CFTC Makes Sure That MF Global Won’t Be Able To Lose Any Customer Money Other Than The $1.2bn It Already LostBy Matt Levine
I find the “MF Global rule” confusing, and to understand it I have to start with some very basic basics. Let’s say I put money in my account at MF Global, which I want to hold as cash or a cash-like thing, because I need it to provide margin for my futures positions. That money is “segregated,” meaning that in some loose sense it belongs to me and not to MF Global, but MF Global can invest it. Now, because I am ignorant, I had to stop here and ask, “why do they do that?” In general, there are two plausible answers:
(1) Because they want to make money for you, or
(2) Because they want to make money for them.
Now I didn’t know which was the right answer. Answer (1) would be like your brokerage firm, which invests your cash sweep into things that make money and then pays you that money. It is a marketing thing: you’re more likely to choose a brokerage that pays you a decent return on your cash sweep. But it turns out that MF Global actually lives in regime (2): they invest your money not to market how they make money for you, but to make money for them. If MF Global puts your $100 into a popcorn popper and $101 comes out, they keep the $1.
That seems sort of risky, so there is a law restricting futures brokers to investing in things that probably don’t have much downside, viz. “in obligations of the United States, in general obligations of any State or of any political subdivision thereof, and in obligations fully guaranteed as to principal and interest by the United States.” And, because that’s no fun, there is a CFTC regulation – the now mildly infamous Regulation 1.25 – that expands that list to include a bunch of other things, including (relevantly for our purposes) money market mutual funds, repo agreements with external banks and brokers, “in-house” repo agreements with the futures merchant itself, and foreign sovereign bonds. Until yesterday, when the CFTC axed in-house repo and foreign bonds from the list as part of their Dodd-Frank get-ratings-out-of-rules push. And also not to make a big deal out of the whole MF Global thing but this is where the CFTC’s heads are at now:
The Commission believes that in-house transactions are fundamentally different than repurchase or reverse repurchase agreements with third parties. In the case of a reverse repurchase agreement, the transaction is similar to a collateralized loan whereby customer cash is exchanged for unencumbered collateral, both of which are housed in legally separate entities. The agreement is transacted at arms-length (often by means of a triparty repo mechanism), on a delivery versus payment basis, and is memorialized by a legally binding contract. By contrast, in an in-house transaction, cash and securities are under common control of the same legal entity, which presents the potential for conflicts of interest in the handling of customer funds that may be tested in times of crisis. Unlike a repurchase or reverse repurchase agreement, there is no mechanism to ensure that an in-house transaction is done on a delivery versus payment basis. Furthermore, an in-house transaction, by its nature, is transacted within a single entity and therefore cannot be legally documented, since an entity cannot contract with itself (the most one could do to document such a transaction would be to make an entry on a ledger or sub-ledger).
The least one could do to document such a transaction, on the other hand, appears to be “nothing,” which maybe was the approach MF took that ended up with them missing gobs of money.
It’s kind of obvious why in-house repo is a good deal for a futures merchant. It’s free money: if you have $100 in custody accounts, and need to borrow $100 to fund your securities positions, you can just do that. Those securities positions need to be in otherwise permissible investments (US or foreign government bonds, and until yesterday corporate bonds too), and maybe you have a bit of a repo haircut (though, weirdly, there’s no minimum haircut in Reg 1.25), and of course you need to “make an entry on a ledger or sub-ledger,” but once you’ve done that, you just take the money from one pot and put it in the other. And you don’t have to pay interest.
You don’t get too excited about not paying interest. The alternative would be (1) lend that $100 in custody accounts to a money market fund and (2) repo your securities positions with a money market fund. That’s still allowed even after the rules change, and maybe it doesn’t cost you that much: you pay a little interest on your repo, but you get a little interest from the money market fund, and since you get to keep the interest it all kind of works out. Kind of. You lose some bid/ask on the interest paid versus interest received, and there’s probably more friction on the repo haircut. It’s not “free money” in quite the same delightfully free way that in-house repo was, but it mostly nets out.
But the big difference surely is that the money market fund is going to be looking at you a lot more closely than, um, you are. The money market fund is going to keep an eye on your ability to pay off the repo, and on the quality of your collateral, and isn’t going to deal with you if you’re a shady character or if you’re hawking 30-year Greek bonds. This is true only to an approximation – you and your bonds can both be pretty highly rated even while being shady, and I suspect not every money market fund is doing immense due diligence on every repo counterparty – but it’s far more true than it is for an in-house repo, where the worse of a credit you are, the happier you are lending yourself your customers’ money.
I’m still hung up on “MF Global gets to keep the interest,” though. A system where MF Global gets to keep interest earned on customer funds is a system where brokers will be tempted to take the most risk possible consistent with the rules. Narrowing the rules to include only really truly we-mean-it-this-time safe securities is a way to address that but, y’know, who knows what’s safe any more? (Munis are still on the list – and, given Dodd-Frank, they don’t even have to be investment grade.)
But the in-house repos also remind me of UBS’s efforts to split its wealth management and investment banking businesses, which is complicated because the “increasingly close relationship we enjoy with wealth management allows its clients to more actively benefit from the full capabilities the investment bank offers,” meaning that the investment bank really enjoys selling to wealth management clients anything that it can’t sell to third parties. But with UBS, it’s not clear that the bank’s “conflict of interest” is all that bad. On the one hand, sure, the bank can sell some bad stuff to some captive clients – but on the other hand, those clients’ advisors have their own incentives to protect their clients, and they have somewhat more leverage to get information out of the bank than third party wealth advisors would have. You’ve got PWM guys who want to make money for their clients, and bankers who want to sell products, and they can fight it out with perhaps unequal power but each with better knowledge about the other’s position than most third parties will have.
That doesn’t seem to have existed with MF Global’s in-house repos: MF had every incentive to raid the customer kitty to fund its positions, and no one with much incentive to fight back. Partly that’s because MF Global had a CEO who tearfully threatened to quit any time anyone asked him if he was sure about taking on all that risk. But it’s also because the investing-segregated-funds function wasn’t a client money-making function. No one at MF Global was looking for ways to make the most interest with the least risk for its custody clients, or getting paid a commission based on how customer segregated funds performed for clients, because that was not a service it provided those clients. The rules say that the segregated accounts belong to customers, but the clients can suffer only grief from them: any profits on that money actually belong to MF Global.
I still don’t have a fully worked out transaction cost economics of when it’s better to let firms screw their clients internally rather than outsourcing it, but surely a basic rule would be “conflicts of interest within firms should only be allowed if there’s a genuine conflict.” When investment banks underwrite IPOs, and bankers want high prices for their issuer client while salesmen want cheap deals for their investor clients, that’s a manageable conflict: everyone represents one firm, but the bankers and salesmen have reasonably complete information and opposite incentives and they can negotiate out a result that sort of protects both clients. But where futures brokers want free money from their clients, and no one has any reason or way to get paid for making money for those clients, then the client has no chance. And if that’s how the incentives work, then banning the futures broker from being on the other side of the trade makes a whole lot of sense.