It’s 4 weeks since Donald Trump won the US Electoral College despite losing the popular vote convincingly. This got me thinking about how other electoral systems could be applied to the Electoral College.
The US Electoral College employs what is known as First Past the Post in the UK (Winner Takes All in American English). That means that in 48 states (and DC) the candidate with the most votes wins all the state’s electoral votes. In two states (Nebraska and Maine) the candidate with the most votes statewide gets 2 electoral votes and candidates also get one electoral vote for each congressional district they win. Even in these exceptions it’s First Past the Post with the by-district method leaving itself open to gerrymandering.
In some parts of the world representation is decided proportionally. Take the assignment of European Parliament seats from the UK. The country is divided into 12 electoral districts electing between 3 and 10 MEPs. The MPs are alloctated by what is called the D’Hondt system (similar in results to the Jefferson system). Put simply, you take each party’s number of votes and divide by the number of seats that party already has plus 1. You then give the next seat to the party with the most votes and then repeat the process till you run out of seats.
The same process could be applied to electoral votes from each US state. Let’s look at the allocation of seats in Alaska (McMullin was not on the ballot),
So Trump wins Electoral Vote 1, Clinton vote 2, Trump vote 3.
This process can be done for all states yielding the following results,
|District of Columbia||3||0||0||0||0|
So we get an inconclusive Electoral College, because neither candidate got close enough to 50% to edge the result. The plots below show the number of electoral votes gained in each state (left) and the fraction of the EV in each state (right) for Clinton, Trump, Johnson, Stein and McMullin. Note only two places (Wyoming and DC) give all their votes to one candidate and that Hawai`i is 3-1 for Clinton and Alaska 2-1 for Trump. Also note that Stein and Johnson only get EVs in larger states. This is because there is a hidden threshold in each state due to a finite number of electoral votes. Bigger states have more votes to give so need a lower percentage to gain one seat.
We can also rerun for other recent elections. For 2012 Obama wins relatively comfortably 274 to 264. In 2000, the last time the popular vote winner didn’t win the Electoral College, it would have been Gore 268, Bush 267, Nader 3 so again tied.
So does this system favour one party or another. Well we can simulate a 50/50 election ignoring 3rd parties) using the R-D partisan bias each state returned in the 2016 election. It comes out 269/269 with Clinton adding an EV in California (from Stein/Johnson) and Trump gaining in California (from Stein/Johnson), Texas (from Johnson) and Utah (from McMullin).
This system opens every state up for campaigning meaning less focus on a handful of swing states. We can also look at how many more votes a candidate would need to flip one vote in each state. From this we can work out which state a campaign would most easily get one extra vote from. This gives the interesting result that the best state to campaign in under this system for Clinton would be Wyoming, needing only a 2167 votes to flip one EV. Trump’s best campaigning location would be Georgia where he’d need 23,604 votes to flip one EV.
So could this ever work? Well it’s not going to be done on a national level but is could be done by red and blue states pairing up so the net effect on the EV numbers would cancel out. This would be pretty good for these safe states as they would suddenly start to matter in presidential elections. However it wouldn’t be in the interests of sqing states to dilute their influence. Unlike the National Popular Vote Compact it would only need to start with two states rather than requiring 270 votes worth of states to sign up.