NYT Business: "Polls Say Kerry, Futures Say Bush"

[swag]

We have had some discussions lately about economic models predicting presidential winners, the relative value of political futures markets (tradesports.com, iowa, etc.), so I thought today's NYT Business section article on the matter was interesting. You can read it here:

http://www.nytimes.com/2004/08/08/business/yourmoney/08view.html

(I'm sticking with truthisall for now)

Polls Say Kerry. Futures Say Bush.
By DANIEL GROSS

Published: August 8, 2004

. . .

Looking at results dating to 1916, Professor Fair has found that incumbents running for re-election have a head start, that Republican incumbents tend to do better than Democratic incumbents and that an incumbent fares better when his party has not controlled the White House for two terms or more. Thus, he said, "Bush has the best possible incumbency situation that you could have." Plug in the variables on growth and inflation, and Professor Fair projects that President Bush will receive 57.48 percent of the two-party vote.

PROFESSOR FAIR'S model has been remarkably accurate over the years, with an average error of only 2.4 percent. And in each of the last two elections, he has come within 1.3 percentage points of predicting the accurate vote total.

Great news for President Bush, right? Not necessarily. Professor Fair says the model can misfire when the equation doesn't include significant economic factors that may influence the electorate. It lacks a variable for job creation, for example, which may be a sore spot among voters this year.

And the model turned in its worst performance when an incumbent named Bush was seeking re-election. In 1992, the model projected that President George H. W. Bush would win 51.7 percent of the two-party vote and retain the presidency. Instead, he received merely 46.5 percent of the two-party vote and lost.

. . .

[napi21]

1. Could this just be wishful thinking?

Watch the futures market for a few days. They are the most unstable, and changing group you can find anywhere. It was a cute blip for the 24/7 media to show, but I'd bet it will change at least 500 times before Nove 2nd.

[swag]

6. Well, I think the bettors on these markets, as the article says,

have an inherent pro-Republican bias and an almost innate sense that nothing can ever take their tax cuts away.

I think tradesports was giving Bush 58% probability when I, pessimist, was giving him 49%. There is definitely a 4 - 10% bias in these "markets" for Bush when compared to the polls.

I wonder how Big John Poindexter's terra sports book is doing these days? I'm sure it exists, albeit on an invitation-only basis.

[TruthIsAll]

18. Swag, I agree. I appreciate your comments.

Did you see my thread re: Princeton Professor Sam Wang?

He's posted that his model produced a 98% Kerry win probability- exactly the same as mine.

tia

[Tarheelhombre]

7. The model has never been right when....

The incumbent is named Bush.

[aquart]

2. And nobody ever lost his shirt in the futures market.

Right?

Wait a minute. "Lacks a variable for job creation"?????????

Who is this dimwitted tuna?

[DemocratSinceBirth]

3. Two Party Vote- What The Hell Is That

Bush got 37% of the vote in 92.... Clinton and Perot got the other 63%....


Those same models had Al Gore getting 63% in 00...

[Julien Sorel]

15. Actually,

Fair's model had Gore winning 50.8% of the vote, as opposed to the 50.5% he ended up winning, which is amazingly accurate. I think there's a lot more on the table this time (Iraq, 2000 election, the fact that Bush is a moron), which vitiates Fair's model somewhat, which is rather simplistic in a lot of ways (ordinarily its simplicity is one of its strong points), but Kerry is still sailing into a headwind.

[gulliver]

4. I don't believe in these models at all.

No variable for "job creation?" What about a variable for starting unnecessary wars? What about a variable for 9/11?

Where we are now can't be modeled easily. It may even be impossible. The whole Bush Administration is anomalous and radical. Anything could happen. Anything.

[unblock]

5. these historical models ignore specifics

they look at things like gdp and the dow and inflation and political control and so on.

but they ignore things like massive change in deficits and massive loss of credibility from lying and massive partisanship that really rile the opposition.

[MidwestTransplant]

8. Also consider that these markets are pricing in a RNC bounce

[swag]

9. Good point. The markets are also pricing in the Machiavellian

factor, I think: the belief that Bushco will and can do anything to win.

Nevertheless, if I didn't have both legal and ethical qualms, I would short the fuck out of that Bush contract.

[TruthIsAll]

10. Kerry stock will rise when the traders realize Bush is TOAST.

The futures markets will correct.

I calculate a 92% probability (as does Professor Sam Wang) that Kerry will win, based on the current polls. This number will slowly increase right up to the election. The probability is 97% right now if you assume that 70% of undecideds/other go for Kerry.

Futures traders are going to have to start reading DU and other polling blogs which have Kerry solidly ahead.

Ray Fair is already qualifying his prediction that Bush will win.
His numbers are way off the mark.

And those wacko economists who predict a 61% Bush popular vote are just drinking BushCo kool-aid.


http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=132x600977

[Liberal Gramma]

11. This one's about as valid as "The Tall Guy Wins"

[Downtown Hound]

12. Sheesh, Economists can't predict job creation,

job losses, deficits, debts, GDP, or anything else they're supposed to be able to predict. I wouldn't trust them with picking a president. These are the same idiots that are astounded that July's job figures were so poor. I have no economics education, never read the business section of the paper, and have no investments, and I'm not surprised. These guys need to start looking for a new job.

[jmknapp]

13. futures=weather

The futures markets are similar to weather bureaus. They are always right because they keep changing as events warrant, right up to the last minute.

[swag]

14. Right. The tradesports.com bush contract will expire at 0 on 11/2

[A_Possum]

16. "remarkably accurate"

Yeah, yeah, that's what all the snake-oil salesmen say.

There's no predicting the future with something as specific as elections. There's only hindsight.

Keep in mind that the weather forecaster who won the prize for most accurate forecasts turned out to have simply always forecasted that the weather tomorrow would be the same as the weather today. ;) That may be a high rate of accuracy, but it doesn't help you much if you don't bring your umbrella and it rains.

"Predictors" like this are bosh, and this one be would be bosh if it said Kerry was going to win, too. But if it makes the Repugs feels better, bring it on.

[Inland]

17. Incredibly accurate, except where it is not, AND

since the futures markets are simply aggregates of individuals relying on polling results anyway, there isn't anything going on except fine tuning of surveys of polls.