Mike the Mad Biologist

It might come as a shock to some readers, but I actually don’t mind investing–that is, long-term value investing–as long as it’s not valued more than labor through the tax code (Got Capital Gains Tax?). But this isn’t value investing, but surfing electrons:

The most striking periodicity involves large peaks of activity separated by almost exactly 1000 milliseconds: they occur 10-30 milliseconds after the ‘tick’ of each second. The spasms, in contrast, seem to be governed not directly by clock time but by an event: the execution of a buy or sell order, the cancellation of an order, or the arrival of a new order. Average activity levels in the first millisecond after such an event are around 300 times higher than normal. There are lengthy periods – lengthy, that’s to say, on a scale measured in milliseconds – in which little or nothing happens, punctuated by spasms of thousands of orders for a corporation’s shares and cancellations of orders. These spasms seem to begin abruptly, last a minute or two, then end just as abruptly.

Little of this has to do directly with human action. None of us can react to an event in a millisecond: the fastest we can achieve is around 140 milliseconds, and that’s only for the simplest stimulus, a sudden sound. The periodicities and spasms found by Hasbrouck and Saar are the traces of an epochal shift. As recently as 20 years ago, the heart of most financial markets was a trading floor on which human beings did deals with each other face to face. The ‘open outcry’ trading pits at the Chicago Mercantile Exchange, for example, were often a mêlée of hundreds of sweating, shouting, gesticulating bodies. Now, the heart of many markets (at least in standard products such as shares) is an air-conditioned warehouse full of computers supervised by only a handful of maintenance staff.

In a related vein, I commented on the observation that seventy percent of stocks are held for eleven seconds:

There is no way the efficient markets hypothesis, in either the weak or strong form, is operating. Eleven second holds can’t be based on information about the quality of the investment because the underlying investment doesn’t fluctuate that rapidly.

Ed Harrison remarked:

The fact that the vast majority of stock market trades are held for 11 seconds shows that the stock market is not a real market with real traders governed by the law of supply and demand, and with no real price discovery.

Years ago, when banking first became computerized, there were apocryphal stories about hi-tech thieves ‘stealing’ the fractions of a penny that’s rounded off on daily interest payments and depositing them in an account–do this at a large enough bank and you could make a pretty penny or two. I’m not sure how that’s different than millisecond-scale trading:

Some, but not all, automated trading strategies require ultra-fast ‘high-frequency trading’. Electronic market making is the clearest example. The ‘spread’ between the price at which a market-making program will buy shares and the price at which it will sell them is now often as little as one cent, so market-making algorithms need to change the quotes they post very quickly as prices and the pattern of orders shift.

On a ten dollar share that’s a 0.1% profit; on a $40 share, that’s a 0.025% profit. I bring this up because a great way to eliminate this utterly useless activity would be a transaction tax of 0.1% (give or take). These guys would be out of business. Then maybe we could reallocate these smart people to do something useful.

By the way, computerized trading is now playing havoc with the commodities markets.

It’s their casino, prepare to be robbed blind in it….

Comments

  1. #1 david
    May 17, 2011

    A transaction tax hurts everyone. A small percentage seems small, but it’s not when it’s not based on the profit made. Whenever you & I buy stocks it would cost us. Even mutual fund fees would go up because of it since they buy & sell huge dollar amounts of stock for the fund.

  2. #2 Kevin
    May 17, 2011

    @ David – When I buy stocks of my random index fund, I already pay a 2-3% fee. Making that 2.1-3.1% isn’t going to affect my future outlook all that much. And if it means more stability in the market, it’s a very small price to pay.

  3. #3 lynxreign
    May 17, 2011

    @david

    And it doesn’t hurt people who aren’t in the market, but fluctuations of stock prices based on nothing can hurt the entire economy.

  4. #4 A
    May 17, 2011

    david: “you & I buy stocks ..” Most Americans own no stocks, though many own them indirectly, through their pension fund or 401k. Would you want to have your pension invested in high-frequency trading?
    david “they buy & sell huge dollar amounts..” but it is a small -percentage- proposed tax. So, if your 401k has $100,000.- in it, 0.1% is $100.-; but I’d hope that your total fund does not turn over every year; if it does, you should change your fund manager (and accuse him of ‘churning’ your account to generate fees for himself). If your $100k were accumulated over 10 years, and 10k/year added, the additional $10 in tax/year would probably be dwarfed by the fund manager’s fees, probably in the 1%+ range (~$500+ over the 10 years). (Notice that many fund fees are hidden in the ‘small print’, and the fund advertises profits after ‘expenses and fees,’ which certainly contain transaction/brokerage fees far bigger than 0.1%, as Kevin pointed out above)

    The lesser volatility, or greater stability of the stock market would make capital allocation more efficient (more going to more productive enterprises), and improve over-all profits, including those of your 401k (unless yours is into high-frequency trading or invests in lottery tickets). That’s a consideration in an economy where ~40% of all corporate profits go to the ‘financial services’ sector (do they deserve so much for so thoroughly wrecking the rest of the economy?)

    A rational government would have such a financial transaction tax, (not only for the revenue, but also the stability), but ours is owned by the banks and traders already. The stock exchange people claim of course, that with such a tax, the NYSE would be at a disadvantage to foreign exchanges, all American companies listed would move to the London or Shanghai exchange and incorporate in the Bahamas. And jobs would be lost. (Especially those of people making campaign contributions, or offering jobs to disgraced congressmen). And small kittens die a painful death.

    I recommend Mike for reminding us of it, and ask all readers to mention it to any congressman/senator they might meet (and ask: could you introduce it, and if not, why not?)

  5. #5 Ben W
    May 17, 2011

    Who cares if 70% of stocks are held for 11 seconds, or trading occurs in milliseconds? Why does this mean we should have a transaction tax? If it’s not broke, why should we try fix it?

    I’d bet that most of these trades are just arbitrage – like setting up a trade between a buyer in Europe and a seller in America, and scraping a fraction of a penny per share off of said trade. Unless you can really show, beyond a reasonable doubt, that high-frequency trading hurts the stock market or general welfare, I don’t see any reason why you should be targetting HFT for special taxes.

    Sure, HFT may have played a part in commodities dropped quickly.. but doesn’t this happen normally, even without HFT? How fast did the stock market crash in the 1920s, or in 1987, or after the stock market bubble? The commodity markets are well-known for being even more bipolar than the stock market. Show me data that HFT is increasing the volatility, and that this short-term volatility is bad.

    Furthermore, how is this evidence that

    “There is no way the efficient markets hypothesis, in either the weak or strong form, is operating.”

    ?

    It’s quite possible the markets are still efficient over longer periods. At the least, that’d mean that you can’t say that HFT is evidence that efficient market theory (EMT) is wrong. More likely, though, is that HFT takes advantage of short-term market inefficiencies.

  6. #6 elspi
    May 17, 2011

    I don’t need to say that HFT proves the EMH is false since we already know that the EMH is false. (See margin call)

    How the hell can a someone who pays 6% sales tax think that a .1% transaction tax would be the end of the world?

    The point of investment is supposed to be that you find a company that you believe in and you invest money in that company.

    The transaction tax would encourage this while discouraging speculation (which is what caused the current mess).

  7. #7 Ben W
    May 18, 2011

    I agree that EMH is (mostly) false, but so what? HFT still does not act as evidence either for or against EMH. We’d all look askance at you if you said “The aether theory of light propagation is incorrect, and here are some tarot cards that prove it”.

    How the hell can a someone who pays 6% sales tax think that a .1% transaction tax would be the end of the world?

    I only pay 6% sales tax on purchases, and those purchases are typically made out of my earnings. But if my job consisted of earning a very small margin on many transactions, then a 0.1% tax could kill my job, obviously.

    The transaction tax would encourage this while discouraging speculation (which is what caused the current mess).

    Ermm, really? How exactly would a 0.1% transaction tax have prevented the subprime mortgage crisis? Would people really not have bought and flipped houses if they had to pay an extra 0.1% (after already paying closing costs of several percent)?

    So I think you’re confusing HFT and speculation. Heck, in some types of HFT, the buyer and seller have already been found, and the trading just fills the gap between them. Obviously this is not speculation*, if you already have a buyer and seller, no?

    Or similarly, arbitrage HFT involves finding mis-priced stocks or bonds and correctly pricing them. Isn’t that the heart of value investing, to find undervalued investments and sell them higher? Why does it matter how long you hold the position – if you can profit from recognizing a 0.1% mispricing, why shouldn’t you?

    Speculation, on the other hand, works quite happily on many timescales. Enacting a transaction tax is unlikely to stop investment banks from speculating on the price of oil as a hedge against inflation and thereby driving the price of gas up. Oddly enough, speculation increases volatility, while HFT usually decreases volatility (with the exception of Flash Crashes), since it increases liquidity…

    But, really, the worst excesses of the markets would hardly be stopped by a 0.1% transaction tax. Hell, they wouldn’t even blink. The classic investment bubble, where people bid internet stocks or railroad stocks or houses or tulips up hundreds of % more than they’re worth, these manias are driven by the zeal of true believers.. and it’s extremely hard to believe that a lousy 0.1% tax will do anything to dissuade them.

    Silver went up 180% from last August to the end of April. Qualcomm’s stock rose by 2800% in 1999, not much different from that of RCA in the 1920s, or of South Sea Trading Company in the 1700s. And today the price of real estate in major Chinese cities mirrors our own recent housing bubble. What difference can a small transaction tax make, when there’s profit of 100s of percent to be had? You’ll do much better to tighten margin requirements or monetary policy.

    *Here I use “speculation” in the common sense of the word, not as the financial definition. If we used the financial definition, most “investors” and homebuyers should also be called speculators.

  8. #8 Wow
    May 18, 2011

    “If it’s not broke, why should we try fix it? ”

    And since the stock market IS broke, we SHOULD fix it.

    “How exactly would a 0.1% transaction tax have prevented the subprime mortgage crisis? ”

    It wouldn’t, but the crisis wasn’t mortgages. It wasn’t sub-prime lending. It was naked trading and leveraging assets to the tune of 100x the GDP that caused the crisis.

    However, the problem for libertarians, randians and most conservatives, this would put the problem at the feet of the Hole Men Of Capitalism: the rich. If they can make it the fault of mortgage purchasers, the problem becomes one of the New Leper Of Capitalism: the poor.

    Much better for their equanimity.

    PS how can you speculate on milliseconds? How much information (or, indeed, thought) can be done in a millisecond? To say nothing of microsecond transactions…

  9. #9 Nomen Nescio
    May 18, 2011

    Who cares if 70% of stocks are held for 11 seconds, or trading occurs in milliseconds? Why does this mean we should have a transaction tax? If it’s not broke, why should we try fix it?

    the fact that stocks are held for mere seconds is prima facie evidence that the stock market IS broken. as has been stated, the underlying investments that stocks represent do not change that rapidly, so if people are holding stocks for such short times and yet profiting from the trades, then stock market values and trades can no longer be based on the companies the stocks represent.

    well, what ARE they based on, then? if the answer isn’t something tangible, solid, and reliable — something with objective existence in the real world — then the stock market has become a shared hallucination of some sort, its prices purely delusionary. that needs fixed.

    i don’t know if a transaction tax would be sufficient to fix things, but hell, it can’t hurt. certainly not hurt any worse than decoupling the stock market from objective reality! at very least, not hurt anyone but the wall street fat cats, and they can afford to be hurt a little by some higher taxation. soak the rich, that part’s long overdue anyways.

    But if my job consisted of earning a very small margin on many transactions, then a 0.1% tax could kill my job, obviously.

    and if your job consisted of earning a very large profit on very few bank robberies, then being thrown in prison for robbing a bank could also kill your job. so what? the fact that people can profit from harmful manipulations of an inefficient market is no argument for letting them continue!

  10. #10 Eric Lund
    May 18, 2011

    Who cares if 70% of stocks are held for 11 seconds, or trading occurs in milliseconds?

    You should, if you invest/gamble in the stock market. The average return of all stock market investors, before expenses, is the return on the overall stock market. The only reason to engage in HFT is if it increases your net return, compared to alternative strategies, by enough to offset the expenses (which are smaller per share than for retail investors but presumably not zero). That extra return, assuming that it actually does exist, comes at the expense of retail investors. I don’t see how this could work other than as a salami scam: some retail investor ends up paying a penny or two more per share when buying, and getting a penny or two less per share when selling. A transaction tax of 0.1% won’t necessarily eliminate the problem (it may still be profitable to do HFT of stocks worth less than $10 per share), but it will reduce the incentive for churning more expensive stocks and the profits from churning cheaper stocks.

  11. #11 Ben W
    May 18, 2011

    Wow says,

    It wouldn’t, but the crisis wasn’t mortgages. It wasn’t sub-prime lending. It was naked trading and leveraging assets to the tune of 100x the GDP that caused the crisis

    We agree, or close enough. Our system blew up because of
    1) Bad investments. Mortgages, packaged into bonds, started defaulting after the housing market was overpriced and overbuilt.
    2) See above, but leverage it 10x or 20x.
    3) Now make the financial system interdependent – the banks buy hedges off of each other, such that their obligations to each other reads like a bowl of spaghetti.

    And when one bank gets taken down by highly leveraged bad bets, it’s a chain of dominoes as others lose their hedge counterparties. Suddenly, they’re exposed to much more risk than they’d counted on…

    But to me, the logical way to fix this is
    1) Decrease the amount of leverage people and banks can use.
    2) Decrease the fragility (interdependence) of the financial system.

    Nomen Nescio says,

    the fact that stocks are held for mere seconds is prima facie evidence that the stock market IS broken. as has been stated, the underlying investments that stocks represent do not change that rapidly, so if people are holding stocks for such short times and yet profiting from the trades, then stock market values and trades can no longer be based on the companies the stocks represent.

    The underlying investment doesn’t change that fast (well, usually), but the price does. I think you’re half-right, half-wrong: even with HFT, the stock market values can still be based on underlying investment, since HFT often doesn’t change the stock market value much. Stock market trades, OTOH, are where HFT makes its money. But who says that stock market trades should always be based on the underlying value of the company?

    Does a grocery store “invest” in the inherent value of oranges, or do supermarkets just serve as a way of getting oranges to us in an affordable manner? It’s just like any of the old traders. Somebody sets up a trade route between you and the next country.. sure, maybe the shipowner who buys your silk and trades it for spices makes a killing, but you’re also able to sell your silk for more $$ because the trade route was set up.

    i don’t know if a transaction tax would be sufficient to fix things, but hell, it can’t hurt.

    Orly?

    First, I’m wary any time someone says “hell, it can’t hurt”. Second.. it’s been tried before, and it didn’t work too great. Liquidity and capital gains tax receipts both dropped. A lose-lose situation.

    that needs [to be] fixed

    FTFY.

    Eric Lund says,

    The only reason to engage in HFT is if it increases your net return, compared to alternative strategies, by enough to offset the expenses (which are smaller per share than for retail investors but presumably not zero). That extra return, assuming that it actually does exist, comes at the expense of retail investors. I don’t see how this could work other than as a salami scam: some retail investor ends up paying a penny or two more per share when buying, and getting a penny or two less per share when selling.

    True enough – if the trade would have taken place anyway, and at the same price. But how do you know that’s so?

    I regularly trade in stocks with a bid/ask spread of more than 2%, and that can hurt. Often, there’s investments I’d like to make, but can’t, because the volume is so low and the B/A spread too wide. If you manage to find some seller in China and shrink the B/A spread for both of us, I’m grateful to you, even if you also profitted.

  12. #12 Eric Lund
    May 19, 2011

    I regularly trade in stocks with a bid/ask spread of more than 2%, and that can hurt.

    So obviously you know that these stocks tend to not make good day trading candidates, because the transaction expenses will eat your profit. (Also, if an 0.1% transaction tax affects your decision whether to invest in such a stock, ISTM that such an investment would be dubious from the get-go.) HFT algorithms won’t touch these stocks either, because they make day trading look like buy-and-hold. The stocks that get do get churned by HFT algorithms are, instead, the shares that get traded on the NYSE. As the linked LRB article explains, the owners of the computers that run the HFT programs pay $10k per month rent to have those computers physically located in the same building as NYSE’s computers. They do this because the speed of light is roughly 1 foot per nanosecond, and nanoseconds matter to an HFT algorithm. It makes no sense for these computers to waste CPU cycles looking at stocks that don’t trade on the NYSE (at least while the NYSE is open), since these computers do not have any particular advantage for trading in other markets. NYSE stocks already have particularly thin margins–these days an $0.01 bid/ask spread is common–and they are the stocks retail investors are most likely to buy. Many NYSE stocks have enough volume to support an $0.01 bid/ask spread even without HFT, so it’s not obvious that there is any benefit to retail investors to having HFT in place.

  13. #13 Wow
    May 19, 2011

    “We agree, or close enough. Our system blew up because of
    1) Bad investments. Mortgages, packaged into bonds, started defaulting after the housing market was overpriced and overbuilt.”

    You start off with “we agree” then go on to say the opposite.

    No, the Bad Investments were the naked shorts. Betting on the stock exchange. Where, since you don’t even have to say you’ll buy ANY stock (unlike leveraged purchasing), means you can “buy” stocks worth 200,000 billion dollars with $0.

    Makes 20:1 leveraging of the fractional reserve banking for mortgate lending of ~10-20 to 1 look puny.

    “2) See above, but leverage it 10x or 20x.”

    See infinity to 1.

    “3) Now make the financial system interdependent – the banks buy hedges off of each other, such that their obligations to each other reads like a bowl of spaghetti.”

    Nothing to do with mortgages. That was due to removal of banking oversight that the bankers SWORE they didn’t need because they knew what to do and were nice people.

    Yah.

    IT WASN’T MORTGAGE LENDING.

    It was rich people betting on HFT and naked shorts, using mortgages to kid on they were doing some real work.

  14. #14 Ben W
    May 23, 2011

    My apologies, this is quite long. It’s also probably my last post on this thread.

    Eric Lund said,

    As the linked LRB article explains, the owners of the computers that run the HFT programs pay $10k per month rent to have those computers physically located in the same building as NYSE’s computers. They do this because the speed of light is roughly 1 foot per nanosecond, and nanoseconds matter to an HFT algorithm. It makes no sense for these computers to waste CPU cycles looking at stocks that don’t trade on the NYSE (at least while the NYSE is open), since these computers do not have any particular advantage for trading in other markets.

    Perhaps. Probably these computers are especially focused on churning NYSE stocks, although they might also gain an advantage in arbitraging between the NYSE and overseas markets.

    Look, if you have a problem (as I do) with frontrunning, which some HFT algorithms do, then by all means, let’s make that illegal. But not all HFT is frontrunning; statistical arbitrage and market making are also common, and quite helpful in providing liquidity. So I don’t see any reason to penalize all HFT with a transaction tax, when only some types of HFT are immoral.

    Wow said,

    You start off with “we agree” then go on to say the opposite.
    No, the Bad Investments were the naked shorts. Betting on the stock exchange. Where, since you don’t even have to say you’ll buy ANY stock (unlike leveraged purchasing), means you can “buy” stocks worth 200,000 billion dollars with $0

    Oops. I guess we don’t agree.

    It seems to me that shorting the stock exchange wasn’t a bad investment at all, seeing as those shorts were quite profitable as the financial crisis unfolded. But I’m guessing that’s not what you meant: You meant that shorting stocks in the exchange caused the crisis, yes?

    I find that extremely difficult to believe. There are very clear, well-understood mechanisms connecting bad mortgage loans with the instability of the banks, and by extension, the rest of the financial system. But the mechanisms connecting stock shorting and the collapse are much weaker, if they even exist at all.

    See, the normal operation of a bank is not materially affected by its stock prices. The stock price is generally only important when the bank is trying to sell stock to raise capital. (It could also be important when shares of the bank are owned by another bank, or if a bank has bets in the form of CDS on the price of another bank; however, these are unimportant/rare compared to the mortgage bonds owned by banks.) There’s no real way for stock shorting to mess up the financial system here. If I’m wrong, please enlighten me, but I’ll expect you to explain the actual mechanism.

    On the other hand, bond failures matter quite a bit. I’ll explain:

    Banks have to keep a minimum reserve requirement: for every ~$15 of bonds (“assets”) they own, they have to hold $1 of their own money (“capital”). This $15 of assets is risk-adjusted, such that higher-risk assets “cost” more to hold than safer assets. This makes sense; a higher-risk bond is more likely to fail, which costs the bank money.

    Ex: You’re a bank with $15 of assets and $1 of capital. Unfortunately, you own $0.25 of a high-risk bond which failed; you’ve just lost $0.25. That loss comes out of your capital, which is now $0.75. Now you have $14.75 assets and $0.75 of capital, and your asset/capital ratio is ~20; too high. You have to sell other assets to raise money and get your ratio back down to 15. (End of example).

    So to keep the ratio in line, you adjust the cost of assets by their risk. If a bond has a 50% chance to fail, it takes up 2x as much space in the asset/capital ratio, if it has a 25% chance to fail, it takes up +33% space, etc.

    Now, there were two problems for banks:

    First was that the junk mortgage bonds were waayy overrated in terms of safety. The bonds started failing way more than they should have (given their risk rating), so they took much bigger chunks out of the banks’ capital than was expected.

    The second thing is an accounting practice called “mark to market”. Keep in mind what I said about assets and capital: profits on assets add to capital, losses take away from capital.

    So, when a bank buys a bond, it had two choices:
    #1) It can keep the bond on the books at the price it paid all the way until the bond expires, and then count its profits or losses.

    #2) The bank can mark the price of the bond to the market price, and adjust its profits or losses continually. When the bond value goes up, this is great! It means profits go up sooner (rather than waiting all the way until bond expiry), and those profits immediately become “capital”, letting the bank leverage up more. As a bank employee who marked profitable bonds to market, you’d probably get a nice pat on the back and a sizeable bonus for increasing bank earnings faster.

    But the caveat is this: once the value of a bond is marked to market, you can’t go back. It remains at the market price; there’s no marking it back to the original price. So if you own a bond on subprime mortgages, and the value drops because people start defaulting on their mortgages, then your bank immediately takes that loss, which comes out of capital, and so you sell assets to bring your capital/asset ratio back into order. This is no big deal until all the other banks are having the same problem, and suddenly they’re all scrambling to sell whatever they can. They can sell safer bonds, but that drives the market values of these bonds down, which means that other banks are now booking losses on the safer bonds, and so they also have to raise capital by selling something else.. etc. It becomes a vicious cycle.

    Ok, so I said that stock price doesn’t matter. That’s not 100% true, the bank could (hypothetically) issues stock to try to raise capital, as an alternative way to get the capital/asset ratio back down. But there’s no guarantee that your bank will be safe even if they can sell enough stock to fix their capital/asset ratio – if other banks are still selling bonds, it’ll keep driving down the price of bonds down at your bank also. All the banks would have to issue enough stock to make up for their losses in the housing market, and good luck with that – that’d be hundreds of billions of dollars.

    In short: yes, this was very much a mortgage bond problem, not a short-selling problem.

The site is currently under maintenance and will be back shortly. New comments have been disabled during this time, please check back soon.