High unemployment gives Romney an edge – but it's complicated

America's unemployment rate rose slightly to 7.9 percent in October, which does not bode well for President Obama's reelection prospects.

The last time unemployment was above 7 percent in October of a presidential election year, incumbent George H.W. Bush  was on his way to losing the White House to upstart challenger Bill Clinton in 1992. So the sad shape of today's economy favors challenger Mitt Romney in Tuesday's election.

But unemployment is not always a decisive factor in elections – indeed, perhaps not a decisive one at all.  October unemployment was even higher in 1984 (7.4 vs. 7.3 percent) and that year President Reagan thumped challenger Walter Mondale with 58.8 percent of the vote. So Mr. Obama's supporters can take heart that elections turn on something other than the unemployment rate.

Of course, these are extraordinary times. October's 7.9 percent unemployment is the worst of any election-year October since 1940, the tail end of the Depression. Back then, it was nearly double today's rate, yet incumbent Franklin Roosevelt handily beat dark horse challenger Wendell Willkie with 54.7 percent of the vote.

So if the unemployment rate on the eve of a presidential election isn't especially decisive, what is? The change in the unemployment rate during a presidency seems to be slightly more predictive, as Nate Silver of The New York Times pointed out last year.

That might explain, for example, President Roosevelt's resounding defeat of Alf Landon in 1936. Even though the unemployment rate was 16.9 percent that year (monthly rates weren't calculated in that era), it had been above 20 percent for the previous three years. By 1940, when Roosevelt beat Wilkie, it had fallen again to 14.6 percent.

By that measure, Obama's reelection prospects don't look good, either. Unemployment is now higher than when he beat John McCain in 2008 (6.5 percent in October) or took office in January 2009 (7.8 percent).

"Economic performance usually is the dominant factor" in elections, writes economist and former University of Gothenburg professor Douglas Hibbs in an article from July, but other factors also play a role. Mr. Hibb's two-factor framework, which he calls the "Bread and Peace" model, ignores unemployment and uses instead the change in per capita real disposable income during a president's term. He couples that with wartime fatalities in an unprovoked war.

Using his model, Obama's chances don't look good, either.  In an Oct. 26 update, he projects the president will get only 46.6 percent of the two-party vote. That would put him on par with George Bush Sr., who in 1992 got 46.5 percent of the two-party vote (excluding the results from third-party candidate Ross Perot) and lost to Mr. Clinton when unemployment was high.

If economic performance is predictive, Mr. Romney has the advantage – but the link between economic performance and elections is complicated.