Liquidity. It was this or a bond cube.

Photographer: Mark Ralston/AFP/Getty Images.

Nobody Knows How Much Bonds Cost

Matt Levine is a Bloomberg View columnist. He was an editor of Dealbreaker, an investment banker at Goldman Sachs, a mergers and acquisitions lawyer at Wachtell, Lipton, Rosen & Katz and a clerk for the U.S. Court of Appeals for the Third Circuit.
Read More.
a | A

As you may have heard, people are worried about bond market liquidity. One problem is that banks don't want to trade bonds any more, because that is not such a good business in the modern regulatory environment. One potential solution for that problem is to have investors just trade bonds with each other, cutting out the banks as middlemen. But it turns out that doesn't work very well, in part because investors don't know how much bonds are supposed to cost, and need banks to tell them:

Many bonds don’t trade for weeks or months, leaving gaps in pricing that historically were filled by banks that had more market information at their command than their customers.

On Bondcube, which announced its closure on July 22, investors who found each other on the company’s system often couldn’t agree on a price, according to a person familiar with the matter, who asked not to be identified because the information isn’t public. So bids and offers were too far apart.


One naïve question here would be, how come big bond investors can't figure out how much bonds are worth? The job of a buyside bond firm is to do research and financial analysis and company meetings to figure out how good a company's credit is. The job of a sellside trader is to balance his book and go drinking with clients and twitch. You would think that the former skill set would be more conducive to knowing what bonds should cost.

And it probably is. It's just that the latter skill set is better for figuring out what bonds do cost. Dealers have lots of information about order flow, market sentiment, which customers want to buy and sell which bonds, etc. They are good at price. Investors have lots of information about companies and credit and economic fundamentals. They are good at value. When they think the price is out of line with value, they trade. If they don't know the price, then they don't know how out of line it is with value, and they don't trade.  

You can learn a few things from poor Bondcube, the bond-trading platform that shut down last week. One is that that sort of pure price information may not look like much, but it has a social value. Sometimes people criticize market participants who act as pure middlemen and do no fundamental analysis. You see this mostly in markets with lots of high-frequency traders : Those traders aren't adding to the information content of markets, the argument goes, so they are probably just skimming money from real investors and should be banned or disfavored or spoofed. Here, though, the real-money investors have done the fundamental analysis, but still can't trade. What they need is someone to do the non-fundamental analysis, to figure out how many buyers there are and how many sellers and what the price should be so that all the fundamental investors can trade. The middlemen aren't adding information to markets, but they're allowing the information that's already there to flow freely. And without them, you can't find the information.

Bondcube is also a nice little lens through which to look at the felony fraud charges against the mortgage traders -- Jesse Litvak, Matthew Katke and potentially others -- accused of lying about this sort of price information. I find these cases very confusing, because these guys didn't lie about any fundamental facts of the bonds they were selling. They never fudged cash-flow statements or bribed home appraisers or anything like that. Instead they (allegedly) just told customers that they'd paid 58 cents on the dollar for a bond, when really they'd paid 57, and pocketed the difference. That seems clearly irrelevant to a sophisticated investor's view of value, and so arguably not material: If the investor thought the bond was worth 59, and paid 58.25 for it, then why does it matter that the dealer bought it for 57 instead of 58? But that sort of pricing information, from a pricing source, is clearly relevant to a sophisticated investor's view of price, and you see what happens to the bond market when it can't get reliable price information. That information is important, though I guess that still doesn't answer the question of whether it's material.

Another thing to think about is the difference between liquidity problems now and a liquidity crisis later. If people can't efficiently trade bonds now, that's a problem. Their returns might suffer; they might allocate capital inefficiently; they will be sad and complain a lot. But if people can't trade bonds at all when bond prices drop, then illiquidity could cause a crisis: Funds will have mark-to-market losses; investors will redeem; funds will need to sell bonds to meet redemptions; they'll only be able to sell the most liquid bonds, and at steep discounts; that will cause further-mark-to-market losses, and further redemptions, and further sales of progressively less liquid assets, and further losses; etc. This might be true. But as I said the last time we talked about bond market liquidity at length, it is hard to see how bank dealers could fix that problem: Dealers are buyers of first resort, not last resort, and they were never in the business of buying bonds all the way down in a crash.

So here you see what it's like when there are no dealers as buyers of first resort: It's hard to get willing buyers and sellers to agree on a price. But that doesn't tell you much about what happens in a crisis. If I think a bond is worth 99 and you think it's worth 97, I may not be able to convince you to sell it to me for 98, and we might miss out on a trade we both would have wanted. But if I think it's worth 90 and you're desperate to unload it for 70, we've got more room to make a deal. The bigger the crash, the more opportunity there is for value investors to step in and provide liquidity that dealers currently don't.

Finally, you could think about solutions to this liquidity problem. One would just be for investors to get better at setting prices:

“The bigger thing really is behavioral change within the buyside trading community. They’re trying to adjust,” he said. “They are hiring more guys with sellside broker-dealer experience, guys that are more comfortable or with experience being market makers, trying to play a similar role now on the buyside. That’s going to take some time.”

Sure, but it's also a bit of an awkward fit. If you're a sellside trader, you have customers. You interact with them all day, and they tell you what bonds they want to buy and sell, and you get a sense of where the market is and what the price should be. If you quit, you can take the same skill set and contacts list and entertainment budget to a mutual fund, but now instead of a service provider to your former customers, you are their competitor. They might still want to hang out and get drinks, but they will be less interested in telling you what bonds they're looking for and how much they're willing to pay.

Of course you could still make a go of it. That's sort of the point of Ken Griffin's op-ed from the other day, arguing that "recent reforms and regulatory pressures have dramatically increased the number of participants who can make prices and provide liquidity across many fixed-income markets." The bio on that piece says that "Mr. Griffin is the founder and CEO of Citadel LLC, a hedge-fund manager and securities dealer," and I suspect there was a time when it would have left off the "securities dealer" part. But Citadel realized that when banks won't make markets, somebody has to, and it might as well be Citadel.  

There are other possible solutions. Many of the BlackRock liquidity suggestions might help with this price-discovery problem. If companies issued fewer bonds in larger sizes, each individual bond would trade more often, and so its price would be more transparent. And bond exchange-traded funds "can help enhance price discovery," though not necessarily for individual bonds. On the other hand, BlackRock has argued for less disclosure of large block trades,  which would reduce the market impact of those trades (and so might encourage trading), but which would also reduce price transparency and so might, judging from the Bondcube experience, hurt liquidity.

But one of the most popular solutions for bond-market illiquidity, urged by BlackRock and others, is more electronic trading of bonds.  At best, electronic venues would aggregate price and order information in a way that increases price discovery and liquidity. Or they might just lead to a bunch of investors twiddling their thumbs and wishing there were some dealers to make markets.

Or they might provide a platform for new, non-bank dealers to make markets. One striking thing about modern financial markets is how many of the transparent, anonymous, all-to-all markets end up dominated by dealers anyway: Electronic trading reduced the dominance of banks in equities and Treasuries, but high-frequency traders came in to fill the void. It turns out that markets like having middlemen to tell them what the price is.

  1. Or, perhaps: Dealers are better at figuring out what the price should be today; investors are better at figuring out what it should be next year.

  2. I mean. This is an idealized toy model. There is ample reason to believe that many professional bond investors are not trading on informed views about value. Some are of course trading for more or less mechanical fund-flow reasons divorced from their own notions of value. Some don't have notions of value. Some have notions of value that are silly, or careless, or overly ratings-driven. Consider Izzy Kaminska's post yesterday about "The bond liquidity 'cognitive bandwidth deficit' problem":

    In other words, it’s all very well depending on the crowd to make judgments about risk, but if that crowd is mostly dumb, doesn’t have the time to evaluate things properly, operates on a massively thin bandwidth or is overly dependent on third-party algorithms that use generic historical patterns that don’t account for externalities or nuance, then, chances are — as per the Goodhart paper — asymmetric information states will manifest and in turn create liquidity crises.

    Or consider the mortgage crisis, etc.

  3. Of course, if two investors have very different views of value, they should trade, but if they don't know the price then they apparently don't. If I think a bond is worth $100, and you think it's worth $95, then you should sell it to me for some price between $95.01 and $99.99, but without someone else to tell us what the price is, how will we trust each other to split up the gains fairly? There is obviously a behavioral-economics element to this problem (any split is Pareto improving!), but then there is obviously a behavioral-economics element to most of life.

  4. Though not exclusively. The famous "Bonfire of the Vanities" speech about crumbs is arguably a criticism in this vein. 

  5. In the Jack Treynor terms that I used my prior post (especially footnote 1), dealers buy and sell at the inside spread, and ultimately lay off their risk to value investors at the outside spread. The failure of Bondcube just means that, when you have no one making markets at an inside spread, you don't get much trading. But value investors still exist, and still should be willing to trade at the outside spread if prices collapse.

    Obviously this is a model that assumes no change in value (i.e., a static outside bid and offer). In a real credit crash or whatever, you'd assume that value would go down. But the theory here is that prices would go down faster than value -- as liquidity illusion, etc., drives funds to sell off bonds -- and would then create an opportunity for value investors to stabilize the market.

  6. It had some practice realizing the same thing about equities market-making a while back.

  7. E.g. page 7 here:

    For investment grade, the TRACE threshold for large block trades is $5 million, meaning that any trade greater than $5 million is publicly reported as greater than $5 million on TRACE and available for the market to see almost instantaneously. The impact is that the marketplace is now almost immediately alerted that a large trade has occurred which reduces market depth by making it more likely that larger trades will have a market impact because large block disclosures are seen as market moving indicators. ...

    We believe that there are two ways that policy makers could address this issue, either of which would likely result in an immediate improvement to market depth where post-trade reporting is already in place. First, reporting for certain asset classes could be delayed – particularly where market depth is more challenged (e.g., high yield) – until the end of the day. Alternatively, the threshold for large block reporting could be reduced. 

  8. I'll outsource the disclosure:

    Bloomberg LP, the parent of this news organization, also has a platform for trading corporate bonds. 

This column does not necessarily reflect the opinion of the editorial board or Bloomberg LP and its owners.

To contact the author on this story:
Matt Levine at

To contact the editor on this story:
Zara Kessler at