(Bloomberg) Democrats may lose control of the U.S. House of Representatives in the November elections after contests that will be closer than polls suggest, according to Yale University economist Ray Fair.

Fair, who has developed a formula using economic data that would have correctly predicted all but three presidential elections since 1916, projects that Democrats will get 49.3% of the national vote in this year's congressional races.

"They're going to get less than half of the two-party vote in the House," the professor at the university in New Haven, Conn., said in a telephone interview. "It looks like the election will be relatively close."

Fair's assessment contrasts with polls that show Democrats trailing Republicans by as many as 13 percentage points when voters are asked which party they plan to support.

In a Washington Post-ABC News poll conducted Aug. 30 to Sept. 2, 53% of registered voters said they would vote for the Republican candidate, compared with 40% who backed the Democrat. A CNN-Opinion Research poll of registered voters conducted Sept. 1-2 gave Republicans a 7-percentage- point advantage.

Still Undecided

Still, national polls suggest voters' minds aren't made up.

An Aug. 23-29 Gallup survey of 1,540 registered voters found Republican candidates were preferred over Democrats by 51% to 41%. The Republicans' 10-point lead, the largest in the history of Gallup's midterm congressional election polling, evaporated by the next Gallup survey conducted Aug. 30-Sept. 5.

The latest Gallup poll found hypothetical Republican and Democratic candidates each getting support from 46% of the 1,650 registered voters surveyed.

Republicans, who need a net gain of 39 seats to take control of the 435-member House, are ahead of Democrats by a little less than 2 percentage points, according to Fair's model.

"That could translate into a loss of the House," he said.

Fair links the decline of Democrats' fortunes to U.S. economic growth that has been less than 3.2% at an annual rate for all but one quarter since the start of 2009.

'Noticeable Drop'

In January, Fair's model projected a 51.6% share of the vote for Democrats as the economy showed signs of improvement.

"It looked like in January we were maybe going to have some strong growth," he said.

Sluggish growth "knocked off a few percentage points," Fair said. "That's a noticeable drop."

The world's largest economy expanded at an annual pace of 1.6% in the second quarter, the Commerce Department reported on Aug. 27. That was down from an initial estimate of 2.4%, reflecting smaller gains in inventories and a wider trade deficit.

"The economy obviously has an effect on votes," Fair said.

Fair measures that effect by plugging the U.S. inflation rate and the number of quarters preceding the election that economic growth was greater than 3.2% at an annual rate.

One Good Quarter

For the current congressional election, Fair counts one quarter of good growth. To track inflation, Fair uses a measurement that adjusts gross domestic product for inflation, which has averaged .77% since the beginning of 2009. Under Fair's equation, low inflation benefits the incumbent party.

Fair, 67, wrote his first paper on an equation to predict election outcomes in 1978. He refined the theory before revamping it in 1992 after failing to predict that year's presidential election.

Since then, Fair said, "the presidential vote equation has done reasonably," accurately predicting each election. In 2008, he came within 1.1 percentage point of Barack Obama's victory margin over Republican John McCain.

During the 2008 elections, Fair also developed his congressional vote equation.

Clinton, Reagan

Low presidential approval ratings in opinion polls can translate into big congressional losses. Former President Bill Clinton's Democratic Party lost 54 House seats in the 1994 congressional elections when his public approval rating was 39% two months before the election. Ronald Reagan's 42% approval rating was followed by Republican losses of 26 seats in 1982.

Democrats in 1982 beat Republicans in the national vote cast in congressional elections by 12 percentage points. In 1994, Republicans bested Democrats by 7 points.

Obama's approval rating has hovered around 50% for much of this year in Gallup's three-day average. It was 46% as of Sept. 6; the record low was 41% last month.

White House Press Secretary Robert Gibbs dismissed the opinion surveys.

"The American people are not concerned about the president's poll numbers," Gibbs told reporters at a White House briefing yesterday.

Economic Numbers Set

Though Fair's latest figures are from July 30, he said he doesn't expect his final pre-election estimate at the end of October to shift his prediction for the outcome of the midterms.

It's unlikely that economic growth in the third quarter will be strong enough to give Democrats a bounce, and inflation won't change enough to affect his formula's result. Voters may respond to worse economic news.

"What happens between now and the election on the economy obviously does have some effect on the vote, it's just that it's too fine of changes for me to pick up," he said.

Political analyst Charlie Cook yesterday predicted that Republicans would gain at least 40 seats and take control of the House.

Fair doesn't try to figure out what his vote breakdown would mean seat-by-seat in the House. He said the vagaries of individual House districts make that difficult. Still, the overall percentage can be useful.

"The percent of a two-party vote of the House is not a bad measure of what I'm after," he said. "It's an indication of what people think of the two parties."