Saturday, August 20, 2016

The United States Electoral College and third party candidates

As most of my readers will know, elections to the Presidency of the United States, as well as approximately a third of the Senate, the entire House of Representatives, and various other state and local offices will be up for election on November 8 (by tradition, the first Tuesday in November).

While primaries for some lower-level offices are yet to take place, Presidential candidates for both major parties have been chosen through a series of primaries and caucuses. The Democrats have chosen Hillary Clinton, the early favourite, though she faced a stiff challenge from left-wing Bernie Sanders. On the other hand, a crowded Republican field narrowed to one candidate; controversial businessman Donald Trump, who had little support from the party leadership, won the nomination.

Trump's views represent a substantial diversion from traditional Republican thinking on a variety of issues. The most notable one of these is free trade; Mr Trump is opposed to it, but Republicans have traditionally supported free trade. Trump's tax plan is also a bone of controversy; while it would reduce tax rates, normally something the Republicans would support, it also does not make cuts to government programs like Medicare (health insurance for the elderly) and Social Security (the government retirement program), thus meaning that the deficit would increase.

These two issues have led a number of Republicans to look elsewhere in the presidential field. While some will vote for Ms Clinton, one other candidate gaining traction is Gary Johnson, of the Libertarian Party. The Libertarians have normally been a very minor political presence in the United States; occasionally winning a seat in state legislatures. The party's best performances have been in Senate and House elections when support for either the Republicans or Democrats in a state or district is so overwhelming that only one of the major parties puts up a candidate; when a Libertarian runs in one of these seats, they normally win a substantial share of the vote.

This year, Mr Johnson is currently polling at around 6-8% in national polls. Mr Johnson receiving support from dissatisfied Republicans is no doubt helped by his credentials as a Republican; he served as the Republican governor of New Mexico, and his running mate William Weld served as the governor of Massachusetts, also as a Republican. While this does not look like much, it would represent the strongest performance ever, by a long way, for a Libertarian in a presidential election.

Some Democrats are also not wholly happy with Hillary Clinton. Some supporters of Sanders, a self-declared democratic socialist, view Ms Clinton as too right-wing, and are also looking elsewhere. This has caused a much more muted increase in support for Green Party candidate Jill Stein. Not all polls include Ms Stein, but those that do mostly show her receiving 2-4% support; a low figure, but a substantial increase over the 0.36% she received in 2012 (though she was not on the ballot in all states at this election).

Minor candidates and Presidential elections

In countries where a President is elected, there are two main ways this is normally done; the plurality system, or the two-round system.

Under the plurality system, voters vote for one candidate, and the candidate with the most candidates is elected. Such a system is used in Taiwan, South Korea, and Venezuela, amongst others. In a close nationwide race, this system has a tendency to lead to two main contenders, as supporters of any weaker third candidate would have strong tactical incentives to vote for the member of the top two they most prefer.

However, when voters do not do this, the system can lead to candidates from one political faction being elected despite a majority of the electorate not supporting that faction. For example, at the Taiwanese presidential election in 2000, the pro-Taiwanese independence Democratic Progressive Party nominated Chen Shui-bian, who won 39.3% of the vote. However, two candidates from the opposing bloc (generally in favour of Chinese reunification) ran; the Nationalist candidate Lien Chan won 23.1%, and independent James Soong won 36.8%. For this reason, despite an apparent majority of voters supporting a pro-unification President, the pro-independence candidate won.

Most countries that have an elected President use a runoff system. Under such a system, voters vote for one candidate in the first round. If one candidate receives a majority, that candidate is elected; if not, a second round takes place, normally a couple of weeks after the first, when voters choose between the top two (in most cases) candidates.

This means that there is no incentive for tactically abandoning your most preferred candidate for the lesser of two evils in the first round, as unless a candidate in the first round wins 50%+1 (in which case choosing the lesser of two evils would not have mattered) you will be able to make that choice in the second round.

However, such a system can have its disadvantages. For example, at the 2002 French presidential election, the centre-right Union for a Popular Movement candidate, incumbent President Jacques Chirac won 20% of the vote. It was expected that he would face Socialist candidate Lionel Jospin in the second round, but a large number of left-leaning voters voted for other leftists in the first round; for example, the Workers' Struggle candidate won 5.7%, the Green won 5.2%, and the Communist won 3.4%. This meant that Jospin received only 16.2%, behind far-right candidate Jean Le Pen, who won 16.9%.

In the runoff, leftists voted for Chirac over Le Pen, resulting in Chirac winning 82% of the vote. These results demonstrate that there are still incentives for tactical voting under a two-round system; a voter may tactically vote to make the opponent of their most preferred candidate in the runoff an unpalatable radical.

The United States has a somewhat different, and currently unique, manner of electing the President. Each state is allocated a number of 'electors' equivalent to the number of members that state has in the House of Representatives and Senate combined. Each state has two Senators, regardless of size, but the number of members each state has in the House is proportional to its population. This method of allocation over-represents smaller States; for example, the state of Montana has 341,333 people per elector, but the state of California has 705,454 people per elector.

This means that on Election Day, voters are not voting for a presidential candidate, but rather electors who pledge to vote for that presidential candidate. There have been cases in the past where these electors have not voted for the candidate they were listed on the ballot for, though this has never changed the outcome of a presidential election.

Potential incentives of the Electoral College

So, what incentives would the Electoral College have, in terms of small party candidates? Well, over time, certain parties have become wholly dominant in states. For example, the Democratic Party has won the state of Massachusetts, in nearly all cases with a double-digit margin of victory, in all elections since 1984. The same goes for the Republican Party in Texas. This is why presidential elections are focused on swing states; those states where the race between the top two candidates is especially close, like Pennsylvania and Florida. Winning extra votes in a safe state is irrelevant, as the number of electoral votes (which are what matter) do not change. 

If you live in a safe state, even if there is a close vote nationwide, you have no incentive to tactically vote to support the 'lesser of two evils', given that the winner of your state is virtually determined in advance. In theory, then, people in safe states should be more willing to vote for third party candidates, given that there is little danger of them 'wasting their vote' and electing someone they dislike.

But does this work in practice? Let's look at two third-party candidates who were both totally unlikely to win; John Anderson in 1980 and Ross Perot in 1996. Mr Anderson was a moderate Republican congressman opposed to the right-wing views of party nominee Ronald Reagan, while Mr Perot was a populist businessman running on a platform of deficit reduction. Perot had run in 1992, but this campaign may not have had the same tactical messaging for the voters, since Perot was in first place in some early polls.

The below charts show a comparison between the vote shares for these candidates by state, compared to the margin between the top two candidates, which represents how close a state was.

There is something of a correlation between the two factors, with the correlation being stronger with Perot than with Anderson (the precise correlation factor being 0.13 for Anderson and 0.30 for Perot). However, there are many other factors which vary these numbers, and it is still quite a weak correlation, which makes it somewhat difficult to draw any conclusions.

It is interesting, though, to consider the states that the third party candidates will do best in. Mr Johnson has public profile in New Mexico, which, though it has been quite close in the past, was comfortably in the Democrat column in 2012 (a margin of ten points to Obama). The state's high Hispanic population is also unlikely to approve of Mr Trump's rhetoric on immigration, which would mean that both campaigns would be sensible to ignore the state.

Mr Johnson's running mate, Mr Weld, is the former governor of Massachusetts, a very safe Democratic state (despite the Republican candidate for President coming from the state, it went Democratic by 23 points). While these two cannot choose their home states, for the Libertarians their states make them sensible nominees, as their home state boosts will not be cancelled out by tactical voting.

The lower statewide profile of the Green ticket (Ms Stein has run for Governor of her home state of Massachusetts, but with little success, and her highest office remains being in the Lexington town meeting, while her running mate Ajamu Baraka does not appear to have run for any office) makes this factor less relevant for them. Nonetheless, it will help Ms Stein somewhat that the state where she has the most profile (relatively speaking) is a safe state, where she will have somewhat clearer air to campaign.