Yale economics professor Ray Fair, who designed a statistical model to predict the vote share in U.S. presidential elections, says the race is in a statistical dead heat.

Drawing on economic figures including growth in real GDP, inflation rate and number of “good news” quarters — those when growth rate of real GDP exceeds 3.2 percent — Fair’s model operates on the premise that the American economy will drive voters’ choices in November. The model predicts that President Barack Obama will receive 49.48 percent of the presidential vote compared to 50.52 percent for Republican nominee Mitt Romney, making a narrow victory for either candidate within the margin of error.

[ydn-legacy-photo-inline id=”1015″ ]

“The hypothesis this model tests is that the economy affects the way people vote,” Fair said. “The economy is growing, but not fast enough to be confident in predicting a winner. The main thing we can take away is that the election will be close — too close to call right now.”

While there is no precise way to quantify voters’ perception of the U.S. economy at the time of the election, Fair maintains that the inputs for his formulae at least do so indirectly, stating that “people’s idea of the economy correlates with these variables.”

With the economy currently performing neither particularly well nor especially poorly, Fair expected his equations to yield largely ambiguous results. Given the study’s margin of error, Fair said the resulting output is somewhat inconclusive beyond indicating that the election should be a tight race.

From November 2010 through April 2012, Fair’s equation predicted that Obama would receive a vote share over 50 percent. Fair said he is hesitant to predict that Romney will win even though his most recent numbers indicate as such, given that the standard error is around three percent.

He emphasized that its inability to predict a winner definitively does not speak to a flaw in the model; rather, the absence of a definite winner based on these numbers reflects the lukewarm nature of the economy. But if one of the candidates ultimately wins by a wide margin, then the model’s validity would be called into question, Fair said.

Historically, the model’s predictions have been fairly accurate, averaging a mean absolute error value of only 3.60 percentage points in the last eight elections, though error reached a full 10.5 percentage points in 1992.

Nonetheless, some experts “do not think highly of Fair’s sequence of models,” said economics professor Douglas Hibbs, who is on the faculty of the University of Gothenburg in Sweden.

“Fair’s model has a lot of statistical ‘junk,’” Hibbs said. “None of the variables in this very peculiar set of economic terms are significant.”

Hibbs said that Fair’s model is too cluttered with variables that do not reflect presidential performance, such as incumbency.

In contrast with Fair’s model, Hibbs’ own equation, known as the “Bread and Peace Model,” factors in the effect of US military fatalities in unprovoked overseas deployments such as those to Iraq, Korea and Vietnam. Furthermore, Hibbs employs the growth rate of per capita real disposable income, a figure that he contends is more telling than the economic variables used in Fair’s work.

As a result, Hibbs estimates that Obama will win only 47.5 percent of the vote, compared to Fair’s estimate of 49.5 percent. In the former case a Romney victory is likely, while the latter is a statistical tie.

Bruce Hansen, professor of economics at the University of Wisconsin, Madison, said he admires Fair’s work.

“[Fair] zeroes in on some key variables and he’s been rigorous about using the same equation over many years.” Hansen said.

He also pointed out that, as one of the pioneers for political econometrics, Fair has done a good job of quantifying the impact of the economy on voting patterns. While Hansen said he has not run his own calculations in anticipation of this year’s election, he agrees that it will be close.

Fair’s model also predicts that Democratic Party candidates will win 46.25 percent of the net vote for Congressional seats.