March 31, 2011
Anne Hathaway and Automatic Trading
A humorous financial story making the rounds concerns the apparent relationship between media mentions of the actress Anne Hathaway and jumps in the stock price of shares of Warren Buffett’s Berkshire Hathaway.
The presumed culprits are algorithmic trading programs, which raises the question, What does Austrian economics have to say about computers buying and selling shares of stock?
The Hathaway Effect
Dan Mirvish at the Huffington Post broke the story, in an article cleverly titled, “The Hathaway Effect: How Anne Gives Warren Buffett a Rise.” Mirvish documents the apparently irrational correlation:
Although it’s always risky to try to explain particular changes in stock prices, Mirvish’s analysis seems plausible. Presumably there are computer programs guiding lightning-fast stock purchases and sales, which scour news sources in order to make “momentum trades.” In other words, if a particular stock is being discussed in the media, then (other things equal) at least some of these programs buy shares, because it’s “hot” and is likely to continue rising as more slow-footed investors read the buzz and want to get a piece of the action.
Of course, the downside of automated stock-trading programs is that they have no common sense (which isn’t to say that human traders necessarily do, either). In order to beat their competitors to the punch, they can’t engage in careful analysis of the news items; they simply look for “Hathaway” and take it as a bullish signal for Berkshire Hathaway A-shares. (I wonder if Mirvish could do another story on the fortunes of the Rand Capital Corporation as the new Atlas Shrugged movie premieres?)
Do Automated Trading Programs Have Any Benefit?
After stories such as these — and certainly after the huge financial collapse in 2008 — many cynics understandably dismiss all the newfangled derivatives markets and financial strategies as a casino for egomaniacs with above-average math skills. Yet there is a danger here in throwing out the baby with the bathwater.
It’s true that economics — as conceived by Ludwig von Mises and his followers — is a logically deductive science. Mises didn’t believe that economists should ape the physicists and develop empirical hypotheses that are then “tested” by the data. Rather, Mises believed careful introspection on the nature of human action could yield a core of economic principles or laws. It was this framework that allowed the economist to then interpret the mass of available data on commodity prices, unemployment rates, and so forth.
For example, Mises wouldn’t “start with a blank slate” and look at the historical statistics to try to develop theories about the business cycle. Rather, he would first reflect on the operation of the capital structure in a market economy, think through the function of market prices and interest rates, and only then be able to start explaining the connection between credit expansion and unsustainable booms. Mises didn’t reject the use of historical data — in fact he helped Hayek found an institute to study the business cycle — but he didn’t fall for the positivist illusion that one could develop economic theories by “letting the facts speak for themselves.”
Having said all that, Austrian economics doesn’t forbid stock traders from using such techniques in their quest for profits. For example, suppose an analyst at a hedge fund starts cranking out regressions on “randomly” selected data. He discovers a startling correlation between the phases of the moon and the NASDAQ index. He shows the other analysts, and they confirm his results. They can only offer the most ad hoc “explanations” to their boss, but the relationship is nonetheless staring them in the face.
Suppose the hedge fund begins trading on the newly discovered relationship, and earns money. Over time, as the formula continues to perform, the hedge fund wagers more and more heavily upon it, and is never let down. At the bar the analysts’ buddies ask, “Why are you guys up 84 percent this quarter?” but the analysts smile and say, “Ancient Chinese secret.”
The question is, are these profits “real” or illusory? Is our hypothetical hedge fund actually doing something useful?
The social function of stock speculators is that they speed up price adjustments. The goal of the speculator is to “buy low, sell high” (or “short sell high, cover low” for an overpriced stock). In this respect, so long as the hedge fund’s moon strategy is profitable, then that is prima facie evidence that it is performing a service to others in the market economy. Specifically, the hedge fund buys into the NASDAQ when its price is about to rise, and it sells when the NASDAQ is about to fall. In this limited yet important sense, the successful trading strategy is a time machine, giving the rest of the world advance access to future information.
In this grand sense, results are what matter. The fact that the hedge fund personnel can’t really explain the correlation is irrelevant. By the same token, the workers at a utility company can’t always explain the laws of physics; they just know that if they repeat certain actions every day, then consumers are able to turn on their lights and run their refrigerators. As David Hume famously pointed out, just because something has happened in the past doesn’t mean it will happen in the future, but in many contexts we benefit from making just such an invalid leap.
What About Bubbles?
The arguments above might make some readers uncomfortable. After all, didn’t the fancy quants on Wall Street look like hot stuff for a few years during the housing boom — until everything blew up in their faces?
Yes, but that is entirely consistent with the position we’ve laid out. Austrian economists do not naïvely endorse the most extreme versions of the “efficient-markets” approach of the Chicago School. Austrians know that investors can make colossal mistakes and that the going market price can be horribly wrong.
If a trading strategy yields profits for a few years, but will eventually bankrupt the company when a “black swan” comes along, then it is an unprofitable strategy — barring government bailouts. For this very reason, our hypothetical hedge fund managers had better be very careful with their uncanny moon-trading strategy. They have no business being shocked if and when the strategy completely backfires on them, and they had better position themselves accordingly rather than shooting the moon (if you’ll forgive the pun) with each new trade.
To correctly assess the value of any entrepreneurial venture, we need some idea of the underlying uncertainty involved. (Note that Mises made a distinction between quantifiable risk and amorphous uncertainty.) To switch away from financial markets to something more concrete, suppose in January a t-shirt manufacturer sunk $1 million into producing shirts saying, “The VCU Miracle of 2011.” After the VCU basketball team made the Final Four against all odds, the manufacturer was able to recover his investment as well as a tidy profit.
Now how should an Austrian interpret this event? Did the entrepreneur see beyond what others saw, and allocate resources more effectively to serve consumers? Or did he take a big chance but “get lucky”? At this point the question is almost philosophical rather than economic, but the scenario sheds light on automated trading programs yielding short-term profits.
Faced with such apparently nonsensical results as the Anne Hathaway effect, automatic trading programs look silly. On the other hand, this is true of any task to which humans put computers; it doesn’t mean computers are useless.
The ultimate criterion for whether automated trading is socially useful is the profit-and-loss test.
If the financial institutions relying on these programs blow up in the long run, we’ll have our answer — if only the government and Fed would stay out of it.