主题：【置底的帖子来乐】经济模型预测，薯仔会赢 -- 西风陶陶

Bush Will Win, Economic Models Say; War, Terror Are Wild Cards

Aug. 31 (Bloomberg) -- President George W. Bush will win four more years in the White House under formulas devised by professors to predict the outcome of U.S. presidential elections. It may not be that simple an equation.

Eight forecasting models are unanimous in predicting Bush will defeat Democrat John Kerry with as much as 57.5 percent of the vote. The models are based on measures including economic- growth projections and opinion polls. Economist Ray Fair of Yale University says his model correctly accounts for the popular vote in 19 of the past 22 elections.

Even so, forecasters such as political scientist James Campbell say the hunt for terrorist Osama bin Laden and discontent with the war in Iraq are among wild cards that may muddle the modeling. Many forecasts are at odds with polls showing Kerry, 60, running even with Bush, 58. The Standard & Poor's 500 Index, down about 1.4 percent this year, isn't providing Bush with an election-year boost either.

``Idiosyncratic events, things that can't possibly be anticipated, might be more likely to occur during this campaign,'' says Campbell, 52, who teaches at the University of Buffalo in New York. ``If they capture bin Laden, or if things take a turn for the worse in Iraq, those are things none of the models can anticipate.''

Conventional Wisdom

Campbell says he isn't surprised that many forecasts show Bush winning, even decisively, on Nov. 2 while public opinion polls indicate a race too close to call. Most assume that an incumbent who returned his party to the White House, as Bush did in 2000, has an inherent advantage, Campbell says.

Campbell's own formula, developed with a colleague, Kenneth Wink, focuses on second-quarter economic growth. The results, to be released after Labor Day, Sept. 6, probably will show Bush, who will accept the Republican nomination Thursday for another term, with a small lead over Kerry, he says.

Michael Lewis-Beck, a political scientist at the University of Iowa in Iowa City, says the discrepancy between polls and economic forecasts might arise because forecasts don't measure how people feel about the economy.

``If you look at economic growth, 3 percent or so, that's not telling you about the increasing job insecurity people have, either because they're unemployed or underemployed,'' says Lewis- Beck, 60.

The economy may expand 4.3 percent this year, based on the median forecast in a Bloomberg News survey of 53 economists taken on Aug. 9.

Polls, Jobs

Using a formula that considers polls, economic growth, employment and the benefit of incumbency, Lewis-Beck and Charles Tien of Hunter College in Manhattan predict Bush will win 51.2 percent of the popular vote. With the model's error margin at 1.5 percentage points, it may mean Bush loses.

The formulas of six professors scheduled to participate in a panel discussion this week at the annual meeting of the American Political Science Association predict Bush winning, with as much as 54.7 percent and as little as 51.2 percent. The models measure the two major parties' votes and generally don't forecast results for third-party candidates, such as Ralph Nader, 70, who has been endorsed by the Reform Party.

Spokesmen for Kerry and Bush didn't respond to telephone messages for comment. Robert Shapiro, an economic adviser to Kerry, said earlier this month that such models ``are fallible and will prove to be so again.''

A look at election-year stock-market performance tells a different story than do the models, one favoring Kerry, a four- term senator from Massachusetts.

S&P 500

Since World War II, the S&P 500 declined from the start of an election year to the date of the presidential vote only twice, in 1960 and 2000, according to Bloomberg data. The incumbent party lost in both contests, as it did in 1992, when the benchmark posted its smallest gain in a postwar election year.

A study by Kenneth Tower, 49, chief market strategist at Charles Schwab Corp.'s CyberTrader Inc. in Princeton, New Jersey, found that in the 10 instances since 1900 when an incumbent-party candidate lost, the outcome was preceded by middling growth in the Dow Jones Industrial Average -- either less than 10 percent over four years or less than 5.2 percent in the election year.

Since Bush took office, the Dow has fallen about 3.8 percent, and it's down 3 percent this year.

Also favoring Kerry's chances: Bush's job-approval ratings have steadily declined from 90 percent less than two weeks after the terrorist attacks of 2001. In a Gallup poll of 1,005 Americans in June, 48 percent said they approved of Bush's performance. The poll had a margin of error of plus or minus 3 percentage points. The last five presidents who won second terms -- going back to Dwight Eisenhower -- had ratings above 50 percent in June of their re-election years, according to Gallup.

49 Percent Approval

The most recent Gallup poll, in late August, put Bush's approval rating at 49 percent. That survey of 1,004 Americans also had a margin of error of plus or minus 3 percentage points.

Forecasting the outcome of presidential races has tested American ingenuity for more than a century. The tools range from the serious to the frivolous, from academic formulas to the collective wisdom of gamblers to how the candidates -- quite literally -- stack up in height.

As early as the 1860s, organized markets accepted wagers on the elections, according to a study by Paul Rhode and Koleman Strumpf, economics professors at the University of North Carolina in Chapel Hill. They examined results from 1884 to 1936.

``The resulting betting odds proved remarkably prescient and almost always correctly predicted election outcomes well in advance, despite the absence of scientific polls,'' they wrote.

Election Futures

Today, traders worldwide can buy and sell futures contracts on the Bush-Kerry contest through the University of Iowa's Electronic Markets. Contracts for Bush's re-election were trading at 55 yesterday afternoon. The contract price represents the chance the market gives of a Bush victory.

Bush was also leading on the Dublin-based Intrade Exchange Internet betting site. Bids last week gave him a 53 percent chance of winning. By the time the Republican convention opened yesterday, his chances had risen to 58 percent in the collective judgment of Intrade traders.

A wave of pro-Kerry bets since the Democratic National Convention has made the race a toss-up at U.K.'s Ladbrokes.com, the world's biggest bookmaker, a division of Hilton Group PLC. For both Bush and Kerry, a correct $6 bet now brings a $5 profit.

``The Republicans opened up as quite strong favorites,'' says Warren Lush, Ladbroke's chief odds maker for political topics. ``But as the Kerry bandwagon continues, we get to the position of 5/6 for each, which for a bookmaker is like the toss of a coin.''

``We can't split it,'' he says.

The Taller Man

One theory, based on softer data, offers hope to Kerry.

In 10 of the past 12 presidential elections, the taller candidate won the popular vote. The edge this year is clear: Kerry, at 6 feet 4 inches (1.93 meters), is 4 inches taller than Bush.

Many of the more serious forecasting models use mathematics and statistics to combine political factors, such as the incumbent's poll numbers, with economic indicators, such as the growth in gross domestic product.

Yale economist Fair weighs growth and inflation. As of July 31, his model gave Bush 57.5 percent of the vote. He doesn't include polling data.

In a note accompanying his forecast, Fair described ``possible pitfalls.''

``The Bush administration has made many large changes in foreign policy, social policy and environmental policy, and it may be that these changes are so large that voters radically change their voting behavior,'' he wrote.

Only Politics

In 2000, Fair's model said then-Vice President Al Gore would win the popular vote with 50.8 percent. Gore ended up winning 50.3 percent in the head-to-head contest with Bush. Bush won the White House by gaining the most electoral votes, which are distributed on the basis of state-by-state results.

A model by Helmut Norpoth, a political scientist at the State University of New York at Stony Brook, uses only political factors, particularly New Hampshire's January presidential primary balloting. His model, revised after the 2000 vote, envisions Bush winning this year with 54.7 percent.

Without a high-profile challenger in New Hampshire, Bush collected more than 90 percent of the Republican primary vote. His father, by contrast, won New Hampshire in 1992 with 58 percent against challenger Pat Buchanan's 40 percent. The elder Bush lost his bid for a second term that year to Bill Clinton.

``The fact that Buchanan was able to get as many votes as he did indicated a very profound weakness of the incumbent president,'' says Norpoth, 60.

He says his model accounts for the winner in 22 of the 23 presidential elections since 1912. Norpoth predicted a Gore victory with 55 percent and made a post-election revision that took Gore's value down to 49.7 percent.

Revealing Vote

Bush gets 52.6 percent of the vote in a projection by political scientists Christopher Wlezien of Oxford University of the U.K. and Robert Erikson of Columbia University in New York that includes leading economic indicators

Wlezien, 43, and Erikson built a model designed to capture signs of late growth in the economy likely to offset what they called Bush's ``mediocre'' poll numbers. The forecast for Bush of 52.6 percent uses polls and economic indicators as of July.

Wlezien says the model correctly accounts for the winner of the popular vote in 10 of the last 13 elections. In August 2000, the model forecast a Gore victory with 55.2 percent.

Decisive Victory

Iowa's Lewis-Beck and Hunter College's Tien predicted a decisive victory for Gore -- 56.9 percent. Back at the drawing board after the 2000 election, they added net job growth to their formula and found it to be ``a potent variable,'' Lewis-Beck says, one that brings the model into better alignment with results in 2000 and earlier.

The economy has registered a net loss of more than 1 million jobs since Bush's inauguration in January 2001, according to Labor Department statistics. The model by Lewis-Beck and Tien giving Bush 51.2 percent also weighs polls, economic growth, employment and the advantage of incumbency.

Lewis-Beck calls the result ``too close to call'' given the model's margin of error of plus or minus 1.5 percentage points.

He says an incumbent who gets 50 percent to 51 percent of the popular vote in his model is more likely than not to fail to gain a majority of electoral votes, the fate that befell Gore in 2000.

``It looks like it may even happen again,'' he says.