Pur Autre Vie

I'm not wrong, I'm just an asshole

Thursday, October 18, 2012

Idea: Intrade Prediction Markets for Commodities

As the 2012 election approaches, I find myself frequently checking the Intrade prediction markets to get the consensus view of the current odds.  Some have criticized the Intrade markets for basically replicating the numbers from Nate Silver's model, which suggests that the markets are not providing any independent information (in fact, maybe they are not providing any information at all).  But I want to mount a brief defense of Intrade, and then to set out a more extended argument for an idea that recently occurred to me:  we should use Intrade-like prediction markets to predict the future prices of commodities like oil and corn.

To defend Intrade, let's ask what it might add to Nate Silver's model even if Intrade rarely or never diverges from that model.  The answer, I think, is robustness.  To people who follow politics, Nate Silver has a lot of credibility because of the sophistication of his model, which doesn't simply aggregate polls but gives them varying weight depending on a number of variables.  But Silver's model is not the only model out there, and one can imagine a sophisticated, well-designed model that nevertheless gets it wrong.  (Moreover, most of us don't have the quantitative skills to assess how well Silver has calibrated his model.)  So Intrade may not be telling us anything that Silver's model isn't, but it is telling us that people who are risking their own money think that Nate Silver is getting it right.  If Intrade prices tracked a different model, like the one published by RealClearPolitics, that would certainly affect my thinking about the race.

But all of this thinking about prediction markets and models led to a "Eureka!" moment for me.  As noted above, prediction markets can be valuable in predicting election outcomes even if they merely lend credibility to a model that is not market-based.  But how much more valuable would they be if they could give us information on important economic variables like the future price of oil?  The idea is startling in its simplicity but has profound implications.  If people could "bet" on the price of oil to be delivered months in the future, we would have market-based information on one of the most important variables for many businesses.  Even if the numbers simply tracked the expert view on the future path of commodity prices, it would be confirmation that the experts are getting it right.

And commodity prediction markets would provide another important service:  they would allow companies to "hedge" their exposure to commodities.  If your business uses a lot of oil, you may be interested in "locking in" a market-based price for oil that you will use months from now.  In the status quo, that isn't possible unless you can find a counterparty willing to take the opposite economic position (that is, someone willing to provide the oil months from now at a price negotiated today).  Good luck with that!

But with an Intrade prediction market for commodity prices, you wouldn't need to go out and individually negotiate a long-term oil supply contract.  Instead, you could just go to Intrade and place a bet that the price of oil will rise.  If oil does indeed go up, you will lose money in your business (when you buy the oil you need on the spot market), but you will make money on the Intrade bet.  You will be "hedged" against changes in the price of oil.  (If oil prices fall, you lose money on the bet, but your business expenses are also lower, so you are basically eliminating both upside and downside risk when you "hedge" in this manner.)  Intrade prices would reflect the "market view" of whether commodity prices are likely to rise or fall in the coming months.

Economists disagree about how important "price discovery" is in the free-market system.  Some think that market prices are an incredibly efficient way to distribute information throughout society.  The baker in Indianapolis doesn't need to know that there has been a drought in Iowa, he just needs to know that butter is more expensive than it used to be.  He will automatically start using less, turning to other fats instead, like canola oil.  He has responded exactly as society would want:  he has economized on a scarce resource, without needing to know why it is scarce.  He doesn't even need to know Iowa exists!

As I said, economists disagree on the exact role of the price system in the economy.  But I don't think anyone would argue that accurate information on future commodity prices would hurt the economy.  So at the very least, an Intrade prediction market on commodities would do no harm, and it would let people .

Now, as so often happens when I have a good idea, it turns out this one isn't entirely original to me.  Intrade does in fact have a "Commodities" section of its website.  But there are only two contracts available for trading!  I would think that there are dozens of commodities that could profitably be traded:  corn, wheat, coffee, oil, natural gas, to name a few.

So I think a big opportunity is being missed here.  If you took the logic of election prediction markets, and applied it to commodities, you could change the world.  If Intrade isn't willing to step up to the plate, then someone else should.  I am so excited by this idea that I would consider quitting my job and trying to set up something like this.  The logic is just incredibly compelling.