Jeremy's Den: February 2020

223 years ago today the Last Invasion of Britain was repealed at the battle of Fishguard and it's also Ed Balls's 53th birthday and Dominic Raab's 46th

It was a dark and stormy night, the rain fell in torrents except at occasional intervals when it was checked by a violent gust of wind which swept up the streets (though it is not in London that our scene lied), and it had been ten weeks since the election, with nine new polls showing the Conservatives leading by 17% average. No buyers' remorse here, contrary to what happened in 2017 when we had eighteen polls in the first ten weeks after the election, sixteen of them with Labour leading by about 4% average. So it looked like as good a time as any to go back to another and possibly game-changing election issue. A year ago I discussed the impact the Sixth Periodic Boundary Review would have had on the 2017 election results. The proposed boundary changes were never sent to Parliament and never implemented, so the 2019 general election was fought on the same boundaries as in 2010, 2015 and 2017. These boundaries are the result of the Fifth Periodic Review that was held between 2000 and 2007. The obvious conclusion here is that these boundaries no longer reflect the current distribution of the electorate, so some sort of boundary changes is both legitimate and necessary. Then the real question remains the same as last year: would a seemingly neutral boundary review result in massive gerrymandering in favour of the Conservative Party? The obvious answer is that we don't know as there is every reason to believe the 2018 review will never be implemented and the Government will request another set of proposals before acting on the matter. But there is also every reason to believe that new proposals would not be massively different from the previous ones, as the Conservatives have a vested interest in passing a very similar set of boundaries, and their massive majority allows them to do just that. So it is quite legitimate to start by assessing how the proposed 2018 boundaries would have influenced the result of the 2019 election. Martin Baxter at Electoral Calculus has published the full notional results for the 2017 election on the initial proposals published in 2016, and the 2019 election on the slightly amended final proposals published in 2018, and I will rely on his data.

The Guardian recently published a reader's letter arguing that first-past-the-post (FPTP) is rigged in favour of Labour. The problem is that the alleged 'evidence' here is based on past elections that were fought on different boundaries and voting patterns, so it basically proves fuck all. What the letter totally misses is that FPTP favours the winning party no matter who they are, and also that the winner's vote share is just one factor. The winner's lead over the second party is just as important, if not more, and the different sizes of majorities are also the result of how the battleground seats (those with the closest results at the previous election) behave under any national swing from one party to the other. Recent results definitely go that way. The Conservatives bagged 48.8% of the seats on 42.3% of the vote in 2017, and then 56.2% of the seats on 43.6% of the vote in 2019. While Labour bagged 40.3% on 40.0% of the vote in 2017, and then 31.1% on 32.2% of the vote in 2019. What the numbers actually say is that FPTP delivered a 42-seat bonus for the Conservatives in 2017 and an 82-seat bonus in 2019, compared to what proportional representation (PR) would have returned. A higher Conservative lead, flawed boundaries and unexpected upsets in marginal seats definitely played a part here and the hypothetical implementation of the proposed 2018 boundaries would only have made it worse. Here are the actual and notional results for the 2017 and 2019 elections, as there is an alternate reality somewhere where the initial proposals were used for May's failed snap election of 2017 and decisively changed the outcome, and the final proposals were enacted for the 2019 snap election and only made an awful result even worse.

In 2017 the new boundaries would have changed the result from Conservatives 9 seats short of a majority to an 16-seat Conservative majority, and in 2019 the Conservative majority would have risen from 80 to 104 seats. So a gerrymandering bonus of 25ish seats for the Conservatives, the biggest notional change since the Third Periodic Review that was implemented for the 1983 election and would have delivered 20 more Conservative seats if its boundaries had been used for the 1979 election. Other Periodic Reviews conducted after 1945 notionally displaced fewer than 10 seats and not always towards the winner of the previous election. Moreover this would have been a massively symbolically significant victory for the Tories as they would have bagged more than twice as many seats as Labour, an exceptional occurrence in the last 75 years. Even the Attlee landslide and the Thatcher landslide did not deliver a similar result (though Attlee missed it by just two seats), only the Blair landslides did. The comparison of vote shares with shares of seats at the last two elections also clearly shows how gerrymandering would distort the results one step beyond the usual FPTP distortion.

Of course the impact of gerrymandering is quite different in Scotland as FPTP massively helps the SNP here, but the changed voting patterns between 2017 and 2019 also deliver some surprising results. The altered boundaries would have worked quite as intended in 2017, protecting both the Conservatives and Unionist Labour while hurting the SNP. Also hurting the Liberal Democrats was probably seen as acceptable, yet not really unwelcome, collateral damage. The situation would have been quite different in 2019, similar to what Americans call dummymandering, one that helps the gerrymanderer at a specific point in time but bites him in the arse when voting patterns change. The main reason why it can happen is when many seats in a given area are marginals and gerrymandering can't significantly lower the number of such seats, a well known situation in Scotland where even smallish swings can switch many seats and bigger swings even more. So the tory dummymandering of Scotland would have increased the SNP's share of seats in 2019 from 80% to 87% and I'm not even counting Neale Hanvey as SNP here though he obviously is in anything but official denomination.

Now it's worth having a closer look at the Scottish Gerrymander. But first bear in mind that the Boundary Review rests on two major principles: reduce the number of seats stretching over several Council areas, draw new seats based on 'community of interest'. The Scottish Gerrymander dismally fails the first test: 47 old seats were fully within one Council area and only 34 new seats do, 11 old seats stretched over two Council areas and 17 new ones do, just one old seat stretched over three Council areas and two new ones do. Then you might wonder what the actual 'community of interests' between Polbeth and Sighthill is, as both would be part of the new Livingston and Edinburgh Pentland constituency which covers a small southwestern part of Edinburgh City Council and a large southeastern part of West Lothian Council, while its predecessor seats (Livingston and Edinburgh South West) were both within one Council area. Then the real rationale behind this was obviously to pit SNP MPs Hannah Bardell and Joanna Cherry against each other and force Cherry out of Westminster. Now let's see what a seat by seat breakdown says (click on the images for larger and easier to read tables).

So seven Scottish seats would be abolished (six SNP and one Conservative, highlighted in red), and it's certainly no coincidence if several high profile SNP frontbenchers would find themselves without a seat, while losing Wee Andy Bowie too would certainly be considered an added bonus by the diehardest Boris cultists. We would also have one 'new' seat (Falkirk South), meaning it has no obvious predecessor seat as it is stitched together from 'leftovers' form other recarved seats, and finally five of the redrawn seats would notionally change hands (highlighted in yellow). For these seats, the notional MPs are the runner-ups (or should that be runners-up?) in the predecessor seats in December. Though of course candidacies would have been different if the boundary changes had been enacted and a couple of newly elected SNP MPs would not even have been candidates. I think we can safely assume Ian Blackford would have stood in the new Highland North and unseated Jamie Stone, and that Joanna Cherry would have stood in the redrawn Edinburgh South and unseated Ian Murray. I also venture Neil Gray would probably have moved some miles to the North East from Shotts to Falkirk and secured the orphan Falkirk South for the SNP. Then I have a hunch John Nicolson would have chosen a rematch against Jo Swinson in the redrawn and more SNP-friendly East Dunbartonshire, saving himself the embarrassing slip-of-the-tongue moment. Just kidding, John, love ya. Finally Philippa Whitford would have found an obvious landing pad in the new Ayr and Carrick seat, notionally Conservative Bill Grant's seat in 2017 and now an open seat after Grant's decision to stand down. So the artfully crafted scheme to unseat prominent SNP figures would certainly not have worked as planned, with only Glasgow North's Patrick Grady left without an obvious Plan B, except now becoming the next MSP for Glasgow Cathcart after James Dornan's retirement.

A textbook case of the Boundary Review's flaws is the situation in Edinburgh, highlighting both the anti-SNP bias and the dummymandering factor. An early proposal, published in 2013 as part of the original Sixth Periodic Review, already reduced the Capital City's seats from five to four but all were neatly within the Edinburgh City Council boundaries. If this proposal had been used at the 2015 election it would have delivered four SNP MPs, as Ian Murray's Edinburgh South was abolished and split between the new Edinburgh East and Edinburgh South West constituencies. Then the resurrected 2018 Review produced two sets of proposals, an initial one in 2016 and a slightly amended final one in 2018. Both proposed four seats within the Edinburgh City Council boundaries but with the added monstrosity of a West Lothian seat protruding into Edinburgh in clear violation of two of the Review's basic principles as I already mentioned. Bear in mind that all Edinburgh seats have been neatly within the city's boundaries since 2005. So the 2018 Review's proposed boundaries are something of a 'once in two generations' monstrosity.. If the 2017 election had been fought under the initial proposal, it would have delivered two seats each for the SNP and Labour, with the new Edinburgh East seat carefully carved to inject enough Labour voters from Liberton and Gilmerton to unseat Tommy Sheppard while protecting Ian Murray in the redrawn People's Republic of Morningside seat. Then the notional 2019 results would have made the gerrymander backfire into dummymander with all four Edinburgh seats going to the SNP. Aye, ye ken, even the best-laid schemes....

The Guardian is now using the Labour leadership race to again push for proportional representation (PR) instead of the centuries-old first-past-the-post (FPTP) and urge Labour to endorse it without actually making a case for it, other than the odd point that it would weaken 'nationalists', insert SNP and Plaid Cymru here but of course mostly the SNP, just as you thought The Guardian had truly renounced their past insidious SNP-baddery. Anyway, and regardless of the likelihood of any system being ever used in the UK, I tried five options.

Proportional representation (PR): using the famous d'Hondt method but not based on a single UK-wide list which would be unacceptable for a fuckload of obvious reasons. Instead I used regional lists in twelve constituencies duplicating the ONS statistical regions and the now defunct European Parliament constituencies. Note that regional PR is more favourable for the two or three biggest parties as it creates a de facto threshold for representation even without such a provision in the electoral law. The first party does not need a majority of the popular vote to bag a majority of seats. Under current voting patterns, around 47% would be enough;

Majority Bonus System (MBS) (which is in fact two options here): the system used in Greece until last year and at some elections in Italy. With two simulations here; one with 600 seats filled on regional PR and 50 allocated en masse to the first party; one with 550 seats on regional PR and 100 allocated to the first party;

Additional Members System (AMS): the one infamously used at the Scottish and Welsh parliamentary elections, providing levelling (or compensatory) seats on regional PR for the losers of the FPTP vote. Here I simulated two thirds of seats filled on FPTP and one third on regional PR, the same proportions as the Welsh Senedd;

Mixed Membership System (MMS): with 325 seats filled on FPTP and 325 on regional PR. I used a true MMS with the PR seats allocated in a totally separate count regardless of the results of the FPTP vote. This is quite different from the allegedly 'mixed member representation' used in Germany and New Zealand, which is tweaked to deliver a similar result to pure PR, or from AMS which factors in the FPTP seats in the allocation of the additional PR seats.

The only options I did not try are Instant Runoff (IR) and Single Transferable vote (STV), as nobody has the fuckiest where the second and further preferences would go. Trying to approximate IR by some odd extrapolation of the winner-runner up combinations on FPTP would only be a very approximate approximation, and certainly widely and wildly off. So here is what the other options would deliver, in numbers and shares of seats. All based on the 12 December results though it's quite certain that voting patterns would have been different under a different electoral system, but that's all we have.

The proponents of PR quite clearly support it only to avoid any party ever getting a majority and to impose coalitions that the electorate don't actually want. For example, in the run-up to the last German general election, only 25% of the electorate actually wanted a CDU-SPD Grand Coalition and yet that's what they got in the end. In the UK last December the only mathematically viable option would have been a Con-Lib coalition. And we all know that, even if we loved to jibe at the LibDems' unique talent for betrayal and their appetite for ministerial cars, it would never have happened under Johnson's leadership. Foreign examples also demonstrate that coalitions can be the recipe for impotence as in Germany where Grand Coalition partners give up on their most divisive manifesto pledges and the Grand Coalition amounts to little better than a long-term caretaker government with no ambition for reform. In other contexts it can also be a recipe for extremism as in Israel where coalescing the fifty shades of racist theocratic far-right around Likud led to each coalition partner fuelling the worst instincts in the others, if they ever needed it. Back to my simulation, AMS here delivers an unexpectedly close result to PR. The reason is that there are many more seats up for grabs on the regional lists than in the real Scottish and Welsh elections, so the FPTP winners have significant opportunities to also bag list seats.

In my view, the best option is MMS as it can deliver a majority if the first party has both a strong plurality of the popular vote and a convincing lead over the second party, which is exactly what happened at the 2019 election. But the Conservatives would have bagged only a 10-seat majority under MMS, meaning the would be in a weaker position than under FPTP and would have to seek compromise in some cases. Bear in mind for example that only 330 out of 365 Conservative MPs voted for Johnson's Withdrawal Agreement Bill at the third reading. Even though there was no open rebellion, 10% of Tory MPs expressed discontent by not showing up to vote despite support for the WAB being carved in stone in the Tory manifesto. If Johnson had had a 10-seat majority instead of 80 seats, he would certainly have been forced to negotiate with his own 'soft Brexit' MPs and amend the WAB. For example by reinstating the provisions that were scrapped between the initial pre-election draft of the bill and the final draft that was tabled after the election. Just sayin'..... In different situations where the results are less one-sided than in 2019, MMS would nit deliver a majority, as the projections for the 2010, 2015 and 2017 elections show. So it can be seen as a perfectly acceptable compromise solution even for PR supporters, as it delivers a majority only if the popular vote goes strongly in one direction, and on more ambiguous results delivers a hung Parliament and a coalition.

For simplification 'Others' above include the 18 Northern Ireland seats, the Speaker and Plaid Cymru. Which is why the numbers are almost the same under FPTP and MMS, as Plaid Cymru is the least impacted by the switch to MMS, bagging three MPs at each election instead of the three or four they bagged under FPTP. I kept separate lines for UKIP and the Greens as both can be significantly impacted by the switch from FPTP to MMS, especially at the 2015 election that saw a more fragmented vote than 2010 and 2017. Under MMS we would still have had the Con-Lib coalition in 2010 but the stronger LibDem representation might have resulted in some changes in policy. Or not... More significantly both the 2015 and 2017 elections would have made the continuation of the Con-Lib coalition the only viable option. And you never know what might have happened then. No EU referendum? No Brexit? We will never know but one can dream, can't he? Anyway it would be daft to expect any significant electoral reform in the foreseeable future, whoever is in power, and for once that really means never-in-a-generation and probably never-in-a-lifetime. But I certainly don't rule out the current Tory government enforcing the 2018 Boundary Review or something very close. They have the votes to pass it no matter what, and losing only 13 seats under the new boundaries is very easy to manage with the natural attrition we witness at every election, with MPs standing down because of age, stress or any of the usual excuses. They could easily promise a seat to all incumbents willing to return and accommodate a batch of newcomers on top of that. So watch out for the new boundaries coming up in Commons, we only don't know when. Time will tell, but when it happens, remember you read it here first.

It's a good rule in life never to apologize
The right kind of people never want apologies
And the wrong kind take advantage of them
(The Orville, episode Pria, 2017)

Jeremy's Den

25/02/2020

The Great Gerrymander of 2020?

We Must Be Dreaming