Is it really possible to manipulate the FX markets? The recent arrest of Mark Johnson, the global head of FX at HSBC, suggests that it’s easy to lay the blame at the feet of a single person. Numerous banks have fired or suspended traders because of the FX fixing scandal that trundles on. But, despite all the controversy surrounding this issue, it’s really quite complex.
As a former global head of FX at Deutsche Bank myself, I have first hand experience of this complexity. For a start, the FX market is huge – a turnover of $5.3 trillion every day according to the last BIS survey (published in 2013), so no private companies or individuals could make the FX market trade at an ‘artificial’ level for any period of time.
Like trying to hold back a stormy sea by building sandcastles, the effects of any feasible private market action would be washed away in minutes or seconds. The only entities that have the materials for a sea wall are the central banks with their ability to print money – and when they do so it is called ‘intervention’ or ‘currency bands’ not ‘manipulation’. Nor are their sea walls necessarily permanent, as those who can recall the destruction of the ERM or the Swiss National Bank’s eventually failure to control CHF appreciation in 2015 can testify.
But can it be done over shorter time periods? There is no denying that if you buy (or sell) enough currency quickly enough you can move the market. If you suddenly bought a billion euros versus the US dollar, say, the ‘stack’ of offers that lie above the current market price will be taken out by your actions and the EUR/USD price will rise. Can this be said to be manipulation? Not really.
“Manipulation”, according to the Securities and Exchange Commission, “is intentional conduct designed to deceive investors by controlling or artificially affecting the market for a security”. The new market price is certainly the result of ‘intentional conduct’ but it is no way ‘artificial’ since it reflects real supply and demand in the market – most specifically your demand. Nor can you really be said to be gaining from it since now you have a risk position on the books, which, when you attempt to get rid of it, will no doubt move the market back down the way it came. You would be subject to ‘slippage’.
To benefit from ‘controlling’ the market like this you would need to be sure of being taken out of your position at a higher level with no slippage. This is where FX fixes come into the story and why – to a large extent – they have become so controversial.
A fixing order is one placed by the customer of an FX market maker (almost always a bank) to buy or sell a currency pair in the future at a specified fixing rate. This rate, in the FX markets, is calculated by an independent firm by means of observing all the actual deals that occur in a short ‘window’ around a specified time. Crucially, there is no element of subjectivity in this calculation. FX fixes are therefore markedly different from benchmarks like LIBOR where estimates of borrowing costs are submitted, not actual dealing rates.
At first sight, this mechanism seems to present a perfect way to benefit from ‘manipulation’. Imagine a bank gets an order at 3:30pm from a customer who wants to buy a billion EUR/USD at the 4pm fix (orders like this are really very rare, the overwhelming majority of fixing orders are rather small).
Let’s say the market is trading at 1.1200. The bank starts buying euros in the five minutes prior to the fix and, by doing so, starts to move the price higher. The bank manages to buy its euros at an average of 1.1207, but because it continues buying until the last moment of the fixing window, the fix is calculated as being 1.1212. It is at this price that the customer will buy its euros from the bank, which, as a result, nets 0.0005 US dollars per euro profit, or $500k on the whole deal.
Could there be a clearer example of bad practice? The bank has ‘front run’ the client order and ‘ramped’ the price higher by buying. Subsequently it has managed to sell the euros it bought at an inflated price. This looks bad. But not so fast! Imagine instead that the bank simply concentrates all its buying into the 60 seconds around 4pm. In these circumstances, the market impact is much greater and the price spikes from 1.1200 to 1.1230 over the minute – price action that results in a fix of 1.1215 (which, for the sake of this example, let us suppose is where the bank has bought its euros – in reality there might be a small difference as the fix calculation doesn’t quite match the bank’s purchases).
There is one particularly striking point to note about this fictitious but entirely plausible set of circumstances: although the second execution methodology matches what the client has implicitly asked for (since deals in the fixing window are how the fix is calculated) both the bank AND the client are worse off. The client is out by $300k (1.1215 vs. 1.1212); the bank $500k. The uncontroversial, but naïve, process of matching the fixing methodology has simply made the rest of the market happier.
Surely, you might be thinking, even if the first methodology was used the client should be filled at 1.1207, the average rate at which the bank bought its euros? But that cannot be the case. By buying the euros ahead of the fix the bank is at risk: if a larger, opposing, selling order was also executed in the five minutes before 4pm, EURUSD could be forced lower and the average price at which our bank bought euros could be higher than the subsequent fix.
No client would accept the loss from buying euros higher than the fix so why should they benefit from the bank buying them at a lower rate? Incidentally, it was to avoid circumstances like the one I describe (two large, but opposing orders) that traders started colluding with rivals at other banks. To be clear, this was behaviour that was unambiguously wrong.
The real problem with fixing orders is that they confuse two distinct ways of dealing FX: agency and principal. Implicitly, fixing orders are a bit like agency deals whereby banks should execute in the window naively. But most banks execute large fixing orders as risk transferring principal deals. The paradox is that this approach – so open to accusations of malpractice – is probably beneficial to both bank and client.
What is the solution? One tempting answer is simply for clients to avoid using fixing orders altogether and to decide upfront with their bank whether they want an agency approach (bank executes the client’s order with no margin in return for a flat fee) or an old school risk transfer price: show an offer; get it paid; live with the consequences.
Another might be to transform the fixings. One plan that I floated to equity and bond index providers in the months before I retired was that they should move to calculate their indexes off an average of hourly fixes each day rather than just 4pm. The idea was that any fund trying to match the index would then necessarily have to split its order into five or six smaller tranches (to be executed at 11am, 12 noon, 1pm, 2pm, 3pm and 4pm London time, say) – this would reduce any market impact. Sadly, my idea was greeted with a polite lack of interest by the index companies despite the fact that it would relatively easy to implement.
But in the absence of any such structural change, the only solution is for banks to be absolutely clear about how they will execute and for clients to be realistic about it too. Let’s hope in the future that better communication before and during deals can replace ugly accusations of impropriety afterwards.
Kevin Rodgers started his career as a trader in 1990 with Merrill Lynch in London before joining another American bank, Bankers Trust. From there he went on to work as a managing director of Deutsche Bank for 15 years, latterly as global head of foreign exchange. His book, “Why Aren’t They Shouting?: A Banker’s Tale of Change, Computers and Perpetual Crisis” was published by Penguin Random House in July 2016.