Is it the Economy, Stupid?

The advent of the internet sportsbook means that for the first time many Americans can participate in betting on the outcome of their own presidential election, a privilege previously only granted to Europeans.

Elections are highly predictable and a potential source of profit for the intelligent gambler, despite the fact you can make this bet only once every four years. Nonetheless there is very little available in gambling literature that will assist you in making a correct prediction.

One book which does make a serious study of presidential elections is Jacques Black's Spread Betting to Win. Black describes a model for predicting elections from a letter to the Economist magazine by a man named Paul Hiniker in 1988.

Hiniker starts by giving 49% of the base vote to each party. Incumbency then adds 4.5%. Each percentage point in the growth of real income adds 1.4% for the incumbent, while each point rise in inflation subtracts 0.6% from the incumbent's vote.

Despite its apparent simplicity, the Hiniker model performs spectacularly well in an analysis of elections in the modern area, not only predicting the actual result of the elction but also the margin of victory to a very high degree of approximation.

Crucially this precision extends not only to past elections but also to the 1988 election (which was some months after Hiniker's letter) and the 1996 election, predicting the victory margin to within 2%.

However, in the 1992 election the model performed very badly, forecasting a huge victory for Bush when it was almost a foregone conclusion that Clinton would win his first term in office.

This illustrates the limitations of any handicapping model.

Hiniker's model is what is known in academic circles as "Vulgar Marxist", a belief that events occur exclusively because of economic factors. Bill Clinton appeared to subscribe to this belief with his famous "The economy, stupid" poster kept above his campaign desk to remind him of the priority in winning the election. Ironic, then, that Clinton should be the candidate to buck the trend and mess up the prediction.

Of course, politics is more complicated than a pure economic analysis would suggest. Nonetheless 100% prediction is not really possible with any form of handicapping. There have been various attempts to create more sophisticated electoral models incorporating other sophisticated factors, such as third party splits and party division. But none of these significantly outperform the Hiniker model when measured against the data of past elections. The problem is that some factors in an election shift constantly. It's unlikely, for example, that the base vote for the two parties is really equal. Moreover, it shifts slowly from decade to decade as demographics change.

It's interesting to see whether this simple system will hold up in the next election, where it predicts a significant victory for Al Gore in a race which currently still hangs in the balance.

My $500 is riding not only on the outcome of the election, but also on the nature of human history. Because if Gore does win by a landslide, the Hiniker model would be vindicated. I'll be a little richer if that happens, but a little sadder too, I guess. Perhaps human nature is controlled by the economy in the same way the moon controls the tide. Or perhaps the spirit of rugged individualism is stronger than we think.