The end of the Trump myth?

Four years ago, after Donald Trump had won the presidency, I wrote the following:

the one thing everyone seems to agree on about Donald Trump’s victory is that it’s a revolutionary development, a complete break with precedent.

And in political terms that’s probably true. But electorally, it’s the one thing that’s demonstrably false. What’s most striking about last week’s result is its normality.

And I went on to show how the 2016 election showed very little change and conformed very closely to recent patterns.

I failed, however, to dent the myth. A torrent of opinion continued to assert, or (more often) to simply assume as uncontroversial background, that Trump was a powerful electoral force in his own right, an agent of change who had rearranged voting behavior and, for good or ill, redrawn the political map of the United States.

It isn’t true, and last week’s election confirmed its falsity. Not only did Trump win office in a thoroughly conventional manner, he lost it the same way.

Four states are still not fully decided, but I’ll assume that the existing leads – Trump in Alaska and North Carolina, and president-elect Joe Biden in Arizona and Georgia – won’t change. On that basis Biden has 306 votes in the electoral college as against Trump’s 232, exactly the same margin that Trump won by in 2016.

The popular vote is less clear because there is still quite a lot of counting to go. At present Biden leads by about 4.4 million votes, giving him about 51.5% of the two-party vote. That number will come up, since the biggest numbers of uncounted votes are in strongly-Democrat California, Illinois and New York, and even in Republican states the postal votes have been trending more Democrat.

Assuming he finishes on, say, 52.5%, that represents a swing of 1.4% from the 51.1% that Hillary Clinton won in 2016. Looking at the pre-election pendulum, a uniform swing of that magnitude would have delivered Biden an extra 76 electoral votes. In fact he gained 74 – almost spot on.

The only state that he missed out on below the 1.4% mark was Florida (0.6%), worth 29 electoral votes. Instead, he picked up two states just above it: Arizona (1.9%) and Georgia (2.7%), with 11 and 16 electoral votes respectively. As Malcolm Mackerras always says, swings are never uniform, but deviations from uniformity roughly cancel out.

Historically, 52.5% is a narrow win. But it’s in keeping with recent patterns: no presidential election has split more than 55-45 since Ronald Reagan’s landslide in 1984. The biggest margin in that time was Bill Clinton’s 54.7% in 1996; Biden’s is right in the middle of the range.

The United States is deeply polarised and very set in its behavior; only small numbers of voters are susceptible to changing their minds. The swing in 2016 was the smallest since the nineteenth century, and this year’s won’t be much bigger. Landslides on the scale of 1964, 1972 or 1984 seem unimaginable today.

While six million or so votes is now a pretty healthy margin, the electoral college is another matter. If Trump had managed to hold on to Arizona, Georgia and Wisconsin he would have tied the college at 269-all. All three of those states are close, so that would have only required a reduction in Biden’s swing of about 0.4%.

In other words, Biden could easily have won more than 52% of the two-party vote and still lost the election. (A tie in the electoral college would have thrown the election to the House of Representatives voting by state, in which the Republicans have a majority.) How is that possible?

At this point another myth comes in: that the electoral college has a structural bias against the Democrats because of its over-representation of small states. The premises are true; small states are over-represented in the college, because they get votes according to their number of senators (two per state) as well as representatives, and the college is biased towards the Republicans, but the two things are mostly unrelated.

You can see this yourself just by looking at the electoral college numbers. Biden has won 25 states and the District of Columbia for his 306 electoral votes; Trump won the other 25. If you take out the two votes per state for the senators (including two notional senators for D.C.), his margin comes to 254-182. Instead of 56.9% of the electoral college he would have won 58.3%: the difference between those two figures is the extent of the small-state bias. In a very close election it might matter, but it’s clearly not the main problem.

The Democrats’ difficulty can instead be explained by looking at just the two largest states, California and Texas, which have 93 electoral votes between them (17.3% of the total). In 2000, the Democrat vote in California was six points ahead of their national vote; by 2016, it was 15 points ahead (this year’s figure is not yet final but will probably be similar). Similarly in Texas: in 2000 the Democrats were 11.2 points behind their national vote; by 2016 that was only 5.8 points, and this year will probably be less.

Those swings represent big gains for Democrats in the national vote (offset, of course, by falls elsewhere). But their electoral college impact is zero: they already had all of California’s electoral votes, and although they are getting closer they have not yet managed to win Texas’s.

Electoral college maths is mostly a matter of luck. If there’s voter movement in big states that are safe for one party or the other, you get a mismatch between the overall vote and the position in the college. Conversely, a small movement in a big marginal state – as sometimes happens in Florida and might easily happen next time in Texas – can have a wildly disproportionate effect on the college.

In recent years the Democrats have been unlucky in that respect; there’s nothing permanent about that situation, but the fact that the partisan divide seems to be becoming more and more an urban-rural divide probably means that they will continue to lock up a lot of votes in a few safe states like California and New York.

As with any non-proportional system, the potential for unfairness is built in. The fact that it may be more random than systematic is small consolation if you happen to be on the losing end of it.