Four More Years

Richard DeKaser The Iowa caucuses are six weeks away, campaign contributions are mounting, and policy debates loom. But if history is any guide, it may all be for naught. Regardless of who represents the Democrats in next year’s presidential election, the outcome looks like a victory for George W. Bush.

Such a prediction may seem audacious, but the track record of election forecasting models is hard to dispute. For example, in what amounts to a statistical version of the adage that “people vote their pocketbook,” Professor Ray Fair of Yale University has developed a simple model that accurately predicts 18 of 22 presidential election outcomes. Combining it with the National City forecast of strong growth and moderate inflation in 2004 implies that a second Bush term is practically a lock.

To be sure, Professor Fair’s model isn’t based solely on economic inputs. Incumbents, for example, are expected to pick up 3.3% of the popular vote by simple virtue of already holding office, and another 3.9% can be expected if a declared war is under way. If the incumbent’s party has exercised more than a one-term hold on the office, however, voter fatigue increasingly subtracts points from its margin. Additionally, Republican candidates can generally count on a 2.7 percentage point edge over Democrats, all other things being equal. In sum, these non-economic considerations slightly tilt the odds in George Bush’s favor.

Economic factors are mixed but on balance help seal the deal. Generally, incumbents rack up points for solid gains in per capita real GDP. And while these were few and far between during the first two-thirds of George Bush’s term, voters are much more sensitive to the three quarters immediately preceding the election, which looks to be very solid indeed. Incumbents also win points for low inflation, which has been the case so far during this President’s term and will be likely to carry into next year. Taken as a whole, plugging in our economic forecasts yields a lopsided 58%-42% victory for George Bush--a comfortable margin comparable to the Reagan and Clinton reelections.

A Republican victory--albeit a narrower one--is also predicted by Iowa Electronic Markets, an online trading system where speculators take positions on election outcomes. Much like a horse race, the odds are determined by the collective investments of many speculators, and research has shown such markets to more reliably predict outcomes than public opinion polls.

At the time this publication s headed to press, a $1 contract on the strongest Democratic contender--Howard Dean--is trading at $0.303, meaning the market places a 30.3% probability on Governor Dean winning.

This overstates the Republicans’ chances, however, since that contract expires worthless if Howard Dean fails to become his party’s candidate (to which, incidentally, the Iowa Markets attach a 37.7% probability). Statistically speaking, the Dean presidential contract represents the conditional probability of the Vermont Governor winning both the Democratic nomination and the popular vote for president. Hence, to access the odds of any Democrat winning the election, one must add in the odds of all the other candidates (including an “other” contract). Doing this still points to a Republican victory, but the predicted 51.5%-to-49.5% split is a much closer contest.

Richard J. DeKaser is Senior Vice President and Chief Economist of National City Bank.

Print page