Invisible hand makes projecting winner dismal.

As long as there are presidential elections, there will be economists projecting the results months in advance. But this year, what appeared to be strong signals pointing overwhelmingly to the re-election of President Bush are beginning to get crossed.

"We're starting to see a bit of pressure from higher gasoline and oil prices and softness in the jobs data that have moved the numbers down for Bush," said Robert Dye, a senior economist with Economy.com. "It's been a very interesting, very tight race."

Economy.com still predicts Bush will win the election with 53.7 percent of the vote, based on an economic model updated last week. That's down slightly from 54 percent of the vote when the consulting firm ran its economic model last month. In February, Bush was forecast to win with 57.9 percent of the popular vote.

James Doti, an economist at Chapman University who also engages in presidential forecasting, conceded that there are all kinds of electoral nuances this time around.

"We're dealing with a presidential election that will increasingly be dependent upon a war in Iraq and it will certainly be a major influence and one that could offset the margin of victory for George Bush." Doti said.

While hardly foolproof, forecasting the rest, It of presidential elections based on economic modeling has been a very accurate gauge over the years. Among the most reliable is Yale University economist Ray Fair, who uses growth and inflation data in making his picks--an approach that would have correctly predicted 18 of the past 22 elections going back to Woodrow Wilson's victory in 1916.

This year, Fair has predicted Bush will win with 58.5 percent of the votes cast, the biggest margin of victory since Ronald Reagan defeated Walter Mondale in 1984.

Normally, data on growth, inflation and employment are reliable indicators of voter sentiment, which is why they tend to get so much attention. But given the recent flurry of sour economic news, "along with public opinion polls generally showing Kerry with a slight lead, forecasters are beginning to hedge their bets.

A poll released last week by the Pew Research Center for the People and the Press found that 66 percent of those surveyed said the economy was just fair or poor, compared with 33 percent saying it was good or excellent. The poll also found that a majority of respondents disapproved of the way Bush was handling the economy.

Sometimes, those trends can be tricky to determine. In the three elections in which first-term incumbents have lost re-election bids--Herbert Hoover in 1932, Jimmy Carter in 1980 and George H.W. Bush in 1992--the unemployment rate rose in the 12 months prior to the election.

Most forecaster had predicted an easy victory for Bush because the unemployment rate has fallen in the past year to 5.5 percent in July, down from 5.9 percent in November. But in the government's separate survey coveting payrolls, the results are far worse, with more than 2 million jobs lost during the Bush administration.

Another imponderable: the Electoral College. Most forecasts are based on who wins the popular vote, which could be less of a relevant indicator than who captures a handful of battleground states such as Ohio, Michigan and Pennsylvania.

And don't forget the stock market. Since 1900, incumbents have been re-elected in years when the Dow Jones Industrial Average posted average annual gains of roughly 6 percent. With the exception of 1932, the incumbent was ousted in election years when the market has averaged losses,

This year, the Dow has fallen 4.9 percent, while the Nasdaq Composite Index is down 11 percent and the Standard & Poor's 500 has dipped 3.2 percent.

Higher voter turnout could radically alter the outcome of the 2004 race. Many of the forecasting models relied on relatively low voter turnout in the 2000 election, when just 68 percent of registered voters hit the polls. Higher turnout could have a major impact on the outcome.

Economists for Bush
Looking at the numbers, re-election is predicted.

Economist                        Projection            Method

Keitaro Matsuda, Union Bank         Bush      Unemployment and
  of California                                 payroll numbers
James Doti, Chapman University      Bush      GDP growth
Ray C. Fair, Yale University        Bush      GDP growth, inflation
Robert Dye, Economy.com             Bush      Unemployment, inflation