This surprises me. Even assuming an Obama victory, I would put the probability much lower. As an economic matter, raising anyone's taxes with the economy so weak seems ill-advised. As a political matter, why not just let the Bush tax cuts expire at the end of 2010? Obama could then claim in four years that he never signed a tax hike. It seems neither economically nor politically sensible for the new President to push for an immediate tax increase, even if an eventual tax increase is his goal.
How then to explain the betting at Intrade? I can think of three hypotheses:
1. The Obama people are not as savvy as I think they are and will push for an immediate tax hike.
2. The Intrade market is so thin that the pricing there does not mean much.
3. Some people are using the Intrade market as a hedge. A high-income person bets that tax rates will go up and bids up the implied probability above the true probability. If the bet pays off, his winnings reduce some of the hit his after-tax income takes by the tax change. It is a form of insurance. Those traders on the other side of this bet--who win if taxes do not rise--are buying a high-risk asset, as measured by covariance with their consumption. They need to be compensated for taking this risk. Under this hypothesis, the Intrade price is not a good gauge of the actual probability but includes a substantial risk premium.
Update: Tony Smith, an economics professor at Yale, emails this comment:
I read with great interest the post on your blog today about how to interpret the prediction market price for the event of a tax hike in 2009. Tyler Cowen made a similar point about interpreting the market price for the event that Congress would approve a bailout before September 30. And, in fact, last May I wrote a comprehensive exam question for the Ph.D. students at Yale that revolved around this same observation in the context of an election prediction market. But I have seen no formal papers that make this point. I think it is a critical one for evaluating the usefulness of prediction markets in aiding decision-making.
The general point that rational investors will use prediction markets to hedge risks can also help to explain an apparent puzzle throughout the recent election campaign: in particular, statistical models designed to predict election outcomes (see, for example, www.fivethirtyeight.com and David Stromberg's website) generally reported probabilities for an Obama victory that exceeded the corresponding market prices on both intrade and betfair. If we take seriously the predictions of these statistical models--that is, if we view them as giving an accurate estimate of the actual probability that Obama will win--then evidently investors seem to think that Obama will be relatively good for the economy compared to McCain, driving down the equilibrium price for an Obama contract (since this contract pays off when marginal utility is low). To help voters make an informed decision, maybe you should post this "evidence" on your blog!