Every time Paul Krugman writes a column about something other than finance, I wholeheartedly agree with him, like I always have. But his columns about finance are downright uninformed and naïve. Unfortunately, today's column is about finance. Krugman claims that both Andrew Hall, the head of Citi's coveted commodities trading operation, and high frequency trading (HFT), are "bad for America." Why? Because, according to Krugman, "speculation based on information not available to the public at large" is socially useless, and could even be destructive. Where to start? First of all, Krugman confuses HFT with "flash orders," which is when an exchange or ECN shows an order to certain algorithmic traders about 30 milliseconds before showing it to everyone else. Flash orders allow trading based on private information, and should legitimately be banned. But HFT isn't the same thing as trading based on flash orders. High-frequency algorithmic trading that isn't based on flash orders (the vast majority of HFT) isn't trading "based on information not available to the public at large." The information has indisputably been released to the public at large. (The fact that algo traders can react to the information faster than Joe Retail Investor doesn't make the information any less public.) The Jack Hirshleifer paper he cites to support his argument is therefore completely inapplicable. Krugman clearly didn't fully understand the concept of "high frequency trading" when he wrote the column. Yet just six days ago, in a post criticizing Martin Feldstein for writing plainly inaccurate op-eds on health care, Krugman said, "Yes, I write about subjects on which I’m not an expert — but I do my homework first." Umm, not this time. Krugman's argument against Andrew Hall is just plain illogical. He starts out by saying:

Just to be clear: financial speculation can serve a useful purpose. It’s good, for example, that futures markets provide an incentive to stockpile heating oil before the weather gets cold and stockpile gasoline ahead of the summer driving season.
But then he claims that Hall's job is "bad for America" because—I kid you not—he makes his money as a speculator:
The Times report suggests that [Hall] makes money mainly by outsmarting other investors, rather than by directing resources to where they’re needed. Again, it’s hard to see the social value of what he does.
It's not possible to reconcile these two statements. It's simply inconceivable that Krugman—one of the best economists of his generation—actually believes his argument against Hall. I'm sorry, but Krugman's columns about finance are just as irresponsible as Feldstein's op-eds about health care.


Don said...

I want to try and put Krugman's concern in a different form:

Does it matter what Joe Retail thinks about the Financial System?
Is it important that he buy into it?
Is it important that he thinks that the system is fair?

Suppose Joe Retail doesn't accept that the govt's actions during this crisis were fair, but that they were skewed towards helping the financial industry?
Suppose that Joe Retail doesn't find the proposed changes to our financial system significant enough?

Then, it might well be the case that an issue in which Joe Retail finds himself at a disadvantage will gain symbolic significance and lead to a further erosion of his trust in the system.

Since I'm a fan of Burke, I take these kinds of social and political justification questions seriously. I tend to agree with you about both HFT and Speculation, but I can also see the point of slightly changing these practices to make Joe Retail feel, and even be, treated better by the financial system.

Some people find such worries silly. Fine.

Don the libertarian Democrat

Anonymous said...

Just to nitpick your nitpicks though, as I understand how some of the algo HF programs work, and the speed at which they do so, don't they move the price before a non-HF entity could? And therefore the information is stale by the time it's effectively public.

Doesn't this just create a millisecond insider trading window kind of problem?

Anonymous said...


I am a professional trader. So I would like to weigh in on this.

First, what Joe Retail and many professional pundits need to realize is that, when there is a crisis, the U.S. govt. will tend to view the U.S. financial system similar to the way it views the U.S. defense apparatus.

In other words, it wants it to be completely superior to all other global alternatives. It does not care about "fairness." It would, if being completely honest, say that Joe Retail and Joe Public are better off being citizens of a nation that has a superior financial system, even if it allows some to benefit "unfairly" vs. citizens of a nation with a second or third rate financial system - but without any incidents of "unfair" compensation or whatever.

A financial system that is superior to the global alternatives means having firms that are not only solvent, but also very strong financially.

Assume that a war has cost the military a more than expected amount of tanks. We both know that there would be a very focused effort to replace those tanks asap. The TARP and the other programs are the financial equivalent of "tank replacement - asap" programs. And yes, I know that we "attacked ourselves" - so the war analogy is not perfect. But if we lost a bunch of tanks to mutinous sabotage - the response would be the same. Replace the tanks and get them battle ready - asap.

Additionally, today's Joe Retail is not tomorrow's Joe Retail. Just as the folks who took on too much leverage and caused this crisis did not learn (or even know of) the role that excess leverage played in the Great Depression or even Long Term Capital Mgt. in the 1990s - tomorrow's Joe Retail will enter the investment arena post stabilization and post "angst."

Re: HF trading. Joe Retail should be more concerned with the long-term prospects of the stocks and bonds in his portfolio. If he wants to be a bit more sophisticated, he can ponder "sector switching" and other such macro view pursuits. Maybe even positions in foreign currencies. None of these approaches will be affected by 30 milliseconds. A momentary hiccup in anyone's broadband connection can eat up 30 milliseconds.

HF trading is to Joe Retail what Formula One racing is to you and I - two "retail drivers."


Don said...


Thanks for the response, which, I have to say, makes a lot of sense to me.

Take care,


Economics of Contempt said...


RBB is right: retail investors don't really matter that much; it's the institutional investors and the market-makers who really matter. And even though HFT seems unfair to Joe Retail — and he may even think it's unfair — it's pretty much indisputable that retail investors have benefited tremendously from HFT. HFT has pushed bid-ask spreads down to are obscenely low levels — measured in pennies now, which wasn't even possible for a long time. Even if retail investors believe that the system is somehow unfair, they quickly realize that the cheap and easily-available liquidity that HFT shops provide is more than worth it. I'm not too worried about retail investors losing confidence in the system en masse, and even if they did, it wouldn't be catastrophic or anything.

Economics of Contempt said...

Anonymous @ 12:27 --

The problem with that argument is that with or without HFT, there's always going to be some entities that are faster than others. Investment banks hire armies of analysts and traders, and invest millions in trading infrastructure, to help them interpret and react to news faster than other investors. The fact that investment banks have the capacity to read, interpret, and trade on news faster than retail investors could ever dream of doing doesn't affect when the information becomes "publicly available" though. Should everyone be required to wait until the average retail investor has read and understood the information before they're allowed to start trading on it? Surely not.

There's always going to be a single point in time at which information becomes "publicly available." You can't control how fast market participants react to the information. Some will always react faster than others. But the speed at which participants react to public information doesn't change the fact that the information was made publicly available.

Anonymous said...

Economics of Contempt,

Never posted before, but really like your blog. Agree with your points here. One thing that I wanted to point out though is that the commentary on flash orders is horribly flawed.

Marking your order as flash is completely voluntary -- the default order type on the exchanges that have been mentioned doesn't flash. NASDAQ, BATS, DirectEdge disseminate the feed as widely as possible given RegNMS. BATS gives their market data away for free, and I'm told that NASDAQ has over 1000 firms that subscribe.

Really the issue here is sub-penny pricing. Since stocks like Citigroup aren't allowed to trade in fractions of a penny, the exchanges have resorted to tryign to come up with inventive fee structures. Since most exchanges charge roughly 1/4 of a penny to remove liquidity, if the market for Citigroup is $3.00 at $3.01, to take liquidity, the market is really $2.9975 at $3.0125 after fees. There's no reason why a firm should be allowed to "flash" a $3.01 bid when a market is $3.01 offered, which effectively allows them to buy for $3.0075 if filled.

Of course, a simpler way to do this would be simply to allow the market to "lock" i.e. stay $3.01 bid at $3.01 offered on the consolidated tape, but currently, that's not allowed under RegNMS.

The irony is that someone can ping a broker-dealer like Getco for $3.01, effectively letting Getco see a locked market for a split second (i.e. they can sell $3.01 without taking the risk of leaving a firm, resting quote that the rest of the market can trade with) but the same opportunity isn't afforded to someone trading on the public exchanges. I think the public exchanges deserve a way to compete.

*Note* It's possible the order originator might not get to set the flash tag if they are using an agency relationship, i.e. sending an order through TD Ameritrade. In this case, I would support requiring an affirmative "opt-in," but if you are going to question flash here, you should question the whole category of payment for order flow relations.


Anonymous said...

Should read

"There's no reason why a firm SHOULDN'T be allowed to "flash" a $3.01 bid when a market is $3.01 offered, which effectively allows them to buy for $3.0075 if filled."

Here is a good article by the BATS CEO on the subject.


Don said...

Economics of Contempt,

I'm late in saying this, but thanks for the excellent response.

Take care,


Anonymous said...

30 ms is less than the time it takes to transmit the information across the country. When you take the discussion to this level you also need to understand information technology to have an informed opinion.


Anonymous said...

Well, yes, I agree you need to understand the details to have an intelligent discussion, but I don't really understand why whether it's 30 milliseconds or 0.5seconds (which it is by the way for some forms of flash, see http://www.batstrading.com/resources/features/bats_exchange_BOLT.pdf) or 5 seconds has much to do with the debate.

So long as exchanges provide liquidity rebates of approximately 1/4 penny to people who post liquidity, and charge slightly more than 1/4 penny to people who take liquidity, then if the "market" is 3.01 bid at 3.02 then it's really 30.0075 bid at 3.02 offered.

And there's no reason not to allow someone to send a "post only" bid of 3.02... so that the market is 3.02 bid at 3.02 offered and really 3.0175 at 3.0225

So flash is really a tool to allow someone to bid the same price as the offer (since they will be different after fees) and a tool to help with subpenny pricing. Currently bidding the same price as the offer and subpenny pricing aren't allowed per RegNMS, even though they would given the investor tighter prices to trade.

Most of the flashes and bolts being sent are actually non-routable orders that stay on the book for 0.5 seconds, and if they are still on the book after getting cancelled, the originator will likely replace them. Ironically, high frequency firms are often the ones SENDING the flash orders because they don't have an extra 0.5 penny of pricement improvement in their economics for the given order. They are providing retail investors a tighter spread to execute against.

Sure, someone in California may not be able to react quickly to them, but if they had a liquidity taking order in flight, they'll get a better price.

If they want to get a better look, there's plenty of datacenter space available in New Jersey with super low latency to the exchanges, and it's pretty cheap relative to trying to play in any other area of the financial markets (try getting dealer access to the corporate bond, CDS, etc markets). You can argue that the cost creates a "two tiered market" but this isn't a flash order critism -- if the market for Google is 25 cents wide and someone improves it by a nickel, the people with the lowest geographic distance and fastest systems will get to react first... nothing to do with flash here.

So it's weird that there is such a public outcry about flash, especially when it's completely optional to mark your order as a flash. If you're a retail investor and scared that someone is going to jump in front of your 200 share order (a meaningless worry in my opinion given that tens or hundreds of thousand of shares are often on the offer for many stocks), then don't mark your order as flash. You'll still have the opportunity to get price improvement from everyone else how submits flash orders and tightens the spread.

Furthermore, once you start to view flash as a means for sub-penny pricing (a result of the super tight bid/offer spreads of late that EOC mentions), the whole notion of front-running doesn't make any sense. If the market for a stock is $110.00 bid at $110.10 offered, and you bid $110.08, and someone lifts the $110.10 offer, that's not front-running. You submitted a public order to the market, and someone reacted to public information. The notion that these flashes are being given out to a select community is completely baseless. NASDAQ disseminates flashes over their normal direct market data feed that over 1000 firms consume. They have no additional charge to view flash orders. BATS gives away market data for free, you just have to write software to interface. The only place these quotes aren't getting published to is the consolidated tape, and that's specifically because RegNMS does not allow the bid price to equal to the offer (locked market). Change the Regs to allow locks, and make flashes go to the consolidated tape. Or move to sub-penny pricing. Either choice seems like it would be much more beneficial to investors than what is being proposed.


Anonymous said...

Seems like everyone is on a witchhunt though. I think it would be much more in the public interest though, to just leave things as is, and work on getting individual investors fair access to corporate bonds (if you want to talk about a two tiered market structure, look no further than here). Or investigate options, in which one firm buys a bin to get special privileges as a designated market maker.

Any way, just my two cents

Economics of Contempt said...


Sorry it's taken me so long to respond; it's been a very busy week. I agree that the commentary on flash orders has been horrible, and even irresponsible. Journalists/pundits have badly misrepresented the motivation of exchanges/ECNs in offering flashes -- they all portray it as some sort of blatant give-away to Evil Wall Street Banks at the expense of Wholesome Retail Investors. They refuse to even acknowledge that there's a non-nefarious explanation for flash orders, because, well, it's extremely complex, and beyond the comprehension of pretty much every mainstream journalist.

That said, I still think flash orders should be done away with, and the parts of Reg NMS that prompted exchanges/ECNs to offer flash orders should be changed so that no one has an incentive to offer them anymore (i.e., allowing the market to lock, and any other necessary changes). Reg NMS is definitely not my area of expertise -- seeing as I was virtually computer illiterate during most of the "computerization" of trade execution -- but in my general experience, Reg NMS is a total disaster zone.

Anonymous said...

A person exhibitedcheap D3 Gold several disgust in which Martin Feldstein held to help his or her particular convictions regarding an AIG bailout. Feldstein basically thought that the organization should assume responsibility Buy Guild Wars 2 Gold because of its activities.

Calvin Brock said...

Because, according to Krugman, "speculation based on information not available to the public at large" is socially useless, and could even be destructive. Where to start? First of all, Krugman confuses HFT with "flash orders," which is when an exchange or ECN shows njtaxpreparation.net

Trần Duy Thuận said...

Trần Duy Thuận Blog đang cần nhận gia công tại nhà. Ai có cần người nhận gia công tại nhà tphcm liên hệ mình nhé. Truy cập vào đặt vé máy bay giá rẻ vietjet để có thêm kinh nghiệm khi đi máy bay nhé.

mua thuoc tot said...

Giải độc gan là phần quan trọng để giúp cơ thể hoạt động tốt hơn. Vậy thuốc giải độc gan nào tốt ? Có nhiều loại tốt , tham khỏa sau nhé. Collagen giúp trẻ hóa làn da , giữ được sự thanh xuân lâu hơn cho người dùng , bổ sung collagen bằng thực phẩm nhiều collagen vừa hiệu quả lại rất an toàn. Để có một trí nhớ tốt thì cần phải làm nhiều việc , vậy làm thế nào để có một trí nhớ tốt ? có thể dùng thuốc bổ não , ăn thức ăn tăng cường trí nhớ ,.. Muốn giảm cân phải làm sao ? thực đơn giảm cân trong 1 tuần sẽ giúp bạn giảm cân nhanh và hiệu quả. Mất ngủ có nhiều nguyên nhân. Vậy bị mất ngủ phải làm sao ? Có nhiều cách trị mất ngủ hiệu quả như dùng thuốc ,..