Fair, calculated look at the next election

Politicians have long recognized that economic prosperity — or the lack thereof — is a powerful factor in politics. Four centuries ago, Henry IV promised “a chicken in every peasant’s pot.” He still is revered as France’s most popular king. Americans also know how Herbert Hoover was trounced by Franklin Delano Roosevelt in 1932 because he could not stem the Depression.

Yale University economist Ray Fair has an equation that predicts presidential races quite accurately using little data. Fair’s model includes some information that is fixed in any particular election, such as whether an incumbent is running and how long a particular party has held the White House. It then factors in three economic variables: the rate of growth of output, the inflation rate and the number of quarters in the incumbent’s term in which real per-capita output grew faster than 3.2 percent.

As of this month, growth is 3 percent, inflation is 1.9 percent per year and output growth per person is expected to exceed 3.2 percent three out of 15 quarters. With these numbers, Fair’s model predicts that incumbent George W. Bush will get 58 percent of the popular vote.

Democrats will find comfort in the fact that while Fair’s equation was far off the mark in only one of the last 22 elections, that one was the 1992 Bush-Clinton race. Fair predicted George H.W. Bush would get 5 percentage points more than the 46.5 percent he actually did.

Stung by that experience, Fair recently hedged the current forecast of 58 percent for Bush, recognizing that, “Perhaps voters look much more now at foreign policy and social policy than they did in the past and less at how the economy is doing.”

To learn more about Fair’s model, go to http://fairmodel.econ.yale.edu/vote2004/index2.htm.

In a different approach, the University of Iowa operates an election futures market. Participants can buy and sell futures contracts that pay out relative to actual votes received by each candidate. The market opened only in 1988, but in the five presidential elections during which it has operated, it has been more accurate in predicting final vote shares than the last national polls taken before Election Day.

Economics and finance professors note that this outcome supports the assumption that those who have a financial stake in some outcome — the minimum amount one can buy in the market is $5 — have more accurate information than disinterested individuals.

As of midweek, the Iowa market gives Bush a 52 percent lead over Kerry’s 48 percent. Information about the Iowa Election market is at http://tippie.uiowa.edu/iem/. Quotes of actual prices are updated every 15 minutes.

The Fair and Iowa models vary sharply. Fair depends on historic relationships and ignores factors many might think relevant. The Iowa model shuns historic data, but simply tabulates the best estimates of thousands of participants who are willing to risk money on the outcome. In this unusual election year, I’d bet on bucolic Iowa over the paneled halls of Yale.

© 2004 Edward Lotterman
Chanarambie Consulting, Inc.