Eric Schneiderman is impressed.

Photographer: Mark Wilson/Getty Images.

Bank of New York Mellon's Best FX Execution Was Pretty Bad

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

Bank of New York Mellon provides custody services to a lot of big investor clients, pensions, endowments, etc. When those custody clients have foreign exchange needs -- to buy or sell foreign securities, for instance -- they come to BoNY Mellon. This creates a nice opportunity for efficiency: BoNY Mellon has a lot of clients who want to exchange dollars for foreign currency, and a lot of clients who want to exchange foreign currency for dollars, so it can pair them off against each other, operating its own little internal foreign-exchange trading venue.

Or it could do this:

Throughout a trading day or session (which could be as long as 24 hours), as each custodial client's account generated FX transactions to be executed pursuant to SIs, the Bank's practice was to aggregate those FX transactions for all SI clients and group them by currency pair. Near the end of the trading day or session, the Bank priced those SI FX trade requests it had received throughout that day or session.

So that seems a little inefficient -- why wait until the end of the day to price the trades? -- but it's basically fine. Sometimes the currency goes up, sometimes it goes down, in expectation clients shouldn't lose much by waiting. Everyone trades at the end-of-day fix, and in the long run it evens out. Right?

Nope!

To determine the price for each SI FX transaction for most currencies, the Bank examined the range of reported interbank rates from the trading day or session and assigned the rate on SI trades as follows: if the client was purchasing foreign currency, the client received a price at or close to the highest reported interbank rate for that day or session (at or near the least favorable interbank price for the client reported during the trading day or session), and if the client was selling foreign currency, the client received a price at or close to the lowest reported interbank rate of the day or session (also at or near the least favorable interbank price for the client reported during the trading day or session).

Because SI clients received pricing at or near the high end of the reported interbank range for their currency purchases and at or near the low end of the reported interbank range for their sales, the Bank was generally buying low from, and selling high to, its own clients. The Bank recorded the difference or "spread" between the rates it gave clients and the interbank market price at the time the SI transactions were priced as "sales margin."

That's probably the best use of scare quotes around the words "sales margin" you're ever likely to see. This is so bad! Come on! BoNY Mellon netted off its clients against each other, giving the buyers the highest price of the day and the sellers the lowest price of the day, and then pocketed the difference. And then, to add euphemism to injury, called it "sales margin."

Now. There are hypothetical defenses of this practice. (I sort of offered one, once, when the case against BoNY Mellon was first announced.) The main hypothetical defense goes like this: The client doesn't have a view on intraday FX rates. In particular, the equity manager or pension treasurer or whoever is actually setting up the "standing instructions" with BoNY Mellon has no view on intraday FX rates. For all he knows, the price when he puts in the order will be the worst price of the day. What he wants is not really to get an efficient instantaneous price. What he wants is to not pay for the service: He is measured, not on getting the best rate, but on his measurable commissions and fees, and if he reports back fees of zero then he wins. He wants "free" foreign exchange trading, and knowingly sacrifices efficiency for a price of zero. 

This would not be a particularly compelling defense even if it were true, but it also turns out to be so, so, so not true. Bank of New York Mellon today settled with federal and New York prosecutors over this stuff, which ran from 2000 through 2011; BoNY Mellon is paying $714 million and will "terminate its employment relationship with certain executives" who were responsible for it. It also agreed to some stipulated facts that are bad. They are bad. The ones above are bad, with an elegant mathematical badness, but they offer a little room for defense. But then you get BoNY Mellon's marketing claims about this product -- the product, remember, that was "we give you the worst price of the day and pocket the difference" -- and they go like this:

  • "Understanding the fiduciary role of the fund manager, it is our goal to provide best execution for all foreign exchange executed in support of our clients' transactions."
  • "Best execution encompasses a variety of services designed to maximize the proceeds of each trade, while containing inherent risks and the total cost of processing."
  • "We price foreign exchange at levels generally reflecting the interbank market at the time the trade is executed by the foreign exchange desk."

And so on. Maybe best of all:

  • "Since The Bank of New York Mellon is one of the largest global custodians, our clients gain the ongoing benefit of aggregation of transactions across our broad customer base; accordingly, we price foreign exchange at levels generally reflecting the interbank market at the time the trade is executed by the foreign exchange desk."

Like: BoNY Mellon called attention to the fact that its big custody business let it match clients up against each other. The obvious implication was: "This lets us give you efficient execution." It would be sort of perverse to read it to mean: "This lets us give you maximally inefficient execution, because we match you up against our other clients but give you each the worst possible price." But that's what it meant!

That's actually a theme of BoNY's claims: They're not exactly lies! The levels really did "generally reflect" the interbank market at the time of the trade, insofar as they incorporated that market into the daily hunt for the worst possible levels. (The formula for "price client gets" involved reference to "interbank market at the time"! Just not in, like, a linear way.) And look at that vague description of what "best execution" meant to BoNY Mellon. Oh, best execution is a "variety of services," is it? The prosecutors absurdly felt the need to spell out, in the stipulated facts, that BoNY Mellon and its executives knew that "Many market participants equated 'best execution' with best price or considered best price to be one of the most important factors in determining best execution." Otherwise BoNY Mellon could have argued that guaranteeing clients the worst price of the day was consistent with "best execution."

A rich vein for financial enforcement is "free" products. Any time a bank is doing something for free, it is somehow making money off of trading dynamics that are not reflected in the sticker price. (Consider the bigger FX manipulation scandal, or, at a smaller scale, the fight over Charles Schwab's robo-adviser.) This is not necessarily a bad thing. Sometimes clients really do want to give up some trading efficiency for a zero price, and do so knowingly. That often makes sense, because the banks are more efficient traders than the clients: The client can't make money on intraday FX volatility, so it give the bank the right to try, in exchange for a lower (zero) sticker price.

The problem is, the bank isn't just better than the client at making money on these trades. It's also way better informed about how it makes money. And sometimes a bank will succumb to the temptation to keep that information to itself. And to cover it with euphemisms and quarter-truths, hoping that to keep the client in the dark. That mostly ends badly.

  1. "SI" stands for "standing instructions," BoNY Mellon's name for a particular category of foreign exchange service.

  2. First of all, because it sort of relies on agency costs within the customer. It envisions a pension manager who prefers to look good to his bosses, over actually providing them good economics. The bank shouldn't help that guy deceive his principals.

    Second, because it is an obviously bad-value trade. If a bank charges you $1 in cash for a service, then the service costs you $1. If the bank instead charges you $0 for the service, but makes it up by trading volatility that you can't monetize, then you can be sure that the bank is monetizing more than $1 worth of volatility. The bank knows the price of volatility better than you do. Meaning that you'd be better off paying the $1. The moral is, never trade options. But certainly never give the bank an option that you don't even know is an option.

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:
Matt Levine at mlevine51@bloomberg.net

To contact the editor on this story:
Zara Kessler at zkessler@bloomberg.net