The Predictor

Yale economist Ray Fair has been predicting presidential-election outcomes for more than 30 years. When asked to simplify his model, Fair cites the mantra popularized in the 1992 race between Bill Clinton and George H.W. Bush: “It’s the economy, stupid.” The theory behind his model is that in a good economy people tend to vote more than they otherwise would for the incumbent party, and in a bad economy they do the opposite.

Fair is an econometrics enthusiast through and through, and he has used quantitative analysis to predict the quality of wine, the correlation between aging and athletic performance, and the traits of people likely to have extramarital affairs. He described his work in an interview with National Journal. Edited excerpts follow.

NJ You’ve had great success predicting presidential elections since 1980—except once. What happened in 1992?

FAIR 1992 was a big mistake for the equation. Voters were quite pessimistic about the economy, although the economic data that existed at the time wasn’t all that bad. So the equation was predicting that Bush was going to win, and, of course, he lost to Clinton by a lot.

NJ Doesn’t that sound a lot like 2012? Economic data getting stronger, people still pessimistic …

FAIR We have some time left. This is only March. So what you have to think about in 2012 is what voters are going to see in early November. If the economy for some reason takes off, then the mood of the voters about the economy could be quite different then than it is now. This is an interesting election for the equation, because it’s an OK, but not terrific, economy. According to my forecast, it’s a close election, almost 50-50. So it really does matter what the economy is going to do. And there still is a reasonable amount of uncertainty about the economy even this late.

NJ Treasury Secretary Timothy Geithner said that we might hit the debt limit before November. Could a repeat of last summer’s debt-limit fight factor into your predictions?

FAIR The standard error of this equation is about 2.5. To the extent things like the debt ceiling matter, they just go into the error term.

NJ It seems like everyone’s an election forecaster these days. How has the field changed over the last 30 years?

FAIR The main distinction between my work and others’ is a methodological one. I’m interested in causality. The theory I’m working off of is that the economy causes voters to vote the way they do. But most of the other forecasters are not causal. They combine economic data with voting surveys and polls, and come out with a prediction. But that’s not causal. If you ask a person two months before the election who they are going to vote for and use that information to predict what’s going to happen two months later, that information is not causal. It’s just a forecasting device.

NJ Should everyone be doing your kind of thing?

FAIR I’m not complaining about the fact that others use forecasting devices. In the end of the election [campaign], you would expect those kinds of equations on average to do better than mine because they have more information about what people are actually going to do. I made my first prediction about 2012 in November 2010. So the advantage of my thing is that two years ahead, if you have some sense of the economy, you can get some sense of what the voters are going to do. But if you want a prediction, probably the best you can do is [check] betting markets near the end of the election. They know what I’m saying, and all the polls, and that all gets reflected in the price of the candidates.

NJ Who will you root for in 2012?

FAIR This work is completely apolitical. So I don’t usually answer these questions. Science is science. You predict what the equation says, and you try not to let any of your political views hinder that. So there’s nothing partisan in this.

NJ So you’ll be neutral?

FAIR As an econometrician, I care about the size of the error term, not necessarily who wins or loses; it’s how close the actual vote share is to the predicted vote share. The metric I’m trying to judge is the size of error, not who wins. But that’s of course the scientific view and not the personal one.

NJ Do you have any election-night rituals?

FAIR I’m not really a political junkie, to be honest. I’m certainly not a pundit. But usually one of my friends has a party, so we go over there and watch the returns. So usually I do it at a party with a bunch of Yale economists.

NJ Do you think your forecast impacts how people vote? You’re a widely cited guy.

FAIR I think that’s stretching the point. I don’t think I’m that popular.

This article appears in the March 24, 2012, edition of National Journal.

Get the latest news and analysis delivered to your inbox. Sign up for National Journal's morning alert, Wake-Up Call, and afternoon newsletter, The Edge. Subscribe here.