Week I of Mifid II: No light yet in dark pools


Mark Baker
Published on:

Some of the blips thrown up by the launch of the new regime look like more than just teething problems.

MB banner 660px

The launch of Mifid II on January 3 went about as smoothly as any mid-week, year-end, market-transforming imposition of thousands of pages of new securities regulations might be expected to do.

Bond volumes cratered, participants struggled to access data and by the end of the first week two crucial parts of the new regulation’s transparency drive had been put on hold.

In the context of grand financial market regulatory projects, therefore, not a total disaster and perhaps even something of a success.

Some individual regimes were certainly keen to trumpet their own efficiency. France’s market regulator, the AMF, noted its pride that its new platform had processed more than one million trades from 50 firms, representing the first day’s trading under Mifid II and filed by midnight.

“The volumes processed demonstrate the marketplace’s overall good state of preparedness, even if further efforts must be made by some firms.”

One small point, though: on the day of launch, exemptions were granted to ICE Futures Europe and the London Metal Exchange from the UK’s Financial Conduct Authority, and to Eurex from Germany’s BaFin, that allow them not to comply with open access requirements until 2020.

So much for regulatory harmony.

But surely the question at the top of everyone’s minds is what life is like now for those instruments being traded via the use of pre-trade transparency waivers? Well perhaps not everyone’s, but it’s interesting to look at, not least because nearly half of European stock trading happens away from exchanges.

For those uninitiated in European Securities and Markets Authority (ESMA) jargon, this is about dark pools, trading arrangements that are not quite in keeping with the spirit of Mifid. Opacity is kind of the point of dark pools, and transparency is kind of the point of Mifid. They’re an odd couple. But they want to find a way to get along.

Mifid II imposed a legal obligation on ESMA to implement by January a mechanism whereby stocks being traded through dark pools are subject to a double volume cap – the DVC. It is supposed to mean that only 4% of all trading in a 12-month period on EU venues in a given stock can take place on an individual dark pool, and only 8% across all dark pools. Triggering the cap means a six-month ban from dark pool trading in that stock.

Inevitably, the DVC doesn’t capture everything. There is a large-in-scale waiver for trades that are big enough to qualify for reduced transparency, but you’re not supposed to add up client trades merely to hit that. Heaven forbid. Trading over the counter is also outside the scope, but you’re not supposed to do that merely to avoid the DVC.

Or you can apply to be treated as a Systematic Internalizer (SI) for specific instruments, which frees you from some reporting requirements some of the time – although it also brings other obligations. Mifid I introduced this for equities, but Mifid II extends it to other asset classes.

There has never been a proper central repository of SIs, which get approved at a national level. But there is now an ESMA datasheet that collates any that it has been informed of. It shows there were 36 of those as of January 3, after a flurry of applications in recent months. It’s fair to assume that there might be a lot more interest in that as time goes on.

Back to the DVC. When Mifid II kicked off, ESMA had intended to begin publishing data on which stocks would be subject to bans on being traded in dark pools. But it hasn’t happened yet. Less than a week into the new regime, ESMA has admitted defeat, saying that it now won’t happen before March. Based on its discussions with authorities and trading venues, it’s confident this deadline can be met.

As easy as DVC

But what went wrong? Well, as one might imagine, operating such a scheme is a Very Complicated Thing. It’s not that ESMA hasn’t tried. It pointedly notes that its IT systems for the DVC have been finalized and open for submissions from October 16, 2017, more than two months early.

In fact, at the ESMA end, it’s all going marvellously. This, from a DVC statement on January 9:

The system is globally operating as designed and is performing calculations based on the data received as envisaged.

That must come as a blessed relief to anyone who felt that there were things in Mifid’s first week that might not have been “operating as designed”.

No, the problem was all at the market end, and here’s how bad it was. For a while now, ESMA has been running transitional calculations. On the basis of the data it received to work those out, it reckoned it ought to be told about some 30,000 instruments come January.

When it came to the final datafiles, it certainly had decent participation by trading venues: about 75% made submissions. But for the DVC calculations to work, the data had to tick two boxes. One, all venues where a stock could be traded had to report, and two, ESMA needed data for the whole period of 2017 to work out what caps should be applied in January 2018.

That simply didn’t happen. The number of instruments for which the data received ticked both boxes was only 650, just 2% of what ESMA was expecting. And surprise, surprise, those 650 are barely traded anyway.

ESMA didn’t want to put all the blame onto market participants, though, pointing out a couple of difficulties that they faced:

The DVC IT system is more complex compared to the other Mifid II IT systems that ESMA is running.

So, think how complex any of the various IT systems are that ESMA uses to process the millions of things relating to Mifid II. The DVC system is ‘more complex’ than any of those.

Then there’s the timing:

ESMA acknowledges that trading venues only had limited time to compute and submit all data as the relevant reporting period only closed on 31 December 2017 and the data had to reach ESMA by close of business on 5 January 2018.

Well, quite.