Marcus Evans Group | Worldwide Headquarters | American Offices | Latin America | European Offices | African / Asian Offices

Obama’s re-election prospects: the hard numbers | Harry J Enten

Economic metrics can be predictive of election results: is the US economy improving enough to keep Obama in the White House?

Unemployment dropped to 8.5% in the month of December, its lowest in three years. Immediately, Republicans and Democrats took to the bullhorn arguing about the apparent jobs recovery and what it might mean for the 2012 presidential election.

The answer for nonpartisan analysts who study elections, such as myself, is that unemployment does not mean a whole heck of a lot. Research has shown that the link between the unemployment rate at the time of a presidential election has literally zero predictive value for whether the incumbent party earns another term in the White House. 

Unemployment forecasts re-election rates slightly better when we concentrate on the change in unemployment rate over the term, and specifically at those presidents who seek re-election (as opposed to the including all elections). But even then, it only explains about 25% of the differences in election results. 

This is not to say that the economy has no bearing on elections. It’s that unemployment is often a lagging variable of economic strength and does not capture the economic situation as well as other variables. 

Perhaps the lowering unemployment rate will precede improvements in other sectors of the economy, but research tells us that while there is definitely a relationship between unemployment and overall economic health, it’s not as concrete as you might think.

The economic indicator that most election analysts rely upon is real disposable personal income (RDPI) growth per capita, which measures the amount of the average person makes after taxes and inflation. The most rigorous election model that takes RDPI into account is Douglas Hibbs’s Bread and Peace model, which explains 86% of the differences in election results since 1952.

The model is quite elegant with only two variables: military fatalities in unprovoked military conflicts and weighted real disposable income per capita over each of the 16 quarters of the presidential term, with the later quarters weighted more heavily than earlier ones.

Currently military conflicts are minimal, which is good for Obama, but RDPI over the first 11 quarters of Obama’s presidency has been negative, at -0.4%. This growth rivals that during Jimmy Carter’s term. If growth continues at this pace, Hibbs’s model would project Obama to garner a little greater than 44% of the vote. [Note that all percentages are of the two-party variety: that is, calculated as incumbent party's % / (incumbent party's % + opposition party's %)].

But what if the decline in unemployment is a precursor to a rise in RDPI? Indeed, Wells Fargo experts predict increasing growth of RDPI over the next year.

The problem for Obama is that he already is in such a deep hole that the expected growth (when population growth is taken into affect) is only predicted to be about about 1%, for an overall weighted growth of +0.3% over the presidential term. And this +0.3% growth would forecast Obama garnering 46.8%, which is obviously not enough to win re-election.

So how bad is it for Obama that arguably the best forecasting model has him losing, with only 46.8% of the vote? It’s worrisome, but certainly not the end of the world. My research shows that even when controlling for DPI growth, presidents (since 1924) whose party is in its first term in the White House usually get a boost of about 2%. My research also indicates that in the past 15 elections, presidents benefit from at least a portion of Congress being controlled by the other party.

Taking into account these two factors not only improves our model’s explanatory power to ~93% of the differences in election results since 1952 (and 80% since 1924), but it also improves Obama’s predicted vote percentage to slightly less than 49%, given the expected improvement of the economy.

But there’s even better news for the president. The standard error, a measure of doubt in estimation, for each model is 1.5-2.2%, which is definitely large enough to flip the result. Further, the ability to predict a future election based on prior elections is not as clean-cut as some would like believe.

Campaigns can – if not as much as some campaign strategists would like to believe – make a difference. Indeed, Al Gore should have garnered 54-55%, instead of his actual 50.2% of the vote in 2000, according to both Hibbs’s original and my amended Hibbs’s model – meaning that the advantage he had going into the campaign, he lost on the trail.

Unfortunately for Obama, RDPI models get a lot more right than they do wrong. The president goes into this re-election campaign as the underdog. © 2012 Guardian News and Media Limited or its affiliated companies. All rights reserved. | Use of this content is subject to our Terms & Conditions | More Feeds