Predicting The Presidential Election

PREV 2 of 4 NEXT

Ray Fair, a Yale University professor of economics, developed the oldest and most (in)famous voting equation, which he has used since the 1980 election. His model pays no heed to polling data or the candidates' identity, which, early in the campaign season, when polls are notoriously unreliable and the primaries are still taking place, is a distinct advantage. Fair's two key variables are the growth rate of U.S. gross domestic product in the three quarters leading up to Election Day and the average U.S. inflation rate during the previous 15 quarters. The idea is that voters reward parties that deliver prosperity and punish ones that mishandle the economy.

Fair's system also takes into consideration the historical success of incumbent presidents and the tendency of voters to tire, eventually, of incumbent parties. The model has correctly predicted five of the past seven elections: 1980, 1984, 1988, the disputed 2000 election, and 2004.

In November 2006, Fair said the 2008 race looked "very close" and avoided picking a winner. Obviously, the Republicans face a tough challenge. Their president is retiring, and they started an unpopular war. But until recently, the economy was strong, and inflation had stayed fairly low—both of which favor the incumbent party. Since October, however, the credit and housing meltdown has soured economic prospects for 2008. To see how recent developments have affected the presidential race, I took the latest forecasts from the Federal Reserve Board, which predict that G.D.P. growth could be as low as 1.8 percent this year, and plugged them into Fair's equation. The result: a clear victory for the Democrats, with 51.4 percent of the two-party vote. (While this is good news for the Democratic field, it isn't definitive, as Fair's margin for error is about 2.5 percent.)

Another vote-predicting equation that emphasizes economic conditions is the Bread and Peace model, which Douglas Hibbs, a well-known political scientist, created in 2000. Hibbs uses a different measure of economic performance: the average per capita growth rate of personal disposable income over the previous four-year term. He argues that this is a better measure of voters' well-being than G.D.P. growth and inflation, because it takes into account all sorts of income, including payments from the government, and is adjusted for taxes.

Hibbs claims that his system, when fed the appropriate data, accurately picks the winner of 12 of the past 14 elections—1996 and 2000 are the exceptions. In 2004, following two years of strong income growth, the Bread and Peace model predicted a victory for George W. Bush.

Hibbs hasn't yet published a forecast for 2008, so I used his approach to create one. The latest version of the Bread and Peace model, which he updated last year, assumes that for each percentage-point rise in per capita disposable income that is maintained over the previous electoral terms, the incumbent party's vote will be boosted by 3.6 percentage points above a benchmark of around 46 percent. Over the first three years of Bush's second term, per capita income grew at an annual rate of around 1.8 percent. If this growth rate were to persist for another nine months, then on economic grounds alone, the Bread and Peace model would put the Republican candidate's share of the two-party vote at about 52.7 percent, enough for a comfortable victory.

PREV 2 of 4 NEXT