Context for anyone reading this from long in the future, or from under a rock: it’s been a busy week in British politics. I’ve torn myself away from watching the endgame of Boris Johnson on Twitter long enough to write a response. It’s not meant as an ideological post, but instead taking this as a more general “teachable moment” (horrible term) of how to think about political evidence. I’ll do this through considering two recently oft-heard claims: ‘Johnson is uniquely popular’ and ‘Johnson will be ousted soon’.
But before I get to that, I have to get my personal views – and my boast – in. I quit No10 because of Boris Johnson before it was cool. I was a civil servant doing social media analysis and communications; I didn’t agree with Theresa May ideologically, but I trusted her sense of public service. Also her team generally left me to do the sorts of social media analysis I felt a good government should be doing. I had zero confidence that Johnson – or Dominic Cummings (for whom I have some respect in other aspects) – wouldn’t use my work for political gain. Also doing counter-disinformation for such a dishonest Prime Minister didn’t appeal.
The moment crystallised for me on Johnson’s first morning, where I was reporting that websites which spout hard-right views and xenophobic fake news were celebrating the new regime. I told my manager I couldn’t do my job in good faith anymore; in the end it took a few months to find a new job, in DCMS. In the meantime, although I can confirm that Johnson in-person is extremely charismatic, everything I saw from the new No.10 administration in the meantime convinced me I was making the right choice.
With that done, let’s think about analysing those two claims about Johnson.
‘Johnson is uniquely popular’
Some Westminster Wisdom can become accepted facts amongst politics watchers, on basis of not much evidence. For instance, the story that Boris Johnson’ 2019 landslide can be explained by his immense popularity with the country, and his unique appeal amongst largely working class voters in ‘red wall’ seats in the North.*
But Johnson has not been popular for a long time. A quick check shows his only net positive rating as Prime Minister was between Dec 2019 and Jun 2020 – a combination of election bounce, early Covid ‘rally round the flag’, and his near death. The 2019 election was two unpopular candidates against one another; Corbyn was just very unpopular. Many of the ‘red wall’ seats had already been shifting away from Labour, and he just finished the job. In the words of the inventor of the term “the 2019 fall of the Red Wall should not have come as a shock – for those viewing the past decade of election data in more objective terms and taking emerging trends at face value, it was evident this was coming.” For all the claims that Johnson won the 2019 majority, I suspect another Conservative leader would also have done so.
Prior to 2019, both the seats Johnson has held as an MP have been solidly Conservative for decades; he inherited Henley from Tory grandee Michael Heseltine, and his most Uxbridge & Ruislip majorities have always been smaller than his 2010 predecessor. He won London Mayor twice; I’ll accept this may be partly due to him and his profile (his Have I Got News For You appearances gave him much more name recognition than the average politician), but he also stood in 2008 and 2012 (hardly great years for Labour) and both times against Ken Livingstone (who brought his own issues). And yes he was on the winning side in the EU Referendum, and he may have played a role; but I’ve seen little evidence that Johnson was a unique and decisive factor (amidst anti-immigration and anti-establishment sentiment, or a more generally vigorous Leave campaign and lacklustre Remain campaign). This evidence is not as clear as the 2019 data, but it also doesn’t exactly support the idea of someone who consistently wins against the odds.
A general point: the above is obviously a simplified summary of how I thought about the evidence (and evidence that I didn’t use). E.g. looking at how someone does in a constituency is swayed a lot by how party does; tracking support amongst classes is complicated by how one defines class, etc. But the evidence is there to check. The psychological impact of Johnson is strong: I suspect people see his eccentric performances (though even there I don’t think he’s that unique amongst some of his colleagues) and think ‘I don’t like him, but I can see how other people like him’. But a much better case needs to be made that this is actually true.
* The ‘red wall’ seats started as a very useful and interesting concept – seats which, given general trends, should have gone Conservative but were still holding out for Labour. It’s since been used quite broadly, and less usefully, as ‘Northern and working class’.
‘Johnson will be ousted soon’
For a long time I have been relatively confident – fluctuating between roughly 60%-90% – that Johnson would lead the Conservatives into the next general election. I held that view for a long time, during various scandals, right up until this week. I’d still argue my reasoning was generally correct looking at broader trends, but was derailed by shock events. This is quite a familiar feature in political predictions; let’s draw it out.
I’ve heard a lot of very sensible arguments as to how the UK’s Parliamentary system comes with excellent reasons why MPs will get rid of unpopular leaders. Most obviously, their jobs depend on re-winning elections. In the meantime, they have to endure haranguing from constituents; and, for ministers, criticism of their government. Eventually enough of them will get sick of that, and will want to find a new leader. But just because there’s good reasons underlying it, “MPs will get rid of leaders when they become a liability” is still just a hypothesis; it needs testing with real evidence.
To the best of my knowledge, most mainstream European political parties do not have nearly as many serious attempts to replace party leaders as the British Conservatives. In many cases that’s due to different leadership systems; but that’s true even in other UK parties (except maybe UKIP). This in itself makes our hypothesis suspect, but let’s just focus on the Conservatives.
The party has a reputation for being ruthless at removing leaders; this is probably a story burned into the brains of political types by the forcing out of Margaret Thatcher, still hailed almost religiously by many Conservatives. For those who don’t know, the Conservative Party can remove leaders by Votes of No Confidence (VONCs). The precise rules have changed over time, but it’s a vote taken by MPs. In other words, it’s a vehicle for MPs to commit the aforementioned fratricides.
Since the Conservatives introduced formal leadership elections in 1965, the only Conservative leaders not to have faced VONCs are William Hague (leader 1997-2001), Michael Howard (2003-2005), and David Cameron (2005-2016). All these three stood down after losing public votes (general elections for the first two, the Brexit referendum for Cameron).
So a very brief whistlestop tour of Conservative Party VONCs:
Ted Heath (1965-1975) – narrowly lost two general elections (February and October 1974). This prompted the Conservatives to introduce rules for challenging sitting leaders, leading to the first such vote in February 1975. Despite the election losses Heath was still supported by the Shadow Cabinet. But in the end there was a surprise victory for...
Margaret Thatcher (1975-1990) – VONC’d in December 1989, following increasing unpopularity over her introduction of the Poll Tax (which led to riots), as well as her increasingly dictatorial leadership style and divisions over Europe. She beat her challenger, a little-known MP called Sir Anthony Meyer, with 90.5% of the votes. However the same concerns continued to mount, and she was VONC’d again in November 1990. She won again, but only 55% to 41% over Michael Heseltine. The closeness of this result led her Cabinet to tell her position was at severe risk, and she reluctantly resigned two days later.
John Major (1990-1997) – Unexpectedly won the 1992 election. Then VONC’d himself in 1995 following anti-Europe mutterings from his own Cabinet, with the great line “it’s time to put up, or shut up”. At this point his standing in the country had been hit badly by the economic catastrophe of Black Wednesday, and he had trailed continuously in the polls behind first John Smith and then Tony Blair. He won 66% against the strong Eurosceptic John Redwood. He would lose the 1997 election two years later.
Iain Duncan Smith (2001-2003) – also VONC’d himself, following discontent about his electoral appeal and some minor scandals. He lost with only 45% of MPs voting that they had confidence in him (by this point rules had changed so there wasn’t an opponent to vote for).
Theresa May (2016-2019) – VONC’d in December 2018 over opposition to her Brexit deal. She won with 63% MPs voting in favour, but after saying that she would not lead the party into the next election. She resigned almost 6 months later, after repeatedly failing to win Brexit votes, a terrible performance in the European Parliament elections, and renewed calls from backbenchers for another VONC.
Boris Johnson (2019-2022) – VONC’d in June 2022 over a growing mass of scandals which he mishandled terribly, massively hitting his popularity in both the party and the country. These largely began October/November 2021, with the Owen Paterson affair and the first reporting of the Partygate scandal. He won his VONC with 59% of MPs voting in favour. A month later, following surprise mass resignations (we’ll come to those shortly), he offered to resign.
Looking at these precedents, the story that the Conservative Party ruthlessly removes leaders when they become liabilities is a bit complicated. Certainly they VONC a lot, and this is a sign that things are going wrong. But when given the chance, the MPs almost always kept the leader – however unpopular. Then eventually the leader themself bowed to pressure (except Major, who made it to the election). But Theresa May hobbled on for 2 years after her catastrophic 2017 election gamble, and another six months after her VONC (with constant Parliamentary defeats). Major lasted for 2 years after his VONC, despite being clearly on course to lose a lot of MPs in 1997. And then there’s the VONCs which didn’t happen – both Hague and Howard were allowed to go into general elections considerably trailing Blair (though admittedly, given Blair’s high bar, the Conservatives may have felt that was unavoidable).
In sum, the hypothesis that MPs will get rid of leaders that are liabilities is a bit suspect. Not necessarily outright wrong; but clearly other factors, whether structural (rules) or psychological (fear of the unknown) also weigh in, and the ‘obvious’ answer doesn’t hold up. It’s for this reason I’ve been arguing against “Boris will be gone soon”. Yes it looked sensible and obvious that he would. But precedent suggested otherwise.
This is an example of the thinking strategy called ‘the inside-outside view’ that I’ve described in another post. The ‘outside view’ – the wider precedents, the bigger picture – suggested that leaders can limp on for a surprisingly long time. Modifying with the ‘inside view’ – the specifics of the case – to me strengthened the case that Johnson would remain. The others had eventually accepted they couldn’t continue; I strongly doubted Boris Johnson would do that.
Harold Macmillan was once asked what the hardest problem of his Premiership was. “Events, dear boy, events” he replied. This is also the hardest problem of predictions. Forecasters, particularly in finance, talk of ‘Black Swan’ events – unpredictable, consequential, but seemingly ‘obvious’ in hindsight. They are the revenge of the inside view against the outside view – a specific story which destroys careful trend analysis.
In this case, the event was the sudden resignation of Rishi Sunak and Sajid Javid. Many people, myself included, had resigned ourselves (pun intended) to the fact that Johnson’s cabinet were staying put. Then two went at once. And then, nearly 60 other government MPs in 48 hours; many of their letters expressing concerns about integrity that could have prompted resignations much earlier. One can point to sort-of-precedents for this; Geoffrey Howe’s dramatic resignation catalysed the second Thatcher VONC. But then again, Theresa May survived Johnson and David Davis resigning together, and then a pretty continual stream of resignations. But looking at a graph of resignations shows that PMs have had quite varied experiences:
(If you'd like the delightful alternative of that chart in cross-stitch, see here)
Fundamentally there’s two difficulties in this kind of political predicting. One is that n, the number of cases we can learn from, is just too small. We can see rough trends – parties tend to stick with leaders when sense would suggest otherwise, winning a VONC is no guarantee of longevity – but nothing precise. Johnson is now another data point to add to that; I will be revising my priors, and be less confident of a liability leader’s long-term survival in future.
The second problem: political data is messy, and in each story the same components can combine in varied ways. But nonetheless the aim, as always, should be to focus on evidence from the wider world more than the contents of our heads. Not to believe that MPs will remove leaders just because it seems sensible; or that Johnson is popular just because he seems funny on TV. And if that seems obvious – well, those claims have stuck pretty strongly.
I will remain uncertain and cautious about predicting the political fates of individuals. But I will also remain utterly unconvinced that Boris Johnson’s perceived ‘successes’ were anything more than jumping on others’ work, and mis-statements both by and about him – accidental and, of course, deliberate. Suffusing political commentary with evidence-based discussion might, maybe, be a weapon against something like him happening again. But quite possibly not.
Fun Fact about: Helmut Kohl
Leaving Boris for another political leader: former German Chancellor Helmut Kohl (Chancellor 1982-1998). This is from the book Drama in Nine Acts by Ashoka Mody, a book about the history of the Euro: Kohl was livid. The Brussels summit was to be his crowning party. But Chirac [the French President] was blocking his nominee for the position of European Central Bank president. The loss of face in the German public’s eyes could prove costly in the elections due in September. Former Swedish Prime Minister Göran Persson later recalled Kohl … In his rage, ate a dozen cubes of butter - and when a new bunch arrived, he ate those too before he started to calm down. Incidentally this book, along with actually dealing with the EU, had an enormous bearing on my attitudes to the EU (which can be roughly summed up as ‘the EU as a political union is often a very bad thing, it probably should have remained more like it was in the 1990s; but Britain probably should still have remained in it’).
If you enjoyed this post and/or following politics with a bit less ‘froth’, I recommend the general output of the Institute for Government, particularly their various Tweeters and the Inside Briefing Podcast. (And for those who like podcast recommendations but don’t yet have a dedicated podcast app, I’d recommend Podcast Addict). I’ve previously plugged Newspeak House. Their residency programme is open, with some scholarships for those who might struggle to afford it. As I write this, news is appearing that Elon Musk is trying to pull out his Twitter deal. Here's the post I wrote last time this was in the news. But for the best commentary, follow Matt Levine. For those who want some poetry alongside politics in their Twitter timelines, follow Nikita Gill and Brian Bilston. Brian’s been doing a lot of good politics poems of late; but lots of his content is also - just - fun.