I love to be wrong in public. For one thing, getting it wrong in public gives pride and ego that little jolt that helps keep one’s mind open and intuition supple. And for another, if I’m wrong with one gloomy-doom-y prognostication, perhaps any of the other looming mishaps I’ve anticipated, like years of sideways home prices, or privately fear, like a premature and hostile resolution of the GSEs, will not come to pass either. Looked at like this, being wrong on the wide and well-lit stages of HousingWire
is a genuine comfort.
Where did I go wrong? I thought mortgage spreads would widen out when the Fed left the market. I wasn’t sure how much, but I thought the limit was the amount of basis points of nominal yield spread between current coupon 30-year pass-throughs and Treasuries Fed purchases shaved off roughly 125bp.
My former colleagues, research analysts with MBS dealers, were anticipating widening more on the order of 30 to 40bp, but in option-adjusted terms (OAS). Given the complex interest rate and credit options embedded in MBS, option-adjusted yields, durations, and so forth are considered far more robust means for determining relative value versus other bond sectors. OAS are produced by models that incorporate sophisticated prepayment models, potential home price appreciation/depreciation, forward interest rates implied by the existing Treasury or swap yield curves and market interest rate volatilities -- so equating a shift in OAS to a shift in spread to Treasuries is a nonsensical project. Suffice it to say, I thought they were whistling in the dark, for courage.
As it turns out, they were closer to right. As most commonly benchmarked, the MBS spread to Treasuries slid up from a 2010 low (actually almost a 12-month low) of 121bp in late March, to 146bp in early May. So, 25bp. (I am grateful to UBS
for MBS spreads and prices.) On an OAS basis (according to Citigroup
’s Yield Book, along with Bloomberg
, a lingua franca
of bond analysis), the current coupon has trended wider trend since late last year, taking on about 56bp by early May. However, in the new Fed-less, flight-to-quality inflected trading environment, many of the parameters in the OAS models have altered to the benefit of MBS. That is to say, OAS are about 10 tighter.
(Refresher course: the “current coupon” is a theoretical security that would sell for par, adjusted for payment delay and forward settlement. It doesn’t exist in the market because liquidity is maximized by issuing whole and half coupons, 4s, 4.5s, etc., with the excess interest stripped off to pay guarantee fees and cover servicing expenses.
The theoretical current coupon is identified by interpolating between the coupons priced above and below par. It is a convenient subject for empirical MBS valuation models because potential loss or gain from prepayments is neutralized when the purchase price is par [reinvestment risk remains however]. The coupon/yield of this theoretical security is then compared to the yield on a similar Treasury, swap, agency or other security. In rallying markets, when expectations of refinancing and housing turnover are high, a 30-year current coupon might be considered more comparable in average life to a 5-year security. In markets with subdued prepayments, as is true currently, a 7.5 year average life is typically assumed, and the Treasury benchmark is an interpolated 5-10 yield.)
What Kept the Lid on Spreads?
Ironically, it appears that the most important technical force acting on spreads after the Fed terminated its purchase program, was dramatic transformation that program wrought on the MBS market landscape. The Fed took the lion’s share (I know this is a cliche, but it seems so right in this context) of supply out of the coupons where trading activity has naturally focused. Prices, as we learned in Econ 1, are a function of supply and demand, and for coupons around the current coupon, both the supply and the demand curves shifted down, to an equilibrium not so far from the old.
I explained this transformation in the March HousingWire
Magazine, “Who Takes the Slack form the Fed.” Briefly, very low mortgage rates were the product of historically low Treasury yields (until supply concerns overtook risk aversion among the globe’s Treasury investors) and tremendous Fed MBS appetite (bolstered by $200bn of Treasury purchases of GSE pass-throughs only). Given historically low mortgage rates, the bulk of new pass-throughs were issued in coupons that previously were very sparsely provided - 30-year 4s and 4.5s. Thirty-year 5s were a big coupon, but 5.5s and 6s were the market’s “center of gravity”. The remarkably low mortgage rates induced a generous wave of refinancings (underwater LTVs and other credit impairments prevented it from being as full a response as would have been seen in better times) out of higher coupons and primarily into 4s, 4.5s, 5s. (For context, consider that the price of 30-year Fannie 4.5s has averaged about 100 3/4 from the first trading day in 2009 to-date).
To ensure that its purchases had the greatest impact on actual borrowing rates, the Fed aimed its purchases at the coupons where new issuance was concentrated. However, it was much more interested in GSE pass-throughs, in 2009 swallowing an amount equal to about 97% of Freddie
issuance, 108% of Fannie
and just 24% of Ginnie
. Bear in mind, Ginnie’s already benefiting from the full faith guarantee whereas investors’ anxiety over the GSE’s uncertain future was a trigger for the Fed (and a $200bn Treasury) buy program. In addition, Ginnie borrowers are supported by government credit support.
By the end of 2009, the Fed had slurped up amounts equivalent to about 76% of 4s, 62% of 4.5s, 59% of 5s. Production of 6s and 6.5s was very light, so the Fed took far less - the equivalent of half the 6s, about two-thirds of 6.5s. (Note, since any pool can be delivered into a TBA trade it is possible that when the Fed bid on a current production coupon, some older securities might have been delivered, even though seasoned securities might not be the cheapest to deliver. However in 4s and 4.5s, where outstanding bonds were scarce, it’s quite likely the Fed only took in new bonds at settlement.)
As a result, the Fed now owns most of the tradable supply (float) in 30-year Fannie and Freddie 4 and 4.5s, almost 2/3rds of 5s and a significant chunk of 5.5s. (Float is defined as total outstanding amount less pass-throughs pledged to REMIC/CMO securities.)
Another Game Changing Global Event
Without the European sovereign debt crisis, it might have been reasonable to overlook the impact of this dramatic reshaping of MBS supply technicals. Earlier this year, when U.S. bond yields were under pressure from anticipated Treasury supply (deficit financing) and expectations of future inflation (economic recovery and/or quantitative easing), it was reasonable to imagine mortgage rates above 5%, even 5.25% with the Fed’s withdrawal. Trading activity would shift up the coupon stack and the supply constraints would have been less, uhm, binding.
Instead, the 10-year Treasury rate has rallied about 80bp since the beginning of April and mortgage rates have rallied about 3/8s of a point, to 4.84% (as of May 20, according to Freddie Mac’s Primary Mortgage Market Survey). And 30-year 4s are the only coupon trading below par (99-00 yesterday, according to SIFMA
). The Fed’s not here to support demand, but supply is so depleted it doesn’t matter.
Why Does MBS Demand Focus Around the Par Bond?
Non-MBS players might be asking, why not just buy MBS that don’t have a scarcity premium? For one thing, in current markets they are priced at a premium. In a non-callable bond, this might be OK, especially if the investor is able to account for the bond at the book yield and amortize that premium into income, but MBS are subject to random partial calls, accounting adjustments for changes in prepayment expectations and so on. And few MBS investors do use investment accounting. Most mark to market. Then too the duration of a premium MBS is shorter than a par or discount security with roughly the same final maturity, because it is expected to pay down faster. And it’s duration shrinks some more in a rally. For these and similar reasons, investors who want to own premiums are picky about which one’s they will buy, looking for characteristics such as loan size, average LTV, geographic concentrations that suggest refinancing response will be moderated in a rally (and, in this credit environment, won’t lead to greater than expected involuntary prepayments).
For similar reasons, investors shun discounts -- the coupon is not full, and the duration is long, so a market backup results in painfully “full” price drops, and so on. These obstacles soften with age - a seasoned discount will begin to pay down more quickly as borrowers begin to move, divorce, etc. But investors have to hold onto the security to get the benefit of a discount or premium “prepayment play”.
Staying close to par and trading TBA reduces the natural disadvantages of MBS. It also comes with certain advantages. Current production coupons, the ones with prices close to par on either side, are sold forward by issuers, for deliveries late in the current month or one or two months forward, but investor demand is focused in the current settlement month. (Now, when they have the cash.) This means investors who own bonds can rent (roll is the technical term) them to dealers who use them to correct the timing imbalance. Rolling can prove to be an important source of return on a position in MBS. It also means that investors need never take delivery of actual bonds (booking MBS entails additional systems and data resources that Treasuries or corporates do not) or, alternatively, that they can finance a significant portion of their MBS position in the roll market. (The Fed, which owns so many of the marbles in this particular game is now rolling some of its MBS to prevent the natural imbalance from resulting in delivery fails. This improves the benefits to the Fed of owning so many MBS).
Of course much, much more could be said on the natural trading strategies that have evolved for this market and for specific institutional investor segments, but the point is that MBS demand strongly tends to move with market prices. When prices go down, investors move UP in coupon, when prices go up, investors move DOWN in coupon to stay where, around par, they believe they will remain most nimble, and MBS cash flows will be cheapest on a risk-adjusted basis.
Back to Positive Technicals
Not quite done with the short list of supply positives that have held MBS prices in line while the Fed retreated. There’s more hangover from the Fed bid still supporting the market - this in the form of purchases made before the end of March, but settling in April, May and June. I haven’t the patience to add them up, but the amount is inside $100bn.
New supply was very light as well. Whether the current rally will extend and trigger a wave of refinancings and perhaps even a merrier housing summer, is still to be determined. But the supply that reached the market in the first four months of the year was described around the Street as light (one house noted a “dearth” of origination).
Another technical force not anticipated when the market began brooding about the Fed’s departure, the GSE’s announced billions of dollars of buyouts from existing securities at the beginning of this year. Freddie led with one big deck-clearing month in February and Fannie has been chopping away at its seriously delinquent securitized loans since March. By June, GSE outstandings will be reduced by about $200bn. This is a softer technical, however, for current coupons spreads, as the delinquencies tend to be concentrated in higher coupons. Also, the buybacks affect other types of MBS, as well as 30-year pass-throughs.
Some analysts also expect a positive demand effect on TBA from the buybacks, but I think it will be small. Investors who were holding higher coupon pass-throughs won’t necessarily think new, low coupon pass-throughs are good substitutes. Many investors affected by the buy backs may not have been 30-year pass-through investors at all, either, but rather held CMOs, strips (IO-PO), or other varieties of MBS. Reinvestment will take some thought.
MBS Trade Like the Good Old Days
Looking ahead, two events are foreseeable (though their likelihood is still debatable) that could turn pass-through markets upside down. The Fed could begin selling its holdings, or a resolution of the GSE situation could be reached that would halt creation of the securities that today trade as TBA. These are topics for another day.
Barring extreme events of this nature, the MBS market will continue to follow its classical pattern: when market yields fall, taking mortgage loan rates with them, spreads tend to rise with expectations of shortening durations and rising supply and, when market yields rise, spreads tend to tighten. It’s subtle and simple. If Treasury trades occur, intra-market, at successively higher prices, MBS trades also occur at higher prices, but the price gains tend to be successively smaller than in the comparable Treasuries. (Usually traders talk about MBS versus a hedge-equivalent Treasury position, an amount that has to be adjusted if yield movements suggest prepayment expectations are significantly changing as well -- it’s a very slippery game. Traders will say something like, ‘mortgages underperformed their hedges’.) As MBS players like to say of this pattern, ‘mortgages trade with the direction of the market,’ or ‘spreads are strongly directional’.
After the buffeting of the last two or three years, just hearing the old phrases back in regular use is a comfort. And taking comfort is where we started this discussion, so let’s end with it as well.
NOTE: Linda Lowell writes a regular column, called Kitchen Sink, for HousingWire magazine.
Editor’s note: Linda Lowell is a 20-year-plus veteran of MBS and ABS research at a handful of Wall Street firms. She is currently principal of OffStreet Research LLC.