ECONOMISTS GIVE BUSH SLIGHT EDGE

		John Kerry in St. Louis; President Bush in Canton, Ohio.
		By Charles Rex Arbogast, AP; Charles Dharapak, AP

From consumer confidence numbers to jobs data to innovative futures markets, economists are using a variety of tools to try to divine the winner of the November presidential election.

But, like the economic outlook, the political projections are mixed.

Iowa political futures markets have President Bush and Democratic nominee Sen. John Kerry in a tight race, with a scant edge for Bush. Consumer confidence data indicate the GOP could hold the White House, though September numbers will be key.

Several comprehensive models combining economic and political data show Bush should win about 54% to 57% of the vote. But others say a GOP victory isn't a lock.

Job growth in the six months prior to an election is the best signal of the outcome, says Donald Straszheim of Straszheim Global Advisors, adding the past couple of months of below-par hiring could tilt the battle to Kerry.

"There has never been a bad economy in which the incumbent has won," says Straszheim, who combed through data back to the 1960s. "There has been a good economy in which the incumbent has lost. It's asymmetric."

Economists say their models help them better understand the intersection between politics and business. They also say developing the ability to predict election outcomes would give them, and their clients, early clues about the direction of tax, spending and regulatory policy.

Many have built on the work of Yale University professor Ray Fair, a pioneer in econometrics, or the use of mathematical and statistical methods to describe economic relationships. Fair's model — which includes such variables as economic growth, inflation, incumbency and duration in office — predicts Bush will get 57.5% of the vote. He has been accurate in five of the past six elections, but missed in 1992.

More difficult to predict this year

But political prognosticating at a time of war could be dicey. Fair's model, which has a standard error of 2.4 percentage points in each election, wouldn't capture possible effects on voting from the Iraq war.

While historical data suggest models aren't tripped up by international conflict, some aren't sure.

"These models are not as helpful this time, because the issues are so different. National security is a tough issue to quantify," says Greg Valliere, managing director of Schwab's Washington Research Group, who has correctly predicted the winner of every presidential election since 1976.

The firm expects a very narrow victory for Bush, based on analysis, intuition and experience.

For those who can't wait for November, here's an advance peek:

Job creation. Straszheim looked at economic data back to 1964 to see which closely correlated with election results. He found job growth in the six months before Election Day was key.

Average job gain over a six-month period before elections is 1%. Anything 25% higher constitutes a good economy, 25% lower, a bad economy, Straszheim says. In an average period, factors such as foreign policy or social issues can trump economics.

The economy would have to create 169,000 jobs a month for Bush to avoid a range that signals a loss. That's happened in three of the 43 months Bush has been president.

Iowa Electronic Markets. Created as a research and teaching tool by the University of Iowa's Henry B. Tippie College of Business, the Iowa Electronic Markets (IEM) let participants invest $5 to $500 in Web-based futures contracts that predict who will win the presidential election or the share of the vote each candidate will garner. Winning contracts pay $1, while losers go away empty-handed.

The markets are open, though traders tend to be younger, more affluent and more educated than the general population, with a heavier concentration of white males. But the markets aren't designed to provide a voter cross section. They let people invest for whom they expect, not necessarily want, to win.

Since the 1988 presidential election, the IEM has had an average election eve prediction error of 1.37%, outperforming many polls. "If your weatherman did that, you'd be thrilled," says Thomas Rietz, associate professor of finance and a director of the IEM.

For periods this year the markets have had Kerry trailing Bush when public opinion polls have shown him ahead. Recently, the markets have tightened.

Consumer confidence. The New York-based Conference Board issues a monthly index of consumer confidence, based on a poll of 5,000 households. For more than three decades, the incumbent party has won when the index was above 100 in September.

While the index hit 106.1 in July, recent economic data have been mixed. And Lynn Franco, director of the Conference Board's Consumer Research Center, cautions that the index can't predict the Electoral College.

"With the 2000 election, what we saw there was (former vice president Al) Gore got the popular vote, but Bush won" with an index over 100, Franco says.

Global insight. The economic consulting firm's model looks at such factors as after-tax income and unemployment. It gives an advantage to incumbents, subtracts for "party fatigue" if one party has been in power a long time and looks at historical links between party affiliation and the economy.

The model works for most elections, though it would have chosen Democrat Hubert Humphrey over Republican victor Richard Nixon in 1968, and predicted a bigger vote for former Republican president Gerald Ford, who lost to Democrat Jimmy Carter in 1976.

Global Insight currently forecasts Bush will get 56% of the vote. The firm says it would be difficult, but not impossible, for the economy to falter to the point Bush would lose. But there are caveats.

"To a large extent, the election may turn out to be a referendum on whether Bush has managed the war on terror or did the correct things in Iraq," says Global Insight economist Nigel Gault.

While the 5.5% unemployment rate is not historically high, Bush has presided over net job loss.

"The model just looks at the absolute level of unemployment and says, 'That's not too bad, Bush is an incumbent, so he should have an advantage,' " Gault says. "The model could get it wrong."

Economy.com. This consulting firm's model looks at changes in state unemployment rates eight quarters before the election and changes in national consumer inflation for six quarters. It includes political factors such as incumbency and the length of time a political party has been in power.

"We wanted to select variables that the consumers look at," says Robert Dye of Economy.com.

The model predicted Gore would win the popular vote in 2000. It expects Bush to win 40 states and 53.7% of the vote.

Dye says the Iraq situation could make modeling tough, adding the election will have a big economic impact on consumers based on how the next president handles war, deficits and taxes.