- - Monday, September 24, 2012


Political scientists and pundits are having a field day with the avalanche of electoral polling data. Every week, there is a new crop of polls eliciting voter sentiments and intentions. Pundits scrutinize the polls to detect post-convention “bounces,” responses to specific events and, ultimately, voter intentions. Polling data are the tea leaves political junkies use to predict election outcomes.

But polling data do not forecast elections very well. First of all, there is a high degree of variability in polling data. As with consumer-confidence surveys, voter surveys are subject to a lot of transitory events and emotions. Bad weather, a down day on Wall Street or a personal crisis can depress consumer confidence.

Some of the same factors can alter a person’s perception of political candidates. Take President Obama’s approval ratings, for example. They started at 68 percent just after he was sworn in (February 2009). They then fell into a death spiral, bottoming out at 42 percent in October 2011. They have recovered a bit since, but still remain highly volatile.

In the past two months, his weekly approval ratings among likely voters have varied by 4 percentage points. Daily polls by Rasmussen reveal twice as much variability. Although trends in polling data might be revealing, the data from any given day or week are clearly not stable.

The second and more serious problem with polling data is their unreliability. Polls taken in the months before an election are not good forecasters of election outcomes. In the last nine presidential elections, the candidate ahead in spring polls lost the November election five times. The most dramatic reversal was in 1980, when Ronald Reagan went from being a 36 percent polling underdog to securing a 50.7 percent electoral victory in a three-way race.

Political scientists try hard to figure out why voter polls are such unreliable election predictors. But the more important question is what better forecasting tools are available. The answer may lie in economics. As James Carville famously observed, “It’s the economy, stupid.” When push comes to shove, people will vote their pocketbooks. If the economy is growing, employment increasing and incomes rising, voters will stick with the incumbent. If the economy is in the doldrums, voters will favor a new “economist in chief.”

Yale economist Ray Fair translated this generic observation into a very specific mathematical equation. Mr. Fair says you can predict the outcome of this year’s presidential election by looking at only two economic variables: 1) The growth rate of the economy as quantified by the gross domestic product (GDP), and 2) the rate of inflation.

Think about this proposition for a moment. Mr. Fair says that just two economic indicators foretell the next president. Convention speeches do not matter. The choice of a running mate does not matter. The amount of money spent on the campaign does not matter. Even the televised debates are inconsequential. All that count are economic growth and inflation. Sounds crazy, but the model works. Mr. Fair has correctly predicted the outcome of all but one modern election — a track record that political scientists and pundits can only envy.

Also notice the omission from the Fair model of the unemployment rate. Although the persistently high unemployment rate has gotten the most media attention, the GDP growth rate is really the critical determinant of our economic welfare. Strong economic growth lowers the unemployment rate, raises wages and incomes, reduces poverty and even diminishes inequality. So, GDP growth alone really conveys how well the economy is doing. The other variables are redundant.

Mr. Fair predicts Mr. Obama goes into the election with an incumbent’s base of 48.39 percent of the popular vote. Then economic growth and inflation come into play, as follows:

GDP growth: For every 1 point of real GDP growth in the first nine months of this year, Mr. Obama picks up an additional 0.672 percent of the popular vote. With GDP growth averaging 1.7 percent, Obama picks up 1.14 points in the popular vote. That takes him to 49.54.

Inflation: Inflation hurts Mr. Obama. He loses 0.684 voter points for every percentage point of inflation during his first term. With average inflation of 2.3 percent, this lops 1.57 points off his vote share. Now Mr. Obama is at 47.97 percent.

Strong quarters: The final ingredient in the Fair model is the number of calendar quarters the economy has grown above the long-term average of 3.2 percent. These are huge pluses for an incumbent, bringing in a whopping 0.990 voter points. Unfortunately for Mr. Obama, he had only two such good quarters in his four years. That brings in only 1.98 percent of the popular vote. That leaves Mr. Obama with only 49.95 percent of the popular vote.

Photo finish: What makes the Fair model so unique is not just its outstanding track record, but also its simplicity. To predict an electoral outcome, all you have to do is plug in a couple of economic numbers and spit out the results. You can “plug and play” this model far in advance of Election Day. Right now, Mr. Fair’s math implies a Romney victory in November. But the difference in vote shares is so small — smaller than the standard estimation error — that Mr. Fair himself predicts a photo finish.

Brad Schiller is emeritus professor of economics at American University.



Click to Read More

Click to Hide