The University of MichiganNews Services
The University Record Online
Updated 10:30 AM October 8, 2004




view events

submit events

UM employment

police beat
regents round-up
research reporter


Advertise with Record

contact us
meet the staff
contact us
contact us
Local economy doesn't drive voters' choices

Is it really the economy, stupid?

Democrats rallied behind that phrase in 1992, but a U-M researcher says that, despite the conventional wisdom of political pundits, voters don't pick a president based on the previous four years of economic conditions or on the local economy.

"We find the incumbent doesn't appear to be held accountable for the economy," says Daniel Eisenberg, assistant professor of health management and policy at the School of Public Health. "Local-level economic performance seems to matter very little."

Eisenberg became curious about the connection between the economy and elections while working on his doctorate in economics at Stanford University. He evaluated data for 18 national elections from 1932 to 2000, and for county-level elections from 1972-2000, looking to see whether the local or national economy predicted an incumbent's votes for re-election.

Eisenberg had his results published in the online journal Topics in Economic Analysis and Policy. His co-author, Jonathan Ketcham, is joining the University of Southern California this year as an assistant research professor in the School of Policy, Planning, and Development.

"National economic conditions do affect the incumbent's vote share," Eisenberg says. "Previous research has shown this and our results are consistent with this finding."

But voters who lived in counties with particularly bad economic conditions were only slightly less likely to vote for a presidential incumbent.

Similarly, a depressed economy for the length of the president's first term did not automatically predict voters turning down the incumbent's bid for re-election. "We find that it's really just the last year of the economic performance that matters most," Eisenberg says.

"Such voting behavior provides incumbents and their parties with incentives to focus on short-term election-year stimulation at the expense of the three years following the election," Eisenberg and Ketcham wrote.

Ketcham connected the research to this year's election: "John Kerry has often pointed out that the economy has fewer jobs than when George W. Bush took office, but this isn't what people have typically looked at," he says. "Voters are more likely to consider the fact that unemployment has dropped to 5.4 percent in August from 6.3 percent in June 2003 than to consider that it has climbed from 4.2 percent when Bush took office."

"Our results suggest that Bush should not be counted out in counties or states that have been hardest hit by the recent recession," the authors wrote. "The most important economic factor appears to be national performance in the most recent year, and these indicators in 2004 to this point are favorable."

Black voters, women and people who are not elderly are more likely to vote in sync with the economic conditions—the converse being that older white men seem to put less stock in the recent economic situation when they vote.

The authors note that local and national economic voting are connected—we form our notion of the country's health by looking around us—but they are not the same. The researchers compare it to the difference between "pocketbook" voting, which is based on personal effects, and "sociotropic" voting, which is motivated by a sense of what is best for society. Though the two are associated, they are not identical.

Of the various economic indicators they considered, national unemployment mattered the most. Each 1-point increase in the unemployment rate nationally corresponded with a 3.2 percentage point decline in the incumbent's vote share. Unemployment only showed statistically significant effects at the national level, not more locally.

To read the full article, go to

More Stories