One way you could describe the debate over high-frequency trading is that it's really a debate over whether U.S. equity markets are too efficient.1 That is: The job of equity markets is to provide liquidity and price discovery. An efficient market will provide liquidity at a very low cost, and will adjust prices very quickly to respond to changes in demand.
That sounds good, but it's not equally good for everyone. Compared to a less efficient market, a very efficient market will be:
- Good for lean savvy technology-enabled liquidity providers who can profitably provide liquidity at low prices.
- Bad for old-school, inefficient liquidity providers who want to charge a lot of money for providing liquidity.
- Good for small investors who want to be able to buy small blocks of stock quickly without paying high commissions.
- Bad for big investors who want to buy a lot of stock without moving the market price.
And all of these things seem to be true. There are electronic market makers who almost never lose money. But old-school sell-side traders complain bitterly to Michael Lewis that their business is endangered by the new market makers. Small retail investors can buy all the stock they want, instantly, with price improvement, for $8. But big hedge-fund managers feel like they're being "front-run" when they try to buy large blocks of stock and the price moves against them -- as, in an efficient market, it should.
There are other problems with very efficient liquidity provision. The lean, savvy modern liquidity providers are different from the old bloated liquidity providers. In particular, they're much more risk-averse: High-frequency electronic market makers make much less profit per trade than old-school New York Stock Exchange specialists did, so they make up for it by taking much less risk. They trade in smaller lots, take smaller positions, and trade out of them within seconds rather than holding them overnight. Put another way, compared to the olden days, the expected profits of liquidity provision are lower, but the distribution is narrower:
This is mostly good -- cheaper trading! -- but there's a social benefit in having liquidity providers who are willing to take risks. The idea -- and plenty of people dispute the facts here -- but the idea is that, in the olden days, NYSE specialists were supposed to step in to keep markets orderly, buying for their own account -- at their own risk -- when everyone else was selling. This reduced volatility and increased public confidence, as people felt that markets would be stable and they'd always be able to trade at a reasonable price.
That system is mostly gone, and that (fake) chart sort of explains why. If you make 5 cents a share in spread on every trade, then that can subsidize a certain amount of market stabilizing. If you make 0.05 cents a share in spread, then you will be roughly 100 times less willing to take risks to stabilize markets.2
This I think is a sensible way to understand the Tick Size Pilot Plan that the Securities and Exchange Commission announced yesterday. Basically, U.S. stocks are quoted in increments of a penny, and sometimes traded in even smaller increments, which makes trading cheap and efficient.3 For some very liquid stocks, this is great: You can buy a stock for $20.01 or sell at $20.00, so trading costs are very low. And the market is deep enough that it's hard for even big institutions to move the price much with their trading, so they get the benefits of cheap trading without the drawbacks of markets moving against them.
But for less liquid stocks, the trade-offs are real. If a stock doesn't trade very much, then you can't make very much money quoting it -- but you can lose a lot of money if the price moves away from you. So you manage your risk by quoting only small sizes, and moving your quote very rapidly if someone trades with you. Or by just not quoting the stock. So low natural liquidity -- low interest from fundamental buyers and sellers -- leads to low profits for market-makers, which leads them to be very jittery in making markets, which leads to many of the evils -- ""phantom liquidity," "front-running"4 -- that critics of high-frequency trading worry about.
So the SEC's proposal is, roughly: Make some markets less efficient, by forcing market makers in smaller stocks to charge more. The pilot program would take 900 stocks, each with a market capitalization of $5 billion or less and an average daily trading volume of one million shares or less, and require them to be quoted in five-cent increments. This would last for one year, and the SEC would then compare trading in the test group with trading in similar stocks that are part of the control group.
The idea is that, if market makers can only buy for, say, $25.00 and sell for $25.05, instead of $25.02/$25.03, then they will, in expectation, make more money per trade:5 No one can compete them down too much on price. And so they will get richer and happier trading small-cap stocks than they would in a penny-pricing world.
And then market makers will use those profits to subsidize other things: They can't compete on price so they'll compete on ... service. What sort of service? Well, here is the service they're supposed to provide, according to the SEC:
From time to time since the introduction of decimal pricing, concerns have been raised that the one penny MPV [minimum price variation] may be detrimental to small- and middle-sized companies. In particular, a few studies have raised questions regarding whether decimalization has reduced incentives for underwriters to pursue public offerings of smaller companies, limited the production of sell-side research for small and middle capitalization companies, and made it less attractive to become a market maker in the shares of smaller companies.
In 2012, Congress passed the Jumpstart Our Business Startups Act (“JOBS Act”), which contained provisions relating to the impact of decimalization on small and middle capitalization companies. Specifically, Section 106(b) of the JOBS Act directed the Commission to conduct a study and report to Congress on how decimalization affected the number of initial public offerings (“IPOs”), and the liquidity and trading of smaller capitalization company securities.
So the idea -- Congress's idea, not the SEC's -- is that the market makers will make enough money trading mid-cap stocks at five-cent increments that they'll write more research, and do more IPOs of mid-cap companies.
Now this is a strange idea. It is probably true that, in the olden days, banks made a lot more money trading stocks, and that this money subsidized research.6 It is less obviously true that the high-frequency traders who took over the market-making business in the last decade -- and who never wrote research or led IPOs -- would, like, get into that business just because they're making a bit more money trading stocks. The arrow of time goes only one way, on this point. The JOBS Act, as is sometimes true of Congressional meddling in market structure, seems to have been driven mostly by nostalgia and confusion.
But the SEC gets that, and its 2012 decimalization report to Congress focused less on IPOs and more on market structure questions: what narrower quoting increments did to market maker profitability, effective spreads, quote depth and liquidity, and how other countries handle tick size questions.7 And it found ... a mixed bag, but enough evidence that bigger tick sizes could improve markets to make it go ahead and try them out.
So we'll see. There's some reason to think that wider tick sizes won't improve trading that much. For one thing, "Commission staff’s preliminary analysis of the Pilot Securities indicates that a significant percentage of Pilot Securities have bid-ask spreads greater than $0.05," so the tick size pilot program just won't have that big an effect: Market makers are already making more than five cents a share on those stocks,8 so they shouldn't change their behavior too much in response to the program. So the program may not do much.
Really, it's likely that the olden days worked the way they did because profits in big liquid stocks -- which "should" have traded in penny increments but instead traded in eighths -- subsidized trading of illiquid stocks, which should have traded in eighths and did. As long as market-making in Apple stock is a cutthroat, low-margin endeavor, nobody's going to spring for research, or at-risk liquidity provision, in small-cap stocks, even if trading those stocks gets a bit more profitable.9
Still, it's a worthwhile test, and a useful symbol. This is a deliberate attempt by the SEC to make markets less efficient, to cut back on competition and frustrate traders' desire to get the best price. If your worry is that markets might be too efficient, then that's exactly the sort of experiment you should be running.
This is an oversimplification. The main debate about the social utility of high-frequency trading is about excessive efficiency (as are the main emotional affiliations in the debate). But there are lots of subsidiary debates about things like maker-taker or direct feeds that don't directly relate to the question of efficiency.
2 There are other trade-offs like this. Efficiency and redundancy are somewhat opposite values: A very efficient market maker will employ just a computer, a cable and a fan, which 99 times out of 100 will be cheaper and faster and more efficient than employing a human in a fancy jacket. But every once in a long while, the computer will go haywire in a way that would never occur to the human in the jacket.
3 And that was its intent. From the SEC's release on the tick size program:
Prior to implementing decimal pricing in April 2001, the U.S. equity markets used fractions as minimum pricing increments. In the 1990s, the Commission began to re-examine the fractional pricing structure, and in 1994, the Commission staff issued a report (the “Market 2000 Report”) on the equities markets that, among other things, expressed concern that the then-existing 1/8th of a dollar minimum pricing increment was “caus[ing] artificially wide spreads and hinder[ing] quote competition,” leading to excessive profits for market makers. In the Market 2000 Report, the Commission staff also expressed concern that fractional pricing put the U.S. equity markets at a competitive disadvantage to foreign equity markets that used decimal pricing increments. The Commission used these findings as part of a public discussion on whether the U.S. equity markets should adopt a lower fractional minimum tick size or adopt decimal pricing.
4 See Cliff Asness on why that term is dumb, but it's a term people use.
5 I'm eliding some important facts here; read the proposal for more details. There will actually be three test groups: Group One will be quoted in five-cent increments but can trade in pennies (or less), Group Two will be quoted in five-cent increments and can only trade in five-cent increments or at the midpoint (with some exceptions), and Group Three is like Group Two but will also be subject to a "Trade At" rule requiring that trades occur at a lit market unless a dark market provides substantial (five-cent or midpoint) price or size improvement.
People are really into "Trade At." The idea behind it is that lit quoting is good: It's good for market efficiency, price discovery, etc., if people display their orders publicly. But posting a public order is writing a free option to the market, and in modern fragmented markets public orders face big adverse-selection risks. One way to reduce that risk is to force more trades to interact with public orders, by preventing them from trading in dark markets.
6 And, sure, maybe IPOs? Though IPOs are plenty profitable anyway.
7 Yesterday's tick size release summarizes:
In the Decimalization Report, the Commission staff also surveyed tick-size conventions in non-U.S. markets. Many foreign jurisdictions utilize a tiered tick size approach that provides greater variability for tick sizes based on the price level of a stock rather than the “one size fits all” approach utilized in the United States. Many countries have tick sizes that are four or more times wider than in the U.S. on a percentage basis. However, a few other countries have tick sizes that are less than half the size of the U.S. on a percentage basis. Therefore, the Decimalization Report stated that the U.S. market would benefit from a broad review of tick sizes, and such review would be informed by the experiences in other countries.
8 I mean, that's not right, that's just shorthand. If a stock trades at $20.00/$20.05 then I guess in expectation you'd make half the spread -- $0.025 -- per trade, and lose some of that in adverse selection and exchange fees and so forth. But you know what I mean.
9 And, again, the arrow of time goes one way: The market makers of 25 years ago were less jittery than modern high frequency traders because they had higher profits protected by regulation, but also because they had human traders and older technology and different business models. A high-frequency trading firm that now holds stocks for five minutes is not going to start holding stocks overnight just because spreads get wider; it's just not set up to do that.
To contact the writer of this article: Matt Levine at firstname.lastname@example.org.
To contact the editor responsible for this article: Tobin Harshaw at email@example.com.