension funds and insurance companies.
The insurance and pension private investors are happy to enhance their return by lending out their MBS (for the right interest rate) to sellers that couldn’t deliver on their sale contracts, which keeps the MBS in circulation for the settlement of trades.
The Fed didn’t do that, so the sellers found themselves in a daisy chain of fails because the Fed didn’t just take a 20% swath out of the MBS market, they also took all those bonds out of circulation.
What has happened is a nearly perfect combination of market forces that kept MBS trading at higher and higher prices. Fear gripped the international markets when a pack of hyenas went after weak European sovereigns using the same “shadow” techniques that served them so well betting with huge hidden leverage against subprime MBS.
Just like it went down with the insolvency of Fannie and Freddie, the profits made by the buyers of CDS protection (credit short sellers) and the eventual cost of the bailout (for Greece, and possibly Spain) were both charged to the taxpayers’ accounts.
Why, you might ask, didn’t the powers that be choose to let the chips fall where they may?
The answer, in shorthand, is that so many “innocent” banks held the paper issued by those failing institutions (or sovereigns, as we saw last quarter) that a failure would have driven literally thousands of banks into insolvency as well. Think of the Crash of ’29 times ten, or think of the Crash of ’29 if more than half the banks held stocks on margin instead of the relative few who did.
So Euroland ended up with Germany backstopping the ECB to make sure Greece, Spain and Ireland didn’t actually default. The US ended up with the Fed buying up more MBS than the country created in 2009.
Last week the “current coupon” fixed rate MBS traded to the astounding yield of 3.43%
With short interest rates in the basement, that means a portfolio manager who owns MBS can (for the moment) collect nearly 300 basis points of yield on every dollar of MBS they finance using three month or shorter repo.
When they use “normal” caution, as a race car driver would entering a curve, those MBS managers hedge the financing risk on the MBS by swapping interest rates. They agree to swaps that are two, three, or even five years in duration to receive LIBOR in the future in exchange for paying a fixed rate for that 2,3 or 5 years. That’s why we see mREITs and analysts saying the net financing spread is more like 150 or 200 basis points than the 300 it looks like on the surface.
However, as the Ginsu knife pitchman used to say “But wait! There’s more!”
The “baseline” assumption is that today’s 4% MBS will have average lives of just five years. That’s the life assumed when the hedges are executed, and that’s the life assumed when the prepayments are expected to average 313% “PSA” (Public Securities Association). Put into practical terms, that says nearly 19% of those borrowers will prepay their mortgages every single year (19% CPR, or Constant Prepayment Rate) from 2012 until 2040.
That is about the same as assuming that mortgage rates will stay under 4.5% forever. Who believes this? Not any seasoned MBS investor. If mortgages go back to 6% or higher from 2012 onward, then the borrowers in today’s Fannie Mae pools will only pay off their mortgages when they have to. Data collected on mortgage prepayment behavior over the past 60 years or so show us that only 5% to 7% of us have to pay off our mortgage loans in a given year. Put in PSA prepayment terms, that’s only 85% to 115% PSA, which translates into 10 to 14 years’ average life.
So what happens when rates go up to more normal levels, and MBS hedged for five years turn into 10 year or longer bonds? In 1989 when mortgage rates hit 10%, 8% MBS traded at just over 80 cents on the dollar, and assumptions about prepayment speeds ranged as low as 80% of the PSA model. It boggles the mind to think how slow the market might think 4% MBS would be.
We do know that a negative funding spread on all the unhedged MBS (all that stuff that might not actually prepay at 19% forever) can put the amREITs out of business, so we have to assume they will sell and take principal losses rather than suffer long-term negative financing spreads.
For the second week in row in her Weekly Crossroads piece, Merrill Ross has included a paragraph in her weekly summary of the mREIT market that lays out the problem of this Hobson’s Choice quite well:
Not For the Faint of Heart. We continue to believe that active portfolio management, normally an anathema to REIT strategies, will be an essential component of earnings optimization during this period of unprecedented volatility in mortgage bonds. We have a strong conviction that there is value in residential cash flows, but that many traditional buy-and-hold strategies could be exposed to erosion of allocated capital, particularly as the Administration experiments with various loan modification programs. Moreover, hedging strategies, generally swapping cash flows against benchmark Treasurys, could suffer from a loss of correlation as the relationship between MBS and Treasurys evolves.
For those who didn’t check the definition, the reason I call this a Hobson’s Choice is that MBS portfolio managers can either choose to buy paper that is way too rich by any historical measure, or they can choose to hold cash and fail to cover expenses. The bigger they are, the worse the choice.
There are tiny pockets of value (all relative) in the MBS market, and there are technical supply-demand issues that occasionally allow a portfolio manager to buy select bonds at a decent risk-adjusted spread. But with all the bonds they look at, they had better have one eye on the exit at all times, and/or be ready to throw on hedges that last for ten years rather than 2, 3 or 5.
It’s a great deal like the choice the mortgage originators faced in 2007 — either they would accept lousy mortgages from flaky borrowers or they would die from lack of business. Stock mutual fund managers faced the same issue in 1999 and early 2000. They could stop being in the mutual fund business, or they could buy stupidly priced stocks, and hope that by choosing slightly better ones they would survive the inevitable correction.
With MBS trading where they are, and the 10-year Treasury Note hitting a yield of 2.61% today, 15% return on capital probably won’t be sustainable. The real issue is whether all the amREIT managers will get out of the way fast enough when the correction comes, be it a spread widening, or a general increase in interest rates out in that 5 to 10 year maturity range.