# Has high frequency trading (HFT) been a net benefit or cost to society?

Various studies have demonstrated the very large and growing influence of high frequency trading (HFT) on the markets. HFT firms are clearly making a great deal of money from somewhere, and it stands to reason that they are making this money at the expense of every other participant in the market. Defenders of HFT will argue that HFT firms provide an essential service to the economy in the form of greater liquidity.

What research has been done on the benefits and costs of HFT? Has any study attempted to measure either the benefits or the costs? How would one attempt to measures these benefits and costs? What would be the effect of banning rapidly cancelled limit orders (see follow-up question), e.g. via a minimum 1-second tick rule?

Any references and professional opinions (backed by research) on this topic would be appreciated.

-
This timely blog post, through commentary on a recent journal article, discusses the effects of algorithmic trading on liquidity. –  Derek Ploor Aug 12 '11 at 5:00
@Derek-Ploor Thanks! I found an electronic copy of the paper, Does Algorithmic Trading Improve Liquidity? The paper claims yes. Nevertheless, it does not answer Buttonwood's concern that the improved liquidity is illusory. BTW, I think it is likely the blogger, Mark Buchanan, was influenced by or involved with the Buttonwood article when writing this post. –  Tal Fishman Aug 12 '11 at 17:24
The debate on this topic continues. Please come to chat to discuss if you're interested. –  Tal Fishman Sep 19 '11 at 15:41
It definitely has been a benefit for exhanges, that have made millions in trading fees –  RockScience Jan 11 '12 at 9:37
and it stands to reason that they are making this money at the expense of every other participant in the market. Disagree. It stands to reason that they make this money off some proper subset of all market participants, not necessarily the entire set of all market participants. –  user2763361 Oct 18 '13 at 14:13

The lead paper in the January 2011 Journal of Finance (Hendershott, Jones, and Menkveld) addresses algorithmic trading (AT). In short, they find that AT improves liquidity as measured by bid-offer spreads. Taking the econometrics as correct (it is in the Journal of Finance) the next question is if bid-offer spreads are a sufficient statistic for measuring liquidity (or any other benefits).

It is a difficult question to answer because, given current market structure, AT may improve liquidity (as measured by bid-offer spreads), but without data on other market structures, it is hard to say that we wouldn't better off with something like on-demand call auctions. I think there's a consensus that opening and closing call auctions have improved market quality as measured by opening and closing volatility, but it is not clear that we'd be better/worse off with completely call on-demand exchanges (although I know of at least call on-demand exchange in the works).

I think at this point it's still a subjective question with smart people on both sides. I tend to think we'd be better with call auctions (in terms of the pure economics of matching supply and demand). Finally, you may find this Big Picture post interesting.

-
but the volume you can trade at bid and ask is lower than in pre-HFT times –  Qbik May 22 '12 at 15:15
@Qbik vacuous statement given that the relative location of the insides is narrow. –  user2763361 Oct 18 '13 at 14:15
@Qbik i can confirm that the available volume between the old spreads have declined (will post study if i can find), but that has had no effect on overall volumes since they have not only continued to grow but have accelerated since the introduction of electronic trading (again, will post study if i can find). is the lost volume between the old spread having an effect in light of the acceleration? –  user6500 Dec 4 '13 at 17:14

This answer is my ongoing attempt to consolidate some recent commentary on this hot topic.

A good place to start for anyone thinking about this question is the Economists's Buttonwood: Not So Fast, which mentions recent research by Biais and Woolley (2011) and Dichev, Huang, and Zhou (2011).

Does Algorithmic Trading Improve Liquidity? This paper claims yes. Nevertheless, it does not answer Buttonwood's concern that the improved liquidity is illusory.

A survey commissioned by LiquidNet (written up in the Financial Times), charges that "high-frequency trading strategies are at direct odds with investors, and it’s incumbent on the institutions to protect the information and the orders of their investors."

Andrew Haldane, Executive Director, Financial Stability and member of the interim Financial Policy Committee at the Bank of England, gave a speech on this topic entitled The Race to Zero. Mark Buchanan is doing a series on HFT based on this speech here.

Institutional equity managers are clearly very worried about the effects of HFT on their costs, according to a TABB survey.

Studies say no link between HFT and volatility [Financial Times]

The Future of Computer Trading in Financial Markets – Working Paper. Commentary by High Frequency Trading Review. MoneyScience has this as just one of a slew of recent papers. Many were commissioned by the UK's Department for Business Innovation & Skills.

-

HFT, when they implement market-making like strategies, are a key element of a fragmented market to build "arbitrage bridges" between trading venues.

There is a cost that for: we are all paying (probably around a fraction of the actual spread) to them, and the resiliency of the order books suffers because of their presence.

As usual, there are positive and negative elements because of their activity.

Moreover, there is a serie of studies, call "Navigating Liquidity" analyzing the change in microstructure in Europe since the fragmentation of equity markets: https://www.cheuvreux.com/pdf/NavigatingLiquidity6_January2012.pdf

Moreover, T Foucault wrote an interesting paper on this in the book: Market microstructure, confronting many viewpoints.

-
Nice paper, thanks for posting (and for writing it, of course). –  Tal Fishman May 1 '12 at 17:35

Two important points going beyond the technical aspect that haven't been mentioned here.

The first is transparency. For example, I sometimes hear the NYSE DMM is an alternative market making model safer than the HFT liquidity frenziness. But what do we now of the additional cost incurred by investors when operating on NYSE DMM-controlled trades? Take a look at the charts of a NYSE specialist company like La Branche to see the evolution of their business as ECNs were growing the last 5/10 years. Dark pools are also sometimes praised for their fairness to investors, but how can we verify the fairness of transactions without order flow data: http://www.tradersmagazine.com/issues/24_331/pipeline-sec-fine-109653-1.html?pg=2

The second point is a pratical consideration on the applicability of regulations aimed at reducing order entry flow. Is it possible to monitor the 1 second delay between order entry and cancel request given the number of transactions taking place in the equity markets? And what guarantees that the 1 second delay will be respected by all participants down to the millisecond precision? A millisecond can make a lot of difference these days.

-

This is a really good question. The media loves to bash HFT and Algorithmic Trading as some kind of evil that steals money from working people. HFT makes a great scapegoat for people who don't understand it and are losing money.

I think HFT really doesn't have a net benefit or cost. It's just another way of using technology to make money. Yes, people make money using HTF. But they also create jobs and build businesses.

Ultimately HFT increases the efficiency of the markets.

-
Thank you for your opinion, but this site in general, and this question in particular, is meant for objective facts. Do you have any research to back up your assertions? –  Tal Fishman Sep 22 '11 at 17:18

the liquidity story is the one all hft use when asked what they actually give back to society and weve heared it all, there is just one more dimension that hasnt been discussed so far. high frequency trading has been expanding the last decade and the financial regulators havent been able to catch up with the developments. The programs have inputs and generate an output based on those inputs. The most popular input is the price(change) itself and there are more like momentum for example. the output is a buy/sell. now we have a system where the output is effecting the input which shouldnt be a problem since one high frequency trader wont make a difference. However if you have thousands of these systems it does make a difference and now you have system interacting with each other. According to the Securities and Exchange Commisson statement put out last year HFT had played a role in the may 6th flash crash. Stability of financial markets is an important aspect to look in to if we are evaluating cost/benefit to society.

-
-1 The SEC statement blamed an unnamed mutual fund (widely believed to be Waddell & Reed) for initiating the crash. –  chrisaycock Jan 5 '12 at 3:30
HFT absorbed much of the selling done by the mutual fund and then stared dumping what they bought quickly. i didnt say they were the main cause i said the played a role. –  JDrama Jan 5 '12 at 11:51
The combined selling pressure from the Sell Algorithm, HFTs and other traders drove the price of the E-Mini down approximately 3% in just four minutes from the beginning of 2:41 p.m. through the end of 2:44 p.m. During this same time cross-market arbitrageurs who did buy the E-Mini, simultaneously sold equivalent amounts in the equities markets, driving the price of SPY also down approximately 3% sec.gov/news/studies/2010/marketevents-report.pdf –  JDrama Jan 5 '12 at 11:52

There has been a significant interest in evaluation of role of HFT. I wonder if this is the correct question.

The study of their roles should be in a scenario where HFT is present and where HFT is not present. Earlier Specialists and Market Makers used to make that profit, now it is shared between them and HFT. So ideally the pie has remained the same (if ceteris paribus is markets is assumed) , the turnover "would" increase because of HFT. (Assumption being HFT turnover is more than Specialists + MM) As a result, more tax is generated for the governments.

-
HFTs generally are market makers. As for "specialists", NYSE disbanded them in favor of "designated market makers" a couple years ago. There's no sharing since everyone is an MM now. –  chrisaycock Feb 5 '13 at 12:15
Thanks for the information. I am not conversant with US markets. Most info I have is from Larry Harris (Quite dated). However my 2 cents on this was that Whatever Specialists were making is now transferred to HFT/MMs. –  shoonya Feb 5 '13 at 12:35

-
Posting a link to a website (especially if it's your website) without any sort of summary or explanation in the context of the question doesn't help anybody. Please edit your answer to include a summary of your findings. –  chrisaycock May 2 '12 at 15:38

we are all from science/engineering background. So let me put it this way.

Algorithmic trading is an engineering product, which bring buyer and seller together. Essentially, there is nothing different between automated trading systems and eBay or Amazon.

Algo trading is not about providing liquidity. There are plenty of liq-taking-only strategies running in the market, and they seem to be still profitable.

Algorithmic trading's contribution is---just like the contribution of eBay and Amazon--- is to replace expensive and unreliable human interventions. Just take a look at trading in Hongkong today, trading in Europe 5 years ago and trading in US 10 years ago. To produce the volume we take for granted today, we used to need tons of well-paid traders sitting there watching screens everyday. Please note that the average monthly cost for a desk (literally a desk with 4 legs) on a trading floor in Manhattan is USD 2000, a bloomberg terminal cost another 1800 a month.

with Algo trading massively implemented these days. How many old school chart-reading traders are still profitable? The number of successful human day traders is going down every months, just like the number of book shops in the street dropped significantly in the past 10 years, thanks to Amazon. Algo trading thus makes the business of trading much cheaper than it used to be.

You might argue that Algo trading is expensive as well. True, all those colocations and MIT CS graduates need to be paid. But you must agree with me that such cost is still much lower than what we used to pay those so-called Big Dick traders, who make a phone call to buy/sell millions shares, who smash monitors with a bats when they make a bad trade, etc.

If you are French, you might want some linear models here to make above arguments looks more 'quanty'. But this is really not rocket science, if we know how Orbitz works, we should also understand why algo trading is the future.

-
-1 Even if it is an engineering product (which is debatable) doesn't mean the economics of it cannot be studied. Economic studies of eBay abound. –  Tal Fishman May 2 '12 at 14:10
-1 Yes, we're all from a science/engineering background. That's why we expect answers to provide solid evidence, not anecdotes. Comparing HFT to Orbitz is what I do when I'm talking to my mother, not other quants. –  chrisaycock May 2 '12 at 15:48
@chrisaycock: There should be a difference between anecdotes and common senses. Without commond senses, no wonder those fancy pricing models built from 'solid evidences' failed miserably a few years ago. –  Fenix Chen May 19 '12 at 7:54
great philosophical answer! i agree with the qualitative assertions of this answer whole-hearted (and minded) ly –  user3232 Feb 16 '13 at 2:03

## protected by olaker♦Feb 16 '13 at 10:08

Thank you for your interest in this question. Because it has attracted low-quality answers, posting an answer now requires 10 reputation on this site.