Augustine Faucher, the director of macroeconomics at Moody's economy.com, has a column in today's Boston Globe reporting on the state-level political forecasting model his organization has developed. He expresses a high degree of confidence in the model, noting that over the last seven presidential elections (back to 1980) the model accurately predicted the winner of each state presidential election 86% of the time, and that its success rate in picking the ultimate Electoral College winner was 100%.
The model is based on economic conditions in each state at the time of the election, noting that these conditions "often vary significantly from coast to coast." It seeks to predict whether the incumbent party will win re-election based on two factors: the national inflation rate over the eighteen months months prior to the election, and the state unemployment rate over the previous two years.
More on the flip.
The model does not consider candidates, the effects of advertising, nor the impact of third parties. It does account for the power of incumbency, previous voting patterns, and what Faucher calls "incumbent fatigue" -- voters' presumed desire for change after the incumbent party holds the White House for two terms.
The 2008 prediction is based on Moody's forecast for economic conditions in November of this year. Faucher notes that in the third quarter:
Overall, rising inflation, recession, and a worsening labor market will make it very difficult for Republican John McCain to win the election, especially on top of the incumbent fatigue problem.
How difficult? This difficult:
We expect Obama to win 30 states plus the District of Columbia for 370 electoral votes, far more than the 270 required. Seventeen states are projected to switch from Republican in 2004 to Democratic in 2008. No states are expected to switch from Democratic to Republican.... There are five states where the forecast for the Republican share of the vote ranges from 45 percent to 50 percent; these can be viewed as "swing" states for the Republicans. Even if McCain were to pick up all five, he would still fall 57 electoral votes short of victory.
I don't offer this as definitive evidence that we will win in November. Even forecasting models with 100% past accuracy are known to fail. An awful lot can happen between now and November.
Still, there is a lot of evidence that economic conditions are strong predictors of presidential elections (e.g. here). Add this data to the spectacular work being done by poblano on the polling front, and it's easy to be confident about our chances.