Thursday, January 31, 2008

Predicting Presidental Elections

Author John Cassidy discusses some of the previous presidential election prediction model developed by other economists in his paper located at: http://www.portfolio.con/views/columns/economics/2008/01/14/Predicting-the-Presidential-Election#

Mr. Cassidy reviews three of the leading models. All of the models based their predictions not on what the presidential candidates were citing as their platforms but upon leading economic indicators such as US gross domestic product over a given time frame, the US inflation rate, the average per capita income growth rate of disposable income, and such non-economic statistics as the number of American combat fatalities in any current conflict or the current president’s approval rating.

Mr. Cassidy’s description of the models leaves the reader to conclude that each of the models that he is reviewing is based upon just two criteria. As examples he states that Yale University Professors Ray Fair’s model used two key variables (growth rate of US gross domestic product in the same three quarters leading up to the election and the average US inflation rate during the previous 15 quarters) to determine the election’s winner.

The winner in all three models is based upon primarily how well the economy faired during the current president’s term in office. If the economy performance was poor, then a change in party was predicted.

1 comment:

Vicki said...

Great! We have models! You have described them, now critique them. How were they developed? How were they tested? How "correct" were they? Do they have "face validity"? How much do the three models agree? Do we need all three models?

If, as these models suggest, the election is a referendum on the current President, it would suggest that we are due for a Democrat in the White House. Would the specifics of the three models predict that? How do these models compare to current polling data?

What is your personal experience and how does it relate to the models?