Predicting The Presidential Election

PREV 4 of 4

I've decided to wait and see who the nominees will be. When one is making bets, waiting often pays off. The best time to try to predict presidential races is in the summer of an election year, when timely economic information is available and opinion polls are more reliable.

Several political scientists have devised voting models that combine economic statistics and survey data. Alan Abramovitz, of Emory University, uses three indicators: G.D.P. growth rate in the first half of the election year, the president's approval rating in June, and a "time for change" factor if the incumbent party has been in office for two or more terms. Michael Lewis-Beck, of the University of Iowa, uses the same economic variable as Abramovitz, the president's approval rating in July of the election year, and a polling indicator that captures the electorate's feelings about noneconomic issues. James Campbell, of the State University of New York at Buffalo, waits until early September before issuing a forecast. His model has just two variables: the growth rate of G.D.P. in the second quarter and how well the incumbent party's candidate performs in polls conducted on Labor Day weekend.

Each of these model builders has had some success, and details of their work are available online. Campbell, for instance, correctly predicted the popular-vote winner in 1996, 2000, and 2004—a feat that neither Fair nor Hibbs managed. But picking the winner a month or two before an election is hardly the same as forecasting the outcome 12 months in advance. If you wait until Election Day is that close, you don't actually need a complicated statistical model to get decent results. Here are three simple rules of thumb that work almost as well:

1. If the annualized growth rate of G.D.P. in the second quarter is 2.6 percent or higher, the incumbent party will retain the White House. If the growth rate is 1.5 percent or lower, the other party will win.
2. If the president's approval rating in the mid-June Gallup poll is below 45 percent, his party will lose in November. If his approval rating is 51 percent or higher, his party will win.
3. Whichever candidate is ahead in the Gallup poll taken in the third week of September will win the election.

But please don't hold me to that forecast, much less bet on it. In case the economy rebounds strongly from the credit crunch in the next six months and the situation in Iraq improves, providing a boost to President Bush's approval rating, I reserve the right to change my mind.

PREV 4 of 4