The race as markets foresee it

By Edward L. Glaeser

Around the dawn of a new year, the urge to prognosticate is harder for a columnist to resist than a glass of champagne. In politics, I look to the electronic prediction markets for a shortcut to insight.

The markets suggest that Mitt Romney will be the Republican nominee, President Obama will win the general election, and Republicans will control the House and Senate. But while centrists might welcome divided government, a closer look at these markets implies that a more extreme outcome is likely: one-party rule.

Electronic prediction markets provide both the advantages and the pitfalls of relying on the wisdom of crowds.

Economists at the University of Iowa started the original electronic election markets. Twelve years ago, Intrade entered the field and became a larger political gambling venue.

These markets harken back to the pre-World War II era, when New York gamblers bet millions on the political future of Woodrow Wilson. Prediction markets can be briefly manipulated, and they suffer the same waves of erroneous groupthink that can afflict any market, but their overall track record is pretty solid.

Economists Paul W. Rohde and Koleman S. Strumpf report that "in the 15 elections between 1884 and 1940, the mid-October betting favorite won 11 times (73 percent) and the underdog won only once." Economists Justin Wolfers and Eric Zitzewitz find that in the presidential elections from 1988 to 2000, the average vote share predicted by the Iowa markets 30 days out was less than 2 percentage points off, which is better than the final Gallup poll forecasts.

The Iowa markets and Intrade depict America's political future similarly. Gingrich-mania having passed, both give Mitt Romney a roughly 75 percent chance of winning the nomination. But his odds are worse in the general election, as the markets give the Republicans about a 45 percent chance of winning the presidency.

The markets give the Republicans a more than 70 percent chance of taking the Senate, and the Iowa markets give them a roughly 75 percent chance of keeping the House. Intrade puts the odds of the latter around 69 percent. The Iowa markets, which allow gambling on whether the GOP will controlling both houses, suggest a 60 percent chance of that.

While Obama seems likely to stay in the White House and Republicans seem likely to take Congress, the markets aren't predicting divided government. Correlation between the contests means that the possibility of one-party rule is larger than it initially appears. If Europe pulls America into deep recession, the Republicans could easily sweep. If the economy roars, Democrats could take all the marbles.

If Democrats take both houses, of which the Iowa markets predict a 15 percent chance, Obama will surely be reelected. If Republicans retake the White House, they seem likely to take both houses of Congress. Taking these markets' implied odds seriously, there is about a 60 percent chance of one-party rule.

This summer's budget brinkmanship illustrates the downside of divided government. At best, political division will skirt the edge of catastrophe, but at worst, we'll go over the edge. Division also means that neither party's pet projects will get enacted, but moderates may prefer that outcome.

Divided government may also enable unpopular but necessary reforms. The country needs entitlement reform, which means less spending on health and retirement benefits. But neither party seems suicidal enough to play Scrooge solo, so entitlement reform will likely require both parties' participation, just like tax reform in the 1980s and welfare reform in the '90s. Those who think America's fiscal health needs tough medicine have good reason to hope that the prediction markets are wrong.