This article appears in the October 15, 2012 issue of National Review.

As economist Ray Fair of Yale University has demonstrated, economic conditions are a crucial factor driving voter sentiment in presidential elections. Throughout history, voters have penalized sitting presidents for a lousy economy. With the election approaching, and the economy in neutral, Fair’s model currently predicts, somewhat unhelpfully, that this election is too close to call.

Ever since his first pioneering paper in 1978, Fair has focused on forecasting the national vote with a measure of how the U.S. economy is doing in the aggregate. The model he came up with accurately predicts 21 of the past 24 presidential elections. In a close race, it might well be that local conditions will be the swing factor that determines the election outcome. If that is the case, then there is significant hope for the Romney team.

While a reliable measure of GDP is unavailable at the state level, the Bureau of Labor Statistics does report state-by-state measures of unemployment. Nate Silver of the New York Times has found that changes in unemployment are highly predictive of voting patterns. If voters see unemployment increasing, they punish incumbents.

With many states solidly in the Republican or Democratic camp, this election will clearly be determined by voters in a few swing states. In the nearby chart, each bar shows the change in unemployment between April and August. The color of the bar illustrates the signal from the most recent polls as calculated by  RealClearPolitics.com. Light blue bars are states that are leaning toward President Obama; light red bars lean toward Governor Romney; and grey bars are a toss-up.

The data clearly indicate that a number of key states are in economic free fall. Since April, the unemployment rate in Michigan has increased 1.1 percent, from 8.3 to 9.4. Wisconsin, South Carolina, New Hampshire, and Pennsylvania have also seen large increases in unemployment. These increases have happened steadily over this time period, and  re clearly not a summer blip.

If voting patterns respond to the data as they have in the past, this is very bad news for President Obama.

On the other hand, there are very few paths to the presidency that do not travel through Ohio, and the unemployment rate has dropped in that key state, even while the rest of the nation has suffered, suggesting that Governor Romney will have a heavier lift.

In 2008, President Obama won 15 of these 18 states, losing only South Carolina, Arizona, and Missouri. While his margins were large in some cases, the deteriorating state economies should make such a record extremely difficult to repeat.

—Kevin Hassett is the director of economic policy studies at AEI