Perhaps the most surprising aspect of the discussions I have been having post the news that Citi is closely studying the number of channels it supports in FX markets is not that several other major LPs have confided in me that they expect, or are currently conducting, a similar review but the number of LPs who expressed surprise that after cutting back Citi is expected to have 15 or so connections. The number of senior e-FX figures who expressed scepticism that there were indeed 45 platforms out there on which to trade genuinely surprised me, because as I think related recently I find that number quite easy to get to.
Part of the issue is, I think, that people are taking a narrow view of what constitutes a platform or channel. As I noted last week, I am told that aggregation providers are firmly in the spotlight and that significantly raises the number, there is also a school of thought out there that some cuts could come with certain channels operated by a company with others left untouched. There are also the prime-of-prime connections, and loosely related firms that have since managed to reinvent themselves as an “institutional” platform, even though that is more window dressing for local regulators than an accurate description of their client base.
Someone made the interesting point that many of the aggregators are actually OMS platforms and questioned whether cutting connections to those made sense given they were operated by clients, but I think this misses the point – which is that Citi, and other major LPs, are paying a small fortune in various costs to connect to, and provide liquidity to, these providers. In a pure aggregation model the benefit to the client would be in tighter pricing and, presumably, the bank would be pricing at a level at which it could earn revenue. In such a circumstance, why should the LP also pay a fee for the privilege? I would argue that by charging the LPs, the provider has moved from being an OMS to being a trading venue – and don’t even get me started on how some of the ‘clients’ using these channels have flow at the very toxic end of the scale.
It just makes absolute sense to at least conduct a survey, and as I noted last week and can reiterate here with a much larger sample size, many other LPs are doing so, or planning to. This is perhaps one area where we can see the impact of AI techniques on the FX business – there is a huge amount of data to crunch and only in the recent past has the industry been able to effectively analyse it. There is something of an irony in the fact that some platforms who also try to sell their data, could be undone by that same data.
I see what is happening as a broader re-evaluation on the part of the LPs’, thanks to the evolution of data and analytical tools, but it is wrong to think this is new thinking – just last year Deutsche’s Roel Oomen completed the publication a series of papers looking at the issue of aggregation and smart liquidity pools and his broad conclusion was that less is more when it comes to LPs in an aggregation pool. If that is indeed the case, then surely a manifestation of this is to cut the number of channels required? I have written and spoke on our In the FICC of It podcast about the flood of data available to LPs, most of which they don’t need, so there is another reason to trim back – there will be no significant disadvantage in terms of data to power the business as a result of cuts.
I also think that the opportunity set for the customer is being overlooked. It would seem inevitable to me that multiple customers connect to multiple LPs across multiple platforms – do they really need to? I accept that certain venues have the advantage of being seen as a specialist in certain currency pairs or products, but I don’t think those channels are under threat. I understand that customers pay very little compared to the LPs when it comes to these channels, but there still is a cost in connecting and maintaining those connections to them. Why keep those multiple connections, it’s not as though the liquidity available on them will be any different.
There is one other aspect of this that has been overlooked and that is the non-e-channels. Voice brokers have lived with pressure on their costs for more years than I care to remember – and more than one person highlighted how the fees charged by the e-platforms are one of the few, if only, charges not to go down over the past decade – but this time for the voice brokers it could be different, thanks to the changing liquidity landscape in FX swaps, NDFs and options.
In NDFs the pressure is on thanks to the growth of electronic trading, but even in the heavily voice broked markets how liquidity moves around has changed – and that could lead to LPs looking at trimming the number of firms they connect to. In the past there were voice broking desks for individual currency pairs and then, these were merged into single currency desks, such as euro-everything. More recently still these desks have been condensed into G10 desks for example, where one desk handles all G10 flow.
This means that unlike in the past, a voice desk no longer builds a strong reputation as a specialist and deep liquidity in one or more currency pairs. Equally, thanks to consolidation and M&A activity, the major players are all connected to all the brokers, but there is less opportunity for unique pricing thanks to the capital rules that dominate the risk profile of most banks.
So the question has to be asked, if there are three or four voice broking firms, all offering the same prices and same connections, why maintain all of them? I absolutely understand the need for competition to keep the others “honest” when it comes to pricing, but does it need four or more? I would argue it doesn’t and again, it costs money to connect to and deal with these voice desks. I would not be surprised to see banks mainly start to trim the voice brokers they support and while this looks at face value a negative for that industry it needn’t be so, for a more stable relationship with a more consistent fee stream could result for those left standing (although in true banking industry style, different players will probably pick different brokers to support, thus not solving the issue at all!)
So the benefit of another week to digest the idea and a host of conversations have reinforced my initial impression that this is a sensible idea that could actually make the FX industry a more efficient place – certainly cutting the number of channels will have, in my opinion, no impact on true liquidity, all it will do is pressure a few liquidity recyclers and to me that is no bad thing. So many things got out of control in FX a decade or so ago, for various reasons, and one of them was a “cheapening” of the value of liquidity. I sense the banking industry in particular is now leveraging the power of AI and taking back some of the ground it lost (thanks in a large part to its own deficiencies). Whether this results in a better market remains to be seen, I suspect it will if the plan to trim the number of channels is actually executed.