High frequency trading as a liquidity tax

By Felix Salmon
July 28, 2009

The high frequency trading (HFT) debate continues today, fueled by a rather credulous Bloomberg article (elegantly fisked by Ryan Chittum) but, more substantively, moved along by Jon Stokes, who has a good article on the subject at Ars Techica. I asked him, via email, where he stood on all these questions; I think his answers are very good. Essentially, HFT turns out to be one of those “financial innovations” which lots of people like in theory but which only seem to benefit financial-market professionals in practice. I, for one, don’t think that there’s $20 billion worth of net societal benefit to it. Anyway, here’s the Q&A with Jon:

FS: Did you see today’s Bloomberg article?

JS: Yeah, and like the NYT article there is a slight conflation of flash orders with HFT. I don’t even mention flash orders in my article… I thought they were the least interesting aspect of HFT, at least to me as a computer guy. But clearly the idea of anyone getting market info ahead of anyone else touches nerves (at least among those who are getting the info second… the folks who are getting it first love it).

FS: Can you translate this, from the Bloomberg article, into English?

At Bats, the third-largest U.S. stock exchange, about half of its customers use flash orders, Chief Executive Officer Joe Ratterman said in an interview yesterday. The system is open to everyone and allows brokers to submit prices that are more competitive because the delay gives them a way to anticipate moves in the market, he said.

JS: This is actually a very good explanation of flash orders and the controversy around them.

As for the above, I can tell you what he’s trying to get across to the reporter, in the context of the currently bubbling controversy over flash orders and whether or not the NYSE will allow them in order to remain competitive with other exchanges and with dark pools: “this is no big deal, and doesn’t create a two-tiered market because any broker can use flash orders on our exchange to get a better price (than the people who aren’t using flash orders).”

Actually, my plain English rephrasing sort of crudely encapsulates the entire HFT debate. I.e., the argument of the “move along, nothing to see here” crowd on almost any HFT-related issue is something like:

“HFT does not create a multi-tiered market because anyone with enough money can move up to the highest tier by simply buying more speed and lower latency. So something with multiple tiers, where you can move to a higher tier if you can afford it, is not /really/ ‘multi-tiered’ because the tiers are open to everyone based on ability to pay. See how that works? No? Then go away, commie.”

FS: Can you tell me whether you think that, at the margin HFT improves liquidity? I’ve heard the opposite argued: that because HFT orders tend to be small and fleeting, they actually act against liquidity. That’s one reason why dark pools had to be invented — they’re the only way of trading in size without moving the market.

JS: At this point, you have to speak in more specifics about what you mean by “HFT.” Are predatory algos improving liquidity? I can’t see how one could claim they are for any reasonably useful definition of “liquidity”.

Are stat arbs and AMMs improving liquidity? Yeah, they are when they’re actually in the market. But they have no obligation to provide liquidity, so when things get choppy (i.e. when you need liquidity the most) they can just bail and take all that liquidity with them.

Are flash orders improving liquidity? I have to think way more about it to answer that.

FS: If HFT makes $20B a year, whose money is that?

JS: On one level, the answer to this question is easy for most of what goes on under the heading of “HFT”: the money is coming from whoever is buying stocks that are marked up a penny or so because they were down a rung on the speed/latency ladder. This could be pension funds, retail investors, or anyone else in the world. So it’s coming from the market participants.

(Of course, on some level, the guy who bought Cisco in March of 2000 is “down a few rungs on the speed/latency ladder” from the guy who bought it in August of ’99. But I think it’s important to draw a line here between what HFT is doing and what went on in a lower-frequency age, the same way that we all recognize a line between “speculation” and “investing” that’s drawn based on the time period that you hold an asset. Much of the debate in HFT is over the drawing of these kinds of lines.)

But to justify this $20B/year “fee” you have to make the case that the market system as a whole is getting something of value to all the payers in return. So supporters will say that it’s the price of liquidity and innovation, and, besides, they’ll argue, everyone who has been participating in the markets for decades has been paying these hidden liquidity taxes (and I’d rather call them taxes than fees) to specialists and any other market maker. But when you see this tax ballooning at Internet speed–much the same way that finance has ballooned as a portion of GDP–you have to take a step back and ask, “what is the real, fundamental benefit that we’re all paying for here when we collectively direct money into this?”


Comments are closed.