The problems of HFT, Joe Stiglitz edition

By Felix Salmon
April 16, 2014

Never mind Michael Lewis. The most interesting and provocative thing to be written of late about financial innovation in general, and high-frequency trading in particular, comes from Joe Stiglitz. The Nobel prize-winning economist delivered a wonderful and fascinating speech at the Atlanta Fed’s 2014 Financial Markets Conference today; here’s a shorter version of what Stiglitz is saying.

Markets can be — and usually are — too active, and too volatile.

This is an idea which goes back to Keynes, if not earlier. Stiglitz says that in the specific area of international capital flows, “there is now a broad consensus that unfettered markets are welfare decreasing” — and certainly you won’t get much argument on that front from, say, Iceland, or Malaysia, or even Spain. As Stiglitz explains:

When countries do not impose capital controls and allow exchange rates to vary freely, this can give rise to high levels of exchange rate volatility. The consequence can be high levels of economic volatility, imposing great costs on workers and firms throughout the economy. Even if they can lay off some of the risk, there is a cost to doing so. The very existence of this volatility affects the structure of the economy and overall economic performance.

The question is: does the same logic, that traders seeking profit can ultimately cause more harm than good, apply equally to high-frequency trading, and other domestic markets? Stiglitz says yes: there’s every reason to believe that it does.

HFT is a negative-sum game.

In the algobot vs algobot world of HFT, the game is to capture profits which would otherwise have gone to someone else. Michael Lewis’s complaint is that if there weren’t any algobots at all, then those profits would have gone to real-money investors, rather than high-frequency traders, and that the algorithms are taking advantage of unfair levels of market access to rip off the rest of the participants in the stock market. But even if you’re agnostic about whether trade profits go to investors or robots, there are undeniably real-world costs to HFT — costs like drilling through Pennsylvania mountains. As a result, the net effect of the algorithms is negative: they reduce profits, for everybody, rather than increasing them.

In theory, HFT could bring with it societal benefits which more than offset all the costs involved. In practice, however, that seems unlikely. To see why, we’ll have to look at the two areas where such benefits might be found.

HFT does not improve price discovery.

Price discovery is the idea that markets create value by putting a price on certain assets. When a company’s securities rise in price, that company finds it easier to raise funds at cheaper rates. That way, capital flows to the places where it can be put to best use. Without the price-discovery mechanism of markets, society would waste more money than it does.

But is faster price discovery better than slower price discovery? Let’s say good news comes out about a company, and its share price moves as a result — does it matter how fast it moves? Is any particular purpose served to seeing the price move within a fraction of a millisecond, rather than over the course of, say, half a minute? It’s hard to think of a societal benefit to faster price discovery which is remotely commensurate with the costs involved in delivering those faster price moves.

What’s more, faster price discovery is generally associated with higher volatility, and higher volatility is in general a bad thing, from the point of view of the total benefit that an economy gets from markets.

HFT sends the rewards of price discovery to the wrong people.

Markets reward people who find out information about the real economy. Armed with that information, they can buy certain securities, sell other securities, and make money. But if robots are front-running the people with the information, says Stiglitz, then the robots “can be thought of as stealing the information rents that otherwise would have gone to those who had invested in information” — with the result that “the market will become less informative”. Prices do a very good job of reflect ignorant flows, but will do a relatively bad job of reflecting underlying fundamentals.

HFT reduces the incentive to find important information.

The less money that you can make by trading the markets, the less incentive you have to obtain the kind of information which would make you money and increase the stock of knowledge about the world. Right now, the stock market has never been better at reacting to information about short-term orders and flows. There’s a good example in Michael Lewis’s book: the president of a big hedge fund uses his online brokerage account to put in an order to buy a small ETF — and immediately the price on the Bloomberg terminal jumps, before he even hits “execute”. The price of stocks is ultra-sensitive to information about orders and flows. But that doesn’t mean the price of stocks does a great job of reflecting everything the world knows, or could theoretically find out, about any given company. Indeed, if investors think they’re just going to end up getting front-run by robots, they’re going to be less likely to do the hard and thankless work of finding out that information. As Stiglitz puts it: “HFT discourages the acquisition of information which would make the market more informative in a relevant sense.”

HFT increases the amount of information in the markets, but decreases the amount of useful information in the markets.

If markets produce a transparent view of all the bids and offers on a certain security at a certain time, that’s valuable information — both for investors and for the economy as a whole. But with the advent of HFT, they don’t. Instead, much of the activity in the stock market happens in dark pools, or never reaches any exchange at all. Today, the markets are overwhelmed with quote-stuffing. Orders are mostly fake, designed to trick rival robots, rather than being real attempts to buy or sell investments. The work involved in trying to understand what is really going on, behind all the noise, “is socially wasteful”, says Stiglitz — and results in a harmful “loss of confidence in markets”.

HFT does not improve the important type of liquidity.

If you’re a small retail investor, you have access to more stock market liquidity than ever. Whatever stock you want to buy or sell, you can do so immediately, at the best market price. But that’s not the kind of liquidity which is most valuable, societally speaking. That kind of liquidity is what you see when market makers step in with relatively patient balance sheets, willing to take a position off somebody else’s book and wait until they can find a counterparty to whom they can willingly offset it. Those market makers may or may not have been important in the past, but they’re certainly few and far between today.

HFT also reduces natural liquidity.

Let’s say I do a lot of homework on a stock, and I determine that it’s a good buy at $35 per share. So I put in a large order at $35 per share. If the stock ever drops to that price, I’ll be willing to buy there. I’m providing natural liquidity to the market at the $35 level. In the age of HFT, however, it’s silly to just post a big order and keep it there, since it’s likely that your entire order will be filled — within a blink of an eye, much faster than you can react — if and only if some information comes out which would be likely to change your fair-value calculation. As a result, you only place your order for a tiny fraction of a second yourself. And in turn, the market becomes less liquid.

It’s important to distinguish between socially useful markets and socially useless ones.

In general, just because somebody is winning and somebody else is losing, doesn’t mean that society as a whole is benefiting in any way. Stiglitz demonstrates this by talking about an umbrella:

If there is one umbrella, and there is a 50/50 chance of rain, if neither of us has any information, the price will reflect that risk. One of us will get the umbrella. If it rains, that person will be the winner. If it does not, the other person will be the winner. Ex ante, each has the same expected utility. If, now, one person finds out whether it’s going to rain, then he is always the winner: he gets the umbrella if and only if it rains. If the other person does not fully understand what is going on, he is always the loser. There is a large redistributive effect associated with the information (in particular, with the information asymmetry), but no real social benefit. And if it cost anything to gather the information, then there is a net social cost.

HFT is socially useless; indeed, most of finance does more harm than good.

As finance has taken over a greater and greater share of the economy, growth rates have slowed, volatility has risen, we’ve had a massive global financial crisis, and far too much talented human capital has found itself sucked into the financial sector rather than the real economy. Insofar as people are making massive amounts of money through short-term trading, or avoiding losses attributable to short-term volatility, those people are not making money by creating long-term value. And, says Stiglitz, “successful growth has to be based on long term investments”.

So let’s do something about it.

HFT shouldn’t be banned, but it should be discouraged. The tax system can help: a small tax on transactions, or on orders, would reduce HFT sharply. “A plausible case can be made for tapping the brakes,” concludes Stiglitz. “Less active markets can not only be safer markets, they can better serve the societal functions that they are intended to serve.”


We welcome comments that advance the story through relevant opinion, anecdotes, links and data. If you see a comment that you believe is irrelevant or inappropriate, you can flag it to our editors by using the report abuse links. Views expressed in the comments do not represent those of Reuters. For more information on our comment policy, see

Excellent speech, and précis of it.

Thanks, Felix.

Posted by crocodilechuck | Report as abusive

Bruno Biais made similar points, but I have yet to see a consensual estimate of what the profit of HFT operators actually amounts to.

Posted by Th.M | Report as abusive

HFT is circular trading. It is propping up entirely fake markets. You n Stiglitz missed that one.

Posted by kiers | Report as abusive

Has Felix Salmon finally read Lewis’ book or does he need someone to interpret it for him to change his mind about high frequency trading? He was so quick to write two articles in a row defending high frequency tranding (and not understanding that the books critique was more of the market structure and the wrong incentives created by this) without having read the book. Real quality journalism…

Posted by aloisk | Report as abusive

Nice post.

The real problem with HTF is that the algobots who are setting prices care nothing about the long term, or even the real world, for that matter. And that creates suspicion about the validity of prices…

Posted by mfw13 | Report as abusive

The answer is to impose a ‘Tobin Tax’ on securities transactions. Securities held for, say, 10 days would be exempt.

Posted by Ed62 | Report as abusive

By all means! Get rid of market liquidity, I say. Too much liquidity always hurts the little guy. Do away with decimals! Back to ‘steenies and eighths!

Posted by Publius | Report as abusive

The umbrella example was given last week by Glasner at Uneasy Money, al-problem-with-high-frequency-trading/, as he noted it is really due to Hirshleifer’s 1971 paper. Stiglitz cites the paper, you should mention that it is Jack’s idea.

Posted by bwickes | Report as abusive

Best piece you have written on HFT yet, Felix!

And Publius, you can have an efficient marketplace and computerized exchanges without HFT. You are swallowing that red herring whole.

Posted by TFF17 | Report as abusive

Stiglitz, Scott Patterson, and every HFT market-maker in the world dislike dark pools.

Stiglitz infers that dark pools exist because fundamental traders’ orders are too easily evinced by HFT algos in the visible markets.

In a fascinating display of perverse logic, Stiglitz argues that HFT’s “sophisticated front running . . . induces those engaged in fundamental research to turn away from markets to dark pools. [And, by so doing] Society loses the advantage of transparent markets… Overall, there is a loss of confidence in markets” (Stiglitz, p. 8).

For Stiglitz, the “loss of confidence in markets” is due to the automated, market-neutral, HFT market-makers who have “induce[d] those engaged in fundamental research to turn away from [visible] markets to dark pools.”

Stiglitz (Sal Arnuk and Michael Lewis) can’t fathom why there is a dearth of very deep, one-penny-wide, two-sided markets in stocks like SPY, GOOG, AAPL, etc.

So, by Stiglitz’ logic, if customers patronize Dark Pool houses of prostitution, the blame for such should be directed at those spouses who “induce” such behavior by refusing to accommodate their counterparty’s fantasies.

And Felix is persuaded by this logic.

Posted by dedalus | Report as abusive

spam/phishing/hacking/malware/virus/crim e are now the norm on computers/internet; the criminals even send you emails from yourself
the market bubble is nothing other than malware/virus programs from the mafia that facilitate their global moneylaundering
why should any stocks be worth more than the net realisable value of the business? because the mafia/insiders says so!
the mafia crime syndicates that control the markets push growth with small dips to give the daytraders daily profits
read it all on equalearth@twitter

Posted by equalearth | Report as abusive

The strategies over the supply and demand curve is now more in the hands of the robots and a little less in the company’s. The algorithm writers are now the ones doing the game playing guided of course by the companies’ forecasts and the real-money investors.

Unless there is a way to discount all these potential efficiency gains then I would leave it alone for now at least.

Posted by abnewallo | Report as abusive

Your statement about it reducing natural liquidity is non-sense. You NEVER put a large order out because of market impact and adverse selection risk. This was standard practice DECADES before HFT.

Posted by NaturalLiquid | Report as abusive

dedalus – can you explain your comment on dark pools another way? i don’t quite follow it, but i’m not that knowledgable in this domain.

Regardless, I don’t think that Stiglitz’s critique rests on dark pools vs transparent markets.

Felix, I appreciate this summary and the direction to an excellent piece.

Posted by coffeebreath | Report as abusive

Thoughtful post. However the zero-sum or negative-sum game line is old and tired. It is not. Wealth effects and inter-market linkages ensure that is the case. Markets have a purpose.

Does HFT help the price discovery? Yes. Does HFT assist in keeping markets in sync and more rational? Yes.

Is there too much emphasis on financial services? Could it be 2% of GDP instead of 7% of GDP? Probably, but that is a different question.

Posted by Matt-Hurd | Report as abusive

Nice post, Felix! Now if only Berkeley economists who have become neo-liberals overnight, e.g. BdL, would see the light.

Even the bitcoin and programmer crowd has similar thoughts on high frequency trading. They call it malicious computing.

Posted by EllieK | Report as abusive

No, not malicious. Developers of malignant computation seek to ensure its survival at the expense of the overall marketplace:

“The Skynet hypothesis is a boogeyman intended to scare the young and the paranoid. The real threat from AI is that it will become so good at the pointless tasks that we have given it that they will become a black hole of resources…This has already happened with high-frequency trading on Wall Street. There is an ongoing arms race between computers that trade stocks to see which one can get the edge over the other. Capital markets serve a function in society. They ensure that businesses that provide value to society will have access to large amounts of capital to invest in otherwise too expensive projects. I have not been able to think of a single way in which the high-frequency trading platforms have improved the markets capacity to serve that function…Malignant computing is a problem in cryptocurrencies too.”
via nt-computation.html

Although over-simplified, even idealized, it is mostly correct. Unfortunately, such sentiments are rarely expressed in ubertech. Perhaps he is also correct about crypto-currency, as a response to the general impression that banks, governments and markets failed to protect the public interest. The crypto-currency concept was extant for more than a decade, but the rise of bitcoin only began in the aftermath of the sub-prime mortgage crisis.

Posted by EllieK | Report as abusive

In the “umbrella” argument, isn’t it kind of the point that the guy who keeps getting rained on now has an incentive to get an umbrella of his own?

Posted by dsquared | Report as abusive

There is one twist to the umbrella paradox–there is an incentive for people to develop predictions about rain. There is even the possibility of people selling their predictions to potential umbrella consumers.

Then when rain is likely the person who values staying dry more will pay more for the umbrella and the other person won’t suffer too much by getting wet.

Of course the person who stays dry may simply have more money to bid up the price of the umbrella; but that fact also provides people an incentive to produce so they can obtain the consumption permits (money) to stay dry.

I know there are plenty of “holes” in these arguments (e.g. perhaps the poor person has a medical condition that will result in his death if he lacks the umbrella); but there is no perfect system.

Posted by Endmathabusenow | Report as abusive

You have the general picture right but you’re missing some of the points as I read the paper. You say, “But even if you’re agnostic about whether trade profits go to investors or robots, there are undeniably real-world costs to HFT”.

If you said “…there are undeniably real-world costs to trading” then I’d agree. The costs are not qualitatively different due to HFT. Steiglitz’s argument, as I read them, are structural and apply to all trading.

You wrote, “higher volatility is in general a bad thing, from the point of view of the total benefit that an economy gets from markets.” What I take Steiglitz to be saying is that people seem to prefer less volatility and therefore we infer that volatility is not a social benefit. That doesn’t make it an economic total bad things, afaict.

“the president of a big hedge fund uses his online brokerage account to put in an order to buy a small ETF — and immediately the price on the Bloomberg terminal jumps, before he even hits “execute”. ” – If he never hit Execute then that order never went to market so I can’t see how any change in the ETF’s price is relevant. The hedge fund president’s system might have been hacked or leaked information or something, but that’s malware, not behavior of markets with respect to HFT.

“markets are overwhelmed with quote-stuffing” and “market makers may or may not have been important in the past, but they’re certainly few and far between today.” It’s harder to market-make in high-volume, highly liquid stocks like the Dow or the Nasdaq 100, but even if you take a broad index like the S&P there’s plenty of market-making going on there. All the big banks have market-makers and there’s still competition for order flow in order to get market-maker rights in different equities. Considered broadly across the 35,000 or so listed equities there’s plenty of market-making going on.

In general I think you miss the mark with your entire discussion of liquidity, which is the ability to transact a symbol when I want to in the quantities I want to.

Finally, a transaction tax would not address the worst problems of HFT, which is using orders for discovery (quote stuffing). IOCs almost never fill so wouldn’t be affected by a transaction tax. If you taxed the placement of orders then you simply drive the activity elsewhere. Markets have been offering rebates for liquidity providers for years because the money follows the money – markets with more flow have more liquidity and attract more business. Rebates are, in effect, paying for flow. You also have to address the question of who gets the tax because the tax is an additional rent; adding a tax increases rent-seeking.

Posted by DrWex | Report as abusive

I have the feeling Michael Stewart doesn’t believe the ridiculous anti-HFT nonsense in his own book. The entire “controversy” is simply Stewart deciding to amplify an irrelevant bandwagon to stir up the politicians and the ignorati and profit.

Stiglitz is delusional so he might actually believe the nonsense he is spouting here.

Posted by Stiglitzator | Report as abusive