The Real Problem with High-Frequency Trading

Everybody seems to be talking about Michael Lewis’s new book (Flash Boys), which has been featured on 60 Minutes and reviewed twice by the New York Times. The book is about something called high-frequency trading, which, I will admit, with some, but not too much, embarrassment, I know almost nothing about. Actually, the first time I heard of the existence of high-frequency trading was from a commenter on a post I wrote almost two years ago, about which I will have something more to say in a moment. Michael Lewis’s book is a polemic against high-frequency trading, alleging that it enables high-frequency traders to rig the stock market and exploit ordinary traders. Lewis makes his case by telling the story of a group of hedge-funds that have banded together to create an alternative trading platform IEX, thereby avoiding contact with the high-frequency platforms, which, according to Lewis and the heroes of his tale, is exploiting everyone else on the stock market.

Lots of other people have weighed in on both sides, some defending high-frequency trading against Lewis’s accusations, pointing out that high-frequency trading has added liquidity to the market and reduced bid-ask spreads, so that ordinary investors are made better off, not worse off, as Lewis charges, and others backing him up. Still others argue that any problems with high-frequency trading are caused by regulators, not by high-speed trading as such.

I think all of this misses the point. Lots of investors are indeed benefiting from the reduced bid-ask spreads resulting from low-cost high-frequency trading. Does that mean that high-frequency trading is a good thing? Um, not necessarily.

To see what I’m getting at, let’s go back to the earlier post I just mentioned. I called it “Soak the Rich?” Here’s what I said then, discussing research by Edward Saez suggesting that marginal tax rates could be increased without reducing economic growth, an idea that, to many economists, including moi, seems counterintuitive.

Is there any way of explaining why raising top marginal rates to very high levels would not cause a loss of real income? Here’s an idea. The era of low marginal tax rates in the US has been associated with a huge expansion in the US financial sector. . . . What has been the social payoff to this expansion of finance? I am not so sure. Over a century ago, Thorstein Veblen wrote his book The Theory of the Leisure Class, followed some years later by his essay “The Engineers and the Price System.” He distinguished between engineers who actually make things that people use and financiers who simply make investments on behalf of the leisure class, adding no value to society. This was a vulgar distinction, premised on the unwarranted assumption that finance is unproductive simply because it generates no tangible physical product. On that criterion, Veblen would have ranked pretty low as a contributor to social welfare. Mainstream economists felt pretty comfortable dismissing Veblen because he was presuming that only physical stuff can be valuable.

However in 1971, Jack Hirshleifer, one of my great teachers at UCLA, wrote a classic article “The Private and Social Value of Information and the Reward to Inventive Activity.” The great insight of that article is that the private value of information, say, about what the weather will be tomorrow, is greater than its value to society. The reason is that if I know that it will rain tomorrow, I can go out today and buy lots of cheap umbrellas (suppose I live in Dallas during a drought), and then sell them all tomorrow at a much higher price than I paid for them. The example does not depend on my having a monopoly in umbrellas; I sell every umbrella that I have at the rainy-day market price for umbrellas instead of the sunny-day price. The gain to me from getting that information exceeds the gain to society, because part of my gain comes at the expense of everyone who sold me an umbrella at the sunny-day price but would not have sold to me yesterday had they known that it would rain today.

Our current overblown financial sector is largely built on people hunting, scrounging, doing whatever they possibly can, to obtain any scrap of useful information — useful, that is for anticipating a price movement that can be traded on. But the net value to society from all the resources expended on that feverish, obsessive, compulsive, all-consuming search for information is close to zero (not exactly zero, but close to zero), because the gains from obtaining slightly better information are mainly obtained at some other trader’s expense. There is a net gain to society from faster adjustment of prices to their equilibrium levels, and there is a gain from the increased market liquidity resulting from increased trading generated by the acquisition of new information. But those gains are second-order compared to gains that merely reflect someone else’s losses. That’s why there is clearly overinvestment — perhaps massive overinvestment — in the mad quest for information.

So I am inclined to conjecture that over the last 30 years, reductions in top marginal tax rates may have provided a huge incentive to expand the financial services industry. The increasing importance of finance also seems to have been a significant factor in the increasing inequality in income distribution observed over the same period. But the net gain to society from an expanding financial sector has been minimal, resources devoted to finance being resources denied to activities that produce positive net returns to society. So if my conjecture is right — and I am not at all confident that it is, but if it is – then raising marginal tax rates could actually increase economic growth by inducing the financial sector and its evil twin the gaming sector — to release resources now being employed without generating any net social benefit.

And here is one of the comments I received to my post.

An example in your favor: the construction of a more direct fiber cable from NYC to Chicago in order to save 20-30 microseconds for HFTs for around $300mm and talk of a similar venture from London (Europe) to Tokyo for five times that amount. I am sure there are sound business reasons for the construction and use of such networks but on a society level a definition of insanity?

So there you have it, high-frequency trading is a new way for traders to exploit — before their competitors can — any slight and fleeting information advantage that they have expended so much effort and so many resources to acquire. In other words, with the opportunity to engage in high-frequency trading, the incentive to search for, and uncover, slight and fleeting information advantages is growing ever larger, and the waste of valuable resources in the quest for such advantages is increasing parri passu.

Felix Salmon nails this point in his review of Flash Boys, observing that Michael Lewis writes his book as a tale of good guys versus bad guys. It’s true the interests of his protagonists and antagonists are diametrically opposed. But his notion, that one side is somehow better than the other, is simply asserted without proof or evidence.

You never know which side Lewis is going to pick in his books. In The Big Short, for instance, he sided with, of all people, the hedge funds who helped destroy the world by making multibillion-dollar bets against the U.S. economy in the highly complex world of mortgage-bond derivatives. And now, in Flash Boys, he sides with a small group of stock traders, funded by some of New York’s most notorious hedge fund billionaires, who have created their own private stock exchange, IEX. Truth be told, the IEX guys are a lot more sympathetic than the guys shorting mortgages. But by creating an oppositional narrative of what he explicitly describes as “good guys and bad guys,” Lewis runs the risk of turning a highly complex issue into an unhelpfully simplistic morality tale.

What Lewis has done is to find a group of traders who find that their attempts to trade on their information advantages are being stymied by the trading strategies devised by high-frequency traders. Lewis’s guys are certainly aggrieved. But just because they have a grievance does not make them any more admirable than the high-frequency traders. Vladimir Putin has lots of grievances, too, but those grievances don’t justify his actions or his arguments. Both sides are engaged in an essentially zero-sum battle for trivial informational advantage that they can exploit at the expense of informationally disadvantaged professional traders. (All traders are sometimes informationally disadvantaged. Their goal is to be informationally advantaged often enough to turn a profit.) Lewis, channeling the story of his IEX heroes, attempts to paint “average investors” as the victims. but Salmon effectively punctures that self-serving pretense.

“[Lewis] interviews a righteous avenger by the name of John Schwall, an IEX employee with justice on his mind:

“As soon as you realize that you are not able to execute your orders because someone else is able to identify what you are trying to do and race ahead of you to the other exchanges, it’s over,” he said. “It changes your mind.” He stewed on the situation; the longer he stewed, the angrier he became. “It really just pissed me off,” he said. “That people set out in this way to make money from everyone else’s retirement account. I knew who was being screwed, people like my mom and pop, and I became hell-bent on figuring out who was doing the screwing.”

Schwall tells Lewis that HFT is “ripping off the retirement savings of the entire country through systematic fraud,” and Lewis just allows the quote to sit there, damningly, even if he would never come out and put it that way himself. After all, the fact of the matter is that of all the various actors screwing your mom and pop out of the money in their retirement account, high-frequency traders are at the very bottom of the list. If, that is, they’re on the list at all.

If your mom has a brokerage account, or a mutual fund manager, or generally entrusts her retirement savings to any kind of intermediary, then the fees charged by her broker or fund manager will dwarf any profits being skimmed from her by HFT. And if your pop invests in the market himself—if he’s among those people with a TD Ameritrade or E-Trade or Schwab account, the “easy kill” for the high-frequency algorithms, then, in reality, he is the one big winner of the high-frequency game.

Of course, the stock market is a game with winners and losers: Every time one person is buying, another person is selling. If you sell before a stock goes up, you’re a loser, but if you sell before it goes down, you’re a winner. And if you’re making your own decisions of what to buy and sell, and at exactly what price, then there is no room to blame anybody but yourself if you make bad decisions. The trading fees and the stock prices, for individual investors, are all completely transparent.

If you’re a big investor, that’s not the case. Brad Katsuyama, when he was at Royal Bank of Canada, would see thousands of shares available for sale at a certain price—but when he tried to buy them, they would suddenly disappear, and he would be forced to pay more. That was the high-frequency traders, front-running his order.

Retail investors don’t run into this problem. If they see a stock available for $50.00, they can buy it at $50.00—not $50.01 or anything higher. They get exactly what they want, at exactly the price they want, which is also the best price in the market, and they get it immediately, in a way that makes big investors rather jealous. . . .

If your mom or your pop buys or sells a stock, that order will almost certainly never make its way to any stock exchange: It will be filled by a high-frequency trading shop that is happy to pay good money for the privilege of doing so. The high-frequency traders do make money from the retail investors—but mainly they do so the old-fashioned way, just by being on the right side of the trade.

If an HFT shop simply fills every single retail order at the best price in the market, then over the course of a day, and certainly over the course of a year, it will make a decent profit. Retail investors, in aggregate, are dumb money: If you take the opposite side of their trades, you’re going to do just fine. Especially when you also buy stock off them for a penny or two less than you will sell the same stock to them. That’s called NBBO—the national best bid/offer—and it simply reflects the fact that there’s always a small gap between the highest price that someone is willing to buy, and the lowest price that someone is willing to sell.

That’s why HFTs love to give retail investors what they want: It turns out that retail investors are very good at making very bad decisions all on their own. What’s more, if you’re an HFT seeing what retail is doing at any given moment, you can use that information to inform your stock-market trades elsewhere. So mom and pop end up making you a lot of money, without your ripping them off in the slightest.

So what we have here is a war between professional traders, the practitioners of l’haute finance. The rest of us are merely bystanders in their battle, with no particular reason to consider one side or the other as representing justice, fairness, or the common good. On the contrary, their perpetual battle for new and better mechanisms for gaining temporary informational advantage make the rest of us worse off, possibly because they are causing greater volatility in them market, though that is a suggestion made by Salmon for which there is no conclusive evidence, but by diverting productive resources into socially unproductive zero-sum activities, using valuable physical and human capital to produce temporary informational advantages with little, if any, net social value, being merely the instrumentality by which to extract wealth from others who are informationally disadvantaged.

I am not a fan of Thorstein Veblens; his celebration of engineering over finance at least partly reflected a crude misunderstanding of the operation of the price system and a failure to grasp the difference between engineering efficiency and economic efficiency. But lurking in his diatribes, there may have been some inkling that much of what financiers do is a waste of real resources in a battle over the surplus generated by the real economy. It is depressing to reflect on the fact that when more than a century ago Veblen was complaining that financiers, though less productive, were more highly remunerated than engineers, the engineers were still out there designing bridges, and railroads, and other wonders of late nineteenth and early twentieth century technology, while, now in the twenty-first century, the engineers are actually employed by the financiers to design complex high-frequency trading systems, connecting New York and Chicago with fiber-optic cable to speed up trading by fractions of a second, and designing complicated software to implement trading strategies designed to exploit socially useless informational advantages. Does that sound like progress?

About these ads

37 Responses to “The Real Problem with High-Frequency Trading”

  1. 1 Morgan Warstler (@morganwarstler) April 8, 2014 at 4:27 pm

    What is interesting is can we engineer a stock market system that has no bid ask spread and offers no advantage to HFT.

    One that either turns it into a bid by the second or better, puts a transaction cost on the trade that chases out even the day traders.

  2. 2 Scott Supak (@ssupak) April 8, 2014 at 4:53 pm

    “Does that sound like progress?”

    I often wonder if terrorists were killing Americans like our crumbling infrastructure is, how much would the GOP be willing to borrow and spend to invade the wrong country.

    The gas main that exploded in Harlem was 127 years old. But safely delivering natural gas to residents of Harlem isn’t as important (read profitable) as moving bits downtown.

  3. 3 John S April 8, 2014 at 4:57 pm

    I know nothing about the HFT debate, but I do feel that, given how inaccurately he reported the Moneyball story, Lewis’ credibility is very suspect. The A’s 2002 success was almost entirely due to contributions of the big 3 starters (Hudson, Zito, and Mulder) plus Chavez & Tejada, all acquired through traditional methods. None of the A’s draft picks featured in MB really panned out except for Swisher (who everyone wanted), and the A’s notoriously passed on Prince Fielder.

    If he’s willing to bend the (rather easily verified) facts to tell a good baseball story, why wouldn’t he do the same for a Wall Street story? (And I say this as someone who enjoyed Liar’s Poker, Moneyball, and The Big Short for their entertainment value).

  4. 4 Lord April 8, 2014 at 6:15 pm

    The other way higher marginal tax rates can promote growth is it is not the rates themselves but expectations of future rates that matter, and higher rates can be expected to be followed by lower ones.

  5. 5 JKH April 8, 2014 at 9:04 pm

    This must be a case of “crossing the line”.

    But “the line” needs to be defined.

    Otherwise, in the limit we’re in a world of barter – with no (competing) advice presumably, financial or otherwise. That seems to be the world that Salmon implies.

    I don’t know how to define that line.

    It probably has something to do with technology and regulation in this case.

    But what is it exactly that differentiates this case from plain vanilla advice – on something other than an ad hoc basis?

  6. 6 Prodipta Ghosh April 9, 2014 at 1:04 am

    of the two points, the second one – HFT is a battle between the biggies is exactly my point ( The first one though is not obvious. If some telco invests 300m for laying fibers, the fibers macnufacturers, their employess and suppliers and everyone else gets paid in the process. So any spending have some good portion. And to dictate if that spending should better be done to fund food coupons of the disadvantaged is quite like a planned economy, in principle

  7. 8 Dan Thorn April 9, 2014 at 4:49 am

    Nice, love that last paragraph. Long live engineers!

  8. 9 dwb April 9, 2014 at 6:01 am

    Capitalism is filled with “wasteful” products that appear to have no redeeming social benefit. Why are the isles filled with 32 kinds of toothpaste? Somewhere, 3 or 4 companies are spending millions researching another kind. Taco Bell spent millions to find a bunch of “Ronald McDonalds” and millions in research to bring you a breakfast sandwich in a burrito. See here if you have no idea what I am talking about:

    Is designing a new toothpaste or breakfast sandwich a zero sum game? If you look at a landscape of products in a competitive marketplace, you might be forgiven if you think they are all the same of similar and developing a new one is a zero sum game.

    But, it’s not. You can only measure productivity *over time* and its the drive to gain market share by competitive firms that each eek out progress one tiny increment at a time. The dental products on the market now are dramatically better than 20 years ago. Had you surveyed them in 1988, you might think they all looked the same too. Who needs another toothpaste? Except now, they are all much better than in 1988.

    Back to HFT. The real issue is not whether someone thinks is worth it to spend $300mn to build a fiber optic cable to shave 30 ms. The issue is whether: 1) the market is competitive; 2) are financiers income mainly derived from rents (oligopolistic market structure); 3) who accrues the benefits of HFT innovation.

    HFT is bad if it is protecting rent-seeking market structure (eg oligopoly). It is good if it is promoting competition and driving down costs. It is wrong, and a fallacy, to look at research expenditures at a static point in time and conclude it is a zero sum game. NO. The marginal benefit of any new product in most competitive marketplaces is very very small – or at least appears so. “Disruptive technologies” are quite rare. Yet, over time, all these tiny innovations add up. I personally have not had a cavity ever, as my teeth were sealed and the electric toothbrushes on the market are as good as dentists had 30 years ago. Who is going to complain about all this expenditure on “zero sum” consumer dental products? Dentists. I only go every few years now. Who actually accrues most of the benefits of all this wasteful research on newer better electric toothbrushes and toothpastes? Consumers.

    If HFT brings benefits of lower costs of trading to retail inventors and is knocking off the rent seeking institutional investors, that’s a good thing.

    And, by the way, I am not saying HFT *is* good. Goldman Sachs, for instance, is well known for locking up HFT software developers contractually and suing for revealing trade secrets. Are there “gentlemans” agreements not to hire from the competition? Do the buyers of HFT software – like GS – collude in the purchase of labor or equipment? Are they pursuing it to cement their position as market makers? Is this a way of stifling the competition?

    In other words, if someone wants to spend 300 mn on fiber optic cable, who cares. When HFT shops cease to use it, somebody else might, like internet providers. What I really care about is the consequent market structure and competitiveness. Yes, a lot of financiers do make a lot of money – but frankly I think a lot of it is rents and I am in favor of anything that makes the landscape more competitive.

  9. 10 dwb April 9, 2014 at 6:08 am

    … when I say “I think a lot of it is rents” that’s because market making is a natural monopoly or oligopoly. Traders tend to gravitate to “deep markets” which means that once a market maker develops a critical mass of liquidity, they have a natural monopoly. I am as yet unconvinced whether HFT is breaking down the natural oligopolistic structure of market making, or reinforcing it.

  10. 11 John S April 9, 2014 at 6:53 am

    As for the societal value of HFT and other modern trading wizardry–doesn’t it at least strongly incentivize Mom and Pop investors to avoid even trying to beat the market and stick to dependable buy and hold indexing? The cumulative time and effort saved among retail investors thus dissuaded from trying to be the next Buffet surely has some societal value, perhaps even exceeding the waste of fiber optic cables and the lost engineering output of the Wall Street quants.

  11. 12 Bas April 9, 2014 at 7:14 am

    “…connecting New York and Chicago with fiber-optic cable to speed up trading by fractions of a second, and designing complicated software to implement trading strategies designed to exploit socially useless informational advantages. Does that sound like progress?”

    Why yes, why wouldn’t it be progress? Perhaps people shouldn’t presume to be able to judge the social, scientific and technological value of any human endeavour, whether it be sending ship to find a seaway to the East, sending astronauts to the moon, or sending information closer to the speed of light.

  12. 13 JP Koning April 9, 2014 at 7:51 am

    “On the contrary, their perpetual battle for new and better mechanisms for gaining temporary informational advantage make the rest of us worse off… by diverting productive resources into socially unproductive zero-sum activities, using valuable physical and human capital to produce temporary informational advantages with little, if any, net social value, being merely the instrumentality by which to extract wealth from others who are informationally disadvantaged.”

    Are more resources entering the domain than leaving it? We certainly see a diversion of human capital into things like HFT design. But at the same time these computer algorithms have also released all sorts of human traders & market makers from the industry, especially on the equity side. After a spell of unemployment, presumably these ex-traders go back to some productive line of work, say drywalling. What makes this story interesting is that it is PhD’s with physics degrees who are displacing those from a job that historically required only high school or college education.

  13. 14 David Glasner April 9, 2014 at 9:44 am

    Morgan, Sorry, I have no clue, and I don’t even know whether to encourage you to try to think of one or just forget about it.

    Scott, Interesting comparison and an important question to think about.

    John S, Thanks for providing that bit of background information.

    Lord, True, but most people would not attach too much importance to that effect. However, in the early 1980s, Art Laffer was warning that the Reagan tax cuts would be self-defeating because they were going to be phased in over three years to reduce the effect on the budget deficit. That was convenient for Laffer because he could explain the 1981-82 recession as the result of deferring economic activity until the tax cuts became fully implemented in 1984. Unfortunately (or fortunately, depending on your perspective), the recovery began in late 1982, so Laffer began downplaying the effect of anticipated future tax cuts.

    JKH, I agree there are problems in identifying particular transactions as socially productive or redistributional. However, the waste of resources is associated with the research that goes into obtaining an information advantage. We don’t want to stop all such research, but the market over-rewards such research. Warren Buffet and George Soros sure that they are wonderful are richer (maybe a lot richer) than they ought to be, and I say that without making any pejorative judgment about their character.

    Prodipta, The problem is not that the telco and the people working for the telco were not paid for laying the fiber optic cable. You could say the same thing about the resources used to make dynamite that one company used to blow up its competitors plant. The fiber optics are being used by some people to make money at the expense of other people by allowing them to exploit a transient information advantage. Democratic societies make decisions about subsidizing some activities and products and taxing or prohibiting other activities all the time. That doesn’t make them planned economies, as Hayek certainly understood and explicitly said.

    Dan, Thanks. But let them know their place. Economists rule.

    dwb, Sorry, but I think you are not grasping the distinction that I am making. Trading on an information advantage is not like selling a cereal. Resources devoted to creating an information advantage are not like resources devoted to making a cereal. Information advantages exist all the time, and people exploit them, but devoting resources to the creation of information advantages and to facilitating their exploitation are a social waste. That waste has nothing to do with oligopoly.

    John S, I agree that most people are better off buying and holding index funds rather than trying to pick stocks. I think that people invest too much in picking stocks, because their returns are an amalgam of increasing the efficiency of markets and wealth transfers from less well informed traders. I also agree that the greatest waste is probably inducing smart people into working on Wall Street and hedge funds instead of doing real productive work.

    Also I can’t help wondering why you didn’t write

    “doesn’t it at least strongly ENCOURAGE Mom and Pop investors to avoid even trying to beat the market”

    instead of using that ugly barbarism “incentivize?”

    Please forgive my crotchetiness. I hope it is not a telltale sign of advancing old age.

    Bas, Well, I try not to be overly judgmental, but as a matter of fact people do make similar judgments all the time about what kind of action is appropriate or inappropriate or whether it should be taxed or subsidized, or whatever. My argument, which could be, but probably isn’t, wrong is not an aesthetic or ethical one, it is an efficiency argument. I claim that there is an inefficiency associated with HFT that is not associated with other activities. You need to address that issue rather than make a plea for being nonjudgmental.

    JP, Good point. It is really all the PhDs working on Wall Street rather than doing productive work that is most disturbing. HFT is just an interesting and timely example of a broader phenomenon.

  14. 15 Frank Restly April 9, 2014 at 4:31 pm


    Would you care to elaborate on this statement:

    “I am not a fan of Thorstein Veblens; his celebration of engineering over finance at least partly reflected a crude misunderstanding of the operation of the price system and a failure to grasp the difference between engineering efficiency and economic efficiency.”

    Where electrical efficiency is a measured unitless value (ratio of output power to input power), the definition of financial market efficiency never addresses how the efficiency of financial markets is actually measured.

    Financial markets have inputs (information) and they have outputs (prices). Information is measured in bits / bytes and prices are measured in the unit of currency (dollars, pesos, etc.).

    In my mind, an informationally efficient market should allow an individual to look at the price of a financial instrument and deduce some portion of the information that was used to arrive at that price. For instance, informational efficiency would be closely related to spectral efficiency:

  15. 16 John S April 9, 2014 at 4:38 pm

    David, thank you for the gentle reproach on my diction. I appreciate any effort to improve the generally low quality of writing in the blogosphere.

    Does anyone have an estimate of how many PhDs are working on Wall Street? I’m not sure that the first generation of quants really would have done much productive work had they not gone into finance; Emanuel Derman was probably saved from a boring life teaching physics undergrads. Does the current generation mostly consist of the “best and the brightest,” or are they largely drafted from the second and third tiers that couldn’t quite cut it in their primary disciplines? I really have no idea.

    My main fear is that Flash Boys and its inevitable movie adaptation will only entrench the increasingly common view that free markets just “don’t work,” particularly in the areas of money, banking, and finance. The IMF recently estimated that the value of implicit too big to fail subsidies in the industrialized world is around $370 billion dollars. If one further considers the social costs of financial instability caused by deposit insurance-fueled risk-taking, the lost engineering and scientific output of a few thousand PhD’s seems quite trivial in comparison.

    It seems like Lewis is repeating his Moneyball error: exaggerating the importance of a fascinating, but peripheral, detail over the main story. When will he write a book about the true danger: government protection and coddling of Wall Street under the cover of promoting macroeconomic stability? Sadly, I doubt such a book is forthcoming.

  16. 17 Kabloosh April 9, 2014 at 4:44 pm

    Thank you for a thought provoking diatribe.

    I disagree with the premise that you need timely information to profit from the stock market. I’ve been investing a long time and I can tell you it takes guts! And discipline. When you buy a stock and it goes down and you gut it out and eventually it goes back up and you cash out you deserve every penny of that profit. You get paid for bearing the volatility. You’re not operating on inside information or prescient information.

    That’s not to say that there is no such thing as information. But the information takes the form of market timing models, well publicized leading indicators and nostrums such as “don’t fight the Fed”.

    I disagree that resources can be redeployed to more socially uplifting enterprise. The reason is that if that were going to happen, all the people or types of people that populate the HFT space would simply migrate to the public sector, being that’s where the money is, and they would intercept said funds and suckle at the largesse.

    I remember reading an interview with multi-billionaire Bill Gates, the owner of the hugely profitable Microsoft and friends with Warren Buffet. The interview took place several years ago but its still relevant.

    The interviewer asked: ” I guess you must get the smartest people in the world to work for Microsoft?”

    “Nah. We can’t compete with Wall Street”

    If we fill the public coffers with rapidly progressive tax revenue, we know where the smarties will project their careers and we know what they will produce: High Frequency Societal Returns. (HFSR) Only it may be high pitched but the returns won’t all that societal. A pig is a pig.

  17. 18 rpl April 10, 2014 at 9:20 am

    David, when you write, “Lots of investors are indeed benefiting from the reduced bid-ask spreads resulting from low-cost high-frequency trading,” it sounds like you’re taking it for granted that the effect of HFT is in fact to reduce spreads and increase liquidity. That claim is hotly debated, and, to me at least, it looks like the people who claim the opposite (i.e., that HFT increases spreads and that the liquidity they purport to provide is largely illusory) have a strong case.

    Here’s one paper on the subject:
    Their findings suggest that markets that clear via auctions at one-second intervals would provide better spreads and would be more robust against anomalies like the flash crash.

  18. 19 dwb April 10, 2014 at 11:30 am

    “Trading on an information advantage is not like selling a cereal. Resources devoted to creating an information advantage are not like resources devoted to making a cereal. Information advantages exist all the time, and people exploit them, but devoting resources to the creation of information advantages and to facilitating their exploitation are a social waste.”

    Wait, why? Example: I just bought a used car, and I paid some money for the “Carfax” Turns out, I found two identical models, nearly the same mileage, one had a transmission repair at 40k miles, one did not. True story. And, the one with the tranny repair was $1500 more!

    Someone spent a lot of resources collecting all this information, so I could refine my search and help distinguish a lemon from an non-lemon. I paid $54.99 for a history report on 5 cars. At first I thought it was a little steep, now I think it was a bargain. Carfax gave *me* one buyer, an informational advantage over many other buyers looking at similar vehicles at that time. So, would you conclude Carfax is a social waste? Did we increase or decrease informational assymmetry net-net with Carfax? On balance we decreased it and reduced the market failure, because the asymmetry is not between buyers, but between seller and buyer.

    By the way, Carfax has annual revenues of $250 million:

    How was “devoting resources to the creation of information advantages and to facilitating their exploitation” in the used car market a social waste?

    Be careful (and rigorous) – just because one party creates an “advantage” for themselves does not mean actual informational asymmetry net-net has gone up.

    I can provide many examples where HFT reduces informational asymmetry, even though only a few people have the algorithm. Here is another. Suppose I am an institutional investor with a big position to unload (say a weeks worth of daily volume or more). I have private information about my demand/supply. I have different people call and spread around the orders – NY, San Fransisco, Chicago, London, etc. This really happens. With a faster fiber pipe now an algorithm can detect my trading patterns and my private information is suddenly much less private. There is a reason people monitor reports of institutional traders like CALPERS for position changes. So what if only one firm has the algo?

    Again, the informational asymmetry is not between different traders, its between a large seller/buyer with private info, and the rest of the market.

    Maybe one or two HFT firms get that information from the HFT algorithm, but before it was zero. It’s not whether one HFT shop can gain an advantage over another, *its whether the market for the creation of such algorithms is competitive itself.* In the Carfax example, for example, anybody can buy a Carfax, and anybody could devote the resources to the same car database. Maybe they do not have actual competition, but they sure do have the threat. Prices stay reasonable.

    Now imagine instead Carfax is a assured monopoly because it owns the service records. What if its an extra service provided by used car dealers? What if dealers were *forced* to divulge the Carfax? See how that changes your view?

    Used cars and market making, very similar. To my earlier point: the social costs of market making are associated with informational asymmetries and the oligopolistic structure of the market. Just because one party gains a competitive advantage does *not* mean we have increased market failure. If one shop discovers the advantage, can another independently create a tool to exploit it? Hire away software developers? If one firm sees outsize profits, they will disappear so long as the market for HFT algos is competitive. *informational asymmetries decline to the actual cost of the tools needed to uncover them* Oligopoly is precisely the point, if HFT ends up further cementing the market makers and those with private info.

    It is definitively *NOT* a social waste to have firms competing to eliminate informational barriers, unless you have this perverse view that there are no barriers or that we should preserve for firms that want to hide information.

    Now, there are definitely some scenarios where HFT could worsen market failures. But, we need to be rigorous about how we think of them not oh noes! trading bad, mkay!

  19. 20 Michael April 10, 2014 at 2:54 pm

    dwb, you’re the voice of reason in this discussion!
    isn’t it sad that people who are great at Macro can get all naive in Micro topics they don’t grok? Of course, we all know the converse, with finance people totally loosing it in Macro…

  20. 21 David Glasner April 10, 2014 at 9:46 pm

    Frank, Engineering efficiency is defined at maximizing the ratio of some output to a particular input. Economic efficiency is measured by the maximization of total output (or welfare). Thus with multiple inputs producing multiple outputs, output is maximized by equalizing ratios of marginal rates of substitution between outputs in consumption and marginal rates of substitution in production. Those marginal conditions can’t be satisfied if one aims at maximizing the ratio of output to any single input. That is a straightforward mathematical implication of constrained maximization, so it should not be beyond the grasp of engineers.

    John, Certainly there are many more issues that ought to be addressed than the over-employment of PhDs on Wall Street. I’m not so sure that Emanuel Derman’s life would have turned out so bad if he had not found a job on Wall Street, but he was certainly fortunate to be able to work with Fischer Black, an opportunity that most quants never had.

    Kabloosh, I did not mean to suggest that timing is the only source of profits in the stock market, but it is an important factor. The volume of stock trading suggests to me that there is a lot of churning taking place.

    Opposition to growth of the financial sector does not imply support for expanding the public sector.

    rpl, I wasn’t taking a position one way or the other, though Felix Salmon, whom I was quoting, was suggesting that that is the case. However, even if high-speed trading does reduce spreads and increase liquidity, there are strong reasons to believe that it is not producing net social benefits. Thanks for providing the link to the paper by Budish, Cramton, and Shin. Perhaps Morgan Warstler, if he is still paying attention, would be interested.

    dwb, Carfax deseminates information that would otherwise be more costly for used-car buyers to obtain, thereby eliminating or reducing the information asymmetry that favors sellers of used cars, so your example doesn’t seem to me to correspond to the situation that I am talking about, where traders are exploiting a temporary information advantage to gain at the expense of disadvantaged traders who haven’t yet obtained information that will eventually be widely known. Carfax gave you an information advantage over other buyers who had a choice about buying the information from Carfax. Wall Street traders obtain information that everyone else will have very soon, but don’t yet have access to, and engage in trade with those who don’t have the information yet. People who buy information from Carfax get information that the sellers of the cars do have, the information advantage over people who choose not to buy the information from Carfax is irrelevant because they aren’t trading with the people who do buy the information from Carfax.

    Your example about the institutional investor is very nice, but what does it prove? If nothing had been done, the institutional investor using his private information would have profited at the expense of traders lacking that information. Not fair, but it’s the way of the world. Now the HFT shop comes along and devices an algorithm that allows it to infer from the institutional investor’s trading pattern, what information the investor is acting on, and drives up the price ahead of the institutional investor, thereby extracting the profit that would have gone to the institutional investor. You celebrate that as a competitive free market. I am pointing out that the resources used by the HFT shop to be able to implement its algorithm and its trading strategy are very costly, so that the transfer of profit from the institutional investor to the HFT shop is problematic not because the institutional investor is a good guy and the HFT shop is a bad guy, but because valuable resources are being used up in the process of transferring rents from the institutional investor to the HFT shop. The issue here is not a competitive issue; it is an externality issue, a divergence between private and social value or private and social cost. This was all explained very clearly by Hirshleifer in his 1971 article, but, as far as I can tell, you have not dealt with his argument.

    Michael, If you think that I am “all naïve in micro topics,” do you think that Jack Hirshleifer whose paper my argument was based on was also “all naïve in micro topics?” If you think that, you are probably the only person alive who holds that view. Certainly no one who knew him or read his work could possibly think so. Perhaps you ought to start reading his work yourself.

  21. 22 Frank Restly April 11, 2014 at 8:04 am


    “Economic efficiency is measured by the maximization of total output (or welfare).”

    Does it not seem odd that economic efficiency is defined without regard to input or input costs?

    Would it be considered an efficient use of labor to dig holes in the ground, bury hundred dollar bills, cover them up, and then dig them up again? Or perhaps paying a burglar to go around breaking windows so that shop owners will pay for replacement windows?

    Output would certainly rise in either case, but efficiency would not.

  22. 23 dwb April 11, 2014 at 11:44 am

    “because valuable resources are being used up in the process of transferring rents from the institutional investor to the HFT shop. The issue here is not a competitive issue; it is an externality issue, a divergence between private and social value or private and social cost.”

    Right, you are making the *assumption* that rents are merely being transferred from one rent-seeking entity to another, from the institutional investor to the HFT shop. If that’s the case, I agree it has really no benefit.

    That is precisely the assumption I question. The correct answer is “it all depends.” There are literally thousands of hedge funds that have been running algos for decades. Some are in your own back yard like Bethesda/DC/Towson, that try really hard to fly under the radar. They do all sorts of algorithmic trading.

    HFT is nothing but a new version of an old me: When Burton Malkiel published A Random Walk Down Wall Street, seasonal patterns in the stock market (“January effect”) were widely discussed. Back in the 70s it was a HUGE deal to compile the daily data needed to uncover seasonal patterns. Now, I can download it on my phone into a spreadsheet. Mysteriously, we no longer see those patterns.

    “HFT” is really two things which I think you are conflating when you say “creating an information advantage”: actual data about bid/offers, depth of market, and so on; With pattern-recognition computer code. Anybody can download or buy daily or now even intraday data to run an algo and detect a pattern. FRED has data I only dreamed about having in one place 15 years ago. There used to be seasonal patterns in short term bond markets (corporate treasurers would stash cash for taxes, etc. the deficit is seasonal!). Those patterns used to exist in the minds of traders that sat on the desk in places like Fannie Mae and Freddie Mac and other institutions, back in the day. It is significantly harder for institutional traders to extract rents based on seasonal patterns, for example.

    The “rents” accrue to the market makers who actually own the depth of market and bid/offer information (and institutional traders). Firms that run dark pools, etc. It is the bid/offer, depth of market, and other order flow information that is critical and the real asymmetry.

    If an HFT shop is merely running an algo on information it bought, it is no different than the Carfax example, or you or I obtaining trading data to (attempt) to trade on seasonal patterns. Like Carfax, it is your choice to ignore it. It’s, there if you want it. Carfax does not *own* the service history, they are merely disseminating it. If I am selling you a stock, it is not whether I happen to know about a pattern, its whether you could go discover it yourself. Carfax is 54.99, market makers charge more.

    However, if an HFT shop actually *owns* the depth of market data, then, yes, we are in your example where rents are merely being transferred from one entity to another. I am selling you stock based on truly private information. The depth of market and order data is the information, not the pattern recognition software. I do not have a problem with HFT per se, I have a problem with Goldman Sachs having an HFT shop because they own the data as a big market maker.

    Market makers should be *required* to disseminate the order flow. After that, there are 5000 hedge funds that will compete to drive out rents.

    In your example of the bigger pipe between NY and Chicago, speeding information flow is unlikely to be a social waste. It is not much different than a hedge fund buying a faster computer. Where its a social waste is when the people building the pipe also run a dark pool and effectively own the order flow information.

  23. 24 kaleberg April 11, 2014 at 10:09 pm

    I’m a retail trader, so I work at a totally different time scale from the HFT traders. I remember my broker once explaining that they were offering execution 1/16 of a point – that long ago – better than competing brokers. I cannot imagine making a trading decision based on that small a margin. Sure, the 1/16 of a point would be nice, but it wasn’t going to affect my buy/sell/hold decision.

    On the other hand, I still consider HFT to be no different from a bait and switch scam combined with a wire con. Still, it is comforting to think that the targets of the scam are the large investors. There’s the Red Queen to be considered as well. The HFT scam really only works before everyone knows about it and countermeasures are developed. Hitler conquered Europe with high speed blitzkrieg tactics which are now considered a major tactical error as they leave one’s flanks exposed and one at the far end of one’s supply line.

    The waste of talent, however, is real. My niece is graduating from Stanford with an engineering degree, but she already has a job in finance lined up, and more power to her. It would be nice if she was doing something better than selling umbrellas at rainy day prices, but she does want to build something real. Besides, if she makes enough money in finance, maybe she’ll get a chance to do something real.

    P.S. When they broke up Ma Bell and I learned about “prime” and “score”, I nearly dropped engineering for finance. Designing financial instruments was seemed almost like writing software, but without the nasty debugging part.

  24. 25 Jack Archer April 13, 2014 at 8:08 am

    It’s been a very long time since I read Veblen’s “Theory of the Leisure Class”, which was presented as an “economic study of institutions”. I looked up its date of publication — 1899, the highpoint of the reign of the robber barons? I think of Veblen as more of a sociologist than as an economist. Certainly he did a kind of economics that resembled nothing that most economists, on the right or left, do today. Nevertheless, his investigations of and comments on the ruling class of his day, and of our time as well, I think, and of the methods of social control and manipulation it used, seem as pertinent now as then. Its consumption is strikingly conspicuous, and on a scale that would have staggered even Veblen. I don’t recall if he considered how such wealth led to influence over and control of governments and legislators, by financing campaigns. I doubt that he would have been surprised that corporations (money) would be recognized as having speech rights, and perhaps religious beliefs. As primitive as his economic research methods may have been, he got a few things about us right.

  25. 26 David Glasner April 13, 2014 at 4:08 pm

    Frank, It is a basic theorem of linear programming that maximizing output and minimizing cost are really equivalent (or dual) problems. Digging holes in the ground is producing no value. Worthless production is valued at zero, so you can’t maximize production by using resources to produce worthless output.

    Dwb, Thanks for your comment which is very helpful and narrows the gap between us. However, I think there is another difference that needs to be addressed, which is that Carfex is, I am guessing, a fairly low cost operation because the information can be relatively easily obtained. So the waste of resources associated with its activity is relatively small. On the other hand, the costs incurred by HFT shops and algorithm writers are, I am guessing again, far greater that Carfax, so the aggregate effect of HFT is much greater than that of Carfax. Ultimately, the answer to the question we are discussing has to be empirically researched and quantified (probably with a pretty large measurement error). My hunch is that the effect, if properly measured, would be really large, but I admit that I am just guessing.

    kaleberg, Thanks for your interesting comments. I wish your niece well, and hope that she succeeds in getting a life of her own.

    Jack, I have not studied Veblen, and have only a vague idea of his work. Harold Demsetz, my thesis adviser at UCLA, once wrote a paper showing that J.K. Galbraith’s New Industrial State – I am not a fan of Galbraith’s either – borrowed liberally from Veblen.

  26. 27 Frank Restly April 14, 2014 at 5:36 am


    I admit my examples were not very good. But consider the window breaking / replacement example. If economic efficiency were solely about output gains (new window construction) without regard to input losses (broken windows), then we could say that any destructive activity is an efficient way to generate output?

  27. 28 dwb April 14, 2014 at 9:14 am

    “On the other hand, the costs incurred by HFT shops and algorithm writers are, I am guessing again, far greater that Carfax, so the aggregate effect of HFT is much greater than that of Carfax. Ultimately, the answer to the question we are discussing has to be empirically researched and quantified (probably with a pretty large measurement error).”

    First, The used car market is significantly smaller than total wealth and trading volume. According to NADAs 2013 report, the used auto market is ~40.5 million cars a year at an average price of ~17.5k. see here:

    That’s three quarters of a trillion annually.

    Household net worth is much larger: about 80 trillion dollars; total stock market volume is trillions *monthly* and bond market trading is measured in the trillions monthly:

    New bond issuance itself is half a trillion monthly.

    Some back of the envelope math: if 40.5 million used cars trade annually on a fleet of about 250 million, that’s 16% turnover. If 16% of household net worth traded annually, similar to cars, that would be over 1 trillion a month. This comparison is somewhat apples and oranges because we are comparing final retail sales to trading volume – trading volume is the equivalent of counting my transfer to the dealer and then the dealers transfer to a new buyer as two sales. If turnover was 16%, I would expect “trading volume” on cars reported consistently to stocks/bonds (which double counts intermediation) to be a multiple of this.

    The numbers in the stock and bond market are staggering. I am non-plussed by $300 million to build a cable. It’s tiny compared to trading volume. Fannie Mae wasted more than that on a stupid multi-year failed infrastructure project called “CORE” (dig into their old annual reports they were wasting $150 million a year at some point which many IT people jokingly people described as upgrading from early-80s MS dos). I think the grand total for CORE came to $500-600 million. A govt granted duopoly with no competition can waste vast sums of money/resources and still make outsize profits.

    As for how to measure it, what I care about are not costs, but profit margins. Are HFT firms earning outsize returns? Earning outsize returns over an extended period in my mind is *prima facie evidence* they are gaining an advantage via market structure or assymetric info. In finance, you measure profit margins on a trading strategy by comparing returns to a dumb strategy of investing in an index like the S&P 500, on a risk-adjusted basis. That is actually pretty easy to do. Hedge funds (“alternative assets”) as a whole generally do not report returns much better than indices on a risk-adjusted basis.

    Some firms like Citadel do publicize returns on HFT. (Google “Citidel tactical trading fund”). Many trading strategies are profitable the first few years when the are discovered, then cease to be profitable as they are replicated. If they continue to earn excess returns for an extended period, long enough that the returns are unlikely to be random chance, then I would conclude they are taking advantage of something or gaining some informational advantage.

    I think we are closer than you think.

  28. 29 TravisV April 20, 2014 at 4:18 pm

    Mr. Glasner,

    I’m a huge fan of yours so I thought I’d run this question I asked Prof. Sumner by you as well!

    “Prof. Sumner,

    Clark Johnson wrote the following (in the paper you linked to):

    “I believe Bernanke means that QE would lower real long-term rates. According to market participants, Fed QE announcements tend to raise inflationary expectations, thereby raising nominal rates on 10- and 30-year bonds.”

    I think this deserves more discussion from you. In particular:

    (1) Do you personally believe that QE3 has put downward pressure on real long-term interest rates (seems doubtful to me)?

    (2) Has Bernanke or anyone else on the FOMC EVER distinguished between the effect of QE on nominal rates vs. its effect on real rates?

    It sure seems like there’s very little actual evidence supporting Clark Johnson’s argument above. Why do Bernanke and Yellen keep saying that QE lowers long-term rates even though the truth is closer to the opposite? My theory: they say it because it sounds more appealing than “increase expectations of future inflation.” It’s all about selling the QE approach in a way that the general public feels like they understand (even though they don’t).”

  1. 1 Wednesday links: simultaneous success | Abnormal Returns Trackback on April 9, 2014 at 9:50 am
  2. 2 Washington Equitable Center for GrowthWashington Equitable Center for Growth | Things to Read on Thursday Evening: April 10, 2014 Trackback on April 10, 2014 at 4:51 pm
  3. 3 Links 4/11/14 (and Programming Note) | Mike the Mad Biologist Trackback on April 11, 2014 at 1:44 pm
  4. 4 Is Bitcoin Like High-Speed Trading? | Matthew C. Klein Trackback on April 16, 2014 at 3:32 pm
  5. 5 What most people seem to have missed about high-frequency trading | Spontaneous Finance Trackback on April 22, 2014 at 2:00 pm
  6. 6 No one really knows if HFT is good or bad | Trackback on April 23, 2014 at 9:08 pm
  7. 7 Book Review: Flash Boys | Trackback on April 23, 2014 at 9:08 pm
  8. 8 Thomas Piketty and Joseph Schumpeter (and Gerard Debreu) | Uneasy Money Trackback on May 29, 2014 at 10:14 am

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

About Me

David Glasner
Washington, DC

I am an economist at the Federal Trade Commission. Nothing that you read on this blog necessarily reflects the views of the FTC or the individual commissioners. Although I work at the FTC as an antitrust economist, most of my research and writing has been on monetary economics and policy and the history of monetary theory. In my book Free Banking and Monetary Reform, I argued for a non-Monetarist non-Keynesian approach to monetary policy, based on a theory of a competitive supply of money. Over the years, I have become increasingly impressed by the similarities between my approach and that of R. G. Hawtrey and hope to bring Hawtrey's unduly neglected contributions to the attention of a wider audience.

Enter your email address to follow this blog and receive notifications of new posts by email.

Join 265 other followers


Get every new post delivered to your Inbox.

Join 265 other followers

%d bloggers like this: