Predicting the Outcome of the Next Presidential Election – A Statistical Approach

by robekulick

Up until now on my blog, whenever I’ve discussed predictions about political events, it’s always been in the context of prediction markets like Intrade. However, an economist at Yale named Ray Fair maintains an econometric model of presidential elections that has been very accurate in the past. Recently, Fair updated his model, which is to say he provided a new prediction of the next Presidential election.

What Fair’s model actually predicts is the share of votes the President will get. As of his most recent update to the model, he predicts a vote share for President Obama of 50%, ie, the next election will be neck and neck.

http://fairmodel.econ.yale.edu/vote2012/index2.htm

Fair’s model uses macroeconomic data to make its predictions about elections. Thus you can kind of think of it as James Carville’s famous quip about the 1992 election “It’s the economy, stupid” transformed into a statistical (econometric) model. This is important because Fair’s model is based on predictions about what will happen in the economy. So because there is substantial uncertainty regarding the economy at this point, the fact that the model is currently at 50% suggests that the outcome of the next election may very well be determined by the path the economy takes over the next year. As Fair explains:

“For a moderately growing economy, which the US model is now forecasting, the election is prediced to be close. If the economy does considerably better, which the US model was forecasting earlier, Obama is predicted to win, although the election is still fairly close. If the economy goes into another recession, Obama is predicted to lose.”

Note that when Fair started making predictions for the upcoming election on Nov. 11, 2010, President Obama was predicted to have a vote share of 55%. Since then as expectations about the economy have declined, President Obama’s predicted vote-share has suffered.

So what’s the take away from this? Well to bring Intrade back into this for a second, President Obama was at 49.8% on Intrade as of my writing of this post:

http://www.intrade.com/v4/markets/contract/?contractId=743474

So essentially the vote share equation and the prediction markets both are predicting a very close election. Given that both of these approaches to predicting the outcome of the next election have a very strong pedigree, I think its safe to say that we’re in for a very interesting election and that either party could easily come out on top.

Advertisements