By Andrew Healy and Gabriel Lenz
In the U.S., we — the voters — elect our presidents using a potentially problematic decision rule: we largely decide who will be president based on the election-year economy (1, 2, 3). If the economy is on an upswing before the election, we usually retain the president or his party’s nominee. If it isn’t, we usually elect the challenger.
Why is this decision rule problematic? Because it means we pick our president in large part by the toss of a coin. Economists have long known that whether the economy is on the upswing or downswing in any given year—that is, where the economy is in the business cycle—is mostly chance. As a result, this decision rule essentially turns the election of the world’s most powerful leader into a game much like, as Chris Achen and Larry Bartels put it, musical chairs: when the music stops on Election Day, voters decide based on where the economy happens to be in the business cycle.
To understand the ramifications, consider the following thought experiment: What if, instead of hitting in 2008, the financial crisis had hit in 2009, after the presidential election? The subprime bubble had been growing since at least 2002, and there’s no reason the music had to stop in 2008. If it had stopped a year later, we may well have elected John McCain rather than Barack Obama. You can conduct a similar thought experiment for many elections. Jimmy Carter experienced a short but severe recession in 1980. If the recession had come a year earlier or a year later, Ronald Reagan may never have become president. If the more severe 1981-82 recession had occurred two years later, Walter Mondale would likely have defeated Reagan in 1984 rather than losing in a landslide.
In “Substituting the End for the Whole,” which the American Journal of Political Science (AJPS) is publishing this month, we investigate why voters follow this rule (press release). The results are intriguing. Our first key finding, based on a combination of national surveys and experiments, is that voters apparently don’t know they are using this decision rule. When we asked voters how they evaluated the president’s economic performance, to our surprise, they said they focus roughly equally on all four years of a president’s term, only intending to put a little more weight on more recent performance. Put differently, voters care about total growth under the President, not just election-year growth.
If voters intend to focus on total growth under the President, why do they actually judge the President on election-year growth and what can we do about it? Using a series of experiments, we investigate these questions by showing participants graphs of economic conditions in each year of a president’s term (they did not know which president). After looking at the graphs, we asked them to evaluate the economy under the president.
After conducting and analyzing more than two dozen such studies and examining several explanations, we ultimately conclude that the election-year focus arises in large part because of the way the news media and the government report economic statistics. Voters want to judge the President on total growth during the term, but the news media rarely report total growth. Instead, their economic coverage generally reports on recent economic news, especially recent trends in unemployment and income growth. Since voters fail to receive the information they desire — total growth over the president’s term — we find that they substitute recent growth in its place without realizing it. If the president’s term ended with strong growth, voters conclude that it went well overall even if it didn’t. If the president’s term ended with weak growth, they conclude that it went poorly overall even if total growth was actually strong.
Voter behavior appears to reflect a pervasive human tendency to inadvertently substitute an easily available attribute for an unavailable one, a tendency that Daniel Kahneman calls “attribute substitution” (4). Studies by Kahneman and many others have documented this tendency across a wide array of experiences, from undergoing colonoscopies to watching TV ads. For example, adding an additional period of mild pain at the end of the colonoscopy can actually make patients recall that unpleasant experience as having been less painful (5). As with economic voting, people judge these experiences not on the whole, but on how they ended.
This explanation suggests a remedy, a way to align voters’ actions with their intentions. In our experiments, providing voters with the attribute they are seeking — total economic growth under the President — eliminates their election-year emphasis. When we provide respondents with total growth, they easily see that strong growth in the first, second, or third year means that presidents can still preside over strong overall growth even if growth is not as strong in the election year. The implications for government and the news media are straightforward. Report not just yearly income growth, but also total growth over a president’s term. It’s what voters say they want to know to evaluate presidential performance. By making that information easily accessible to voters, the government and the news media could help end our game of musical chairs with our highest office.
Crossposted from the AJPS blog