Election Forecasters Using Unusual Methods to Predict the Presidential Winner

PAUL SOLMAN:

Want to know who'll win in November? Here are some competing indicators you may not be aware of. You can check out the markets on the Internet, says this financial journalist, the betting markets for president. No, says this professor, try econometrics, which assumes we vote our pocketbooks, and so plugs economic data into math equations to predict the outcome. Finally, there's this political scientist, who also uses equations, but finds voters economically irrational. Our story was prompted by the journalist James Surowiecki and his book "The Wisdom of Crowds," which argues that markets are better than polls — or anything else, for that matter — at calling elections.

JAMES SUROWIECKI, Author:

In 1988 a group of people at the business school at the University of Iowa decided to set up something called the Iowa electronic markets. And the idea was that you would set up a market where people could essentially wager on the outcome of the presidential election. And from '88 to 2000, the IEM basically has outperformed polls three-quarters of the time.

PAUL SOLMAN:

Iowa actually runs two markets, one for which candidate will win the election, the other for the share of the vote.

JAMES SUROWIECKI:

What percentage of the vote, of the popular vote, Bush and Kerry will get. And historically the election eve forecast in this market has only been off by 1.4 percent, which is better than any poll.

SPOKESMAN:

13,000 here!

PAUL SOLMAN:

Surowiecki likens the Iowa Electronic Markets to a stock, bond or commodities market, where the prices at any given moment are the public's best collective guess. At tradesports.com, based in Ireland, you can get even more specific.

JAMES SUROWIECKI:

You can bet on the performance of Bush and Kerry in each individual state in the union; you can bet on Bush in different regions; you can bet on senatorial candidates; you can bet on whether or not the GOP is going to sweep the House and the Senate; and you can bet on things like will George Bush be elected and Osama bin Laden be what they called "neutralized" by Nov. 2.

PAUL SOLMAN:

That's right, the old Bush wins, Osama loses parlay, paying 5 1/2:1 as of Tuesday. Now, this bet may sound outlandish but, says Surowiecki, the crowd, so long as it's diverse and acting independently, has a wisdom greater than any of its members. And you can actually see this in a mundane experiment that's both short and sweet…

JAMES SUROWIECKI:

We're just doing this experiment for television.

PAUL SOLMAN:

…How many jelly beans do you think are in this tub?

MAN:

Five, six, seven…

PAUL SOLMAN:

Out where things were hip-hopping, on the sidewalks of New York, we popped that question.

JAMES SUROWIECKI:

Good morning. If you could tell me how many jellybeans you think are in this jar?

MAN ON SIDEWALK:

800.

JAMES SUROWIECKI:

800?

PAUL SOLMAN:

800.

MAN:

500.

JAMES SUROWIECKI:

500? 500.

PAUL SOLMAN:

The average of the first two guesses: 650.

PAUL SOLMAN:

Your guess now?

MAN:

Let's see. I'd say there are 550 jellybeans in that jar.

JAMES SUROWIECKI:

550.

MAN:

1475.

WOMAN:

500.

PAUL SOLMAN:

500, and you say?

JAMES SUROWIECKI:

How many did he say?

MAN:

10,000.

JAMES SUROWIECKI:

10,000.

PAUL SOLMAN:

The average after eight responses: 2,109. Is your guess going up, going down?

MAN:

How many?

JAMES SUROWIECKI:

How many? One, two…

MAN:

Oh, how many. Okay…

JAMES SUROWIECKI:

Yes, yes.

MAN:

250?

MAN:

352?

WOMAN:

700?

WOMAN:

1,526. ( Laughs )

PAUL SOLMAN:

And now that we have 20 people, we're at, what?

JAMES SUROWIECKI:

1,419.

PAUL SOLMAN:

1,419. And the actual number is?

JAMES SUROWIECKI:

1,350?

PAUL SOLMAN:

1350

WOMAN:

1353.

JAMES SUROWIECKI:

1353, which is…

PAUL SOLMAN:

And there is no person, not a one…

JAMES SUROWIECKI:

Not one did better…

PAUL SOLMAN:

…On all this list did better than the actual collectivity.

JAMES SUROWIECKI:

…Than the group did. That's right. So that's it in action.

PAUL SOLMAN:

And you yourself guessed?

JAMES SUROWIECKI:

I guessed 450. So I could not have been less expert at it than this was. But the group knew.

MAN:

I bet with you it's 4,500 pieces.

PAUL SOLMAN:

Small wonder Surowiecki thinks the betting crowd, putting its actual money where its mouth is on the presidential election, has the best collective guess as to who's going to win. But economist Ray Fair says his mathematical model would have called the winner at least as well in every election but one since 1916. It called the last election 50- 50. The model's based on just three economic factors, or as Bill Clinton might have said, "the economy, stupid."

RAY FAIR:

The growth rate of real GDP in the year of the election.

PAUL SOLMAN:

Per capita because you're not… you're then including population growth?

RAY FAIR:

Yeah. Yes.

PAUL SOLMAN:

Okay, good. And so then, what's the second one?

RAY FAIR:

The second one is the inflation rate. That's number two. The third variable, the number of quarters within the administration have been really strong growth, what I call good news quarters.

PAUL SOLMAN:

Add a few political variables– "is the incumbent running?" For instance– plug in the numbers, and you've got an early peek at Nov. 2, based mainly on voters' views of the president's handling of the economy. Then there's a third way to look at the election: That voters are largely irrational. Political scientist Larry Bartels' research suggests voters may confuse acts of God, like droughts and floods, with acts of the president.

LARRY BARTELS:

We looked at data from the weather service over the whole 20th century, and there's a pretty strong pattern of incumbents doing less well in particular election years in particular states where things have been either too wet or too dry.

PAUL SOLMAN:

Too wet or too dry?

LARRY BARTELS:

Yes.

PAUL SOLMAN:

In fact, Bartels thinks the extreme inclemency of the year 2000 cost the Democrats the White House, much as President Woodrow Wilson's reelection was threatened by the famous New Jersey shore shark attacks of 1916. The government had moved quickly to counterattack and keep people out of the water back then, but…

LARRY BARTELS:

People along the shore were unhappy about the fact that Wilson wasn't doing anything to control the shark attacks and were more likely to vote against him as a result.

PAUL SOLMAN:

Absent an outbreak of drought or sharks, then, what would Larry Bartels look at to predict this year's election?

LARRY BARTELS:

It makes sense to look at lots of different models rather than any one, because all of them have more or less plausible sets of factors that they include; all of them do more or less equally well at accounting for what's happened in the past.

PAUL SOLMAN:

So now, to review the competition, there's Bartels' consensus forecast approach; Ray Fair's econometric equation; the markets as monitored by James Surowiecki; and, of course, the good, old polls, which we'll leave to other reporters. The picks?

RAY FAIR:

57.5-something percent for the two- party vote for Bush.

LARRY BARTELS:

The range is from about 50 percent of the popular vote for Bush up to about 57 percent.

PAUL SOLMAN:

Right now you can buy a Bush contract for about 60 cents, and that means the market thinks there's about a 60 percent chance of winning.

PAUL SOLMAN:

In Ireland, meanwhile, Bush's price has really soared since the republican convention.

JAMES SUROWIECKI:

These are the numbers. So he's up around 70 percent basically.

PAUL SOLMAN:

So if I'm looking at this right now I'm thinking…

JAMES SUROWIECKI:

The election is over? No. Like any market it is possible that we're witnessing a sort of speculative frenzy, much like we saw say in the late 1990s with Internet stocks, it's possible that that is just people feeding on themselves and everyone thinking that they have to get in now before it's too late.

PAUL SOLMAN:

So this might be the Bush bubble?

JAMES SUROWIECKI:

This might be a Bush bubble.

PAUL SOLMAN:

What James Surowiecki means is that there are reasons to doubt all of the current predictions. For example, the Iowa vote share market is still very close.

JAMES SUROWIECKI:

And it's still showing Kerry getting 47 percent, 48 percent of the vote –

PAUL SOLMAN:

Of the two-party vote.

JAMES SUROWIECKI:

— of the two-party vote.

PAUL SOLMAN:

That means excluding Ralph Nader.

JAMES SUROWIECKI:

Right. But so that doesn't suggest that he's completely out of the game.

PAUL SOLMAN:

As for Ray Fair's model, well, when we interviewed him in the early '90s he picked the first George Bush over Bill Clinton. He was wrong in part because of dissatisfaction with the economy that the model didn't seem to pick up, but our own Robin MacNeil did on this very program.

ROBERT MAC NEIL:

The phrase I keep hearing and reading is that you simply don't get it, that you deeply do not understand the depth of unhappiness in the country.

GEORGE BUSH:

Yes, I've heard those charges, but I don't think they're true.

ROBERT MAC NEIL:

You do get it, you're saying.

GEORGE BUSH:

I think I get it.

PAUL SOLMAN:

But the first President Bush didn't get the votes. Professor fair thinks he might be wrong again in 2004.

RAY FAIR:

In my equation I'm only using the growth rate of GDP. So I don't have any income distribution thing. Income distribution this time because of the tax cuts has been… become much more unequal under the Bush administration.

SPOKESMAN:

My friends call me Rob. Rob Moore.

PAUL SOLMAN:

To the extent then that inequality is a major issue this year– and the president's opponents have certainly been emphasizing it– Fair's model may be missing something that really matters. Similarly, there's what's been called the jobless recovery. Economic output has been growing. But almost everyone agrees that it hasn't been generating the usual number of American jobs. Why would that throw Fair's equation off?

RAY FAIR:

Because I'm looking at output growth, not job growth in the equation. And that could be another reason it would be off at least somewhat.

PAUL SOLMAN:

A third reason, says fair: The ongoing troubles in Iraq, totally outside his model. As for Larry Bartels, well, there might be consolation for Kerry supporters from the recent hurricanes in Florida. Voters' adverse reaction to wet weather, that is, might pose a problem for the incumbent despite his efforts. And finally, there's this to consider: Those markets picking Bush only give you a guess as to what's going to happen, not the actual outcome.

JAMES SUROWIECKI:

So say in tradesports, they're now saying George Bush has a seven in ten chance of winning. Well, that means John Kerry has a three in ten chance of winning, and we know that long shots come in. I mean, that's what happens, right? Long shots win.

PAUL SOLMAN:

Put it one last way: From bettors on the presidential election to the crowd watching the break dancers of Times Square, what people think they see at any given moment doesn't always wind up jibing with the ultimate reality.