Don't make eye contact.

Pimco Bought Some Little Bonds

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

Say you run an S&P 500 exchange-traded fund. At the end of each day, you value all your stocks, apply the patented S&P 500 formula, and you get the value of your ETF; yesterday it was about $198. But hang on, you say. You know what's a great company? Apple. They make big phones, they're doing a watch, just crackerjack stuff. Right now the stock trades at about $102, but that just seems crazy to you; it's gotta be worth at least $150. So you adjust your books to reflect Apple's real value of $150, not the silly $102 that it's trading at. And since Apple is about 3.5 percent of the S&P 500, that adjustment adds about $3 to the value of your portfolio. So now your ETF is worth $201. You've just made a 1.5 percent extra return in a day. Your fund is great!

What stops you from doing this? Well it is boringly illegal, but on the other hand the Securities and Exchange Commission moves slowly, so you could do it for a while before the law catches up to you. What stops you quicker is that you're running an ETF. The way your ETF works is that anyone can come to you, hand you a big basket of the underlying stocks in the index and get back shares of your ETF (this is called "creation"). Or they can hand you shares of the ETF and get back the underlying stocks ("redemption"). If you're pricing Apple at $150 when it's trading at $102, then someone will go buy all the shares in the index, including Apple, for $198, hand them to you and get back shares of your ETF that you say are worth $201. Then they'll sell the shares of the ETF, make a profit and arbitrage the difference down to zero. So your claim that Apple is worth $150 in your portfolio has no effect. Apple is worth what the market says it is worth, and Apple in your portfolio is worth what the market says it's worth elsewhere.

That is kind of the point of ETFs. That's how the value of an ETF mostly stays in line with the value of the underlying stuff. If the ETF gets too valuable, someone will buy the stuff and sell the ETF until the prices come back into line, and vice versa.

Here is a delightful story about Pimco's Total Return Exchange-Traded Fund, an actively managed bond ETF run by Bill Gross. The SEC thinks it may have mis-valued some bonds. Here's the theory:

The investments believed to be in question, such as small amounts of mortgage securities -- or "odd lots" in the terminology of the financial markets -- tend to receive lower prices because of their small sizes or because they are backed by smaller institutions, among other factors.

After the launch of the ETF, Wall Street traders were encouraged by Pimco to offer these small securities to the Pimco ETF, according to some of the people familiar with the matter. Mortgage bonds with a relatively small $500,000 face amount, for example, might have sold for only $480,000, because few investors wanted them, due to the small size.

But when Pimco, shortly after purchasing the bonds, placed a value on them, it typically used outside pricing companies that often assigned higher valuations because they used a similar, but much larger, pool of mortgage bonds to compare them with, according to people close to the firm.

Well that is charming isn't it? One thing to say about it is that the job of an investment manager is to find undervalued securities in the marketplace and then buy them. So to the extent that Pimco found securities that were undervalued and bought them, that was good. If they were undervalued because investors who are not Pimco are skittish or have short time horizons or have trouble getting a bid on small sizes from their brokers, and Pimco can solve those problems, then it should buy those undervalued securities all day long. Of course, whether to buy those securities is a separate question from how to value them; every investor thinks that his holdings are underpriced, but you still gotta report the market price. But for stuff that doesn't trade every day, "the market price" is a bit of a fiction anyway, and Pimco is upfront that it will use pricing services to value its assets if they don't trade much. If Pimco thinks the price it bought at was wrong, and the price its pricing service provides is right, then why shouldn't it use that price?

Another thing to say about it is that "undervalued" is a dumb word and nobody knows what anything is worth, and you should buy things that are worth more to you than they are to the guy selling them. That's a good reason for Pimco to do this trade: If it can buy Bond XYZ for 85, and its valuation service says that Bond XYZ is worth 90, then Pimco has created value. It has done this by shifting the value system: Bond XYZ was worth 85, where "was worth" means "traded in this specific lot size at." It now is worth 90, where "is worth" means "is valued by an outside pricing company." If you're the prior owner of Bond XYZ, you're (presumably) compensated based on your ability to trade Bond XYZ at the market price. If you're Pimco, you're (apparently) attracting investors based on the outside pricing company's price for Bond XYZ. So moving the bond from a market-sensitive investor to a pricing-company-sensitive investor creates value. It's going to its highest and best use! That use being, like, attracting investors or compensating managers or whatever. You should only object to this if you think that values of financial assets should be commensurable, which is a somewhat humorless thing to think.

But another thing to say about it is that this is sort of not how ETFs are supposed to work. Remember that fable I told in the first two paragraphs. How does it apply here? I'm going to tell you an imaginary story because I like it, even though it is clearly wrong. The story goes like this:

  1. There are hundreds of small orphaned mortgage-backed securities, each of which is hidden away in a few small chunks. There's Bond ABC and Bond DEF and Bond QRS and so forth.
  2. Every one of those bonds is "worth" 90, in credit models or comparisons to similar less-orphaned bonds or whatever. But because they're small and illiquid and lame, they all trade at 85.
  3. Pimco goes out and says to anyone who'll listen, "Hey, sell us your orphans, we'll pay 85."
  4. Some people respond, some don't, and Pimco ends up buying a chunk of ABC but no DEF or QRS.
  5. Then Pimco marks its Bond ABC up to 90, increasing the net asset value of its fund.
  6. You notice that Pimco's fund is overvalued, and you decide to arbitrage it.
  7. So you decide to put together a basket of stuff to hand in to Pimco and get back ETF shares.
  8. The idea is, you will buy all the liquid easy-to-value stuff for $X, buy Bond ABC for 85, hand in your basket for $X + 85, get back a share marked at $X + 90, and then sell it for $X + 89 or whatever, making money for yourself and arbitraging away Pimco's scheme.
  9. Good idea!
  10. Except a funny thing happens. This strategy requires that you buy Bond ABC for 85. But no one's selling Bond ABC for 85 anymore. Anyone who was selling it for 85 sold it to Pimco. And anyone who still has it knows that it is no longer just a random orphan: It is the bond that makes this Pimco arbitrage work. You can get DEF or QRS for 85, but you don't want DEF or QRS. You want ABC. And ABC is no longer weird and orphaned. It is scarce.
  11. So if you can find any Bond ABC, you'll have to pay, like, 92 for it.
  12. Pimco has created market value in Bond ABC.

This story is surely not true. First of all, you create shares of Pimco's ETF by giving Pimco a basket of securities specified by Pimco every morning that "will generally correspond pro rata, to the extent practicable," to the stuff that's in the ETF. But only to the extent practicable. You create Pimco ETF shares in units of 100,000 shares; there are 32.8 million shares outstanding, so each creation basket has about 0.3 percent of the portfolio. If the ETF owns $500,000 worth of Bond ABC, then a creation basket should have $1,500 worth of it. It's tough to deal in bonds in units that small, so some small positions don't seem to get included in the basket. They're just rounded out of the arbitrage; you pay cash for them instead.

Even if they're not rounded out -- and the creation basket includes some wee positions -- it's still a huge pain to collect them all. A pain in the sense I mentioned above, maybe -- for the scarce ones, just the need to collect them drives the price up -- but also just an administrative pain. Even finding $1,000 face amount of a liquid bond is terrible. So in practice you'd be nuts to attempt this arbitrage. Which means that Pimco can basically value its bonds however it wants, until the SEC gets annoyed.

One thing that people worry about is that bond ETFs are a risk to financial stability. For instance, here is Canada's representative on the Financial Stability Board, worrying about that:

“They look right now that they have the liquidity of an equity, but they have the diversification of another type of asset,” she said. “The thing about them is the underlying assets can sometimes be very illiquid. What’s unclear is whether or not these ETFs in a time of stress, actually have the liquidity that’s there.”

Another thing that people worry about is that bond markets are illiquid and inefficient because there are too many little bonds issued and they're not interchangeable with one another. Here is BlackRock worrying about that.

Those two worries come together here, in a quiet way. The point of a bond ETF is, in large part, to make the illiquid liquid: to make it easy for small investors to buy and sell diversified bond portfolios in small sizes. The point of the ETF structure, on the other hand, is to use the market to prevent mispricing: The market in the underlying acts as a check on the valuation of the fund. And the point of the bond market sometimes seems to be to slice credit into tiny weird units that trade in idiosyncratic ways and reward cleverness. Those three things don't really go together. It sounds like the SEC's worry is that Pimco's ETF made the illiquid liquid, but at the cost of losing the check on its valuation. Which then provided idiosyncratic opportunities to reward cleverness.

  1. Not really, but it is copyrighted. And Pimco actually has some patented indexes, good times.

  2. Most obviously because your prospectus probably says how you'll value your holdings, and it's not like this, so if you value your holdings in a way different from what you disclosed then That Is Fraud. But even you've disclosed that you might make up the valuation, it's probably not a great idea to do this.

  3. Not really, just Authorized Participants, big financial institutions who do this on behalf of customers. But for no-arbitrage purposes, "anyone."

  4. In fact, some ETFs (though not SPY, the big S&P 500 index fund) allow creation and redemption to be in-kind (as described in the text) or for cash (participant gives sponsor cash equal to the net asset value and gets back ETF shares, or vice versa). So the easy arbitrage is:

    • buy underlying basket for $198,
    • create a unit in-kind, getting back one ETF share, and
    • redeem that share in cash for $201, the stated net asset value.

    So if you misstate your net asset value, and you allow creations and redemptions in cash and kind, then you've sort of arbitraged yourself. Pimco's Total Return ETF, unlike SPY, actually allows cash and in-kind creations so this mechanism would apply.

  5. See page 11 of this prospectus-y thing:

    For purposes of calculating NAV, portfolio securities and other assets for which market quotes are readily available are valued at market value. Market value is generally determined on the basis of last reported sales prices, or if no sales are reported, based on quotes obtained from a quotation reporting system, established market makers, or pricing services. Domestic and foreign fixed income securities are normally valued on the basis of quotes obtained from brokers and dealers or pricing services using data reflecting the earlier closing of the principal markets for those securities. Prices obtained from independent pricing services use information provided by market makers or estimates of market values obtained from yield data relating to investments or securities with similar characteristics.

    Presumably if you buy a thing at 85 then the "last reported sales price" is 85, at least initially. But then if you wait a while then "no sales are reported" and you can use the pricing service. I assume that's how that works.

  6. Also, randomly: What if they bought odd lots here and there until they added up to a round lot? That really would create value, right? Doesn't sound like that's what happened, but still.

  7. Or, since Pimco's Total Return ETF allows cash creation and redemption, you'll create shares in kind and then redeem them for cash at the NAV, i.e. at $X + 90. See pages 54-61 of the prospectus-y thing.

  8. The Bloomberg function BOND <Equity> MHD <Go> 97 <Go> will (I think!) show you today's creation basket. The biggest position seems to be $858,000 face of the 2.5 percent Treasury note due May 2024. So for a 100,000 share creation unit ($10.9 million worth at a share price of around $109), almost 8 percent is that note. But it gets as small as one bond of things like HCA's 6.375 percent of 2015, worth about $1,022.5. You also have to stump up $1.04 million in cash, presumably in part because the fund holds some cash but also in part to account for positions that are too small (or are short!) to be included in the basket as securities.

    I have to say, it looks like a huge pain to create Pimco Total Return ETF units.

  9. Especially since this is an actively managed fund and can change its holdings every day; imagine spending three days finding a bond and then realizing it's no longer in the creation basket. Basically you have a few hours in the morning to round up all the bonds you need before you run the risk that the list will change.

  10. It can do this in its -- much, much bigger -- Total Return Fund too, of course. The ETF mechanism only works on ETFs! It's hard to imagine MBS odd-lots doing much good for the much bigger fund, though. But on the other hand, do managers of smaller funds with illiquid holdings sometimes value them too aggressively? Why yes, yes they do, a whole lot!

This column does not necessarily reflect the opinion of Bloomberg View's editorial board or Bloomberg LP, its owners and investors.

To contact the author on this story:
Matthew S Levine at

To contact the editor on this story:
Zara Kessler at