Bond markets don’t fall out of bed every day. It just feels a little like it right now. But the reasons can vary a lot — and they’re not always the obvious ones.
When bond yields rise/prices fall, it’s usually because investors are collectively re-evaluating their expected path for growth, inflation and central bank rates. But it can also happen when investors are shaken out of their positions by more technical forces. In fact, last Wednesday, MainFT reported investors warning that the US Treasury market “has been gripped by a ‘vicious loop’ of selling”.
So maybe this is a good moment to post about one longstanding loopy bond market accelerant that is pretty mechanical: US mortgage convexity trading. In normie terms, holders of US mortgage bonds often have to dump bonds when the bond market comes under pressure, in practice pouring gasoline on a bond market fire.
Alphaville skipped this in bond boot camp because it seemed like a bit of a sidebar. But the US mortgage-backed securities market backed by one of the big government-related agencies — primarily Fannie Mae and Freddie Mac — isn’t exactly small.
In fact, at about $7tn the agency MBS market is only a touch smaller than the entire US investment-grade corporate bond market, if respective weightings in the Bloomberg US Aggregate bond index are anything to go by.
So MBS market foibles do have repercussions. Let’s explore just how it works.
US mortgages 101
FT Unhedged’s Robert Armstrong proudly outed himself on Friday as the obligor to a 30-year mortgage with a fixed rate of 2.75 per cent. Our guess is that it was credit-wrapped by Fannie Mae and bundled — together with a host of other mortgages, into an MBS that sits happily in a bond portfolio somewhere.
Rob lives in Brooklyn rather than London or Frankfurt, so he didn’t have to faff about guesstimating the likely path of future near-term central bank rates, or even whether — if he just waited another six months — bond yields would fall further to give him a rate even lower than the one he locked in. He just took the rate.
Because in America, the mortgage market is structured in a way that minimises regret. Long-term mortgage borrowers can ditch their old high fixed rate any time a new lower mortgage rate comes along.
Sure, refinancing fees don’t pay for themselves. So doing this every time bonds rally a few basis points makes no sense. But it’s really through this mechanism that a big bond market rally (remember those?) passes directly into a real boon for households, especially those with the bigget mortgages.
Who was dumb enough to give Rob and his fellow countrymen this free option? No one. They pay for the option in the interest rate that they lock in. That 2.75 per cent mortgage was written back in the days when 10-year US Treasuries yielded a mere 0.5 per cent. And a good deal of the difference in the interest rate Rob pays over that government rate works out as payment for the pre-payment option.
Back to bond portfolios
The flipside of Rob’s optionality is that portfolio managers who own the US agency MBS that holds Rob’s mortgage are short the pre-payment risk.
If bond yields shoot up — like they have recently — then Rob’s mortgage and others like it look a lot like long-dated bonds, albeit weird amortising ones. No one with one of those ultra-low rate beauties will want to move house, let alone refinance, because the cost of doing so would be just too high. And if bond yields head to (or through) zero then MBS dissolve into cash payouts as everyone refinances out of their old high-rate mortgage deals and gets newer cheaper ones.
All this means that MBS holders take all the (substantial) downside of a longish-dated bond in a rising-rate environment, but lose the upside that long-dated bonds might bring when bond yields are coming down. In bond market jargon, MBS have “negative convexity”.
So, if a bond portfolio owns this slippery stuff, how much interest rate risk — aka duration — are they actually running? It depends.
Most normal bonds’ duration falls as yields rise, and rises as yields fall. But the fancy mathematical models that fixed income types build to try to answer this question for mortgages all estimate that MBS duration increases as yields rise.
In other words, if you hold fixed-rate MBS, the amount of interest rate risk in your portfolio goes up at precisely the time you least want it.
OK but what does this all have to do with mechanical bond selling?!?
Let’s imagine that you’re running a bond portfolio. You want to produce income for your client so you buy a ton of MBS because of that sweet extra yield they offer over Treasuries.
But then bond yields rise, pushing down prices. Your total return sucks. Moreover, your portfolio now has a lot more interest-rate risk than you’d planned, because the duration of all those MBS has increased. No worries: to get your duration back to where it was yesterday, all you have to do is sell some bonds.
If everyone is in your boat, this means that when bonds are sucking hardest, everyone will be scrambling to offload them at the same time — making them suck more. Conversely, everyone will be rushing to buy them when bonds are the plat du jour, turbocharging the rally.
Importantly, this isn’t because portfolio managers are chasing returns or because they have some new view about how the world might pan out that they want to pop into their portfolios. It’s precisely because they don’t have a new view of the world; they just want to run the same amount of risk in their portfolios as they did yesterday.
This is the so-called ‘negative convexity trade’. In other words, bond investors who have a lot of MBS in their portfolio often feel forced to dump Treasuries precisely when bonds are already getting clobbered, potentially worsening the rout further and causing another round of bond selling. (It also works the other way around, causing investors to buy bonds when they’re rallying.)
An important further wrinkle
The way we’ve described it might sound like a never-ending doom loop of a trade, but it’s actually not.
Firstly, because while mortgage convexity trading can be big, it’s not the be-all and end-all. There are a LOT of factors that at any given moment can push bond yields up and down.
Second, the accelerant only keeps trucking until the moment when embedded pre-payment optionality is more or less worthless. That is to say, if the cost of a new loan is radically higher than your old one you’re not going to be tempted to refinance.
Take Rob Armstrong and his fabulous 2.75 per cent mortgage. Without knowing the particulars of his personal creditworthiness, the average cost of a brand-new 30-year mortgage is now 7.28 per cent, so he’s not going to be keen to move and refinance any time soon, perhaps ever.
And when the temptation is close to zero, their mortgages look a lot more like normal bonds, with so-called positive convexity — where duration falls as yields rise.
And according to ICE BofA index data, we’re within a hair’s breadth of that moment:
Of course, this doesn’t mean that there’s a top for US Treasury yields around the corner. And the convexity of fixed-rate agency MBS still sits some way below bonds that don’t have any embedded optionality, which means that relative to other bonds in a portfolio it could still be a thing.
But the kind of forced selling mechanically attached to holding US MBS and targeting duration might soon be over.
We understand that any bond-curious readers who have made it through to the end of this weedy post might be disappointed to find that they’ve learnt about a big important driver of the bond market, just at the moment that it threatens to become irrelevant.
But on the bright side, it looks like there’ll soon be one less forced seller out there.
Further reading:
— Everything you always wanted to know about bonds (but were afraid to ask) (FTAV)
— Everything you always wanted to know about credit (but were afraid to ask) (FTAV)