How electoral systems keep fooling betting markets


The US Election votes are still being counted but with each day that passes, this result looks ever more like the greatest electoral anomaly in living memory. Hillary Clinton is on course to win by the popular vote by around 2M votes, yet suffer a resounding 306-232 defeat in the electoral college.

This isn’t an excuse. I argued many times that the electoral college favoured the Democrats nowadays and am happy to fess up to being completely blindsided by this result. Trump redrew the map in a way other Republicans have only dreamed – that fact is unarguable.

Nevertheless, we need to understand why the overwhelming majority of pundits – and betting markets – were proved so spectacularly wrong.

Pollsters are predictably taking flak and differential turnout seems almost certainly to have been a factor – just as it was with Brexit and the 2015 UK General Election. We know that older voters turnout far more reliably than younger ones – favouring the Right.

However these don’t tell the whole story, nor really vindicate talk of a ‘silent majority’. If that was so, Clinton wouldn’t be winning the popular vote. Pollsters weren’t so far off the mark.

One area I believe requires further examination is the electoral system, and the effect it has on undecided or voters that are less than enthusiastic about the main options. It almost certainly applies to UK elections too, and have long suspected goes a long way towards explaining how almost everyone called our 2015 General Election so wrong.

Unlike countries that use proportional representation, both the USA and UK have first-past-the-post (FPTP) voting systems, with results awarded on a winner-takes-all basis by constituency or state. The effect is to create the sense of a binary choice – despite the political preferences of both electorates being increasingly diverse.

Could it be that respondents answer polls honestly, listing their favourite choice, yet end up going for a tactical option on the day, having considered the realistic effect of their vote?

While FPTP worked perfectly in the 1950s, when Conservatives and Labour shared over 95% of the vote, it was inappropriate last year, when our TV debates included seven different parties. The big-two haven’t even scored 70% between them since 2001.

Likewise, whilst the 2016 election was always in reality a Clinton v Trump head-to-head, polls consistently showed the public were interested in other options – either via a historically high number of undecided voters, or Gary Johnson and Jill Stein scoring double digits combined.

Yet in both cases, the main two parties were miles apart on policy, the population increasingly partisan and the polls pointed to a very tight contest. Whatever voters felt in their hearts, they knew that a vote for Gary Johnson, Jill Stein, the Lib Dems or UKIP would feel like a wasted one if the ‘wrong’ side won.

Both shock results, I believe, were driven a significant rise in what effectively amounts to tactical voting. Trump owes his electoral college victory to the fact that he dominated among late deciders and the collapse in both Johnson and Stein’s support.

Likewise, the Tories owed their majority in 2015 to almost wiping out their Lib Dem coalition partners – something which was not predicted by constituency markets or polls. A squeezing of Lib Dem or UKIP voters may also have swung several key Con-Lab marginals in favour of David Cameron’s party – again in defiance of polls and markets.

When polled several times during the course of the 2010-15 parliament, the Lib Dems were consistently close or ahead in the constituencies which they already held. The market factored in they’d lose around half of their 57 seats, but nobody saw them getting just 8. Virtually every Lib Dem seat targeted by Labour or the Conservatives fell – and the latter won many more, dramatically altering the electoral maths.

There was simply no polling evidence for anything like the scale of swing in Lib-Con marginals. Nor was there much evidence on the ground or sense among activists. Yet there was undoubtably a big, late swing in these places to the Tories, without which they would never have won a majority.

These voters are often labelled ‘Shy Tories’ or ‘Shy Trumpers’, but I’m not sure that is accurate. They might ideally be Johnson or Lib Dem voters (at least on a local level), but went for the practical option in fear of helping Clinton or Labour.

This would be far less likely to be an issue under a proportional representation system, as widely used elsewhere. They would have no need to switch, as it is usually clear which way a candidate will swing after the election. Hence people are free to vote their conscience.

It is way too early to predict the effect of Trumpism on US politics. There is definitely a yearning for extra choices and in some respects the rise of Trump and Bernie Sanders reflects that. Both parties could undergo ideological transformation, but we can only wait and see how that affects voting behaviour.

So the UK will provide the next test for this theory, whenever the election happens. Ours will remain very much a multi-party system and the fate of both UKIP and the Lib Dems will have a pivotal effect.

Interestingly, Lib Dem performance has regularly blindsided betting markets. In 1997, 2001 and 2005, they considerably overperformed expectations. The logical explanation was that this small party could focus all it’s resources effectively on key targets, yet could never compete on a national scale.

In 2010, they underperformed after a national polling surge, for the same reason. They didn’t have a nationwide organisation, or the resources to compete in inner-city Labour-held seats that had suddenly appeared within range.

Next time could see the trend reverse. Unlike 2010 and 2015, it will probably not be close, with the Tories expected to win big, just as Labour did between 1997 and 2005. The motivation for voters in those (mostly southern) Con-Lib marginals to ‘stop Labour’ may no longer exist, leaving them free to switch back.

The first test of that is looming at the Richmond by-election, which the Lib Dems are trying to turn into a contest about Brexit. I’m not convinced they’ll oust the sitting MP Zac Goldsmith and suspect it is too early for this new dividing line to truly reap dividends. It is, nevertheless, an interesting work in progress.


4 responses to “How electoral systems keep fooling betting markets”

  1. uhmmm how else are you going to elect a President other than by fptp? are they going to job share in proportion to their share of votes?

    I find this statement very strange ;
    “This would be far less likely to be an issue under a proportional representation system, as widely used elsewhere. They would have no need to switch, as it is usually clear which way a candidate will swing after the election. Hence people are free to vote their conscience.”
    given the Lib dems role in the coalition and their voters response.

    I think a plausible narrative is that a lot of these alternative voters don’t actually want their party in power, and have minority/protest views and are somewhat indifferent between the main parties (eg the antitrident libdem voters).

    • I think we’re talking at cross-purposes mate. I’m not suggesting anyone change their system or making a judgement about the validity of these voters’ views. Rather, for the purposes of prediction, I’m trying to explain why there may have been such dramatic late swings away from and underperformance by 3rd party options in these two elections. In the Lib Dem case, it is the opposite of what happened between 97 & 05.

  2. I disagree, and with some of your arguments I do so emphatically. I have data to back me up.

    It simply is not true that Trump’s victory was a due to late swing; there were several reliable polls they consistently had Trump ahead. Rand / USC Dornsife and people’s pundit, and also a proprietary poll of which I was apprised. The US corporation controlled press chose to assume – based on no evidence whatsoever and against decades of evidence that they were completely wrong- that the demographic makeup of the Americans who decided to vote would be exactly the same in 2016 as in 2012. Never mind that for blacks in particular it was a matter of pride to reelect Obama in 2016 (while they had very mixed feelings about Hillary “Superpredator” Clinton & more of Obamas’s for them disastrous economic policies, while the white working class that Romney had essentially told to shove off and go die quietly somewhere was extremely keen to vote this time around. Obviously this was ridiculous right from the beginning.

    Initially the idea was to try to create a self-fulfilling prophecy, and then down the line as this didn’t happen, to not discourage Hillary voters from going to vote (turnout for a presidential candidate spills over into downballot races) and thereby prevent and even worse bloodbath for the Democratic Party.

    Moreover, if you looked into the internals of many of the polls, you saw that they were complete garbage, and skewed in such a way as to raise legitimate doubts of their good faith. Samples of 58% female, 42% male. Or 50% university graduates simply don’t represent a cross section of voters.

    In probably the most polarising election in 70 or more years, where outright violence against signs and even people, not to mention ostracisation were hardly unknown, one needs to consider the possibility that many of the Gary Johnson votes were “parked” Trump votes. If I loath Hillary, will never vote for her, but don’t want to worry about my workmates hating me and my car getting scratched up, I’m not going to tell them I’m voting Hillary, but Gary Johnson, and they’ll leave me alone. But neither my workmates nor my wife will be peering over my shoulder when I’m in the voting booth. I am a bit surprised that a blog of “neutral analysis” of the race has never mentioned the sometimes immense social pressure and even fat from rare acts of violence that was proven to influence what voters told pollsters.

    Your assertion that Hillary’s having gotten more votes means that she was the more popular politician overlooks the fact that a good portion of voters only bothers to go vote when the they believe their vote could matter, myself included. In California, for example, with its 38 million voters, not one republican had a chance of winning state wide office, and it was clear that Hillary would carry the state. For any republican voter not living in a district where a republican state representative was running reasonably close to his democratic challenger, the time taken to vote was completely wasted. The same calculus would apply to democrats in heavily republican states like Utah, but the significantly larger number of electoral votes from states that are locked up for Hillary (California, Massachusetts etc) than vice-versa practically guarantees that she would not have kept her lead in a popular vote.

    The argument that a campaign that outmobilised its opponent, winning the biggest victory in 28 years in the states in which it bothered to compete, would not have been able to mobilise even many more voters if it had competed in states where it spent no money at all on campaigning strikes me as ridiculous.

    And this doesn’t yet take in to account that in the United States merely ticking a box on a driver’s licence registration that you are a citizen is all it takes to register to vote, no checks done, unless a candidate starts a Florida 2000 style legal showdown. Throw in that Barack Hussein Obama pointedly refused to condemn noncitizens from illegally voting and explicitly pointed out that they should fear not any legal consequences (is the country a banana republic or what?), and the Hillary’s lead looks even more dubious.

    This is no different than arguing after having your team lose the World Cup final that since your team far outscored its rival in goals kicked with the left foot your team is the actual winner.

    • Frank, there is ample evidence of a late swing. He closed the gap significantly over the last couple of weeks, primarily at the expense of the third candidates and undecideds. Likewise many of the mainstream polls taken afterwards showed him winning big amongst late deciders and among election day voters. For all the stick they’ve taken, the polls were closer than in 2012. This isn’t to make any sort of comment about either candidate’s strengths, weaknesses or legitimacy. Maybe she did win the popular vote off the backs of voters in states where Trump didn’t campaign – though equally one could say she took MI and WI for granted and may have fared better if spending more time and cash there. My instinct is that campaigning made much less difference than usual, and that most voters had a clear sense of which candidate they *wouldn’t* vote for.

Leave a Reply

Your email address will not be published. Required fields are marked *