The Great Gerrymander of 2019

Almost quiet on the polling front for almost two weeks. Even YouGov poll all sorts of topics and the kitchen sink, but not the next GE. So let's take a few moments to explore the long-delayed reform that might alter the shape of things to come: the Boundary Review, aka the Great Gerrymander of 20-something. Or not. Or just a little bit of it. Or totally so.

Boundaries bound to change?

A lot has already been written about the Boundary Review so I will not bore you with the details of the legislation. It has been a protracted process that started with the legally mandated Sixth Periodic Review of Westminster Constituencies, triggered by the Parliamentary Voting System and Constituencies Act 2011. Ironically the Act also included provisions for the Alternative Vote Referendum that failed so massively in May 2011, the only time David Cameron won a referendum. The Boundary Commissions then published proposals (known as the '2013 proposed boundaries') for each nation.

Nothing came out of it as both Commons and Lords voted to postpone the boundary changes to 2018, which de facto killed the Sixth Periodic Review. The Electoral Registration and Administration Act 2013 amounted to starting the whole process from scratch, now known as the '2018 Review'. The new Act reaffirmed the basic principle: reducing the size of Commons from 650 to 600 but with a slightly different apportionment of seats between the four nations and regions within England.

All four Boundary Commissions issued their initial proposals in October 2016. Then revised proposals were issued between October 2017 and January 2018 after public consultations. These proved controversial again and a 'final' set of proposals was published in September 2018. These too attracted criticism from all points of the compass except the Conservative Party, but were soon set aside while the last stages of the Brexit clusterbùrachfuck unfolded. For further and detailed information about the content of the current proposals I suggest you read Martin Baxter's excellent article on the Review(s). Another analysis by UK Polling Report is also of interest.

Closer to home the Boundary Commission for Scotland also published detailed papers about the original Sixth Periodic Review and the 2018 Review. They also made an interactive map available online, with all details of current constituencies and how boundaries evolved though the three 2016-2018 sets of proposals.

Are the proposals fair?

This is a technical question that points at the reapportionment of seats by nation and region, not the boundaries themselves (more on this later). Also quite rhetorical as the allocation of seats is required by law to be based on an 'electoral quota', that is the average size of the electorate in each constituency must conform to an uniform standard all across the UK. The only exceptions to the rule are the Island Constituencies (Na h-Eileanan an Iar, Orkney and Shetland, Isle of Wight) who have a life of their own.

So it's pretty much down-river from there as it's pure math, and simple one at that. Really no way the apportionment can be unfair as electorate data are easily available from the Office for National Statistics and anyone can check the results. So far so good the but the 2018 Review is actually the first to fully enact the rules. Current boundaries (used since the 2010 GE except in Scotland where they were enforced at the 2005 GE) come from the Fifth Periodic Review which perpetuated a major exception to the rule: blatant over-representation of Wales.

Scotland's representation then dropped from 72 seats to 59 based on the electoral quota (ironically defined back then as the 'English electoral quota' as it was based on the average electorate per seat in England and not on the UK-wide average). But Wales kept their 40 seats when simple math says that the electoral quota entitled them to 32 only at the time. Now the comparison between shares of the electorate, share of current seats and share of proposed seats by nation and English region:

Bear in mind that the 2013 Act requires the current review to be based on the 2015 electorate while the Fifth Review was based on the 2002 electorate and in best cases the 2004 electorate. So discrepancies between current electorate and the current apportionment of seats are easily explained by demographics. In the 10 to 15 years between Reviews the population (and hence the electorate) has increased faster in England than in the other nations, and also in London's suburbia within England.

Insofar as only the first stage (apportionment of seats by nation and region) is taken into account, this Review is fair. It can even be argued that the end of the Welsh exception also makes it the fairest ever since Periodic Reviews exist. So it boils down to a philosophical debate: is electorate-based apportionment a better approach than population-based apportionment as used in the United States, France and Germany?

Basically Americans and French believe every citizen has an unalienable right to parliamentary representation even if they don't bother to register to vote. While the British system sends the opposite message: if you don't register then you don't 'deserve' representation. Which can be interpreted as an incentive to register as areas with higher registration will be better represented. Exhibit A: Scotland with 8.7% of the electorate on 8.2% of the population. Just do the math: 0.5% higher registration implies a 3-seat 'bonus'. Fine with me.

Are the proposals biased?

That's the key issue as fair apportionment does not mean there is no gerrymandering in the second stage of the Review when individual boundaries are redrawn. The only way to assess partisan bias is to compare the actual GE result with what it would have been under new boundaries (the infamous 'notional' results that are often challenged). Here is what I find, based on reconstituted constituency data readily available online:

First spectacular result is that the first party in every area systematically does better on the new boundaries than on the current ones, and I will go into more detail on this later. Then only the massive weight of English seats turns the overall result into a significant bonus for the Conservatives. Even more obvious if you translate the whole thing into a bar chart comparing the shares of votes at the 2017 GE and the shares of seats under current and proposed boundaries. Net result: a hung Parliament turns into a Conservative majority thanks to the magic of redrawn boundaries.

Common wisdom including mine has it that Labour suffers from a competitive disadvantage under FPTP because of differential turnout and amount of wasted votes in deep sinkholes. This is not readily apparent in the chart above that shows LibDems as the ones getting crushed way below what their share of the votes would get them under proportional representation, and also hurt the most by the new boundaries. The point is made though if I replay the 2017 GE with a Con-Lab tie, both on 41.3% instead of the actual 42.5%-40% split.

Simulation on current boundaries has Tories 16 seats ahead and it rises to 41 seats ahead on the proposed 600-seat boundaries. Guess that makes the case that current voting patterns on current boundaries already favour the Conservatives and that the proposed redrawn boundaries would make it worse at the expense of both Labour and LibDems. Circumstantial evidence that the 2018 Review does include gerrymandering indeed. Of course there is more here than meets the eye when looking at the big picture only. A strange brew with a pattern of 'fuck Labour but fuck LibDems even more'. As always the devil in the details. Which deserve a closer examination nation by nation.

Northern Ireland

Just one seat down and no obvious specific purpose other than solidifying the DUP-Sinn Féin polarisation seen in 2017 and the communitarian split that so flabbergasted Karen Bradley. Only collateral damage is Sylvia Hermon vanishing from the scene as her North Down seat is grafted enough new territory from DUP stronghold Strangford to switch it to DUP with a 10% lead. Somewhat of an overkill but that was probably the desired effect: getting rid of a wildcard MP who wouldn't fit into the prewritten Westminster powerplays.

Fun part of the Northern Ireland Review is that an earlier version of the proposed boundaries made a complete pig's ear of the whole process with a notional nine seats going to Sinn Féin and eight to DUP. Must have taken some very creative cut-paste to achieve this but that was swiftly corrected with a return to the 'natural order' of things.

Wales

With nearly 30% of seats woodchipped everybody is bound to lose. Weird thing here is that Labour would actually win with a higher share of seats. Only notable Labour loser would be Owen Jones with his Pontypridd seat neatly cut in two and fed to the redrawn Cynon Valley and Rhondda.

There is also some creative recarving at work here. Only in Wales can you take bits of one Plaid Cymru seat and one Conservative seat and turn the resulting new seat into a Labour notional (see Carmarthen). Or add enough odd bits from other seats to a former LibDem stronghold turned Plaid-LibDem marginal in 2017 in a way that makes the new seat an unpredictable four-way marginal (see Ceredigion).

Scotland

Here the comparison delivers surprising results. You would probably expect new boundaries to hurt the SNP but they don't. That was not always the case but I guess the consultation phase helped the Boundary Commission come up with an interesting updated scenario. So now Scotland loses 6 seats out of 59. To which SNP contribute 3 out of 35, LibDems 2 out of 4, Conservatives 1 out of 13 and Labour…. none. An earlier version was even worse, with some creative redrawing in Inverclyde and Ayrshire pushing the SNP down to 30 notional seats and Labour up to 9. Was probably a bit too osé to last. 

The new boundaries here would actually trigger some interesting cases of musical chairs. Seven of the 2017 MPs would find themselves without a seat while one newly-carved seat would find itself without a 'natural' MP. The 2017 result would be reversed in seven seats compared to their notional predecessor seats. Nuff said fur noo. I will devote another post to the detailed oddities of the proposed Scottish boundaries. Including bizarre recarvings across Council boundaries and strange assemblies of wards within the same Council area. Stay tuned.

London

There's a pattern so far and not the most obvious one: fuck LibDems and don't fuck with the SNP, be not too harsh on Labour we'll get even with them later. So you'd think something different would emerge in London as a few seats can be scraped off Labour in one of their strongholds. Only it doesn't and LibDems are once again the ones hurt most.

London is a strange beast indeed with prominent figures of all three main English parties homeported there. Which offers a golden opportunity to score a couple of points in the Tory Civil War. Both Boris Johnson and Iain Duncan Smith are elected from seats with strongly Labour areas in the immediate vicinity. So enough Labour-leaning territory has been carved into both to turn them into Con-Lab marginals notionally. Unsurprisingly BoJo is the most endangered with a 10% lead in 2017 shrunk into almost zero. IDS could still manage to save his sorry ass as his lead has shrunk from 5% to 2% 'only' on the new boundaries.

But playing a nasty trick on Labour was probably too tempting to miss. And that's what happens with some creative redrawing across Borough boundaries in the Hackney-Islington area. Starts with four seats and ends with three with one abolished in favour of a new seat overlapping Hackney and neighbouring Tower Hamlets. Current MPs: Jeremy Corbyn (Islington North), Emily Thornberry (Islington South), Meg Hiller (Hackney South), Diane Abbott (Hackney North). Corbyn and Thornberry would naturally fit into the two successor seats in the Islington area. But odds are Hiller and Abbott would find themselves pitted against each other in the surviving Hackney seat. Unless one of them agrees to move to the new and so far orphaned seat stitched together from bits and pieces of Bethnal Green and the two current Hackney seats.

The Rest of England

Now's when the chickens come home to where grass is greener. If you have read through the Three Celtic Nations and London and kept tabs then you know the body count so far is: Tories -6, Labour -8, LibDems -3. Not quite yet what's in the big picture we saw at the beginning though the LibDem Culling is already quite well underway. Now it's Labour's turn on the battlegrounds of Little Egland. Just watch the regional summary of seats that will go down there:

Conservatives managing to gain seats in two English regions is quite a feat when all regions are bound to lose seats. With the proverbial 20-20 hindsight LibDems must now regret having supported the 2013 Bill as part of the Coalition agreement. Probably they did not expect their constituencies to be butchered in such an extensive and creative way, reducing them again to fringe status with fewer seats notionally than the DUP and level with Sinn Féin. Too late to whine though.

Dressed in the skin the Wolf strolled into the pasture with the Sheep

Soon a little Lamb was following him about and was quickly led away to slaughter

In the East Midlands two great classics are at work. Pack Labour votes from two neighbouring seats into a safer Labour new one so that the other turns Tory (see Derby and how the new East-West split instead of the current North-South one strengthens Chris Williamson). Or add enough Tory-leaning rural territory to a marginal urban Labour seat to switch it (see the new Lincoln and North Hykeham). Of course the combination of both is a sure winner and that's how Margaret Beckett's 24% lead in the old Derby South turns into a 4% Tory lead in the new Derby East.

It's even easier in the South East. Allocating two seats instead of one to the 'special status' Isle of Wight automatically creates one more safe Tory seat. Then creative recarving and restitching around some large cities (Brighton, Canterbury, Oxford, Portsmouth, Southampton) does the rest with four Labour and one LibDem seat notionally switching to the Conservatives. Added bonus is that the Brighton recarving both kicks Labour out of Brighton Kemptown and makes Caroline Lucas' seat Brighton Pavilion competitive and a possible target for Labour. Pitting opponents against each other is always a good idea, isn't it?

Incidentally and certainly not coincidentally one of the few unchanged seats is Amber Rudd's Hastings and Rye. Which leaves her with a majority of only 349 votes and a very obvious first tier target for Labour in any future GE.

A case for electoral reform?

The consequences of the boundary changes might offer an opportunity to revive the call for electoral reform, i.e. changing the electoral law and scrapping FPTP. Last time it was put to a vote was in 2011, remarkable because it failed by a wide margin and Labour sat on the fence as they were unable to agree on which side to stand. Labstain already a trademark. The choice then was between FPTP and the so called 'alternative vote' more commonly known as Instant Runoff Voting. Which is absolutely not a variant of proportional representation as some might have believed at the time, but basically just a multiple round variant of FPTP, the big difference being the winner is elected on a majority and not a plurality of the popular vote. Might have been an improvement on basic FPTP though.

As you already know I do not support Proportional Representation (PR) but a straightforward Mixed Member System (MMS) similar to what I suggested for Scottish Parliament elections. Which of course will never happen as no political party will ever support it. Anyway I made a simulation of what PR and MMS would have delivered based on the actual 2017 vote, on both the current 650 seats and the proposed 600 seats.

Similar to my outline for Scotland the proportional part of MMS and the full PR are based on separate national lists for the Celtic Nations, and regional lists within England. On current boundaries neither option would improve Labour's result as their number of seats on FPTP is already roughly what they would get on PR. In all cases the main 'victims' of the change would be the Conservatives and the SNP as they are the beneficiaries of the biggest 'FPTP bonus'.

As can expected the main beneficiaries of any change would be the LibDems. Just bear in mind that they would need about 25% of the popular vote on current voting patterns and boundaries to get close to 50 FPTP seats. They do less well on MMS simply because the FPTP component still favours the two main parties in England and Wales, and the SNP in Scotland. Both UKIP and Greens would bag some seats though nothing spectacular because of their low vote shares at the 2017 GE.

Interestingly such changes in the electoral law could revive a Con-Lib coalition which would get a workable minority on PR and 650 seats, and a majority on all other options as long as Sinn Féin don't take their seats. Labour couldn't counter this even with the unlikely Lab-Lib-Green coalition that would anyway fall short of a majority except on 650-seat PR. Just sayin'.

It's not over 'til it's over 🔊

So there is enough circumstantial evidence to conclude the proposed boundaries are definitely a pro-Tory gerrymander, mostly concentrated on England outside London. Though with some bizarre traits. What is the incentive for crippling LibDems even more than Labour? Why go soft on the SNP? Guess we'll never know. At this point it's fun to compare what we have now with what the original Sixth Periodic Review would have delivered, based on the 2013 final proposals versus the actual 2010 GE results. 

There are also signs of some gerrymandering but to a much lesser extent than with the 2018 proposals and the visible transfers from Labour to the Conservatives could have been brushed aside as just a stronger by-product of the 'FPTP bonus'. Obviously back then LibDem Culling was out of the question thanks to their Protected Species Status as the Coalition's Junior Partner. Just note that the Conservatives on just 50% of seats would in fact have had a majority (300 seats out of 595) with Sinn Féin sitting out. Which was probably just a safety net in case relations turned sour within the Coalition.

Now what might happen next with the 2018 Review is as usual anybody's guess. Tories and DUP have a vested interest in seeing the revised boundaries approved by Parliament as soon as possible and all it takes is an Order in Council as the core legislation has been passed long ago. The Order could probably be drafted in a matter of days. If I understand the conventions correctly all is required then is that 'each House of Parliament by resolution approves the draft' Order, meaning no debate or division as the principles behind the proposals have already been approved and the Order in Council is basically just a technical formality. So I expect the revised boundaries to be approved some time this year and take effect at the next GE.

The only alternative would be to scrap the whole process altogether as Parliament did in 2013. But there is surely no appetite for starting from scratch again, mostly because of the long delay and additional costs involved. So brace yourselves for the next GE being fought on the final 2018 proposals.

In the meantime don't forget to stockpile all the basics including pet food. And keep the cooshite detector handy.

Saor Alba Gu Bràth