Did Turkey Mayhem Expose FX Market Structural Flaws?

The arrest last week of a likely rival to Turkish president Erdogan prompted a sharp rally in USD/TRY, which hit a new all-time high, with the pair shifting more than 14% at one stage, before settling just the 3.5% lower on the news. The events, however, have prompted some questions over how the FX market handled the mayhem.

Dealers spoken to after the event said that liquidity was, naturally, at a premium, but, as one put it “there was always a price”. Spreads blew out – again as they should – and depth of book collapsed, all of which are “normal” reactions to a bout of extreme volatility. This suggests that the market behaved in as good a fashion as could be expected, however that has not stopped some raising issues.

Inevitably, the first questions raised were from the retail world, where punters were exited from their positions as brokers made margin calls. While nowhere near as bad as SNB-Day, one source at a broking firm says “several” accounts were “wiped out”, mainly because they were highly leveraged on one strategy – Carry. “The Turkish lira is popular with clients because of the 35-40% carry,” the source says. “Some have been put off by the steady decline in the lira over the past couple of years, but it has never been enough to cause a rout – until now.”

It does seem as though margin calls, along with the exiting of other positions and speculative activity, drove USD/TRY to a reported 42 high, before settling back down to the 38 area as local banks, including the central bank, bought lira to bring some calm to markets. Rather than highlight a structural weakness, however, other observers argue this event serves to reinforce the importance of retail traders in some less-liquid markets, especially where the carry trade is in play. “A news event, such as that seen last week, serves to create a bottleneck through which too many punters are trying to exit the same position, and the market sees a spike,” says an analyst at a bank in London. “As to whether this is a structural fault…I don’t think so, it’s just a function of market positioning and the vulnerability of capital-poor retail traders to ride out a brief storm.”

One observer points out that if a retail trader had been able to ride out the storm, they still would not be suffering due to the enormous carry available. “At one point it moved 14.5%, but on the day it was only 3.5-4%, that’s nothing compared to the carry available,” the source says. “It was only going to be a problem for those who were highly-leveraged.”

The problem may have been, therefore, that too many players were highly leveraged, because, as noted earlier, prices did seem to be available throughout. The Full FX has not been told of one liquidity provider that stopped pricing, even for a second, during the move, although spreads were “significant” according to the source at the broking firm, “we saw a couple of instances where they were extreme, but luckily other prices came in and we were able to exit our risk. It could have been a different story had we had to hit the really wide spreads, we would have been buying around 45 and even 50 in USD/TRY.”

It’s like the philosophical debate about the tree falling in an empty forest, if no-one had bought USD/TRY, as was the case for an hour or more, the price would have stayed exactly where it was because there was no data for the machines to adjust it

The move happened at the European open, which saw LPs cope with a pent-up demand for dollars (the news of the arrest occurred three hours earlier). Interestingly, this means that for three hours, aside from a tiny blip higher, nothing happened in the market until Europe came in. Some suggest this is due to Asian players’ unwillingness to price TRY until local markets and Europe more generally is open, and the broking source says that their firm hunted liquidity ahead of the open, but “there wasn’t much on offer”. Equally, the source says that demand from clients was low until Europe opened, suggesting a great number were based in that geographical zone. Once European traders reacted to the news, the snowball started rolling and LPs were under pressure, and the true scale of the event was realised.

This does raise the interesting question, would this have happened in a less-automated world?

Firstly, it is fair to say that yes, the move would still have happened – indeed the TRY market is not highly-automated so there are still a lot of manual interactions. Secondly, this does, to a degree, give lie to the fact that the machines have spread liquidity throughout the day. Again noting that the TRY market is largely manual execution, it seems clear that the pricing engines were not streaming TRY in any size until they had a local market and a handle on how the news would go down.

Where it is fair to say that a less automated world would have reacted differently is in the timing. Several sources spoken to at banks have said their pricing was in very small amounts, but even there, they were not being hit aggressively or persistently. One quant analyst at a bank says this indicates that the machines that were pricing had little or no data to go on. “It’s like the philosophical debate about the tree falling in an empty forest,” the analyst suggests. “The machines carry on pricing until they are hit, then they adjust the price. In this case, you could argue that if no-one had bought USD/TRY, as was the case for an hour or more, the price would have stayed exactly where it was because there was no data for the machines to adjust it.”

Some insight into market events can be gleaned from data provided by benchmark and execution analysis firm Siren FX, which offers a benchmark Fix. Every month, The Full FX publishes data comparing the outcome of the Siren 20-minute calculation methodology compared with the five-minute version from WM, and Siren FX has provided data for last week’s USD/TRY event.

At the key London 4pm Fix, the potential savings from using a longer window were about average for a month-end at $1,444 per million. This is about average for the emerging market pair that is selected randomly each month-end, but there are also some interesting insights from other fixes during the day.

Fixes at 8am, 10am, midday and 3pm London time all provided savings in the region of $1600-$3500 per million, but what happened at 7.30am London, when European markets opened and the move really took place, are insightful.

At a high level the potential savings from using a longer calculation window were immense – over $14,000 per million – but that should be seen through the prism of the circumstances and the exact time. The Siren calculation largely took place before most of the trading took place, whereas the WM window would have seen a great deal of actual trades. Equally, the Siren Fix is based upon mid-rates provided by New Change FX, whereas WM uses, where possible, actual trades.

The Full FX has been unable to ascertain the actual WM Fix at 7.30am on 19 March (the data provided is a New Change FX-calculated proxy), but it is fair to assume that in these circumstances, it is likely that the WM rate provided a more accurate observation of market conditions. This is because it would have taken in multiple actual trades as the buyers got into their work post the European open.

This reinforces the sense that there was pent up demand for USD/TRY into that open because players were unable to exit the risk in Asian hours, or simply were not aware of the implications of the arrest in Istanbul until local and European traders started reacting.

Guarantees

Another area provided lively conversation over the past few days following the TRY move – what happened with those banks that offer fixed exchange rates for the day?

It seems to depend upon how the bank offers the fixed rates. According to one source, at least two banks were operating on a “set and forget” basis, where the rates were published at the Asian open on the premise they were good for certain clients for the whole day.

Those banks offering a guaranteed “fixed rate for the day” would probably have had a nervous day, not least because of the question of where to set the price

These players may have been hurt by the move, unless they are offering more than 3.5-4% spreads, in which case the spread on these fixed rates is somewhat outrageous and akin to the airport exchange bureau, but several others set the rate when asked by the client. This means that they would have been asked for USD/TRY, taken the events into account, and then provided a rate that was good for 24-hours.

In both cases the risks were high on both sides. For clients – typically sources say they are small and medium corporates with little or no interest in the market – they may have found themselves paying above the odds for USD/TRY if their timing, not something for which they are noted in FX terms, was off. For the banks, the risk was, if they have a “set and forget” regime, they could have been hurt by clients opportunistically taking advantage of events. That presumes the clients would have both had the USD/TRY interest and the knowledge of market events to take advantage. As one source notes, “the banks are very careful about who they let have these fixed rates”, hinting that they are largely for ill-informed clients who have neither the ability or permission to trade speculatively.

Those banks offering a fixed rate upon request would probably have had a much more nervous day, not least because of the question of where to set the price. Assuming they did not, as was the case apparently with some streams, provide a 10 big figure spread, actually setting a rate for the next 24-hours would have been challenging.

Ultimately, given the nature of those clients trading on fixed rates, the damage (or benefit) is unlikely to have been notable, and as such banks offering the model can continue to do so. They may wish to look at their disclosures around how those rates can be changed, but largely this is likely to be dismissed as a rare event.

Structural Faults

As always happens in these events, questions are inevitably asked about how the FX market structure held up. Often they are asked by people with a vested interest, and The Full FX has been told of at least one incidence where a client asked for a re-papered trade after hitting a price at the high.

Re-papering trades is a thorny issue, and in this instance, because the trade was under or around the ‘traded high” published by LSEG FX (as opposed to the market high which needs to satisfy certain criteria), no action was taken.

There may be questions as to whether we need to revisit, as an industry, how high-lows are calculated, especially given the growing influence of smaller trades, but in general the structure seems to have held up well. Retail traders, as often is the case, were the most badly hurt, and liquidity did thin out dramatically as spreads widened in the early stages of the move, but the reality is these are “normal” symptoms of an “unknown-unknown” event.

Overall, however, this appears to have been just another day in a volatile year, where specific local events served to send local markets into a tailspin. As to whether Turkey can ride out the storm remains to be seen, and it seems certain that liquidity will be “twitchy” in the days ahead as dealers look for the slightest sign of trouble (or settlement of the issue).

The institutional market is well-versed in coping with these events, although it is fair to question whether a stronger risk absorption capacity might have capped the move out slightly lower than it did, but it is, inevitably, in the retail sector that most pain was felt. This is a function of the carry trade, and for these traders, they could be either grateful that they are still intact (and nervous of a real blow-up in the strategy as happened with Japan last year), or they will be regretting their level of leverage.

Either way, what happened in USD/TRY last week was nothing extraordinary – at least in terms of how the FX market reacted.