RBS Promises Not to Trade Against Clients Too Much

A thing that happens a lot is that a client comes to a bank and is like "I would like to sell you some X," and the bank says OK, and then the bank goes and sells some X to other people before or after or at the same time as it buys the X from the first client.

A thing that happens a lot is that a client comes to a bank and is like "I would like to sell you some X," and the bank says OK, and then the bank goes and sells some X to other people before or after or at the same time as it buys the X from the first client. That schematic generates a surprisingly rich set of possible scandals and non-scandals, and it can sometimes be difficult to understand which is which. Sometimes the bank's selling is called "market making" or "hedging," and that's more or less fine. Sometimes it's "front-running" or "manipulation," and that's more or less not.* Other variants are possible.**

Here's a Bloomberg News story about WM/Reuters foreign exchange benchmark maybe-manipulation, which is a lovely example of this fact pattern and a genuine puzzle. Here's the story. Clients want to trade currencies with banks at the WM/Reuters fix, a benchmark price set at fixed times each day. So they come to the banks before the fixing and tell them how much they want to trade. This makes sense: A bank's not going to let you trade at the 4 p.m. price at 4:30,*** and if you call them at like 3:59 and 58 seconds they'll probably be busy, but if you come to them at 3:30 then they'll agree to lock you in at the 4 p.m. price, whatever it is.

Now the bank wants to hedge. The simplest way to hedge is to trade the opposite way -- if the client is buying Swiss francs from you, you buy Swiss francs from the market -- at exactly 4 p.m., so that you buy at the WM/Reuters fixing price and sell to the client at the same price.

There are two problems with this. One, while the bank is guaranteeing the client the WM/Reuters fix price, nobody's guaranteeing the bank anything. The bank has to actually go trade and try to hit that price. The way the fix works is that it's set by sampling trades over a 60-second window, so you have to have a certain amount of skill and luck to trade at (or better than) the official price. It's risky.

Two -- and this is important, too -- banks are in the business of making money, and the trade they want is not "buy at WM/Reuters fix and sell at WM/Reuters fix" but rather "buy at less than WM/Reuters fix and sell at WM/Reuters fix."

If your client tells you at 3:30 that they want to buy francs at the 4 p.m. price, there are fairly straightforward ways to try to ensure a profit. Like: If they're buying a lot of francs, then you'd expect that to push the market for francs up. So what you could do is start buying francs at 3:30 when they're trading at $1.1150 or whatever,**** and keep buying steadily until 4 p.m. when they're trading at $1.1200, for an average price of like $1.1175, and then sell to the client at the 4 p.m. fix of $1.1200. You buy francs for $1.1175, you sell them for $1.1200, you make $0.0025 per franc, you do it a lot, boom, good business.

This is not necessarily easy because there are no guarantees that the franc will go up; maybe another bank's client is selling a lot of francs or whatever. Maybe your best bet is to try to do all your buying (hedging) right at 4 p.m. Maybe it's to wait until after 4 p.m. because you think the franc will weaken. I don't know, that's why you're paid to trade, right? You're supposed to have a sense of supply and demand and short-term price moves and how best to make a market for clients while making a profit for yourself. That's what they pay you your fractions of a penny for.

But that aside, it's easy to make it look bad. Because look at what I described: Client comes to you at 3:30, when francs are at $1.1150, and says he wants to buy francs at 4 p.m. You go buy a lot of francs before 4 p.m. What is that? Well, I mean, you're buying the francs for your own account -- you're not selling to the client at the price you paid, you're selling to the client at the WM/Reuters fix***** -- and you're doing it after you get the client's order but before you sell to him. That looks like front-running, no?

Or: You're buying a ton of francs just before the fixing, and the price is going up, partly because of your buying. And then the fixing -- the price at which you're selling to the client -- is high. Your buying has pushed up the price at which you sell. That looks like market manipulation, no?

And, I mean, no, it's hedging your market-making obligations, it's totally legitimate, is a thing you can say, but I doubt that there's a neat answer. It's a question of disclosure and conventions. If clients understand that the purpose of telling you their order half an hour in advance is to let you get set up for the fixing -- that is, to let you hedge -- that is, to let you trade ahead of them -- then it's fine. If clients think that you are putting their order in a secret locked box and not doing anything with it until 4 p.m., then you're deceiving them.

Also some things are right out:

In June, Bloomberg News reported that dealers pooled information about their positions through instant messages, executed their own trades before client orders and sought to manipulate the benchmark WM/Reuters rates by pushing through trades around the 60-second windows when the benchmarks are set.

The sharing client data among dealers seems self-evidently bad. The "executed their own trades before client orders and sought to manipulate the benchmark" is either bad, or just legit hedging that looks bad, or something in between, or some combination thereof, it is hard to know.

Anyway today's story is about how RBS is trying to help clients understand:

RBS's foreign exchange sales team contacted some clients, pledging the Edinburgh-based lender won't share details of their orders or use them to make proprietary bets, according to an Oct. 23 e-mail, which was read to Bloomberg News.


RBS told clients it would start "pre-hedging" orders up to 15 minutes before the benchmark is set to protect itself against market movements.

"You should also be aware that this could potentially result in the market moving against you," the bank said in the e-mail. "However, we will take steps to minimize the market impact within the above context notably by executing the order over a longer period of time."

So I mean. That's a solution. It's a good one, even. Tell your client when you might be in the market ahead of them, make that period not too long (front-running!) and not too short (manipulation!), set some clear guidelines, and try not to do anything too obviously shady. But it's also a good illustration of how thin the line between scandal and normal -- between manipulation and hedging -- can sometimes be. Here, it's apparently the 15 minute mark.

* A helpful but fallible guide is: If the bank trades before dealing with the client, that often looks shady; if it trades after dealing with the client, that often doesn't. That guide would I think lead you astray in this particular story but depends on where you think "astray" is.

** Abacus, for instance: John Paulson comes to Goldman Sachs and is like "I would like to sell you some mortgages" (synthetically). Goldman says fine, pre-sells the mortgages (synthetically) to ACA and so forth. Scandal, to a jury, though reasonable minds disagree.

*** I mean, y'know. Unless the price has moved in your favor. But probably trading on a stale price that favors the bank isn't a great idea either.

**** I realize this is not the quoting convention but bear with me, it's easier to read this way.

***** Which by the way is what he wants, because it's the official-ish benchmarky price. He could just ask you to sell at your execution price plus a commission if that's what he wanted.

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

    Before it's here, it's on the Bloomberg Terminal.