Can economics predict presidential elections?
By Mike Walden
N.C. State University
The presidential election is still months away, and one of the major parties hasn’t even picked its candidate. Nevertheless, this doesn’t prevent people from speculating about who might be the eventual winner.
Of course, anyone can make a forecast, or guess, and make that prediction based on anything or nothing at all! Las Vegas will even take bets on the race. But if you want to make an informed forecast, you have to have some model — or theory — about what determines how people vote.
Certainly this is a broad topic that can be approached from many angles or disciplines. Historically, however, analysts have found that economic conditions are among the more important set of factors influencing voting behavior. And in this year — when economic challenges are still obvious — the idea that economics might prevail in the election is even more compelling.
Indeed, for several decades one economist has been tracking the relationship between the economy and presidential elections. His name is Ray Fair, of Yale University, and his updated book on this work, simply titled “Predicting Presidential Elections,” just hit the bookstands.
Now Fair doesn’t claim that only economics matters to how we vote. But he does argue that economics is very, very important, and he has put his model of presidential election predictions to the test for almost of century of contests.
Before I tell you how good Fair’s predictions have been, let’s see what economic factors he deems most important in influencing our voting behavior. Fair thinks there are two economic biggies when we pull the election lever (or connect the lines on North Carolina’s ballot): growth and inflation.
For growth he uses two specific measures: how much the economy has grown on a per-person basis in the first nine months of the election year, and in how many of the 15 quarters (three-month periods) of the presidential term up to the election has the national economic growth rate per person exceeded 3.2 percent.
Fair found these two growth measures to be the most predictive and reliable among many he tested. Yet some may question why job market data, like the unemployment rate or speed of job growth, weren’t included. Again, Fair found the two broader economic growth measures to be better at forecasting. Also, job growth and broader economic growth are highly related.
For inflation, Fair reasoned that most people like price stability. We don’t like prices rising rapidly, but we also don’t like them falling fast because price drops are usually related to lower pay. So Fair used the average rate of price change (without distinguishing between price rises or price drops) for the 15 quarters of the presidential term up to the election.
Professor Fair also included a couple of non-economic factors related to the person and party holding the presidency. First, a sitting president has an enormous amount of power, and Fair’s statistical analysis found that an incumbent president running for re-election is a plus for that person in receiving another term.
But second, people may become tired of the same party holding the presidency. Fair’s work found that the number of terms the incumbent party has held the presidential office was related to lower chances of that party continuing in office.
So how good has Ray Fair’s model of predicting presidential elections been? Very, very good! In the 24 elections from 1916 to 2008, his model correctly predicted 21 of the contests and incorrectly predicted three. The three bad calls were 1960, when John Kennedy defeated Richard Nixon; 1992 when Bill Clinton beat George H.W. Bush; and 2000 when George W. Bush won over Al Gore.
However, it can be argued that all three of those elections had special circumstances. In 1960, the margin between Kennedy and Nixon was razor thin. A strong third party challenge by Ross Perot made the 1992 election difficult to predict. And in 2000 Al Gore won the popular vote but George W. Bush prevailed in electoral votes.
So what can the Fair presidential election predictor tell us about the 2012 outcome? Incumbency gives the president an advantage, as does the fact his party has held the presidency for only one term.
Also, so far the president holds an advantage on inflation, with the average rate for his first three years being a modest 1.5 percent. But recently, the rate has been ticking higher, and if gas prices continue to soar, by election day people may perceive inflation to be much higher, and this would favor the president’s challenger.
The real wild card is economic growth. Average economic growth per person for the first three years of the President’s term has been very slow, as a result of the deep recession and slow recovery. But growth appears to be picking up in 2012, and if it accelerates, voters could enter the election booth with more optimism about their economic future.
So at this point, Fair’s model is pointing to a close election!
Any economic model is a simplification of reality, and certainly our country is about much more than economics. So you can throw out Ray Fair’s presidential predictor, or you may use it as a guide, but of course, you decide!
• • •
Dr. Mike Walden is a William Neal Reynolds Professor and North Carolina Cooperative Extension economist in N.C. State University’s College of Agriculture and Life Sciences.