20071018 dan drezner 18
Dan Drezner - 


BOB MOON: And now let's do some more numbers. John McCain is up one and a third percent heading into the South Carolina primary, and shares in the leading Democrats are trading flat. No, the last we checked GOP was not a ticker symbol on the New York Stock Exchange, and Democrats weren't trading on the Nasdaq. Those numbers are from Intrade's political futures market, but you can't necessarily take them to the bank.

Commentator Dan Drezner says that as indicators go, political futures markets have a mixed record.

DAN DREZNER: Pssst . . . want to make some extra money and contribute to the public good? I've got a plan. Let's play the political prediction markets. Websites like the Iowa Electronic Market or Intrade have been the trendy oracle in recent election cycles among pundits in the know. Participants in these markets trade in the likelihood of a particular candidate winning the nomination. The market price for each candidate essentially represents that person's probability of winning the nomination.

If one believes in efficient markets, then prediction market prices represent the best forecast of the future. Wharton School professor Justin Wolfers recently argued in the Wall Street Journal that prediction markets outperform contemporaneous polls on a consistent basis. As he put it, markets are forward-looking while polls are backwards looking.

The problem is that it's not at all clear whether prediction markets are efficient. Prediction markets have gotten past elections wrong. On Election Day in 2004 the Iowa Electronic Market priced John Kerry as the likely winner. On Election day in 2006 Intrade put an 85 percent probability on the Republicans holding the Senate. Ironically when Wolfers wrote his column a few weeks ago the prediction markets priced Hillary Clinton as the likely Democratic nominee. Since then the Intrade share price for Clinton dropped by 40 points after Iowa before recovering its value after New Hampshire.

In the future, as prediction markets become more mainstream, they will expand in size and scope, and their informational efficiency might improve. And to be clear, political prediction markets are far from useless. Think of them as representing the purest, most distilled snapshot of the conventional wisdom's expectations of the upcoming election.

So back to us making some money. Political science research suggests that in the past a sophisticated analysis of polling could outperform the market. The problem, even for savvy and sophisticated public radio listeners, is that in such a wide-open election year, with polls fluctuating wildly from day to day, there is not a lot of sophisticated analysis to go around. Writing about Hollywood, screenwriter William Goldman famously said, "nobody knows anything." Remember that line when looking for your next political oracle.

BOB MOON: Dan Drezner is a professor of international politics at Tufts University. His most recent book is called "All Politics is Global."