Jeff Horwich: Polling organizations like Gallup serve us up a new poll on the presidential race almost every week. But some economists and others set the polls aside and head for the data. Their goal is to isolate which critical factors determine who wins presidential elections. And then they use these forecasting models to project a winner. This year, two of the best-known models -- both with very strong track-records -- come to different conclusions. We're going to do one today, and the other tomorrow.
Today I've got Douglas Hibbs, father of the so-called "Bread and Peace" model. He's recently retired as chair of the economics department at Sweden's University of Gothenburg. Good to talk with you.
Douglas Hibbs: Happy to be here.
Horwich: You have one of the simplest models, but it's been very successful. What do you factor in?
Hibbs: Well, my model asserts that the only systematic factors persistently affecting presidential election results are the weighted average growth of real disposable personal income over the term of office, and the number of fatalities of American forces involved in unprovoked, foreign conflicts.
Horwich: Just those two factors? What about everything else?
Hibbs: Well, everything else potentially can matter, but not in the systematic or persistent fashion. The other stuff tends to be particular to individual elections.
Horwich: So, Romney versus Obama, 2012. What does the Bread and Peace model say?
Hibbs: Well the point estimate is 47.5 percentage points for the incumbent Obama of the major party vote. So in the aggregate of idiosyncratic factors, say, favored Obama by two standard errors, he could comfortably win. He'd be up above 52 percent. The likelihood of that, I think, is small, about 1 in 10 perhaps, a little less than 1 in 10, something like that.
Horwich: So setting aside the electoral college, which can do weird things to anybody's model, your model suggests that Obama is likely to lose?
Hibbs: Correct. With quite high probability.
Horwich: You believe pretty strongly that your model's going to get it right this time?
Hibbs: No. If I get it exactly right, it'll be a stroke of great good luck.
Horwich: Of course, if you get it exactly right. But let's talk about the outcome that matters: winners and losers.
Hibbs: Well as I said, my model implies that Obama will lose with fairly high probability. But it's probabilistic. One in 10 or so chance of him winning is, you know, it's not a negligible chance, after all, right? It's not negligible at all.
Horwich: Since disposable income growth is so important to your model, how strong would that growth have to be, say at the time of the election, for the president to prevail?
Hibbs: It's late in the day and of course the economy has performed very poorly under Obama, and rightly or wrongly, he owns the economy. In order for him to eke out a victory, growth would really have to ratchet up to 6 percent or higher.
Horwich: Economist Douglas Hibbs, inventor and maintainer of the Bread and Peace election forecasting model. Good to talk with you.
Hibbs: Bye bye.