Elections

Yes, the party in power tends to lose votes over time. But how often?

Oct 4 '16
Rutherford B. Hayes
Rutherford B. Hayes

Evidence shows that incumbent parties lose vote share over time (as University of Texas at Austin professor Chris Wlezien writes in a recent article, “governing parties tend to lose vote share the longer they are in power” in the United States and elsewhere). Two-term parties tend to lose vote share when they seek a third term. (See this article by Emory University professor Alan Abramowitz from 2008.)

But this is only one of many factors, and, in 2016, factors such as the economy and presidential popularity suggest a close election, possibly favoring the Democratic nominee.

I was thinking about this the other day after reading this from Gail Collins in a New York Times op-ed:

Americans have always been pretty pragmatic about the presidents they pick. Mostly, they go for change or not-change.

Given all the complications out there, it’s a pretty pragmatic approach. If the same president has been in office for eight years, they pick his successor from the other party. Give or take a few vice presidents, it’s a rule that goes back to Rutherford B. Hayes.

This claim seemed surprising to me, so I looked up all of the elections that follow a president’s second term, starting with Rutherford B. Hayes, who was elected after the two terms of Ulysses S. Grant. There have been 11 such elections, six of which have been won by the candidate from the opposite party of the sitting president.

It turns out there’s some flexibility in this calculation, as one has to decide how to include cases like that of Lyndon B. Johnson, who served two terms but not the full eight years. If you exclude those presidents, we then have seven elections, four of which have been won by the candidate from the party opposite the incumbent president.

Or we can follow Collins’s lead and exclude cases where sitting vice presidents were running for reelection, in which case the out-party rule again works in four out of seven cases.

No matter how you slice it, the rule doesn’t work. Or, to be more precise, it works about half the time.

This is important. We can only learn from history if we know what the history was. It can perhaps be frustrating that political scientists speak in terms of percentages and tendencies rather than deterministic rules. But ultimately, the hedging we do is necessary, giving us a better understanding of the many different factors that affect voting decisions.