Wales Special: Snooping on the Senedd Reform

As something a little bit different, Ballot Box Scotland is taking a look at some interesting developments in Wales. Whilst Scotland has been mired in constitutional deadlock for the better part of a decade now, the Welsh have been engaged in some fascinating constitutional discussions, and getting rather more support across the constitutional divide whilst doing so.

As part of a general policy of co-operation in this Senedd term, the Labour Welsh Government and Plaid Cymru have been pushing forward some fundamental reforms to the Senedd itself. Meanwhile the Independent Commission on the Constitutional Future of Wales, appointed in 2021 following the deal between the parties, published their final report last month (January 2024), finding that all of Devolution, Federalism and Independence were viable options for Wales. For those of us in Scotland, this may look a bit bizarre.

A Labour government appointing a constitutional commission and allowing them to investigate Independence, rather than insisting the concept be banned from discussion, and that commission saying “yeah, sure, it’s a workable option”? Can you imagine that happening in Scotland? I certainly can’t, and it didn’t either pre-Devolution or during the rapidly-overtaken Calman Commission. You may likewise recall that the SNP refused to take part in the pre-devolution Constitutional Convention precisely because it wouldn’t consider Independence. Our parties have often been reluctant to engage with anything bar their own preferred outcome.

Perhaps it’s simply that Welsh Independence is seen as less likely and thus less politically defining than Scottish Independence, or that Plaid Cymru have typically proven a weaker force, but that’s a political maturity sadly absent up here. We should be able to disagree on the constitution whilst still engaging in useful, meaningful and fruitful discussion about the various options, and perhaps if we did we’d be further towards settling the issue one way or another.

I still haven’t gotten around to reading that report in full, and as you can see by the read time indicator at the top of this piece, we’re already in Long Read territory. I’m going to focus on the proposed reforms of the Senedd in the Senedd Cymru (Members and Elections) Bill. At the core of this bill is a plan to expand the Senedd and overhaul the voting system, though it’ll also effect a return to four-yearly elections as standard rather than five.

Expanding the Senedd

More Members, More Money

If you’ve heard anything about the reforms this is probably it, with a 60% jump from 60 to 96 Members of the Senedd (MSs). In a country where democracy is often victim to a “cost of everything, value of nothing” mentality, the cost of more politicians has proven controversial. The Conservatives in particular have been very upset, which made it to the Commons last week when Penny Mordaunt pointed out that the Commons “would have to accommodate 2,058 Members of Parliament” were it to have the same constituent-to-representative ratio. This was, frankly, a very stupid thing to say.

The Senedd already has a higher constituent-to-representative ratio, with 60 seats versus the 40 current Welsh MPs and 32 at the next general election. That’d be a not much less shocking 1,286 MPs if replicated in the Commons! But more local decision making bodies always have more representatives relative to the population they are representing. Mordaunt is one of two MPs for Portsmouth, is she next going to suggest the City Council should drop from 42 to just two councillors? I’d certainly hope not!

That said it’s understandable why the public dislikes the idea of more politicians. Nobody likes politicians very much, especially not these days. Your average voter’s ideal world has rather fewer politicians in it, thank you very much, not more. Yet politicians are a necessary component of representative democracy, the only nationally practical form of democracy.

Two Decades of Change

The Senedd needs more members simply because it is doing more – a lot more – than it originally did. You may assume that Devolution is the same for everyone, differing only in what subjects are devolved. In reality, Scottish and Welsh Devolution started out with a vast gulf in fundamental structures. This was as wide as the gulf in popular support in the 1997 referendums, nearly three-quarters (74.3%) in Scotland but barely over half (50.3%) in Wales.

The Scottish Parliament has always been able to pass primary legislation, or Acts of Parliament. The National Assembly for Wales could only pass secondary legislation, tweaking the Welsh implementation of Acts passed at Westminster. Devolved powers in Scotland were defined expansively via “Reserved Powers” listing what the Scottish Parliament didn’t have power over, automatically giving it power over everything else. In Wales, powers were defined restrictively via “Conferred Powers” listing what the Assembly did have power over, automatically keeping everything else with Wesmtinster.

Finally, in Scotland we had a proper Executive (now Government) legally constituting a separate body from the legislature, just like the UK Government and Parliament. Wales didn’t, instead having the “Executive Committee of the National Assembly for Wales.” This was headed up by a First Secretary, not a First Minister, who was thus a glorified committee convener.

All of this has changed radically in the years since. In 2006 the Assembly was given the ability to pass “Measures”, a weaker form of primary legislation than an Act. At the same time they got a true First Minister heading up a legally distinct executive arm. A second referendum in 2011 gave a nearly two-thirds favourable vote for the Assembly to assume full primary law making powers in its areas of control.

A series of further commissions and following Acts of the UK Parliament have further deepened devolution. Powers over taxation, the shift to a Scotland-style reserved powers model, and the ability to control its own form have also since been granted, leading to the National Assembly for Wales formally becoming the Senedd Cymru (or Welsh Parliament) in 2020. Given how much more the Senedd of the 2020’s has to do compared to the National Assembly of 1999, more members is an entirely reasonable idea.

Overhauling the Voting System

Six, Six, Six, the Number is the Best

The other major component of the proposal is a complete overhaul in how those MSs are elected. The Senedd currently uses the same system as the Scottish Parliament, the Additional Member System. AMS combines UK-traditional First Past the Post constituencies with a regional layer of members elected proportionally, aiming to balance local representation with fair representation.

This isn’t a bad system relative to pure FPTP but it’s not perfect, not least because any element of FPTP can impact proportionality. This is especially true in Wales, where the balance of constituency seats (67%) is a fair bit more than in Scotland (57%). This makes Wales especially vulnerable to “overhang”, when a party wins more constituency seats than its fair share of seats in the region overall, thereby depriving other parties of their fair share.

In place of FPTP a purely proportional system is being proposed. Six MSs are to be elected from each constituency through closed list PR using the familiar D’Hondt formula. The positive aspect of this proposed system is that it completely removes FPTP from Senedd elections. Long-standing followers of BBS will know that I have a deep-seated loathing of FPTP.

The very principle of saying only one group of voters in a given area should have a voice in an elected body is unrepresentative, which means it’s unfair, which means it’s undemocratic. Anything that weakens the vicelike grip that FPTP has over the UK’s politics is a step forward in my opinion, and a step towards a hoped for future where we don’t use it for anything. Unfortunately however, my list of negatives is a fair bit longer.

It's Not Properly Proportional

Firstly, six seats per constituency simply isn’t proportional enough. The more you have of something, the easier it is to split into smaller portions. 10% of six seats is 0.6, and you obviously can’t get 0.6% of an MS, though as Scottish results for the European Parliament have shown in a highly fractured voting space you can just about squeak a seat in those circumstances. In general though, this will be a significant barrier to smaller parties winning seats in the Senedd.

The collective 12.5% of the vote won by the Greens (4.4%), Lib Dems (4.3%) and Abolish the Welsh Assembly (3.7%) in 2021 would ideally have entitled them to eight seats, three apiece for the Greens and Lib Dems and two for Abolish. In reality, AMS meant the Lib Dems elected one MS and that was it. Some might argue that in many countries, such as Germany, Czechia and Lithuania, those parties would be out on their backside for missing a 5% threshold anyway. Even the more generous 4% of Norway, Sweden and Slovenia would bar Abolish, and hey, they hate the place so much they refused to even update their name, so maybe that’s fair?

Obviously, I don’t think so. My own work on BBS advocating for electoral reform uses a 3% threshold, which is itself a recommendation from the Parliamentary Assembly of the Council of Europe. Sticking to Wales and this proposed system, whilst we don’t have final boundaries that anyone has been able to model from, Make Votes Matter came up with a suggestion here that gives an indication of a possible 2021 result under the new system.

This does show a weakening of Labour’s position relative to AMS, exactly what a more proportional system should do. Note though that it still only has one Lib Dem and that’s it from the smaller parties, even though the proportional entitlement for all three would grow to 12 out of 96. We don’t need to know the new boundaries to be sure this is side of things is pretty spot on.

With the Greens peaking at 8.2% and Abolish at 7.0% in their best constituencies, there’s nowhere they’d be into the double digits needed to pick anything up with only six seats per constituency. The now-known final shape of Cardiff constituencies almost certainly carves up Cardiff Central too much for the Lib Dems’ 15.9% there to combine with another (all 5% or less) in a way that’d give a second seat. 

When I said this on Twitter a while back I did get a proper academic “Well, Actually”-ing from someone pointing out that the literature suggests after about 5 seats, constituency size doesn’t significantly impact on the number of parties elected. That may be true on the balance of academia, but I’m interested in practicalities. Practically speaking, we know this system wouldn’t have done anything to fix under-representation in 2021.

Being represented also isn’t the same as being fairly represented. The Lib Dems and Greens won 11.6% and 2.7% respectively in the 2019 UK general election and both won representation, but at 1.7% and 0.2% of seats it was pretty paltry. At best this represents good intentions with shoddy implementation, but at worst it’s a deliberate stitch-up by two larger parties (Labour and Plaid) keen to burnish their democratic credentials without actually risking their position.

I regret to inform Single Transferrable Vote fans that basically all of this applies to STV too, which delivered similarly disproportionate results in councils across Scotland in 2022. For example, see how there were no Conservatives elected in West Dunbartonshire despite 9% of votes and only two out of 85 seats (2.4%) in Glasgow despite getting 10%.

Similarly, the Scottish Greens failed to win any seats despite vote shares that would justify that under most other forms of PR in Midlothian (8%), Dundee (6%) and Aberdeen (5%). No, it’s not made okay because these voters got to transfer their vote, not least because STV would still guarantee a minimum of 14.3% of votes in a six member constituency would not be “useful”. A first preference is a first preference for a reason, and that is what we should represent insofar as is possible.

It Keeps the Lists Closed

Just like Scotland, Wales currently uses “Closed” lists, meaning the order of candidates is entirely chosen by the parties themselves. This will continue under the new system, despite pretty much everyone else recommending what in Wales seems to be termed “Flexible Lists”. I’m not sure why they’ve had to come up with a special Welsh name for what is otherwise known as an “Open List”, but they have. Regardless, open list systems give voters the most choice over candidates within parties of any voting system. Despite frequent claims to the contrary, neither FPTP nor STV can compare.

FPTP comes with the very simple problem that each party stands one candidate. We know fine well the majority of voters are voting on the basis of party, meaning they either vote for that party and its own candidate or neither. STV hardly fares any better because the optimal approach to STV is to stand as many candidates as you think you can get elected. In many cases that’s one candidate, presenting the same dilemma as FPTP. Even where it’s multiple candidates, voter power to meaningfully choose between them is limited by the fact not giving a preference to one or more candidates from a party risks reducing that party’s overall representation.

It's Geographically Inflexible

Using equally sized constituencies guarantees that some of those constituencies will look awkward and unnatural. No human geography cleanly splits into roughly evenly sized chunks whilst still maintaining clear and well understood local connections. Keeping the current Senedd regions at roughly equal sizes (from 11 to 13 seats overall) has meant that the Gwynedd local authority is split between two regions. In Scotland (from 15 to 17 seats) we have a Lothian region which is missing most of East Lothian and a Glasgow region which extends outside the city, which is part of why South Lanarkshire splits between three regions.

Equally sized constituencies also require regular redrawing of boundaries in order to account for a shifting population. This further adds to the inconsistency and confusion, as places that may have been part of the same constituency for many years end up separated at future elections.

It Starts Out Tied to Westminster

In order to get these reforms up in place for the 2026 election, Senedd constituencies will simply be Westminster constituencies stuck together. There are 32 new Westminster constituencies which will be paired to create 16 Senedd constituencies. That means taking the problems with awkward, overly artificial boundaries outlined above and making it even worse by using FPTP constituencies drawn to a very strict electorate limit.

This aspect is at least temporary. The bill ensures that boundaries are properly redrawn to the Welsh electorate and with a wider 10% variance from quota ahead of the 2030 election. Although that’s the right approach if moving forward with this system, if you’re going to end up with lack of alignment between the Senedd and Westminster anyway, you might as well try and ensure the Senedd aligns with something instead. And if you are wanting to avoid disruption, why not get the constituencies right now rather than use three different sets in three successive Senedd elections?

Preserved County Proportional Representation

For my criticisms of proportionality, boundaries, and the tie to Westminster, there is one solution: Preserved Counties. These are purely ceremonial modern iterations of the eight counties local government counties that existed from the mid-70’s to the mid-90’s. There have been slight tweaks to account for changes to the modern local authority boundaries, so every council is entirely within one Preserved County.

As there are eight of these, versus 16 proposed constituencies, it doubles the average size of a constituency to 12 seats. By coincidence that’s the same average as current Senedd regions, albeit entirely proportionally elected rather than distorted by two-thirds of seats being elected through FPTP. That alone would significantly improve proportionality.

They wouldn’t all be 12 seats due to population divergence. This is one of the inbuilt advantages of PR though, and one that I find it odd the Welsh Government have refused to work with. The ratio of voters to representatives matters, not meeting an arbitrary number of representatives. A differing number of seats per constituency means you can be more flexible in shape than you can when you insist they must all be the same size. In this case that means using existing, and thus comparatively more “natural”, boundaries rather than merely replicating bizarre FPTP creations to start with.

It also means instead of regular boundary reviews, all that needs to change is the number of seats per constituency. Given that the argument in favour of using the existing Westminster seats for the 2026 election is convenience and minimising disruption, I’d argue that the Preserved Counties serve exactly the same purpose, following as they do already familiar boundaries.

Seats per Preserved County

To see what this might look like, we first have to work out the number of seats per Preserved County. For simplicity’s sake I’ve used population rather than electorate. This differs from the actual initial basis (Westminster electorate) and proposed future basis (Senedd electorate, which includes votes at 16 and non-citizens resident in Wales), but is fine for an example. As with my Scottish equivalents, I’ve used the Largest Remainder method to apportion seats between constituencies.

This comes out with a very neat distribution where every constituency has an even number of seats. Most constituencies are at or above the average size, with the three largest each having 16 seats. There are then two much smaller constituencies, Powys especially notable for only being entitled to four seats. In my Scottish reform proposals I do include a slight boost for the most rural parts of Scotland, and you could definitely argue me up to six seats for Powys (entirely for rurality) and eight for Gwynedd (in part to reflect its status as the Welsh language heartland).

Votes per Preserved County

Next, we need the votes in each constituency. Again I’m taking a cue from my Scottish reform examples here in specifically using the regional vote and ignoring the constituency vote, on the basis this is the proportional vote that best reflects true voting intention. Some voters certainly would vote differently if this was their sole vote versus the current two vote model, but it’s the best proxy we have at the moment.

Note there there is one break in Preserved County lines at the moment as Merthyr Tydfil and Rhymney covers parts of both Mid Glamorgan and Gwent. As the balance of the population lies with Merthyr Tydfil I’ve counted it fully as part of Mid Glamorgan here. In reality there would be a slight difference in the votes for those counties due to a correct boundary.

With boundaries set and votes carefully tallied up, we can get down to the business of working out results. In true BBS style, I’ve not just done one voting system using these, but a few variations using slightly different approaches to demonstrate the importance of how exactly you do proportionality. Unlike my Scottish models, I’ve simply ordered my “seat bubbles” in the graphics by party strength in each constituency rather than showing them in order of election, to save myself hassle.

Basic D'Hondt - Gallagher Index 6.6

This simply uses the D’Hondt formula, which is known for favouring larger parties. That’s why although there is significantly better representation for the smaller parties under these boundaries than with six member constituencies, they still come up short. Both the Greens and Lib Dems still only have about half their fair share of seats, whilst Abolish end up with about one-third to one-quarter. All three of the big parties benefit as a result, though Labour the most.

Basic Sainte-Laguë - Gallagher Index 3.6

Here, we use the Sainte-Laguë formula instead of D’Hondt. This doesn’t give a radically different outcome, but it does give an edge to smaller parties. As such, all three of them end up on 4 seats here. This is roughly their fair share, but it should be noted that sometimes if a party gets their vote shares just right, Sainte-Laguë can actually go too far in the opposite direction and over-represent smaller parties.

The fact each constituency is still entirely self-contained also preserves a key element of disproportionality. Small disproportionalities in one constituency can sum up to big disproportionalities nationally, sometimes to the benefit of smaller parties as indicated above. However on the whole it helps the largest parties still, as demonstrated here by the fact Labour are still seeing the biggest seat share edge over their vote share, whilst Plaid Cymru get no advantage.

Nationally Allocated Sainte-Laguë - Gallagher Index 2.9

The final model effectively follows my preferred Scandinavian-style system, though I’ve used “national allocation” here so it’s clear how the final tally is arrived at. Most seats in each constituency are still elected directly using only the votes in that constituency, again using Sainte-Laguë. However a small proportion are held back to be allocated using the national vote share for all parties crossing a 3% threshold. This just follows Sainte-Laguë again to determine who gets each seat, allocating it to the constituency with seats left to fill where that party was closest to another seat.

This has the advantage of being maximally proportional for parties above the threshold, which is why this gives Labour the fewest seats. The oddity here is that sometimes the national allocation process jars with local results, as you can see two constituencies here where the Conservatives end up with more seats than Labour despite fewer votes. That is equally possible under First Past the Post, where a party can win more seats with fewer votes so long as all of its wins are narrow and their opponent’s wins have large majorities.

Open Lists

One final tweak that could apply to any model is to use open lists, giving voters more control over who is elected in each constituency, rather than leaving it entirely to parties. I recognise that there is a balance needed here, with fully open lists potentially vulnerable to deliberate or unconscious biases that may disadvantage women, young people, or people from ethnic minority backgrounds.

I’d suggest following something like the Dutch model, where parties set the initial order of candidates, but any candidates who receive personal votes equivalent to at least one quarter of a seat in that constituency move up the order accordingly. I’ve come to quite like the elegance of the Latvian approach to open lists, where for the party they’re voting for, voters put a little tick next to any candidates they especially like and cross out the names of those they don’t. This would also allow candidates to be moved down the list, using the same threshold.

Comparisons and Conclusions

That’s a lot of numbers to have waded through, I know, so here’s how these models stack up against one another and the actual result of the 2021 Senedd election:

That’s a really clear visualisation of how much more representative any form of Preserved County Proportional Representation would be, but especially the two versions rooted in the Sainte-Laguë method. A truly fair democracy requires as many voters as possible to be meaningfully represented in our decision-making bodies, and these proposals go further towards that goal than the Welsh Government’s bill does.

As much as I’ve been critical of the detail, whilst enjoying the opportunity to truly nerd out about it, on the whole I definitely think this is a step in the right direction for the Senedd. It clearly needs to be significantly expanded to most effectively reflect the growth in its powers, and abandoning FPTP entirely is very welcome. If done just a little bit differently though, it has the potential to make the Senedd the most democratically representative legislature anywhere in the UK. Wouldn’t that be a nice claim to fame, and something to brag that you’ve done better than the rest of us?

If you find this or other Ballot Box Scotland output useful and/or interesting, and you can afford to do so, please consider donating to support my work. I love doing this, but it’s a one-man project and takes a lot of time and effort. All donations, no matter how small, are greatly appreciated and extremely helpful.
(About Donations)