What the hell is going to happen in the 2019 General Election?
Introduction
I have used a Uniform National Swing (UNS) model to predict the outcome of the upcoming 2019 UK general election. I find that given currently polling the Conservatives are on track for a small majority. However, a plausible path exists for Jeremy Corbyn to enter №10 Downing Street.
Background
Britain is going to the polls… again. On Thursday December 12th, the UK will hold its third election in four years. At long last the end of Brexit is rearing its head. If Boris Johnson can win the Conservatives a majority, expect him to push through his hard Brexit. If the Liberal Democrats and the SNP are able to install Jeremy Corbyn in №10, expect a second Brexit referendum with remain on the ballot.
The two leaders’ visions for the future of the United Kingdom could not be more opposed. Johnson and his motley crew intend to re-orientate Britain towards America and to mirror its low regulation economy. While Corbyn intends to re-nationalise the trains, the postal service and other utilities. More importantly, if Corbyn makes it to Downing Street, Britain will likely remain in the EU.
National Polls
As of November 3rd, the Financial Times’ poll-of-polls has the Conservatives leading Labour by 12 points. However, they are both currently polling below what they achieved during the 2017 general election: Conservatives (42%) and Labour (40%).
Modelling Methodology
I have employed a Uniform National Swing (UNS) model to predict the outcome of the election. The model works by taking the total national number of votes achieved by each party at the previous general election (2017) and adjusting it based upon current national polling data. I then distribute each parties’ national votes to constituencies taking into account that not all parties compete nationally (e.g. the Scottish National Party, Plaid Cymru, etc.).
The FT’s poll-of-polls was the source used to estimate current national polling data.
Results
Since Sinn Fein do not take up the 7 seats that they usually win, a government needs 321 seats to have a majority. Consequently, the model suggests that the Conservatives are on track for a small majority of 17.
However, their path to this majority if fraught. The model predicts that the Tories will lose 8 seats to the SNP in Scotland and a whopping 33 seats to the Liberal Democrats in London and the South of England. Consequently, to form a majority the Conservatives will have to smash a hole through Corbyn by picking up 45+ Labour seats across England and Wales.
The implication of this dynamic is that Jeremy Corbyn’s path to №10 is certainly not closed. Corbyn realistically needs Labour, the Lib Dems, and the SNP to pick up 10 to 15 seats to form a minority government. The model already predicts that the Lib Dems and the SNP could pick up 41 Tory seats between them meaning that if Labour can limit its losses to 25 to 30 seats, Jeremy Corbyn could well be the next British prime minister.
Some final thoughts
- The Lib Dems are on track to achieve close to their best ever general election results (62 seats in 2005)
- Labour could achieve their worst ever election results (206 in 1983)
- The Brexit party could take more seats from Labour (10) than from the Conservatives (1)
- The method employed is very blunt, it does not take into account the fact that most voters are aware of the limitations of FPTP and consequently do not waste their vote on a party that is not locally competitive even if it is their favorite party
- UK general election polling moves very fast. Expect the Labour party to rise significantly in the polls between now and election day
Here are the results updated for the polls 08–12–19:
Here is a comparison of the final model results versus the actual election results:
So what went well and what went poorly. First off, my prediction for a large Conservative majority was accurate. In addition, my results predicted that Labour would have a disaster, which clearly happened.
As can be seen, the big miss is between the Conservatives and the Lib Dems. To be fair, when I ran the model for the final time I knew 31 seats would be a real stretch for them. That said, I am disappointed that the model was off by 20 seats.
So the key question is: was my model poor or were the polls very different from what happened on election day. Here is a comparison between my model with the actual vote numbers and the actual results. As can be seen, the Lib Dems are still way off and Northern Island is a bit of mess too. What does this tell us? Nothing that isn’t already well established: (1) modelling elections in Northern Ireland is incredibly hard, (2) uniform swing models is not very helpful for modelling the Lib Dems as the distribution of their support can vary widely between elections.
