The Buy-side: Transforming FX Through TCA
Richard Coulstock, Head of Dealing, Eastspring Investments Singapore.
Developing TCAs beyond equities
We began our equity TCA project about seven years ago, so we’ve been working at it for a while. It has been in place for two or three years now without much in the way of further adjustment, so it’s almost running on a care and maintenance basis. We have subsequently been asked by clients, senior management, audit and compliance, “OK, you’ve done this on the equity side, why don’t you try and replicate it into the non-equity asset classes?” and “how do you monitor custodial FX and your FX processes in general?” One of the outside influences for our looking at foreign exchange were the court cases a couple of years ago in the United States into custodial FX transactions. So there have been both internal and external forces that have worked on us to look into the non-equity side of things. I believe it is far harder to do FX TCA than it is on the equity side.
If you think about equity markets, everything’s pretty transparent. You have the exchanges which provide price and volume information and everything is time stamped, and if you do your equity trades through an order management system, you can do a like-for-like comparison at any point in time and get an idea of performance against whichever benchmark you like. In the foreign exchange world, we don’t have the benefit of equivalent transparency and it is far harder to do efficient comparative measurements. Generally speaking, foreign exchange has always been seen by long only asset managers as a bit of a nuisance, but something that has to be done. We have been trying to work through that and to change our thinking about FX, to give it the respect that an asset class deserves. I think that FX trading is generally inefficient because there’s been no measurement and there have been issues with custodial FX, and it’s imperative that we work towards getting this right. I think there are huge savings to be made for every fund.
There were no external products available to buy that would do the measurements for us, so we began by asking our custodians and counterparties from the FX side to do some self-analysis. In the beginning we didn’t know quite what we wanted and the counterparties didn’t know how to do the analysis that we were after. But after a while we started receiving monthly and quarterly reports from each counterparty about FX trading — this meant that we had a process in place to measure and monitor FX that we could explain to regulators and to clients. There is a psychological benefit to this, as soon as you tell someone that they’re being measured and that those results are being analysed, even if the results don’t tell you an awful lot, people do work a little bit harder and put a bit more care and effort into what they’re doing.
You also have to think about what benchmark you use in the FX world and also the fact that the benchmark might change depending on each individual funding you’re trading for. Our spot FX trades fall into three broad categories:
- Currency trades in unrestricted currencies where you can trade with any counterparty you like.
- Trades in restricted currencies, where you go through a custodian, but you get interaction at the time of trade; a phone call, Bloomberg message etc.
- The third category is the most problematic; the custodial trades – where the asset manager only finds out what’s being done on a post-trade basis – so there’s very little transparency. You may just be given a rate and the trade date.
So that third category has been the main issue, but we will be looking at each of the three in our TCA project.
There are data cleansing issues, in particular with custodial trades. If we only receive advice of those trades on a post-trade basis and they are keyed into our order management system, we have to be aware that timestamp data in the OMS will not relate directly to the execution time and date. For example, if we get a trade that the custodians executed on the first of the month and we enter it onto our system on the second of the month at 2pm, you have to be aware that the particular FX trade wasn’t actually made at 2pm, it was actually traded at some point on the first. So the issue of data cleansing requires a lot of work that will take time and a coordinated effort between us and our counterparties.
Benchmarks are more complex in the FX world and I think that the benchmarks we use will vary depending on the fund for which we are trading. So for a custodial trade where all you know is the day you traded, then all you might need to look at is to ensure that they’re in the range of the high and low of the day or you might compare against the midpoint of that high and low and then look for trends over a period of time. If you are constantly, or 90% of the time, on the wrong side of that midpoint or heading toward the worst trade of the day, then that gives you an indication that perhaps the execution is not as good as it should be. But if we go back to the first category above, with a non-restrictive currency, we can trade whenever we like (and we will have clean time-stamped data), then we might start looking at TWAP or VWAP-type measurements as well as the range of the day.
We are not sure yet which is going to be the clearest and the most appropriate benchmark to use. But over the coming months, we will discuss this with both our provider and each of our counterparties. Benchmarks must be realistic, attainable and measurable and they must be fair to everybody concerned. I don’t want to come up with a measurement process that none of our counterparties are happy with, so we will engage with them to see what they think is realistic but also relevant in terms of improving FX executions for our clients.
Using the reports
I want to get to a position where I can share a monthly report that shows performance on a rolling 12-month basis, split by counterparties and split by currency pair with each FX counterparty. I want an idea of trends over time, who’s continually strong or weak in certain currencies, and then divert our trades according to those results. We already do this on the equity side and it works well and hopefully, we can transfer this to the FX world too. It will take time to get to that stage because we need to build up a period of clean data.