Pimco Sold Itself Some Bonds
Thursday morning's Pimco cross-selling story is a fascinating little study in bond market liquidity. There are two stories of bond market liquidity. One is a story of costs: As dealers retreat from trading risk, it becomes harder to move bonds between end users. Fund X wants to sell a particular sort of bond right now. Two days later, Fund Y wants to buy a slightly different sort of bond -- maybe a bond from the same company but with a different maturity, or from a different company in the same industry. In the glory days of liquidity, Fund X would sell its bond to a dealer bank, and two days later Fund Y would buy its preferred bond from the dealer's vast inventory of bonds. The dealer overall would be flattish on the trade: It would increase its bond exposure by buying from Fund X, and reduce it by selling to Fund Y, and take a small amount of risk on the two days' delay and the mismatch between the different bonds that it was trading. The dealer made its money by intermediating those time and bond-flavor mismatch risks.
Now, the story goes, that doesn't happen: Dealers have shrunk their inventories and are less willing to do these sorts of trades with investors. Now, if you want to sell a bond, the dealer will moan and whine, and charge you a huge bid/ask spread, and only do the trade in small size, and only do it if he can immediately find a buyer for the exact same bonds, and take a week to get back to you, and so forth.
The other story is a story of risks: When bond prices go down, the story goes, everyone will have to sell, bond funds will liquidate in a vicious cycle, and dealers won't be willing to take any risk to support the market. This is a somewhat untested theory, and also a somewhat weird one: Why would dealers buy tons of bonds when everyone else is selling? The dealer's job is to intermediate bonds over time, not to stand on the opposite side of the whole market.
Bill Gross's departure from Pacific Investment Management Co. was a good test case of, I think, the first story, but at scale. Bill Gross is certainly bond royalty, but it is not obvious that Bill Gross leaving one employer to go to another employer is a good reason for bond prices to crash. Like, at time zero, people own a bunch of bonds through the Bill-Gross-run Pimco Total Return Fund. At time one, people own roughly the same amount of bonds through some combination of the (1) non-Bill-Gross-run Pimco Total Return Fund, (2) non-Bill-Gross-run other Pimco funds, (3) Bill-Gross-run Janus Global Unconstrained Bond Fund and (4) whatever other bond funds they move to. Bill Gross's departure may be a reason to get out of his old fund, but it is not obvious that it would be a reason to get out of bonds, permanently.
But in between time zero and time one, someone has to intermediate the flow of bonds out of the hands of the Total Return Fund and into the hands of whoever ends up holding them. That's an intermediation in time -- Total Return isn't necessarily selling at the precise time that someone else is buying -- and also in flavor, as you'd expect bonds that Bill Gross idiosyncratically favored to lose value during this migration to other, less idiosyncratic managers.
So who intermediates those flows? The obvious, and probably mostly correct, answer is "dealers." But dealers are, so the story goes, expensive and annoying to deal with. They charge bid/ask spreads, they take time, and their balance sheets could be overwhelmed by the tens of billions of dollars that were leaving Gross's old fund to go elsewhere. So there was some economic incentive for the end users to cut them out of the equation.
So they did! Or Pimco did. Or Pimco did a little. Pimco was a big bond seller, out of Total Return. It was also a buyer, in non-Bill-Gross-run funds to which investors were re-allocating money. For Total Return to dump bonds to dealers, paying bid/ask and pushing down prices, and for other Pimco funds to buy back those bonds, paying bid/ask again and pushing prices back up, would be a little nuts. So Pimco intermediated a portion of the trading in-house. How big a portion depends on how you count, but it's not trivial. Total Return sold about $260 billion of bonds in the six months ended March 31, 2015, but it also bought about $224 billion, for net sales of $36 billion. It sold about $18 billion of those bonds to other Pimco funds, and bought about $4 billion from them. So something like 7 percent of Total Return's gross selling, or 39 percent of its net selling, was to other Pimco funds. 1
By selling to itself, Pimco avoided having to pay dealers to intermediate its trades in time. It also, more subtly, avoided having to pay dealers to intermediate between categories of bonds. So Bill Gross was a big fan of Treasury Inflation Protected Securities, and people worried that his departure would hurt TIPS prices as people reallocated from his fund, which was idiosyncratically overweight TIPS, to other funds, which weren't. The Pimco Income Fund was historically less of a fan of TIPS, which makes sense, because it's an Income Fund, and it's not obvious that a fund whose "primary investment objective is to maximize current income" would be all that attracted to a 10-year bond with a 0.25 percent coupon. 2 But at the end of March that was the largest holding of the Income Fund, which "previously held little or no TIPS since the middle of 2012."
Because that bond looked like an okay trade, even if it wasn't quite an income-y trade, and more to the point there was economic value to be added by absorbing it internally rather than paying dealers to deal with it. Something like $1.5 billion worth of TIPS migrated from the Total Return Fund to the Income Fund. Just regular quoted bid/ask spreads on those TIPS would be something like 0.2 percent of par, 3 so you save $3 million by doing it internally. But that is a very, very low lower bound on the savings: Those quotes are for $1 to $10 million sizes, not $1.5 billion. In a market in which dealers had to intermediate $1.5 billion of TIPS over time, you'd probably pay at least an order of magnitude more than that in price slippage. And in a market in which dealers had to intermediate TIPS between a guy who loves TIPS and a market that was relatively indifferent to them, they'd charge even more. By perhaps stretching its mandate a bit, the Income Fund was able to avoid those intermediation costs. The Securities and Exchange Commission rule covering trades between funds in the same family requires trading at mids, so the Income and Total Return Funds pretty much split the savings here, which surely amounted to tens of millions of dollars. 4
There have been more, you know, formal efforts by the buy side to cut out some of the costs of dealer intermediation and let funds trade between themselves, but they have gone sort of poorly. BlackRock's comically recurring efforts to make electronic trading of corporate bonds a thing have been stymied by the fact that everyone wants to trade the same way: Mostly, among big asset managers and insurers, there are plenty of buyers and no sellers, so the electronic platforms are sort of boring. There just aren't that many days when a big bond manager wakes up and decides to sell billions of dollars worth of bonds, while other managers wake up and decide to buy billions of dollars worth of the same bonds. Bill Gross Departure Day was that sort of day, but you don't get many of them.
One thing to consider here is that this ought to be an advantage of giant fund complexes. One service that big fund complexes should be able to offer to their investors is avoiding market illiquidity. Funds within the same complex that want to buy and sell the same bonds can trade with each other without paying the costs of market intermediation. And funds within the same complex that want to buy and sell slightly different bonds can, you know, call each other up and talk it out and agree to a mutually beneficial deal that also avoids the costs of market intermediation. The more expensive market intermediation is, the more valuable it is to have lots of different funds under one roof, trading with each other away from the market.
Another thing to consider here is: The poor dealers! When balance sheet is expensive, risk is restricted, and the main bond-investor strategy is to buy and hold new issues, there are not a lot of ways for dealers to make money. 5 They must have been salivating over Bill Gross's departure, knowing that tens of billions of dollars worth of bonds would have to migrate from time zero to time one, in a way that looked likely to temporarily dislocate prices without a fundamental crash, and that they would get to take a cut of all of those bonds. But it turned out not to be all of them: Some bonds moved silently out of Gross's fund into their new homes, without the dealers ever getting their hands on them.
You don't have to feel bad for the dealers, but the cost story and the risk story of bond market illiquidity are in some ways opposite stories. If you want dealers to be willing to take risks to cushion price drops in bond markets, then they need to be able to make money trading bonds. Right now investors worry about liquidity but also push to keep trading costs low. If you can't make money trading bonds in normal times, it's harder to justify risking your balance sheet to support bonds when they crash. And if you're cut out of profitable trades in the most lucrative times, it's even worse.
This column does not necessarily reflect the opinion of Bloomberg View's editorial board or Bloomberg LP, its owners and investors.
Up, incidentally, from 4 percent and 18 percent, respectively, in the previous six months. You can get annual and six-months-ending-September-30 numbers from Pimco's annual (pages 76-77) and semiannual (pages 83-84) reports for the Total Return fund. The math is my own, though the $18 billion and $4 billion numbers check out with Miles Weiss's Bloomberg News story.
That's the 0.25 of January 2025, apparently a big part of the cross-selling:
For example, Pimco Total Return Fund sold almost $3.8 billion of its inflation-linked Treasuries maturing in July 2021, with a coupon of 0.625 percent, and about $1.6 billion of 2.375 percent TIPS that come due in January 2025, according to holdings reports published on Pimco’s website.
The Pimco Income Fund, run by Daniel Ivascyn, acquired 2021 TIPS and 2025 TIPS with face values of $527 million and $984 million respectively during the same quarter. The 2025 TIPS ranked as Pimco Income Fund’s largest holding at the end of March, according to data compiled by Bloomberg.
Looking at Bloomberg's ALLQ screen I see something like 4/32 of a point for the 2025s and 8/32 for the 2021s, for $10 million sizes, but good lord is this unscientific. Here's a 2011 paper finding average bid/ask spreads of about 3/32 for 10-year TIPS, but for $1.3 million quote sizes.
The rule also imposes other requirements, including that "The transaction is consistent with the policy of each registered investment company and separate series of a registered investment company participating in the transaction." Pimco describes its policies on, e.g., page 76 of the Total Return Fund annual report.
I mean, besides underwriting new issues.
To contact the author on this story:
Matt Levine at email@example.com
To contact the editor on this story:
Zara Kessler at firstname.lastname@example.org