Formula has knack for picking presidents
Using two economic indicators and three political factors, one equation has predicted presidential winners with uncanny accuracy
Five bits of data. That’s all it takes for IHS Global Insight to predict with uncanny accuracy who will become the next president.
The Lexington economic forecasting firm takes two economic indicators plus three political variables, plugs them into a formula, and out pops the percentage of the popular vote the incumbent party is expected to win. The formula has correctly predicted the outcome in 14 of the last 16 presidential elections, in most cases projecting the vote within a couple of percentage points of the final results.
The IHS formula aims to quantify the old political principle that Americans vote their pocketbooks — or as Bill Clinton’s campaign advisers put it in 1992, “It’s the economy stupid.” With Labor Day marking the start of the final and most intense stretch of the presidential campaign, the just ended Republican convention has left little doubt that the economy will be at the center of the election.
The most recent update of the formula, using July economic data, shows President Obama losing decisively, with just over 45 percent of the popular vote. But Nigel Gault, chief US economist at IHS, is hedging this time, because other factors could play a large role in swaying voters. He noted that Obama is running with unemployment at the highest level in an election year since 1948, but is still neck-and-neck or slightly ahead of his opponent, former Massachusetts Governor Mitt Romney, in the polls.
“Voters are not yet convinced the alternative to Obama is better,” Gault said.
The economy, nonetheless, has been a strong predictor of how elections will turn out. Former presidents Gerald Ford, Jimmy Carter, and George H.W. Bush all ran for reelection when unemployment was more than 7 percent and lost. But there have been exceptions.
Ronald Reagan won reelection against Walter Mondale in 1984 when unemployment in the months leading up to the election year averaged 7.4 percent. Indeed, Reagan won a landslide victory, securing nearly 60 percent of the popular vote.
The IHS formula correctly predicted all those outcomes, including Reagan’s win. In fact, it came within five-tenths of a percentage point of the 1984 result. That’s because the formula includes the growth in per capita income in the year ending in the September before the election, as well as unemployment.
While a high unemployment rate creates uncertainty and unease, for the vast majority of voters who have jobs, income growth more directly affects the view of their personal circumstances — the proverbial pocketbook. And Reagan benefited from a 6 percent rise in per capita disposable income in the year ending in September of 1984.
“The unemployment rate on its own helps, but historically it has not been enough” to predict an election, Gault said. “If everyone believes the economy is improving — if people think we’re on the right track — then an incumbent president facing high unemployment can win.”
The formula also includes political factors, including incumbency, which typically helps a candidate, but can count as a negative when unemployment rises to about 6.7 percent, according to the IHS formula.
Party affiliation is another component; historically Republicans tend to do better in presidential elections, a possible sign that the US voters lean conservative, Gault said.
The final factor is voter fatigue, which works against an incumbent whose party has been in office for consecutive terms. At a certain point, voters start to get fed up with the party in power, Gault said, and are more likely to vote for someone new.
IHS credits Yale University economist Ray Fair for pioneering the forecasting model in the 1970s. Fair, who still forecasts presidential races and recently authored the book, “Predicting Presidential Elections and Other Things,” said the current race is “too close to call” and advised against relying on any prediction too closely.
“It’s not like its perfect,” Fair said of the formula. “There’s a lot of other things that affect votes. It explains a lot, but not all.”
The formula, which IHS began using in the 1980s, was tested against past elections to see how well it could predict outcomes — correctly predicting Harry S. Truman’s surprise victory over Thomas E. Dewey in 1948. But it did fail twice, both times in close races.
In the 1968 election, IHS predicted that Richard Nixon would lose to Hubert Humphrey. In the 1976 race, the formula calculated that incumbent Gerald Ford would beat Jimmy Carter. Ford lost with 48.9 percent of the vote.
(The formula also predicted that Al Gore would win in 2000. Gore won the popular vote, which is what the formula measures, but lost in the Electoral College. “We count that as a success,” Gault said.)
In those close races, outside factors played a determining role. Gault said the Vietnam War factored into Nixon’s win and the Watergate scandal — and Ford’s pardon of Nixon — helped swing votes to Carter.
Which brings us back to this fall’s presidential prediction. Obama inherited the recession, but unemployment has remained high, and disposable personal income has stagnated, growing just 1 percent over the year. That’s hardly a winning combination for the president, who is expected to receive just 45.4 percent of the popular vote, according to the latest IHS calculation.
Yet there are factors the formula cannot measure that could make a difference this year, such as likability, which appears to be helping Obama, according to Gault and recent polls.
Gault, who updates the formula every month with the latest economic data, will do it a final time in October, after September data are released. The gradually improving economy has helped Obama’s chances a bit; a year ago, the formula projected that Obama would receive only 43.5 percent of the popular vote, 2 percentage points below the latest prediction.
But the economy, which moves slowly, will probably not change enough to alter the outlook for the president significantly, he said.
“We’ll have projections at the beginning of September, October and the very beginning of November, one day before the election,” Gault said. “But the forecast will change just marginally.