There seems to be general acceptance that last week’s flash
crash in sterling merely highlighted what we have known for so long – there is
a growing structural problem in FX markets.
While, as is the case with so many issues, accepting the
problem exists is vital, finding a solution is more important. Already there
are those bemoaning the reduction in voice desks, but it is pretty clear to me
that there is little chance of banks in particular rebuilding their voice
businesses – over-zealous regulators have taken care of that.
I am not sure that a voice business is the answer anyway,
the trouble is more the risk-averse environment the fines, suspensions and
court cases have engendered. Speaking with a very senior figure in the industry
recently we agreed that the ideal FX employee is a RoboCop type, someone who
uses the technology that is provided well, but can overlay that innate human
sense of when something is wrong.
So rebuilding voice businesses is not the answer, rather I
feel it should be building this hybrid business model.
This, however, will not reduce the risk of flash crashes,
for as we saw last week, the manual traders will wait for the move to get
extreme before stepping in – their presence does not, per se, eliminate the risk.
So what can reduce or eliminate the risk? One solution is to
get more resting orders into the market, but just as rebuilding voice
businesses is not the answer, nor should we expect customers to be comfortable
leaving resting orders. Best execution rules have changed and we live in a
world in which sharp moves mean stops can be triggered easily.
Clients want more control (for the time being at least) of
their orders, which means they will not be contributing to what is the
essential element of depth of book. More pertinently, the last place these
customers want their resting orders is in the public markets that inform so
many of the price engines in the modern market.
One answer that I believe could help is perhaps unpalatable
to many in authority –the central banks stepping up to be liquidity providers
of last resort – and I do mean “last resort”.
Clearly these moves take the central banks and other
authorities as much by surprise as they do market observers but there is a way
they can at least get an early warning and, possibly, make a quick decision.
Simply speaking, if central banks acknowledge the liquidity risk inherent in
foreign exchange markets they can place resting bids and offers a certain
distance from the market.
I would suggest that an 800 point band in an exchange rate
would suffice – it would mean “natural” moves, on the back of news or the sort,
can happen without interference, but if things get out of control they will be
the first to know because their ticket feed/alert system will notify them they
are being given/taken. Obviously this can vary by currency but generally
speaking in the G10, a 400-point either side band would seem reasonable. This
band could be a moveable feast, perhaps it is reset every 30 seconds around
mid-market to ensure that a flash move doesn’t drive the band away.
Equally the presence of decent bids/offers will mean the
machines won’t get spooked in the fashion they clearly can now. And if the move
is “genuine” or “normal” and the central bank in question believes there is no
point in stepping in, they can merely reset the bid/offer.
Whilst I accept that this is anathema to some, so too should
flash crashes be, and at the moment I don’t see an alternative. Most central
banks are committed (often legally) to maintaining a stable market, this is one
way they can do so.
Of course many central banks have desks that do just this,
but they are not populated on a 24-hour basis and, crucially in the modern era,
they rarely use the machines. Mainly, according to my understanding of the
situation, this is because they are trying to gather market intelligence to
inform policy making and machines can’t really give you that – but of course
the people watching the machines can and so too can the central bank’s
participation (at a distance) in electronic markets.
Just as I believe the future FX market banking participant
is a hybrid, so too should a central bank desk have dedicated e-intelligence
teams as well as their traditional voice desks. If nothing else this would help
alleviate flash crashes and it would also help them understand the situation.
One observation from last Friday’s events to back this
suggestion up. People have asked me why, on Twitter mainly, I was quick to
dismiss the suggestion the move was algos gone wild on the back of the Financial Times headline. My information
then and now is that the sell off was almost exclusive to Cable and that there
was little action in the crosses.
If this was indeed a reaction to Hollande’s comments then
the news-reading algos would have been having a real crack at EUR/GBP as well
as other crosses, and they apparently didn’t. An e-intelligence team at a
central bank would hopefully be able to spot the circumstances early and make
an informed decision that could stabilise markets.
This debate will naturally run and run, and the
investigation into events may, and I stress the word may discover exactly what happened. There are many that see the
growing number of events in FX markets ending with the exchange model, but in a
global market that model has shortcomings. One way to avert that is by local
authorities working with their local market participants to build confidence
and restore decent liquidity to markets.
It may be a short-term measure to have central banks use
price bands (personally I don’t think it will in the electronic age) but it
would most undoubtedly help.