Inaugural financial markets research from the JP Morgan Chase Institute studies trading behaviour around three major market events, and while the findings will not come as a surprise to most FX market participants – active traders were much more involved in the market than passive investors or corporate hedgers – they should prove useful to central banks as they come to terms with a changing market structure.
The research, FX Markets Move on Surprise News, was written by Diana Farrell, Kanav Bhagat and Chen Zhao at the Institute and looks at three specific surprise events, the Swiss National Bank’s decision to remove the EUR/CHF floor in January 2015, the Brexit vote in June 2016 and the 2016 US presidential election.
The paper notes that all three events shared one important quality – they had unexpected outcomes that led to the largest one-day moves in sterling, the Swiss franc and the Mexican peso – that made them ideal candidates for research aimed at building a better understanding of institutional investor trading behaviour.
The research examined institutional investor trades in FX markets in the days and hours leading up to, during, and after each event and analysed total trading volumes and net flows, the latter being a measure of risk transferred across all institutional investors; and then examined net flows at the investor sector and region level. The paper also looks at transactions within each investor type to observe within-sector variation in trading behaviour.
The paper uses data from 18,000 spot and forward trades by 500 institutional investors during the SNB event, 99,000 trades from 1500 institutional investors during Brexit and 4,000 trades from 500 investors during the US election. JP Morgan was a market maker in all trades used and the paper does not include positioning information pre- and post-trade. The research also converts all volumes to dollars and uses a scaling factor to normalise risk across the currencies involved.
Breaking each event down into a pre-event, repricing and stabilisation period, the research reinforces the “flash” nature of the Swiss franc move compared to the more elongated repricing period during the two public votes in the UK and US.
The first finding from the paper is that while overall FX trading volumes spiked by four or five standard deviations, not all investor sectors increased their trading activity relative to their average daily volume. Trading volumes from hedge funds, asset managers, and banks (the active investor sectors) were particularly elevated on the event day itself.
In contrast, corporate, pension/insurance, and public/other investors (the less-active investor sectors) showed little change in trading volumes on the three event days relative to their average daily volumes, despite the largest one-day moves in the relevant currencies in 20 years.
Total volume for hedge funds was more than 3.5 standard deviations above the ADV across all three events. Asset managers traded nearly four standard deviations more than their ADV during the SNB event and the US Election. Bank trading volumes were well above average for the SNB event and nearly 12 standard deviations above the average for Brexit.
Less active investors, meanwhile, saw activity rise by less than two standard deviations, an outlier here being Public/Other investors, which probably includes the aforementioned SNB and, probably, the Bank of England.
The paper also notes that while investors traded “significant amounts” of FX risk during the events, their net flows alone cannot explain the sharp moves. This should be unsurprising because in such events market makers adjust their pricing according to the news as well as to trade flows.
The authors say that in some instances, the direction of overall net flows is consistent with the prevailing changes in the FX rate, but not always. For example, in the three minutes following the SNB press release, they see buying of nearly $1 billion scaled CHF against other currencies. The large purchases of CHF during this three-minute period coincide with a 2.5 percent drop in EUR/CHF. In contrast, they see net buying of MXN against other currencies at the start of the US Election repricing period, just as the odds tilt towards a Trump win and MXN depreciates. “Therefore, there is evident variation in how much observed net flows impact market prices, and we explore this variation and potential explanations below,” they say.
Pre-event positions held by market participants influence net flows after an event, and this may in part explain the variation in the relationship between net flows and exchange rates described above for the SNB event and the US Election. It is likely that the surprise nature of the SNB event did not allow for pre-event position adjustments, while the highly anticipated US Election event did. The lack of net flows in the SNB pre-event period relative to the pre-event periods for Brexit and the US Election support this idea, the authors suggest.
The research also finds that the net flows during the volatile repricing period were not much larger in size than the net flows during the stabilisation period, when exchange rates were steadier.
For the SNB event, there was 2.9 times the average risk transferred every three minutes during the repricing period compared to the stabilisation period, whereas the velocity of the exchange rate move in the repricing period was 16.2 times greater. For Brexit and the US Election, the risk transferred per 15 minutes was similar during the repricing and stabilisation periods, however the velocity of the exchange rate moves in the repricing periods was three to four times larger than in the stabilisation period for both events.
Another finding of the research is also unsurprising – hedge funds transferred risk much more quickly on the news breaking that other investors, who waited for markets to stabilise. The hedge funds’ search for Alpha would have meant they were much more actively involved in the market, whereas institutional investors often trade according to a pre-arranged schedule.
The research finds that overall, banks did not transfer nearly as much risk as hedge funds because bank net flows were largely offsetting. The data for banks during the Brexit repricing period provide a good example, the authors say, noting that banks had the highest trading volumes in this period, but the activity generated very little net risk because they were buying and selling in roughly equal amounts during each time interval in the period.
Asset managers did not transfer significant risk during the repricing periods. In fact, the cumulative net flows for the asset manager sector did not become material until three to six hours after each repricing period ended. Among the less-active investor sectors, only the pension/insurance sector transferred material risk during one repricing period (Brexit).
The paper also finds that active investors played different roles in each event, during the SNB event they all, unsurprisingly, bought Swiss francs, but highlighting the less absolute nature of an election results sequence, their net flows were mixed during Brexit and the US election. Interestingly, the research also finds, supporting the finding that hegde funds were all buyers of CHF after the event, that hedge funds transacted as if they expected the 1.20 EUR/CHF floor to remain in place right up until the SNB announcement.
The paper also refutes the theory that during the events all participants in the various sectors traded in the same direction. Banks and hedge funds were active in transferring risk, and in the time intervals during which they were active, the buys and sells are considerably larger than the net flows?for that period. When net flows were large, there was actually activity in both directions, rather than just in the direction that aligns with the net flow, the research finds.
Equally, Asset managers were less active in transferring risk than banks and hedge funds, but in the periods when they were active, some asset managers were buying and some were selling. This behaviour was particularly evident during the Brexit event, where risk transferred was more consistent over the full event timeline.
The study also looks at the willingness of investors to trade outside of their normal working hours, noting that while the SNB event took place during local trading hours, the outcome of both Brexit and the US election took place in out of hours trading.
Bank and hedge fund trading volumes spiked compared to average volumes during APAC hours for both the Brexit and US election events, a marked change in trading behaviour, the authors say. In contrast, asset managers were less inclined to trade currencies outside of the currency’s home market, and their trading volumes only spiked during US trading hours, 10 to 20 hours after the news broke.
This held true for the SNB event as well – despite the fact that the event took place during local trading hours for CHF, asset manager trading volumes were highest during Americas trading hours. “While one might expect this result for asset managers based in the Americas, it is also true for asset managers based in EMEA – for all three events, asset managers based in EMEA traded their highest volumes during US trading hours,” the report states.
The authors conclude that the findings in the report are informative along two dimensions: financial market stability and central bank communications.
“Our analysis shows that the institutional investor reactions to major market events, as reflected in trading volumes and risk transferred, varied in pace and size across sectors,” the paper states. “During the three events we studied, banks and hedge funds increased their trading activity immediately, but only hedge funds transferred risk soon after news broke and as exchange rates were repricing rapidly. Asset managers increased their trading volumes and transferred risk, but only after exchange rates had stabilised. Investors in the corporate, pension/insurance, and public/other sectors waited for a day or two after the events to transfer risk.
“The slower risk transfer response of asset managers, corporates, pension/insurance companies, and public sector investors to the three news events that changed perceptions of the fundamental value of each currency suggests these four investor sectors did not participate in the price discovery process,” it continues. “Furthermore, these same four investor sectors did not transact against the prevailing move in exchange rates during the volatile repricing periods of these three events, contradicting the popularly held narrative that long-only investors with long-term investment horizons act as a stabilising force during market dislocations.”
The authors further observe that Hedge funds and market makers played an “especially significant” role in the market ecosystem during these three events, and argue that company policies or regulations that limit the trading activity of institutional investor sectors to their normal business hours or the local market of a currency may prevent these investors from accessing liquidity and mitigating their risk during market-moving events.
“Our results could be helpful to central banks as they pursue the appropriate balance between their increasing tendency toward transparency in communicating policy actions and other critical factors, such as maintaining their credibility,” the authors say. “Our data demonstrate the extent to which announcing unexpected policy changes at a previously scheduled event (i.e., similar to the Brexit referendum and the US election) may allow investors to prepare for the range of possible outcomes and produce more balanced post-event flows, while enacting unexpected policy changes via a surprise announcement (like the SNB event) may not allow investors to adjust their risk in advance which in turn leads to directional net flows that could amplify price movements.
“When choosing the most appropriate method to communicate policy changes, policymakers can use our results to help weigh market expectations with respect to both the timing of announcements and the outcome in the context of other pertinent factors and their desired market impacts,” they continue, adding, “When deliberating unconventional policy measures that directly set the price of financial instruments, policymakers should carefully consider how they will unwind the policy. For example, policies such as the SNB’s minimum exchange rate take pricing power away from the market and therefore can distort the incentives and, in turn, the behaviour of market participants.
“We found direct evidence of this, as hedge funds traded in a manner consistent with the exchange rate floor and with the expectation that the policy would remain in place up until the SNB announced its removal,” the say. “To the extent that policymakers want to unwind such a policy and return pricing power to the market with minimal unintended market impacts, the behaviour induced by the distorted incentives can become a complicating factor.”