* our two-party system is deeply entrenched and inherently symmetric

* the parties are more alike than different in the most important way: they are both organizational vehicles designed to help broad coalitions get power, and each can be very flexible with candidates and policies as necessary to win

* with a full 2 years to go, there’s plenty of time for the ‘median voter theorem’ process of crafting candidates to win the median ’swing’ voter to work

Until the election is a lot closer — and parties start to lock-in on issues of that future day — almost any exogenous info shock can be absorbed, by varying the future candidates/policies/spin/campaigning, restoring a highly-competitive expectation.

And of course the current market-implied odds mean more than “nothing”: if cocky partisans of either side offer significantly better odds, bet them and hedge for free money.

Would’ve been really nice if GDP markets were available.

And I don’t really know which way it should have gone (I’m not yet calling ‘08 for the Dems). For instance, Bush winning in ‘04 provides the Republicans with an incumbency advantage. Countering that, the last two years have provided the Dems with an advantage in the midterms, which you might think could persist to ‘08. And the cast of possible candidates is also starting to take shape, and the absence of Warner, the rise of Barak, and the continuing dominance of both Hillary and McCain are all important factors that we didn’t know about two years ago. All of this is hard to reconcile with the odds remaining unchanged.

I response to Mike’s point: he is exactly correct to emphasize the difficulty in discerning the direction of causation between election outcomes and economic outcomes. But that is precisely the point of my forthcoming QJE paper with Snowberg and Zitzewitz (available at: http://bpp.wharton.upenn.edu/j.....ctions.pdf).

In that paper (which is what I describe in the WSJ), we look at stock market reactions to what are clearly random (or exogenous) shocks to the expectations of Bush’s re-election - the leaked exit polls, and the subsequent vote count. These experiments allow us to draw inferences about how changes in electoral prospects drive economic outcomes.

I only skimmed the Wolfers/Thoma conversation and haven’t read the papers referenced, but one thing missing from the exchange concerning GDP growth and presidential party is that causation may run in both directions, e.g., if people are more likely to vote for a party when growth is slowing, presumably it will appear that having that party in office is correlated with slower growth to a greater extent than having that party in office actually slows growth.

Of course this is #5 on Wolfers’ list of five open questions for prediction markets, but it has been bugging me. It seems that any interpretation like the above growth/party correlation or conditional PM prices need to carefully explain why causation runs primarily in the implied direction or come with a huge disclaimer.