Verizon: Attractive bond deal, ugly building. Photographer: Ron Antonelli/Bloomberg
Verizon: Attractive bond deal, ugly building. Photographer: Ron Antonelli/Bloomberg

If capital markets were totally efficient there probably wouldn't be much work for investment banks. For instance, if you were a company and you wanted to sell bonds, you'd just go to the efficient bond-selling place -- it'd probably be a computer system, realistically -- and put in an order to sell bonds, and then whoever was willing to pay the most for them would buy them, and that would be that.

That is not that! What you do if you want to issue bonds is, you go to a bank, and the bank markets the deal and negotiates a price with a bunch of investors, and then you issue the bonds to investors selected by the banks, and you pay the banks a chunk of money, and then the bonds tend to go up in value, so you've effectively paid the investors a chunk of money. You effectively sell bonds to the banks at 99 and they sell them to investors at 100 and those investors trade them a bit and they settle at 101.1 And you're like, why didn't I just sell them at 101, which is the market clearing price?

Actually, that's mostly not a question you're asking, if you're a bond issuer, though it occasionally comes up. But it is a question you're asking if you are, for instance, an investor not selected by the banks to pay 100 for bonds worth 101. And it is apparently also a question that regulators are asking. Actually a huge portion of writing and thinking and regulating about financial markets consists of asking, "well, why are there banks?," though the question is phrased more neutrally as, "Why do banks do this archaic and inefficient thing instead of just letting markets sort themselves out?"

It is not, per se, a terrible question. In this particular case, the regulators are worried that banks are allocating big chunks of exciting bond offerings to the banks' good clients, instead of to the banks' bad clients. This either is, or is not, okay:

Smaller money managers that trade less frequently are often excluded from gains that new deals can generate, according to interviews with more than a dozen investment firms. The U.S. Securities and Exchange Commission is investigating the way the biggest banks allocate corporate-bond offerings and whether they are giving preferential treatment to certain clients, according to a person with direct knowledge of the matter.

This matters because new bonds tend to go up. Peter Tchir looked at the numbers and found that 87 percent of new bond issues went up, with an average one-day return of 0.5 percent. He concluded that U.S. bond syndicate desks have allocated about $800 million to $1.5 billion of wealth so far this year, so figure at least $5 billion annualized that banks can just give to favored clients in exchange for ...

In exchange for what? I have to say that I do not entirely understand this investigation, which as far as I can tell is still in the vague stage.2 My understanding of how bond allocations work is that the banks and, sometimes, the issuer, get together and divide up the bonds, giving lots of bonds to good guys, fewer bonds to less good guys, and no bonds to bad guys. Also few or no bonds to unimportant guys. So if you're a good guy you get your share of that $5 billion a year; if you're not, you don't.

This is sort of by its nature a fuzzy process.3 What makes you a good guy? One set of good-guy things you can do is be good for the issuer of the bonds. So bond issuers want some things from their investors: To be relatively supportive of managment, say, or not a pain to deal with if there are any problems in the future. Also, they typically want their investors to hang on to their bonds, or even buy more, so that the yield on the bonds stays low. Because bond issuers tend to issue more bonds, and the lower the yields on their old bonds, the more cheaply they'll be able to borrow in the future. So long-term, buy-and-hold, management-friendly investors who are likely to buy more bonds in the aftermarket tend to get better allocations than irritating activist fast-money investors.

Another set of good-guy things you can do is be good for other issuers of bonds. Most new bonds are good for investors. Some are bad. The more helpful you are with the bad ones, the more issuers, as a class, owe you:

“Allocations always come down to favours,” says one former banker. “If an investor comes into [agrees to buy] a private placement that helps get it over the line, then you hook him up on the next few hot deals.”

So if you help a shaky issuer do a dog of a deal, you should be rewarded by a good issuer giving you a big allocation of an attractive deal. This is not a request you can make directly of an issuer: What does Verizon care that you helped some crummy company get a bond deal done last week? (Verizon's $49 billion bond offering in 2013 may have triggered this investigation.) But it is a request that you can make of a bank: Verizon's banks care very much that you helped out on the last deal.

Now this arguably rips Verizon off a bit -- they're effectively subsidizing someone else's deal -- but they don't seem to be the ones complaining. And it's not like they've never issued bonds before. They understand that they're paying banks, and also paying a new-issue concession to investors, in exchange for being a part of this system of favor exchanges. Verizon's 2013 deal went great, but it was also the largest corporate bond deal ever, and you could imagine the company being nervous about getting it done. Some deals are Twitter, others are Facebook, and you don't necessarily know which is which until after you're done. If you turn out to be Facebook, you'll be glad that your bank can call in some favors to get the deal done. If you turn out to be Twitter, well, there are worse outcomes imaginable.

A final set of good-guy things you can do is be good for the banks. The banks make a lot of money on secondary trading of bonds, or used to anyway. The more of that trading you do, the happier the banks will be with you, and the better your allocation will be:

“Investors aren’t being treated equally,” said Jeffery Elswick, director of fixed-income at San Antonio-based Frost Investment Advisors, which oversees about $5 billion in fixed-income securities. “We’ve had a broker-dealer tell us there’s nothing we can do. We’ve had another say that unless we did much more volume with them in the secondary market, there was very little we could do.”

All three of these good-guy things tend to favor big investors, by the way: Pimco and BlackRock tend not to be annoying activists, are always on the list of people you'd call to get the next difficult deal done, and also trade a lot. So it's no surprise that they get big allocations: In the 2013 Verizon deal, Pimco got $8 billion and BlackRock $5 billion of the $49 billion total.

There are various possible reactions you could have to this list. One would be to say: Look, this is a web of relationships. It is not entirely efficient or fair, but the mutual back-scratching is what allows deals to get done with much more certainty, and much less risk and volatility, than a purely "efficient" and transparent auction-y system.4 The banks are in a relationship business, and the relationships are an important part of getting the actual business done.

A second would be to say, okay, I get the web of relationships where banks allocate deals to help their issuers. I even get the web of relationships where banks allocate deals to help other, future issuers. But the thing where banks allocate deals just to help their own secondary trading operations: That seems wrong. And if you think that, I mean, okay, fine. That's not totally unreasonable.

A third reaction would be to say, no, no, this is all corrupt medieval nonsense, get rid of all of it, allocate bonds to whoever is willing to pay the most for them, the end, give the little guy a shot. You could imagine that being the SEC's reaction, since it tends to like level playing fields. The fact that bond issuers tend not to choose the pure-auction model, but instead let those corrupt banks go around allocating their bonds, suggests that there's something wrong with that analysis.

The interesting thing is that the law doesn't give you much guidance on which reaction is right. The IPO allocation process went through a similar investigation a decade ago, and the settlements from that investigation are about general violations of rules like the old NASD Conduct Rule 2110, which requires banks "to observe high standards of commercial honor and just and equitable principles of trade." That can cover a lot of ground, and in fact the IPO settlements forbade conduct that was purely bank self-serving (requiring aftermarket trading through the banks), conduct that served issuers broadly (rewarding investors who bought "cold" IPOs with shares in "hot" ones), and conduct that might even have served the specific IPO issuers (buying more shares after pricing). The law depends on what you -- what the SEC, I mean, and maybe a judge -- think is fair and just and equitable. And that depends a lot on what you want to get out of your capital markets.

1 Those numbers are made up for ease of use but they're sort of the right order of magnitude. Peter Tchir calculates that the average 10-day gain for recent new issue bonds is 0.91 percent (of price), so 100 to 100.91, round to 101. And banks' fees are all over the place, with investment grade gross spreads typically under 1 percent and high yield fees typically over. The Verizon bonds, which are of particular interest to regulators and whatnot, came with underwriting discounts of 0.3 to 0.75 percent. And, per Bloomberg/TRACE, they traded up a lot on the first day -- to 103 and change, for instance, for the 10-year, or to 105.7 for the 30-year. So Verizon effectively sold the 30-year to the banks at 99.25, which sold to investors at 100, who ended up with bonds worth 105.7. (Actually the public price was 99.883 but we are just going to let that go.)

2 The world became aware of it because of the insertion of the phrase "allocations of and trading in fixed-income securities" into an already hilariously enormous list of investigations and lawsuits in Goldman Sachs's 10-K last week (page 224, if you're interested).

3 I would not go so far as to say it's "one of the few unregulated areas that remain within banks," but this guy would, and I don't entirely disagree with him:

"This is one of the few unregulated areas that remain within banks, especially post-financial crisis,” said Adrian Miller at GMP Securities. “Scrutiny on all sorts of dealings within banks has increased, and it will be interesting to see what happens if regulators start to increase oversight of syndication desks.”

4 I've endorsed that Epicurean Dealmaker post before but it is a great explanation of what the bookbuilding process is. It is perhaps more relevant for equity IPOs than for bond deals, though. Bond deals you often kinda know what you're getting.

To contact the writer of this article: Matt Levine at mlevine51@bloomberg.net.

To contact the editor responsible for this article: Tobin Harshaw at tharshaw@bloomberg.net.