From Joy Rajiv, who formerly worked in the Foreign Exchange trading divisions at both Morgan Stanley and Deutsche Bank. He left the industry in early 2013 for personal reasons unrelated to the current regulatory probe into the FX industry, and writes in a private capacity here drawing on his experience in the industry.
As someone who has worked in FX trading for three years at Morgan Stanley and Deutsche Bank from 2010 to 2013, I have been dismayed and discouraged by the recent coverage of alleged manipulation by FX traders at major banks. Traders have been fired, lawsuits have been filed and comparisons to Libor have been thrown around without much concern for detail. Media reports have focused their attention on abuse of a daily benchmark, called the WM/R (World Markets/Reuters) fix.
They’ve also been quick to use terms such as “banging the close” to explain the nature of the manipulation. My goal in writing this article is to explain why such comparisons are unjustified and why many of these allegations are unfounded.
WM/R rates for active currencies are published by Reuters every 30 minutes. Out of all of these, the rate published at 4pm London-time is the most used and most important. Unlike Libor, which is based on trader submissions about where banks are prepared to trade, the WM/R fix is based on real trades that happen in a one-minute time frame around the 4pm fix time.
To come up with the rate, exchange data is collected at one second intervals from two exchanges, EBS and Reuters and the median bid and offer observed during that time becomes the WM/R fix rate. Bank clients conducting FX operations like to use this rate because it means they get the same deal as everyone else trading at that time.
When a bank accepts a fix order, they do so as a market maker. That means they are not just matching the buy or sell order with a similar order on the other side, as a broker might, but acting as a principally-exposed go between. This involves taking some degree of risk on their part.
If, for example, they accept a ‘fix’ buy order from a client before the fix, all they know is that they will have to provide the client with the WM/R rate at the point of settlement, regardless of whether they are able to execute that order in the market before the fix, at the fix or even after the fix at the WM/R rate, worse, or better. These conventions mean that fix trades are very different to the FX trader’s usual business of making markets to clients by quoting a spread around the market mid rate.
The all important quoted spread in those trades compensates traders for the risk they’re taking by accepting the flow onto their own books. In contrast, for fix trades traders do not have the freedom to quote a spread. From their point of view, the fix is a bit like agreeing to buy bananas from someone at a market determined rate, which they do not know yet.
Technically there is a fixed spread that the trader theoretically makes on a fix trade since the published WM/R rates has a bid and offer side. But unlike a non fix trade, on a given day this spread is always the same regardless of trade size, market volatility etc. So the trader is not adequately compensated for the risk he is taking on a fix trade.
The traders try to control risk on a fix trade by processing the trade within the fix time window to ensure the slippage between their executed price and the published WM/R price is minimal. This explains the reason why large FX volumes are executed around the fix time window.
Traders are not ‘banging the close’ as is being implied in some reports. Their behavior is fundamentally different from typical banging the close behavior where there are no clients involved and a trader deliberately (and illegally) executes a large futures trade to influence a closing rate that benefits his proprietary position.
The trader’s inadequate compensation on fix trades starts to matter when the size of the fix trade becomes large. For example assume client A has 1m EURUSD to sell and client B has 1bn EURUSD to sell at the fix. From the trader’s viewpoint client B’s order poses much larger risk than client A. To make this difference explicit assume it’s a volatile fix and the trader misses the WM/R rate by 10bps.
This implies the trader incurs an instant loss or gain of $1,000 on client A and $1m on client B. The two trades pose different risks and yet the trader has no freedom to charge different prices to each client reflecting the asymmetry. This is not a fair deal for the trader and clearly client B has a large edge over the trader on the fix. Given this, it’s natural to wonder about the trader’s options if he has a large fix order to execute. Not surprisingly he does not want to play a modified version of Russian roulette and execute all of it in the fix window. The only option to control risk around the fix window is to execute part of the fix order in advance.
Allegations about FX traders colluding to buy/sell fix orders in advance of the fix deserves to be studied in this context. Given the size and depth of the FX markets it would take collusion of inordinate magnitude for such a strategy to work reliably and consecutively. The spot EURUSD market is the biggest market in the world. A conservative estimate using volumes from banks and exchanges alone implies that the daily volume turnover in EURUSD is in excess of €200bn. In comparison the largest future (e-minis on CME) and stock (perhaps BofA) have much, much smaller turnovers of $4bn and $2bn respectively.
The closest competitor in size across all markets to the EURUSD market is the USDJPY market. So someone who has a billion EURUSD to execute can perhaps influence the market for five minutes on some days and for a few seconds on other days. Skilled traders at large banks maximise the likelihood of success with their market awareness of flows, positioning etc. But like everyone else, whether they collude or not, they too are taking a chance if they trade ahead of a fix.
They will give up quickly when they sense a larger unknown presence in the market.There are no market guarantees meaning unlike with Libor, there’s everything to lose.
What these investigations really show is that traders who execute large sized fix orders face serious challenges. On some days the currency rate can swing aggressively during the fix window, meaning traders can’t even guarantee that they will break even on the trades they accepted and executed during the window.
An FX prisoner’s dilemma
Considering the chances of a large difference emerging between the fix and the execution price, every trader who enters the fix window with a large order to execute is automatically exposing himself to risk. It makes much more sense to break these orders up. Deliberate or unwitting collaboration of the prisoner’s dilemma variety focused on minimising risks is arguably very different to explicit for-profit manipulation. After all, if a trader is not being paid the correct risk premium on a large fix order, how else can he manage the risk in question?
Can this behaviour be termed manipulation in the same sense as the Libor case? Or is it simply prudent trading? Surely the size of the FX market, the large market risk the trader is taking and his reasons for doing so have to be taken into account.
The investigations also show a need to re-evaluate the benchmarking process. The popularity of the WM/R fix and the current controversy has unexpectedly highlighted its weaknesses. The time is perhaps right for clients, banks, exchanges and the benchmark setters to discuss overdue options to improve the benchmarking process designed in the 90s.
Worthwhile topics for discussion may include: a) using multiple benchmarks across the day to minimise the importance of one benchmark, b) widening the fix time window from one minute to say 10 minutes to reduce slippage risks for the trader, c) laying down guidelines for acceptable trader behaviour across benchmark windows etc.
Having said all of this, any claim of manipulation is serious and deserves to be investigated to preserve market integrity. If the current investigation results in improving benchmarking processes and in defining best practices for a trader it will be a good end result. But if it results in a litigation and regulation overkill and a few careers being sacrificed, then the end result will be a lower quality FX market.
The inevitable consequence of that is anyone who uses FX will pay higher costs due to wider spreads, lower liquidity and lower quality of personnel in the industry. Traders will have to factor the cost of processing the fix into the spreads they charge elsewhere. Worse than that, possibly stop providing the fix service altogether?
Is that satisfactory? If the answer is no then it may be sensible for everyone to dial it down, drop the automatic presumption of guilt, stop making ill informed comparisons to previous scandals and wait for regulators to investigate matters quietly and thoroughly so that facts not rhetoric dictate the outcome.