16th
Good research, bad interpretations
In another life, I might have been a scientist instead of a business guy. I was convinced I was going to be a physicist through most of junior high and high school; I loved Feynman’s books and Real Genius and did various geeky Science Olympics things; and I was preparing to go to MIT right up until the moment I visited Princeton’s campus a couple of weeks before I had to choose a college.
In college, I majored in physics even though I was starting to realize that I liked economics even better, figuring that I could satisfy my science jones as an undergrad and then do an econ Ph.D. (Absolutely the right choice, by the way.) I got elected to Sigma Xi, and although I may be the least-active member in that entire organization, they still send me a daily digest of top science stories, and reading them is a fun and efficient way to hear about some cool research from various fields. Some bore me (ecology, pharmaceutical stuff) and others are more to my taste (physics, evolutionary biology & some anthopology, mathematics, some neuroscience), but I typically learn 1-2 new things every day and get to “feel like a scientist again” for a few moments.
So, as with other activities where I invest some emotion (c.f. Stanford basketball), I also get aggravated when I see something that’s sloppy, misguided, or just plain wrong. Today’s example: Traders said to be driven by hormones. The over-simplified version of the story:
- Traders with more testosterone are more likely to make an unusually big profit on a given day
- Testosterone levels are correlated with willingness to take risks
- At some point, the traders “over-reach” and things fall apart
… with the following conclusions:
- “high testosterone levels can lead to irrational risk-taking”
- “trader behavior… did not make sense in terms of economic or game theory”
- “…it would be good for both banks and the financial system as a whole if we had more women and older men in the markets… such a change would produce a much more stable financial system”
I’m going to assume that the researchers did their research properly - that is, that they set up appropriate controls, measured well, and so on. Furthermore, I recognize that newspaper articles 1) simplify for the sake of brevity and comprehension and 2) can be a place for a scientist to do a little marketing by making provocative statements.
That said, these interpretations seem woefully flawed. Going from data (even clear data) to findings or principles is often the trickiest part of science (and business, too), and in this particular case the researchers seem to be showing that they don’t “get” economics. (Economic understanding, for some reason, seems to be something that’s in particularly short supply in the general population, but that’s a topic for later.)
Let’s start with what I see as the obvious, conservative finding from the reported data: higher testosterone levels cause increased appetite for risk. (Causation, instead of just correlation, is supported by studies that show that inducing higher testosterone levels - e.g., by showing the subject an erotic picture - leads to increased risk-taking.)
What are the implications of this? Well, let’s see: if you’re taking bigger risks, you’ll have larger variance on your performance - which means you’re “more likely to make an unusually big profit on a given day”. You’re also more likely to experience an unusually large loss as a result of this risk-taking, which looks an awful lot like ”at some point, the traders over-reach and things fall apart”.
None of this supports the conclusions drawn by the researchers. Let’s see:
“High testosterone levels can lead to irrational risk-taking.”
Why is the risk-taking irrational? Because, at some point, the high-testosterone traders suffered large losses? Economic theory assumes that different individuals will have different tolerances and appetites for risk; there’s nothing irrational about this. In some cases, (e.g., lottery tickets) people will even sacrifice expected return in order to increase risk; no cosmic rule says this is irrational. In order to claim irrationality, you’ve got to meet a pretty high bar - basically, sacrificing an absolute good (money) for nothing, or making a set of mutually inconsistent decisions under near-identical conditions. Mere appetite for risk doesn’t qualify.
In addition, the language used by the researcher makes me strongly suspicious that he’s conflating outcomes with rational behavior. Making an unusually large profit doesn’t mean you’re behaving smartly or rationally and losing a lot of money doesn’t mean you’re over-reaching or being irrational - they’re both consistent with and typical of a large appetite for risk. (For a later post: one good thing about poker players is that, much more than the general population, they understand that good decisions are made ex-ante - that is, before any uncertainty is resolved - and that good (or bad) ex-post outcomes can very often result from bad (or good) decisions.)
Another problem with this conclusion: there’s no discussion of the incentives faced by the traders. Presuming that they’re each trying to maximize expected return, modified by risk or not, is a very specific assumption, and one that often doesn’t hold in real-world conditions. Maybe traders that make unusually large profits get more glory and are more likely to be promoted, leading people to take large risks in order to maximize their chances of winning (as people do in NCAA tournament brackets and some business situations). Maybe there’s some sort of safety net (bankruptcy? a government bailout? the ability to quit trading and take another job?) that effectively limits downside. There are often many factors outside the primary gamble that affect the actual impact to the risk-taker, and if you don’t take those into account, it’s really hard to draw good conclusions.
(Very brief aside: a famous game theory experiment was done on the “ultimatum game”, where player X gets to make one proposal to player Y about how to divide, say, 20 dollars, and player Y can accept or reject the proposal; if she rejects the proposal, neither player gets anything. Basic game theory suggests that player X will offer a pittance to player Y - e.g., one penny - but under ”normal” conditions, player X tends to offer half, or maybe slightly less than half„ and player Y rejects offers that are too low. The obvious problem here is that of social reputation between the players, but when the players are made anonymous to each other, the “mostly fair” behavior remains. However, when all social effects are removed - in particular, when the players are made anonymous not only to each other but also to the researchers - behavior comes very close to what’s predicted by basic theory: player X offers something like $0.25, keeping $19.75 for herself, and player Y accepts. The moral: figuring out key elements in a decision process can be tricky and non-intuitive.)
“Trader behavior… did not make sense in terms of economic or game theory.”
Economic theory and game theory are pretty broad, and they allow for a very wide range of individual behaviors. If you’re assuming that risk-taking and losses indicate behavior that’s not supported by theory, you’re not on thin ice - you’re already at the bottom of the pond.
“It would be good for both banks and the financial system as a whole if we had more women and older men in the markets… such a change would produce a much more stable financial system.”
Okay, I recognize that this has all the signals of a statement made mostly to get attention. That said, there’s still a fundamental flaw in the thinking: putting a population of more conservative/less risk-seeking traders in the financial markets won’t necessarily result in a more stable economy. One of the benefits of a highly liquid financial system is that it provides, via the prices of securities, both the incentives and the signals for important business decisions. Less risk-tolerant traders will tend to create and respond to those incentives/signals more slowly. Slower price movement is probably what the researchers mean by a “more stable financial system”, but, at some point, if market adjustments are too slow, the economy will adjust itself, violently and abruptly. (Metaphor: not allowing enough small, natural fires in a national park, which eventually leads to a massive wildfire.)
(This leaves aside the point that slower price movement may hamper economic efficiency, but the researchers’ statement was about stability, and one can debate whether/how to trade between the two.)
Morals of this story:
- Check to see whether a simple finding (testosterone -> increased appetite for risk) can explain most/all of the observed phenomena. (Embrace Occam’s Razor.)
- Be careful about saying data is “inconsistent with X” when you don’t know that much about X - restrict your claims to things you understand well, or give appropriate caveats.
