Primary Games


By Josep Colomer.



It’s supposed that the U.S. primary elections are a mechanism to simplify the political supply to only two candidates, which will produce a single majority winner in the real election. The primaries are a substitute for the formation of a single majority winner in parliament, usually by forming a multiparty coalition, as is usual in Europe and elsewhere. In Europe, the mess is after the election, while in the U.S. it’s starting more than a year before voting day.

In comparison, the primaries mechanism can be distorted by the failure to achieve two objectives: coordination and convergence.

Coordination means that a varied array of primary candidates should be reduced to only one from each party, two in total. Only in a two-candidate competition plurality rule can guarantee that the winner in the real election will receive a majority support. In fact, multiple candidates run within a political and ideological range similar to multiparty systems in European countries: from social-democrats to liberals within the Democratic party and from moderates to populists within the Republican one.

The point is whether, at the end, only two will run in the election. In the first Republican debate, Donald Trump, who had identified himself as a Democrat in the past, didn’t pledge to abstain from running as a third candidate in the election if he didn’t win the Republican primary.

This looked like a threat in the intention to attract more voters in the primary who would fear the breakup of the Republican side if he didn’t win. But the maneuver soon backfired and a few weeks later Trump held a theatrical press conference to publicly sign a loyalty pledge to the Republican party.

What’s more intriguing is why nobody has yet asked Bernie Sanders, who was a mayor for a ‘Progressive party’ against Democratic rivals and ran and stayed as ‘independent’ in the House of Representatives for 15 years, whether he can also pledge not to run as a third candidate next year if he doesn’t win the Democratic primary.

In fact, third candidates are a very common feature in U.S. presidential elections. As a consequence of multi-candidate races, more than one-third of the presidents have won the election on the basis of a minority support in popular votes. Third candidates have also spoiled many elections by not winning but indirectly favoring the likely loser between the other two.

Egregious cases include: Southern Democrat John C. Breckinridge favoring the victory of Republican Abraham Lincoln in 1860, Progressive Theodore Roosevelt favoring Democrat Woodrow Wilson in 1912, Southern Democrat George Wallace favoring Republican Richard Nixon in 1968, Independence Ross Perot favoring Democrat Bill Clinton in 1992 and 1996, and Green Ralph Nader favoring Republican George W. Bush in 2000.

It can always happen again.

Convergence means that if coordination achieves to select only two candidates, the best strategy is approaching the median voter’s preference, which is usually a moderate one. Yet on a number of occasions highly divided party primaries have produced extreme candidates who have experienced huge defeats in the real election.

Memorable cases include Republican Barry Goldwater (who was derided by the Democrats because “he was right, yes, extreme right”) who lost to Lyndon Johnson in 1964, and Democrat George McGovern (who was supported by the hippy students) who lost to Richard Nixon in 1972, both by about 38% to 61% of popular votes.

Something of the sort could happen again if the two candidates were, say, Donald Trump against Hillary Clinton, or Bernie Sanders against… hmmm, is still Jeb Bush around?

But if even if there is coordination and convergence, the Downsian model of competing for the median actually predicts a tie. In fact, there have been numerous presidential winners with less than half a percentage point advantage over the runner-up.

They include Rutherford Hayes (who actually obtained fewer votes but one more elector in the College) against Samuel Tilden in 1876, John Kennedy against Richard Nixon in 1960, and George W. Bush against Al Gore in 2000. In those cases the actual tie is broken by the way a few votes are counted –as happened in three Southern states, Chicago, and Florida, respectively in the cases mentioned. Anything may happen on occasions like these.

The coming presidential election is going to be interesting because the result is always contingent on surprises, be it by lack of coordination, non-convergence or peculiar tie-breaks. But these types of surprises can happen because the institutional mechanisms enforced: primary elections and plurality rule, are highly imperfect, to say the least.

What Next?

Recent Articles