Continued from page 1

In Ohio, for example — a critical swing state needed by Mr. Romney to win in most forecaster scenarios — Mr. Obama is getting a lot of mileage from his 2009 bailout of the auto industry, which Democrats argue helped to preserve jobs in the state, Mr. Cheng said.

In all, Moody’s expects Mr. Obama to win 10 of the 14 swing states. It predicts Mr. Romney will pick up Florida, which has an above-average jobless rate of 8.8 percent, and three other states — Indiana, Missouri and North Carolina.

Others have different points of view. Douglas Hibbs, a retired Swedish economics professor, has developed a “bread and peace” model that gives the election to Mr. Romney, based on the level of casualties in U.S. wars and the depressed growth of incomes in the past four years. His model, which he said is accurate in explaining every presidential vote since World War II, projects that Mr. Obama will get 45.4 percent of the popular vote.

By Mr. Hibbs’ estimate, disposable-income growth, which is running at an annual rate of less than 2 percent after adjusting for inflation, would have to soar to 4 percent in the month before the election for Mr. Obama to have a chance of re-election. However, Mr. Hibbs conceded that his model is not in sync with other forecasting models this year.

“Every election is affected to some degree by idiosyncratic factors, which at times are important enough to overwhelm the persistent influence of fundamentals,” he said. He suggested that, given the close race this year, perhaps the “best prediction of election results” will come from the Iowa electronic futures market. That investor-driven index puts Mr. Obama’s re-election chances at about 75 percent.

Adjusting for polarization

One of the most closely followed forecasting models, developed by Alan Abramowitz, a political science professor at Emory University, is predicting a narrow victory for Mr. Obama based on the 1.3 percent level of economic growth in the second quarter, combined with the president’s 49.8 percent approval rating and a 2.5 percentage point advantage he enjoys as an incumbent. That would translate to a popular vote for Mr. Obama of a little more than 50 percent, he estimates.

Mr. Abramowitz has adjusted his model to reflect the increasing polarization of the electorate, which has resulted in fewer swing voters in the past four presidential elections. He said the adjustment makes his model more accurate than ever in predicting the outcomes of presidential elections.

“With the American electorate both closely and deeply divided along party lines, we can expect another close election this year — probably closer than the 2008 election and possibly as close as the 2000 election,” said Mr. Abramowitz. “There is only one prediction that seems very safe right now — it’s going to be a long election night.”