If You Win A Nobel, But No One Reads Your Work...

I just asked Joe, “Who is going to win the World Series?”

He answered, “Oakland, of course.”

I knew Oakland was his favorite team since the days of Canseco, McGwire, Stewart, and Eck, so I replied “No, who do you really think is going to win?”

Was Joe's answer biased? Or did he answer truthfully. Does it matter? Are there any true population parameters to subtract our sample parameter from? When we say something is biased, aren't we guessing about a guess? My advisor explained bias as systematic pressure in one direction. Everything had to be either biased upward or downward or towards Oakland. But like with externality theory, someone has decide that Truth exists, someone has decide on social optimality or a true population parameter. Saying something is biased upward implicitly assumes your sample results are higher than some unknown Truth. So it seems to me that bias is as an arbitrary concept that has no real objective meaning.

So Forsythe et al in “Anatomy of a Political Stock Market” explain the assimilation-contrast effect and the false-consensus effect, but I do not understand how they jump to bias. People see through rose-colored glasses and think everyone else is like themselves. Of course, but how is this bias? Don't the people think that their candidate is going to win? They might be fooling themselves, but who am I or the authors to say what people are thinking? I see bias as a slippery slope that can be used to explain every result a researcher dislikes. (Footnote 22 is an example which does not deal with bias, but I never like when researchers throw away inconvenient observations.) He voted for Joe, but he was biased and really wanted Jack. Let's make Jack Senator. He chose Oakland but he really thinks Detroit is going to win. “Really thinks” and “really wants” scares the hell out of me. I see Hitler and Stalin. I see Orwell's “stopthink.” I see my Dad telling me to eat my vegetables because “they will put hair on my chest” (like chest hair was important). I just do not get it. Maybe I am biased.

This is a good paper, but one error is so blatant, so absolutely disgusting, it cannot go without discussion. Forsythe et al. comment:
“Our findings lend support to the so-called Hayek hypothesis which asserts that markets can work correctly even if participants have very limited knowledge about their environment or about other participants. How markets can work under such circumstances was never clearly specified by Hayek, and the hypothesis is not easily tested.”
The second sentence shows the authors' complete ignorance of Hayek's work. He won a Nobel Prize, and no economist outside of George Mason or Auburn could tell anyone why. Hayek and the Austrian School's greatest contribution to economics was explaining that we can never explain how markets work (especially with calculus). But they do work, and economists should spend their time letting them work and defending them against the collective. And that is exactly what this research shows. All of the statistics, all of the money spent on polls, and all the political “science” professors on TV cannot predict as well the Iowa Presidential Market even with all of its possible judgmental bias (whatever the hell that means) and its misrepresented sample. Hayek would be proud. It is a shame the authors do not know who Hayek is. (I am biased. Hayek is one of my favorite economists and currently I am reading Hayek' s Counter Revolution of Science.)

To summarize, the authors set up an electronic betting market on the 1988 Presidential election. Since it is specifically for research purposes, it can legally compete with state-run lotteries, Wall Street, Las Vegas, and Atlantic City. Since it is electronic, all trade information can be saved and analyzed. The effects of different news events, debates, and polls could be teased out the market. It is a wonderful set-up.

The authors throughly analyze their data. I doubt if their statistics could withstand a barrage of misspecification tests. But even if the statistics do not hold up, it does not change the significance of their results. The whole thing worked. Make people put money where their mouth is, and the market will do the rest.

The authors go on to discuss that the market works because prices are determined on the margin while bias only affects averages. They separate the markets into arbitragers who look for profit and political “homers” who only buy shares in their candidate. The arbitragers set the prices while making money off the “homers.” This idea pulls their work back into economic theory and makes sense. Every grocery store manager knows that the “cherry pickers” (those customers who only by items on sale) keeps stores afloat by increasing cash flow but homers who buy everything at the store regardless of price generates profits.

But it greatly simplifies what is actually going on in the market. Hayek and the Austrians describe the market as an “information discovery process.” Prices convey information. The Iowa Presidential Market allowed individual traders (and researchers) to discover information and to use this information as they deemed fit. This discovery process led to the final market price giving a good prediction of the actual election. Is it because “markets allocate scarce resources to their highest value”? Well, that depends on how you define value. Economists have been trying to pin that definition down for 200 years. But in the end, the whole decentralized process worked. Why it worked can be partially explained by saying some traders understand calculus and throw their personal preferences out of the window, but there is a whole lot more to it than that. A whole lot more than we will never be able to explain. But markets work.