This has been an occasional theme for us, but after some conversations last week, think it warrants a deeper look. The various parties interacting in the market have become more correlated and the relationships more intricate. The growing use of ETFs by institutions is a key part of that self-reinforcing feedback loop.
A now regular occurrence is a bond trader gets “lifted” in the morning or sells bonds and is left short. That trader waits all day hoping to find a seller of those bonds. At the end of the day, rather than lifting an “egregious” offer, or going home short, they will buy some ETFs to cover the market risk. Of course the next morning, everyone sees the spike in the ETF and views it as a sign that the market got really strong into the close, so they try to buy more bonds. This behaviour first became noticeable in the credit markets with the creation of CDS indices but is now occurring with more frequency in the ETF space.
The “arbitrage” community also plays a role in these loops, especially when quoted bond “prices” don’t reflect the reality of where the bonds would trade.
To understand how the feedback loop can create problems, particularly in a down market, it is worth going back to seeing how credit traded in the “old days” like 2007.
Bond Sell-Offs in the “Old Days”
Joey, the trader at a big market maker sends out a run on ShmoGo bonds. It is early in the morning, nothing much going on, futures a bit lower, but nothing that made the trader put much effort into the generic bond run.
A client sees the run, and wants to sell. They call the trader directly, asking him to refresh the ShmoGo 7’s. Joey immediately guesses the client is a seller, so rather than “repeating” the 99.5/100.5 market from the run, quotes the bonds 99/100. The client is a bit annoyed as he clearly hoped to sell at 99.5.
The client that proceeds to explain that everyone and he means “everyone” is 99.25 bid, won’t the trader take a couple at 99.25 to save him from having to call around? By this time, the customer’s salesperson has gotten involved and clearly thinks this is generous of the client, since the trader just flaked on the price and pushes the trader. Now feeling trapped and getting sucked into buying bonds he doesn’t want, the trader protests a little, asking what the client is doing with their overall position.
The trader knows the client has a big block of these bonds and is legitimately concerned that if the client is exiting, then this purchase of a couple million will end very badly. The client insists they are just “trimming” the name. ShmoGo’s are a core part of our strategy, we still love the name, but are just lightening up to cut back from an aggressive overweight to a more normal position. At this stage, trader finally accepts and buys $2 million ShmoGo 7’s at 99.25.
By this time, it’s off to the morning meeting. The trader brings up the “axe” of having to move some ShmoGo 7’s at 100. They spend a half hour going through axes that are unlikely to be filled and listening to a couple research people drone on about some other company that looks cheap and another that is lucky not to be filing tomorrow.
Meanwhile, finally back on the desk, having missed some weak economic data at 8:30 while in their meetings, futures have declined a bit more. There is a slight “risk-off” feeling.
Sales are on the phones with their clients. Senior sales figuring out what their clients want to do and how to get it done, and junior sales trying to flog the morning axes, not completely aware that non of the bids are likely to be there now that the weakness has accelerated.
A little while later a salesperson asks the trader where he is on the ShmoGo 7’s. The trader, a little exasperated by now, shouts out that he can offer them at par, and asks just under his breath, “weren’t you at the morning meeting?” This gets a little chuckle from the entourage of traders around him. Sales is less impressed, but they were expecting this. The respond with, “OK, that’s what I thought, but my guy is telling me that’s pretty generic as its is 99/100 everywhere, and he thinks he could get the bonds at 99.5”. The trader asks the salesperson if their client is a buyer, and after a brief pause while the salesperson checks with the client, he gets the answer that no, they aren’t a buyer, were just giving some colour.
As the morning drags on, a couple of other traders manage to get hit on bonds and it is clear that the market tone today isn’t good. Not horrible, but certainly no strength.
The trader gets to work. He sends out a fresh run ShmoGo 7’s 99/100* (where the * indicates “better” seller). He calls one of his interdealer broker, and after chitchatting about last night’s game, yet again, he tries to get some sense on ShmoGo’s). The broker said he’s had a few traders come in checking on the name, thinks they were sellers, but has nothing live. The trader says to work a 98.75 bid, but to work it lightly. The broker can’t help from smiling, so you’re there, but not really looking to be worked?
So the trader has done his best to create a “picture” that the bonds are 99/100 and the street is “bid without”. He knows the 98 ¾ bid isn’t really going to help the picture, but he’s got a bad vibe about the situation and is getting concerned the original client might actually be out there spraying the street with these bonds.
One of the earlier salespeople comes back asking for a refresh on the ShmoGo 7’s. The trader immediately knows the client isn’t looking to buy bonds. They would have engaged him much differently, asking him “what was the best level on the offer?” or something similar. This is clearly someone trying to get him to make a two-sided market so he can hit the bid. Ever careful, the trader responds, “I went out 99/100 but am a seller only right now”. The salesperson immediately frowns, because they too knew what the client was trying to do. After a short conversation with a disproportionate amount of nodding, the salesperson walks over to the trader. This isn’t a good sign.
The salesperson is looking for a firm bid. The client wants to sell a couple bonds, and wants to know the real bid. The trader, doesn’t want to buy the bonds, but also doesn’t want to lose control of the situation, so they say they would buy pay 98.5 for a couple. Since he had bid the street 98.75 earlier, he figures the client has probably seen a lot of 98.75 and 98.5 bids and wasn’t going to hit his 98.5 bid “on the wire”. Sales trudges off to make the call, acting as though he had been horribly wronged, making it obvious to everyone that he was embarrassed by the traders “lowball” price.
As sales gets on the phone, the face changes, a big smile, he has found something, or at least he thinks he has. He shouts across, so everyone can here, “my guess says he would sell us $2 million at 99 and leave us an order for 3 more”. Sales is quite happy as 99 was the bid from the last run, and an order on a few that they could work riskless has to be great news. The trader is far from happy. Now there is a block seller, below where he bought bonds in the morning. He says the best he could do is pay 98.5 for a couple and work something higher, but, he tells the salesperson, they are working on developing a block buyer (not a total lie, but not completely true either, depends on your definition of “developing”). He tells sales, that if the client is patient, they would take his bonds along when the block buyer came in.
He looks down at his phone to get research on the line, to help this situation and sees the broker’s line flashing. He doesn’t even have to pick up the wire to know that the little spiv is trying to hit him on his old 98.75. Sure enough, “hey, um I know you didn’t want to be worked, but you know I’ve got a guy who said if he can get the bid back, he’d sell bonds there”. No. That trade is easy to duck, but now what.
The situation isn’t great and the client didn’t hit the 98.5 bid, so now is about as good as time as any to grab a coffee. If he’s on the desk, someone might try and hit that bid, and certainly no one is coming to buy bonds, so it is time to hide mode.
The firm spends the rest of the day trying to find a buyer. People aren’t really talking about a specific price, but “in the context”. Everyone knows the bonds are probably 97.5/98.5 right now. The top salespeople don’t really focus on the axe, because they know the trader bought bonds higher and isn’t about to sell them in the right context, yet. He would need some more pain before that happens.
They do have one client, who they are pretty sure is short the bonds, but that client is as sketchy as they come. They manoeuvre the world of opaque, over the phone bond trading, finding the mistakes and taking advantage of them. This is a dangerous call to make, but they risk it. The client claims that not only do they not want to cover, but they might short more. Not good. It is too bad the market maker can’t hear the laughter at the client. One analyst asks the PM why aren’t you covering the short yet? It has gotten to the price target. The PM just laughs. He has watched this trader for years, and knows that if he hit a 98 bid, the bonds would be coming out at 97.5 tomorrow when the trader got the tap from management.
So, by the end of the day, the trader goes home with a couple million of bonds bought at 99.25, that are realistically 97/98, though mostly quoted 97.5/98.5 since another dealer had managed to get hooked and bought some bonds at 98.5, probably to work an order on 3 more from the client who had passed on that deal earlier with this firm.
The next morning, the firms that bought go out as 98/99 seller only. Others are 97/98, but you can’t hit a 97 bid, when there is a 98 “bid” even though there really isn’t a 98 bid. Yes, this is weird, but true. If 2 dealers are sending out runs saying 98/99 seller only, all you really know is that bonds are offered at 99.
Yet, somehow, it is hard to sell bonds below that fictitious 98 bid. Especially since the sellers could have sold much higher than 97 the prior day.
So now the market just sits there another day as everyone tries to protect their position. While this may be an exaggeration to make a point, there will be a lot people out there who think I’m talking about them (I’m not). ETF’s, just like the CDS indices, change this.
How is it different now?
The ETFs, similar to the CDS indices, can create a feedback loop.
There is no place to “hide” anymore with one sided runs. The ETF’s show you where the market is heading. CDS did as well, but not everyone gets CDS pricing, and many traditional bond investors, choose to ignore CDS when it moves opposite to how they want the market to move. Then the CDS market is all about “technicals” and hyper trading macro people who “know nothing about credit”.
It is harder to ignore it when the ETF’s are moving. If any bond you wanted to sell is either offered only, or down more than a point on the bid side, but HYG or JNK have barely budged, you sell them. They are “beta” management tools. At some price you will attempt to manage market risk rather than name specific risk.
So far so good, but the real feedback loop occurs when the ETF goes “cheap” to NAV. If the ETF is outperforming, it is easy to see why someone would sell that against their bonds, but why would you sell if the ETF was “cheap”?
Because it isn’t really cheap, it is actually expensive still. If all bonds started the day at 99/100, the NAV would be 99.5. If the market weakens and all bonds are quoted 97.5/99.5, the NAV is 98.5, so only down 1%, but in the real world, if you are a seller you are down at least 1.5 points (from 99 to 97.5). So now you can sell the ETF down more than 1% because although NAV moved only 1%, the bid side dropped more. Also, with bid/offer increasing in the bond market, the cost of selling a bond and having to buy it back has skyrocketed, making the ETF’s (or CDS indices) more appealing.
So now the ETF leads the way and shows the real flows. Any chance a dealer had of selling bonds at a high price is pretty much gone. The transparency makes it harder to hide, and brings down offers quickly, making the market seem “heavy” causing bids to pull, and ETF’s to sell off more.
That is risky enough, but the ETF “arb” like index “arb” is what typically creates the next wave of the feedback loop.
While a real bond investor is unlikely to hit the “low” bid, an ETF arb client might. A real investor will see how quickly the bid has dropped and determine what they think is the best course of action. They might not sell here thinking the market has gotten ahead of itself. They might sell something else. They might short the ETF (yes, there is a propensity to short something you don’t own, even if it is “cheap” rather than taking the hit on what you own, I understand it, I’ve done it, but it is unhealthy).
So while traditional bond investors can commiserate with their salespeople over how quickly things gapped down and strategize about how to play it, and traders can try to avoid having long pushed in their mush, there is a group of traders who don’t think like that.
All they need to know, is if they can hit 98 bids on X number of bonds that the ETF’s are looking for, they can hit those bids, buy the ETF, do a redemption, where they exchange ETF’s for the bonds (to get net flat) and take out a profit if the ETF is trading cheap enough.
This group of traders isn’t concerned about the absolute price of the bonds, because they didn’t own them before, and won’t own them again in a few minutes (slight exaggeration). All they care about is the relationship between the bonds and the ETF.
In credit markets, and high yield in particular, this puts the single name risk in the exact wrong hands, at least if you are looking for stability. It means market makers are getting hit on bonds on levels they didn’t think possible, in a falling market. At some point, any hopes of greed and crossing the bonds for some decent coin get replaced by fear. What does someone know? How much is coming out.
Intellectually they might know it is the arb, but they won’t be able to stop themselves from getting some protection such as shorting the ETF. The market maker stops worrying about making and losing eighths and quarters and worries about losing the notional amount. At that moment, they need to get out.
So far we haven’t seen a full feedback loop on the downside in the ETF space. We came close a couple of times, but it never quite seemed to crack. We have seen it on the upside. I would argue that much of the rally in the first part of this year, was because the ETF’s were trading at a premium and the self-fulfilling feedback loop was in play. But on the upside, the fear, even of being short, never hits the same way as the fear of being long. Being short, especially at record prices on bonds, doesn’t have the same danger as being long. So that feedback loop isn’t as dangerous.
On the other hand, if you looked at CDS, we had a big feedback loop this year. We saw it happen with the JPM IG9 trade unwind.
Just a Theory, Not a Warning
I don’t see anything to trigger this loop right now. Europe is doing some things to reduce risk. The Fed is pumping money into the system. I’m not concerned about this occurring anytime soon, but while we have such a quiet day, I figured it was worth explaining, because if and when we get a loop like this on the downside there will be no time to explain.
In the meantime, one reason I’m comfortable being neutral on high yield up here and don’t think there is any urgency to buy is because the premiums of the funds don’t indicate strong demand or that the arb activity will be there to support the feedback loop. I sent out to some people last Wednesday why I thought the CDS market would outperform ETF’s, and that is still my view, and has a lot to do with the bonds that make up the high yield index and their rate risk exposure for some, and horrible convexity for others.
Business Insider Emails & Alerts
Site highlights each day to your inbox.