Putting a Value on Pre-Hedging
Posted by Colin Lambert. Last updated: July 10, 2023
With The Full FX View
As with so many issues in foreign exchange, indeed in financial markets generally, the issue of pre-hedging is mired in grey areas, emotions and, often strident, opinions. At one extreme it is viewed as an effective risk management tool that enables a dealer to more effectively mitigate the risk from a large trade, at the other as front running.
The answer is, again as is often the case, probably somewhere in the middle, with the vast majority of pre-hedging instances having outcomes somewhere between those of a highly-beneficial execution outcome for the customer, or one where the market impact cost is equally high.
Although controversial in its nature, pre-hedging is not explicitly frowned upon by codes of conduct, albeit some regulators apparently take a different view. The FX Global Code, for example, recommends that any pre-hedging is commensurate with the total risk the dealer is expecting to assume and is not detrimental to the client.
The challenge, given the uncertainty around what is, and is not, a “good” pre-hedge, is that views can be wildly different. A dealer hedging with a trending market, for example, will have greater market impact, which will look bad on paper, however the reality may have been the client would have paid an even higher cost without the pre-hedging.
Equally, a client may see minimal market impact but not realise that a dealer had been overly-aggressive in its pre-hedging and actually created outsized market impact. Unless there are pre-defined expectations – that are shared by both parties – no Transaction Cost Analysis (TCA) report is going to be able to adequately quantify the benefit, or otherwise, of any pre-hedging. More pertinently, the TCA report may not even pick up the pre-hedging trades if it is focused on a specific window of activity.
A new research paper, authored by Roel Oomen, MD of FIC quant trading at Deutsche Bank, and Johannes Muhle-Karbe from Imperial College, London, presents a new theory designed to quantify whether pre-hedging activity is beneficial or detrimental to a client. Acknowledging that opinions on pre-hedging among market participants is “divided” on the subject, the authors say their results demonstrate that the success or otherwise of a pre-hedge depends upon specific circumstances, but can be beneficial to both parties when done patiently and for reasonable size.
The paper effectively codifies the truism amongst dealers that the more aggressive the pre-hedging, the worse the outcome for the client. That said, it observes that by imposing a limit on the dealer’s pre-hedging activity and/or creating some uncertainty over the exact timing of any transaction, the temptation to pre-hedge over-aggressively is dampened, thus benefiting the client.
The paper neatly lays out the issues surrounding pre-hedging – on one hand it notes the dealer has an inherent conflict of interest because pre-hedging can improve its inventory positioning and facilitate the impending trade but the resulting market impact adversely affects the execution price for the client. A related challenge, the paper points out, is what spread the dealer should charge the client.
The Full FX View
Anything that helps take the emotion out of a debate is a good thing, and as such it is good to see a proposed model that could, in theory but probably also in practice, provide a decent guideline for market participants involved in pre-hedging activity.
The question of pre-hedging is never going to go away – there remains too much uncertainty over the outcomes for that – but at least with an agreed framework in place the outcome can be fairly judged by both parties.
Pre-hedging has become a problem in markets due to changes in market structure and behaviours – effectively in the past a dealer, typically a bank, “worked” an order for their client on a “best efforts” basis. This too was fraught with conduct risk, not least how careful was the dealer with the order given they had no market risk?
This saw more clients looking to execute large trades putting dealers in competition – evolving best execution policies that were too rigid also were a driver – and as such pre-hedging became “a thing”. It is interesting that the paper’s authors also argue that their proposed model works when dealers are in competition, although here I am little more wary – not least because of the old spies’ maxim “a secret shared is a secret squared”.
The paper acknowledges the need for a limit on the pre-hedging activity and states outright that hedging more than the proposed amount of the trade would most likely be detrimental to the total cost to the client, but what happens if, for example, four dealers are put in competition and each pre-hedges to, again for example, 30%?
That is being overly-critical, however, for what this paper represents is a genuine method for improving the environment around pre-hedging generally. It could also be observed that if a client offers a large trade to four dealers, then they are creating the market impact themselves through information leakage and by creating the conditions for excessive pre-hedging in relation to the trade itself.
The paper does also, briefly, touch upon trading ahead of fixings, however it sticks with the standard market explanation, as supported by the Global Foreign Exchange Committee, that what dealers are doing ahead of the fixing window is hedging actual risk rather than anticipated risk.
There is some irony in this, in that the phrase “pre-hedging” largely came into the public domain when discussing activity and allegations of market manipulation around the London 4pm Fix. Notably, however, the paper does state that any market impact in the pre-fixing window is “effectively paid for by the client”.
Smarter minds than mine will dissect this paper in more detail and be able to find any holes in the theory should they exist, but either way, assuming this model does work mathematically, the paper should be required reading for the heads of dealing businesses on both sides of the divide. If some form of framework can be introduced – a form of “super TCA” perhaps, then the FX market in particular, but also other asset classes, will become safer environments for all.
With pre-hedging, the paper cites the rationale of proponents of the practice, namely that the dealer may be able to quote tighter and increase their chances of winning the trade. It also points out – and again this is argued by proponents of the practice – that if the client decides not to trade the dealer will need to unwind the acquired pre-hedge position and incur trading costs without any compensating revenues.
“Our analysis shows that – barring some specific scenarios discussed below – the dealer’s optimal pre-hedging strategy is to gradually build up a position in the direction of the anticipated client trade, which is proportional to the expected trade size,” the authors write. “In doing so, the window over which they manage risk is effectively expanded and this reduces transient market impact costs.”
The key to the issue is effectively the arrival price for the trade. Assuming the dealer(s) is willing to provide a reference price, the pre-hedging activity – and presumed subsequent reduction in spread quoted – can be calculated. As the authors observe in the paper, this “re-enforces and provides a mathematical foundation for recently issued industry guidance which stipulates that pre-hedging should be conducted in a manner that limits market impact, for an amount that is reasonable relative to the anticipated trade size, and be designed to benefit the client”.
On the subject of detrimental pre-hedging, the authors write, “When permanent impact is large (or transient impact decays slowly), dealer’s risk aversion low, and the time at which the client will make a trade decision is known and fixed, then it may be optimal for the dealer to charge a zero spread to maximise the probability of winning the trade and then pre-hedge aggressively for an amount that far exceeds the client’s potential trade size. The dealer’s revenues now come from the market impact that is generated by pre-hedging and is paid for by the client when they execute the trade against prevailing market rates. Despite the client being quoted a zero spread, their all-in-costs of execution then can sharply increase compared to when the dealer does not pre-hedge.”
To control the pre-hedging activity under the formula proposed, two non-academic conditions can be imposed. Firstly, the dealer’s trading and compliance function can impose a limit beyond which the trader is not allowed to pre-hedge, thus limiting, hopefully, market impact to a certain level (although there is, of course, no guarantee that will happen, the initial pre-hedging could trigger a broader move by other dealers). The proposed approach captures this by including an instantaneous cost of trading which penalises excessively erratic or “elastic” hedging activity, which the authors say is “effective”.
Secondly, the paper says the clients themselves can constrain activity by introducing some uncertainty about the actual timing of the trade. Its model has a fixed and an exponentially distributed decision time. “We show that even a small amount of uncertainty in the decision time is effective at mitigating any harmful pre-hedging that arises in some of the most extreme and contrived model configurations,” the authors write.
“These points provide guardrails that are easily monitored in practice for any pre- hedging strategy,” the add. “It can strengthen confidence that the activity is legitimate and designed to benefit the client even when individual trade outcomes may vary.”