The Last Look…
Posted by Colin Lambert. Last updated: October 31, 2022
Do we need to change how the BIS Triennial Survey data is collated and published? I am always reluctant to advocate for change in this area because temporal comparisons become so much harder, but it strikes me that how the data is published fails to reflect the modern FX market and may be missing an opportunity to provide end users of the FX market with some genuinely valuable data.
At the heart of the issue are my doubts that the survey truly reflects the FX market structure in the era of electronic trading. The BIS allocates firstly according to the location of the sales desk, and then to the trading desk, which is where the differences between the regional semi-annual surveys and the triennial event occur. In the modern era there is a tremendous (and growing) amount of volume that is not intermediated by a sales relationship. Some banks will allocate a customer to a sales desk, but when the relationship is global – as so many big ones are – it gets tricky when allocating multiple trades.
There is also the question of whether this skews the local liquidity picture – for example a sales desk in Asia could book a trade late in the day, which is managed by a desk in London, or, if it’s a one-centre risk engine, New York. To me, the liquidity from this trade is in London or New York, not the regional centre where the local market may have closed, or is edging towards the end of day.
There is no doubt that if flow is allocated according to the location of the trading desk, the data will look significantly different, but I feel this is a more accurate way of estimating where volume is done. If the BIS feel this is a step too far and would lose regional nuances, perhaps a secondary report could be published along three geographical lines – Asia Pacific, Europe and the Americas, indicating where the risk is actually managed? More on this later.
The BIS report is very valuable for managers looking to allocate resources, and has become even more so since Brexit and the need to start looking to build a serious EU presence, but in broader terms – knowing where liquidity really sits – it is less relevant than it used to be. As an example, look at AUD/USD. A customer with an admittedly large four yards of spot AUD to trade would be just under one-third of average daily volume in Australia. In the UK it would be (according to the regional FX data) one-twelfth – and remember the Australia data is measured by the sales desk, some of the flow could be going to the UK.
There is also the question of categorising market participants, and here it gets really tricky, because while the headline from this year’s survey is the rise in volume between Reporting Dealers, how much of this is between the genuine top tier players? More pertinently, and this is the tricky bit, how do we decide who consistutes the top tier?
The regional surveys that break out such measures often indicate that the top five or six players in the major centres handle about 75% of spot flow each day. The UK’s survey, however, has 27 banks providing data. This includes some regional specialists, but I doubt there are 27 institutions active, as a principal market maker, across a broad range of currencies, every day. Several of these banks are treated as customers by the top tier institutions and – another issue that probably needs dealing with at some stage – are also taking liquidity from non-bank firms that are registered under the hedge Fund/PTF category.
If there was genuinely more volume between the top tier players, I would have thought this would have shown up in the volume data from EBS and Refinitiv – and although there is a lack of transparency as to where the activity takes place within those firm’s platforms, compared to April 2019, EBS ADV was actually slightly lower, while Refinitiv’s was some 25% higher (again though, this could have been on FXall, it’s impossible to know).
Either way, across the two platforms, spot volume went up some 20 yards, whereas globally between Reporting Dealers it was up $249 billion – a lot of flow going elsewhere between counterparties that are, in spite of the existing methodology, unlikely to be dealing direct with each other.
The most common use of the data is to analyse the number one concern of all traders – access to liquidity. To which end – how about additional analysis that is temporal?
This is not to take a pop at the two venues concerned, more to point out that a lot of the flow allocated to Reporting Dealers is actually seen by the top tier as customer flow. Several of the Reporting Dealers use top tier aggregators, for example and have, with the exception of specific currencies, a one-way relationship.
The triennial and semi-annual surveys are very important studies of the FX industry and conditions therein and I find a tremendous amount of value in them, as do so many business managers. Now may not be the time to think of a radical overhaul of how the data is collated and published, but I think there should at least be a discussion about the possibilities down the line.
Some firms use the data to ascertain market share, others to allocate, as noted earlier, regional resources, but I suspect the most common cause is to analyse the number one concern of all traders – access to liquidity. To which end, and to end with my idea – how about additional analysis that is temporal?
Rather than worry about the location of the sales desk (irrelevant for many trades) or even the trading desk (equally irrelevant if they are run 24 hours), perhaps we need to start reporting turnover data by the hour? We can still have regional and product data, but the headline act is when the trade goes through and how it is executed.
The data can still be broken down by counterparty type (although I would combine the Reporting Dealers/Other Banks, or reduce the former to 10 or 15 names) and by product. This means we would know when liquidity is at its best, typically, and in what currency pairs and products. We would also garner interesting information on when different client segments trade. Of course, for institutional investors’ flow there would be data mayhem around London 4pm, but if nothing else, it would show up the impact of (pre) hedging! (sorry, couldn’t resist…)
p.s. Can we have some NDF data as well please?