Prediction markets for government program budgets

I maintain that I'm still largely opting out of the healthcare debate (although not totally successfully). But I found this suggestion from Robin Hanson quite interesting (in response to Megan McArdle on the CBO projection of the healthcare legislations budgetary implications):

The alternative is prediction markets. Compared to the value of making good decisions on these bills, or to the effort spent in “rage and anguish” on them, the cost to create prediction markets giving quality unbiased estimates of actual bill budgets would be small.

This is a clever idea, but I'm afraid the major obstacle is that it would be very hard to create prediction markets large enough that they couldn't be manipulated by those who had a high stake in the outcome of legislation. To take one example, Democrats would have you believe that insurance companies stand to lose billions in profits if this legislation passes - you think the industry would hesitate to spend a few million dollars influencing the budget forecast of said legislation, were it in their power?

We've even had a recent example of this, first spotted by the hawk-eyed Nate Silver at fivethirtyeight.com: a McCain supporter investing "hundreds of thousands of dollars" in pumping up McCain presidential victory contracts on Intrade in September 2008.

Overall, if the trader’s motive was to influence the Intrade market, he was remarkably successful, Rothschild said. The trader’s actions help keep the probability of Obama winning the election on Intrade about 10 percent lower than Betfair and IEM for more than a month.

Enabling deep-pocketed stakeholders to influence the budget forecasts for major legislation by 10% would be downright disastrous.

P.S. Not only would potential arbitrageurs be seriously outgunned by insurance companies, but the long time horizon (when would the contract be closed out – 2019?) makes the arbitrage risky in other ways reduces the attractiveness of the return on an annualized basis.