The conventional wisdom among the chattering class about the Republican field is that voters face a choice between “electability” and “ideology.” But a careful look at elections since the end of World War II suggests that is not the case. What most pundits think of as “electable,” a safe candidate attractive to moderate voters, has historically been highly unlikely to unseat an incumbent president. In the five elections since World War II in which the party out of power has picked a “safe” candidate to take on a sitting president, the result was defeat for the supposedly safe, electable challenger.
To understand why this is not always the case, consider the math of an election as composed of two parts: the “normal” or “expected” result and the variability or chance part around that normal result. The term used in markets to denote this chance or variable part is “beta.” A high-beta stock—or a high-beta candidate—might do a lot better or a lot worse than what would normally be expected, whereas a low-beta candidate (or stock) is pretty much going to do as expected, or, in the case of a stock, as the rest of the market does.
If there is an election a party would normally be expected to win, the smart thing to do is to nominate a low-beta candidate. If, on the other hand, a party is thought likely to lose, it might select a high-beta candidate. That would increase its chances of winning, though it would also increase its chances of losing big.
Statistically, the odds of winning depend a lot on whether there is an incumbent seeking reelection, as has happened in 10 of the 16 presidential races since World War II. Incumbency offers tremendous advantages, by some estimates adding at least 3 points to the “normal” result, turning a 50-50 election into one that is 53-47. Job performance matters, so a bad economy tends to subtract from this edge. With subpar growth, but no recession, President Obama might normally be expected to win narrowly, say by 1 percentage point.
If that model is right and the Republicans ran a “zero-beta” candidate, one with absolutely no variability around the “normal” result, he would perform quite respectably but lose. To be precise, he would lose by one point, 50.5 to 49.5. A very low-beta candidate, say one with a variability equivalent to half a point, would have an equal chance of producing a “tie” or losing 51-49. A high-beta candidate, with 10 times the variability of that very low-beta candidate, would have an equal chance of winning 54.5-45.5 or losing big, by 55-45. But if a party really wants to win—and doesn’t care about how badly it might lose—it should pick a high-beta candidate.
Stated simply, given the incumbent’s built-in advantage, the opposition party must nominate someone who will “shake things up” in order to win. After all, if an election was simply going to be a rerun of the previous election, the incumbent would win again. Indeed, running the same opposing candidate as the last time is the ultimate “low beta” strategy—the candidate has been vetted, is therefore “safe,” has good national name identification, and obviously has the support of the party machinery. But this recalls Einstein’s famous definition of insanity—doing the exact same thing and expecting a different result. Actually both parties have tried this—Dewey ran in 1944 and 1948 and garnered just as many votes both times; Stevenson ran in 1952 and 1956 and got clobbered by roughly the same margin.
There were three other postwar incumbent elections in which the opposition party tried the next safest thing, picking a party stalwart who had experience running for national office. Bob Dole, Senate Republican leader in 1996 and a runner-up in the 1988 presidential contest, was one such example. So was Walter Mondale in 1984, who had been the vice presidential candidate in the two previous contests. John Kerry was also “safe” in 2004, a war hero running in a nation at war. Kerry, with the least experience, probably had the highest beta of these candidates and, by the way, came the closest to unseating the incumbent. And unlike his lower-beta colleagues, Kerry actually increased his party’s popular vote over its previous performance. Still, low-beta choices have been zero-for-five at unseating incumbents.