Is redemption in reach for FX?

For much of the post-financial crisis period, participants in the foreign exchange (FX) market looked at the fate of other asset classes and counted their blessings. Unlike their peers in fixed income or equity trading, they found themselves unaffected by the massive new capital requirements installed by regulators. FX remained a largely unfettered activity, free from the taint of Libor-like scandals.

In the past year or so, however, it has been brought back down to earth with a bump. The prolonged low-interest-rate policies pursued by many major central banks have depressed volatility and profits for market-makers, and the growth of high-speed trading technologies has squeezed margins even further. Allegations of FX benchmark fixing have cast a pall over the business, leading to calls for improved conduct and greater transparency from market-making banks.

Push for new rules

FX is now firmly in the regulatory crosshairs, and will form a major part of the Bank of England’s fair and effective markets review (FEMR). On a global level, manipulation of the benchmark rate set everyday at 4pm Greenwich Mean Time by WMR/Reuters is still the main target. Dozens of traders have already lost their jobs as a result of investigations into this matter by multiple regulatory agencies. In November 2014, six global banks were hit with collective fines running into the billions of dollars by UK, US and Swiss regulators.

In September last year, the Financial Stability Board (FSB) published a series of recommended reforms to the FX benchmark system. These included widening the window during which the ‘fix’ is set, introducing a pricing system for benchmark orders, which banks have historically executed for free on behalf of clients, and improving the transparency of banks’ benchmark order execution processes.

For Guy Debelle, assistant governor of the Reserve Bank of Australia and co-chair of the FSB review into benchmarking practices, progress on some these reforms has not been swift enough.

“There has yet to be significant progress in terms of the pricing and execution of the fix business within institutions,” he said in a speech on February 13. “There is a strong expectation that these recommendations will be implemented to deliver an improvement in the execution of FX transactions referencing FX benchmarks and the integrity of the benchmarks themselves.” He went on to warn that if progress is not made quickly, the FSB’s recommendations could soon turn to set-in-stone requirements.

Mr Debelle’s remarks are echoed by buy-side users of the fix. “The change has been slower than I expected. The FSB recommendations came out seven months ago, which is long enough for significant steps to have been made,” says the head of trading at a large European asset management fund.

Asset managers, faced with the problem of tracking errors between valuation rates and actual transaction rates when rebalancing or transferring portfolios, are invariably the biggest users of fix execution. Banks receive benchmark orders in advance, and guarantee to transact them at the exact valuation rate calculated during the window itself. Tracking error is therefore theoretically eliminated.

Price hike

Duly chastened, the industry has begun to get in line over the past couple of months. On February 16, WMR/Reuters moved the 4pm London fix from a one-minute window to a five-minute window. A number of banks have also made moves towards a proper pricing policy for benchmark trades.

“Banks currently differ in their approach to pricing. Some still offer execution at the mid-price, effectively for free. Some now charge a bid/ask spread, as is generally applied to non-benchmark orders. Others have decided to charge a flat fee for any benchmark orders. We have no specific preference for a particular approach, we’re just happy to see momentum gathering at last,” says the European asset manager.

It may seem odd for clients who were previously getting something for free to now be keen to be charged for it. Research conducted by independent benchmark provider New Change FX shows that banks pricing at the bid/offer adds a cost of $280 per $1m traded. That does not sound like a great deal, but fix orders from individual clients often run into the hundreds of millions of dollars or more. If a single client executes $500m of fix orders per week, that equates to an annual cost of $7.28m.

However, for Trevor Charsley, senior advisory consultant at AFEX, a London-based non-bank FX solutions provider, a pricing change makes perfect sense for participants on both sides of the trade. “Offering benchmark execution as a free service never made much sense. It gave market-makers an incentive to manipulate the fix for their own gain as they weren’t making any other profit from it. In fact, it often cost them money to offer it free. Asset managers are getting a valuable service from banks during the benchmark window, so that value should be reflected by some kind of fee.”

The existence of legacy benchmark execution contracts between banks and their clients makes this pricing transition a more drawn out process than some would like. A senior manager at one European bank admits that conversations around benchmark pricing with some clients have been hard, with many resisting the change. “The discussions have often been as difficult as those we have with clients at the peak of the crisis around liquidity funding costs,” says the banker.

Window of opportunity

Alongside these structural changes, regulators are expecting significant improvements in bank conduct around the fixing window. For instance, some banks are now constructing ‘Chinese walls’ to prevent collusion between spot traders executing benchmark orders and those looking after ordinary client flow or the bank’s proprietary trades. Benchmark traders will be physically sealed off on the trading floor, with their own execution systems that have no access to information from the rest of the bank’s order book.

Software can also be introduced to the order execution system that denies access to non-benchmark traders during the period of the fix.

“More than any specific structural changes, we need full disclosure of banks’ behavioural changes to the benchmarking process, such as the ring-fencing of benchmark execution. That will provide a better insight into the value delivered by the fix,” says the head of trading at a European asset manager.

James Kemp, managing director at the Global Financial Markets Association’s global FX division, is open to the idea of regulators stating more clearly their preference for specific conduct changes. “Given the fragmented different approaches to 'separation' of benchmark execution services currently developing across banks, it would be helpful for supervisors to provide further interpretive guidance over what its final form should ideally look like,” he says.

An unfixable fix?

For some, the entire concept of a point-in-time FX benchmark has had its day. No amount of tinkering with behaviour or processes will change the fact that the massive improvements made to FX trading speeds in recent years leave asset managers executing at the fix defenceless against other participants who can make money out of the market impact of benchmark orders.

By joining liquidity pools provided by individual banks or interbank trading platforms, high-frequency algorithms operated by hedge funds or currency managers can detect pricing trends produced in the wake of large benchmark orders as they hit the market. They can then ‘front run’ these orders, fractionally pushing up the price of certain currencies before a more cumbersome asset manager has finished buying his selected amount.

The technological disparity between the two types of participant borders on frightening. The latest FX trading software makes it possible for advanced algorithms to execute a trade every 740 nanoseconds (a nanosecond being one-billionth of a second). That produces 1.35 million trades per second, and 405 million trades in the new five-minute window allotted for the 4pm WMR/Reuters fix. Most asset managers, on the other hand, are running bank-provided algorithms that buy or sell a chunk of the benchmark order once every second across the five-minute window. The information leakage between these second-by-second executions is massive, allowing faster algorithms to move in and move the market against a large order.

“It seems that a lot of asset managers are unaware of this issue,” says Andy Woolmer, managing director of New Change FX. “Widening the window to five minutes has not reduced the odds of manipulation, as it has increased the opportunity for high-frequency traders to make money at the expense of asset managers. The slower the client is forced to transact, the better for the high-frequency trader. From the asset manager's point of view, this reform has made an extremely poor benchmark even worse.”

Analysis put together by New Change FX for one pension fund estimated that, independent of any extra price charged by banks, it was losing money on the fix at a rate of 8 basis points per trade. Armed with this information, asset managers appear to face a choice – either fight back with high-frequency trading software of their own, or ditch the traditional point-in-time benchmark in favour of a live benchmark constantly updated with fresh market prices.

Looking for trouble

The spotlight placed on FX by the benchmarking scandal has illuminated other practices that concern regulators. Once an unremarkable part of the spot FX market architecture, the practice of ‘last look’ is now under scrutiny by the FEMR, amid claims that it systematically advantages price-makers at the expense of price-takers.

Last look evolved in response to the vastly differing quality of technology used by participants in the early days of electronic execution. If a client used slow trading software, or was on the other side of the world, an appreciable gap could develop between the confirmation of a trade and its actual execution. The same problem can occur when a bank streams prices to two different trading platforms if one operates at a slower speed than the other. Last look allows the market maker to cancel a trade if the pricing moves in an adverse direction during this period.

In the modern trading world, with trading latencies no longer counted in minutes or seconds but in milliseconds, this practice strikes many participants as antiquated at best, and outright unfair at worst. Banks or other liquidity providers that systematically use last look can theoretically provide tighter prices because they limit their exposure to random market-moves, mid-trade. However, they also complete a smaller percentage of client orders and can offer prices purely to seek market information, without any intention to trade. Clients with rejected trades, having already revealed their intentions to the market, can find that prices move against them when they re-submit orders to different counterparties.

The FEMR recognises the practice as potentially problematic, and many in the industry believe it is extremely prevalent among the biggest FX market makers. Some buy-side firms say they would be prepared to foreswear last look even at the expense of increased pricing. “Our preference in the FX market would be to a view on the liquidity in which we can deal, even if this comes at a higher cost compared to the ‘phantom liquidity’, which can be removed at short notice,” said Blackrock in its submission to the FEMR’s consultation process.

Others throw their weight behind a more extreme course of action. “Our view is that last look should be banned outright. With trading technology as it is these days, there is no need for latency protection,” says Paul Chappell, chief investment officer at currency management firm C-View. “If banks can't protect themselves from movements across multiple different trading platforms, they shouldn’t be streaming prices to those platforms. That would lead to a less fragmented market and might actually improve overall liquidity.”

More transparency 

Regulators have so far shown no inclination to take this radical step. Neither is there much enthusiasm for a micro-management of last look practices. Supervisors would be left with the unenviable task of working out whether a bank was using last look legitimately, with an intention to trade that was genuinely scotched by unfortunate market moves, or whether it was cynically dangling prices in front of clients’ noses then retracting them at the last minute once market information had been acquired. Setting a latency threshold after which last look could not be used would be similarly problematic – would 200 milliseconds be a suitable span of time? Five hundred milliseconds? Eight hundred milliseconds?

Some FX platforms, such as ParFX, have taken these decisions out of regulators' hands by not allowing participants to use last look. Others, such as KCGHotspot, have introduced maximum response times for users of last look.

The Association for Financial Markets in Europe's James Kemp pushes the idea of regulatory ‘guidelines’ over the use of last look, which would be included in a global code of conduct for FX market practitioners. “The sense I get from the market is that last look is a valuable tool that allows prices to be streamed tighter. It is obviously not right to use it purely for information gathering, and that is where efforts toward changing conduct need to be focused,” he says.

For many, the best solution is simple transparency. “If a bank uses last look, it should disclose this fact. All market makers should also publish their ‘hit ratios’, ie, the number of trades they accept versus the number they reject. Then clients could make an informed decision on who to trade with,” says the European asset manager.