Tyler Cowen just linked to a comment left by Robin Hanson on this blog. I agree with Cowen’s comment left on the same post here: “Robin is awesome, enough said.”
Hanson’s writing never disappoints, even when he’s claiming that medicine is useless (the statistical argument is strong).
On the other hand Cowen is one of my most eagerly read bloggers (and semi-frequent provider of fodder for my comments), but sometimes Cowen says the darndest things, like this from the post linked above:
The very reason we resort to a firm, rather than the market, is to build consensus and morale, not to forecast the truth.
Consensus I’ll buy, as shorthand for lower in-firm transaction costs. Morale? He’s got to be kidding (note that the only instance of “morale” in the Wikipedia article on theory of the firm is immediately followed by “-damaging”).
Cowen continues:
Prediction markets would tend to break down firms, but of course they still can flourish in Arrow-Hahn-Debreu space.
My guess is that in the short term adoption of prediction markets will favor firms that have access to specialists needed by early adopters to succeed and layers of management that can be made redundant without immediately threatening the authority of the top, i.e., large firms.
I have no idea what Arrow-Hahn-Debreu space is, other than that it has something to do with competitive equilibrium. If I had to take a wild ignorant guess at the import of “but of course…” I would say it is arguably a tautology.
Well, I don’t know either of these people. But Cowen’s statement is silly, with or without that gratuitous finance reference. The idea that prediction markets are shunned because they would lead to a breakdown of morale, or loss of respect for one’s bosses, or whatever you want to call it, is conspiratorial thinking and sounds like the writing of someone who’s spent very little time in an actual corporation. If markets could really help improve corporate decision-making, you’d see them getting implemented more; even if Cowen’s conspiracy theory was right and employers feared them, workers could be setting them up as skunkworks projects, a la how internal blogs got their start. I haven’t seen any evidence of that happening, and I think I’d be in a position to see it.
I think the reason that they haven’t been used more for corporate decision-making is, basically that… they’re not that useful for corporate decision-making. There just aren’t that many predictive questions you can ask to help in deciding on a course of action. The most obvious question, “will project X succeed?”, is not as appropriate as it sounds: to start getting meaningful numbers on it, you have to already start project X, which means it’s in a sense too late to make your decision. And, of course, you run the risk of creating negative incentives.
There are more pedestrian questions you can ask: how much a revenue will a competitor’s product make? Will certain needed components from another vendor be delivered on time? But these are relatively mundane, and they’re not really going to revolutionize the decision-making process.
As time goes on, I become more convinced that decisions and predictions are two things that usually should remain separate; even the phrase “decision market” is one I tend to avoid.