Hefyd ar gael yn: Cymraeg

Reshaping the Senedd

Electoral Reform Society,

Posted on the 1st October 2016

About Us

Wales Governance Centre at Cardiff University

The Wales Governance Centre is a Cardiff University research centre undertaking innovative research into all aspects of the law, politics, government and political economy of Wales, as well the wider UK and European contexts of territorial governance. A key objective of the Centre is to facilitate and encourage informed public debate of key developments in Welsh governance not only through its research, but also through events and postgraduate teaching.

Electoral Reform Society Cymru

Electoral Reform Society Cymru is an independent campaigning organisation working to champion the rights of voters and build a better democracy in Wales. We offer an independent voice, and work to shape the democratic debate at all levels. We put the interests of the citizens within our democracy first.

We believe:

  • Every vote and every voice has value and should be heard
  • Everyone should be able to shape the decisions that affect their lives – Our institutions should refect the people they serve
  • People should be able to hold those in power to account
  • Politics should offer people real alternatives

For more information, or to join the Electoral Reform Society Cymru, please visit https://www.electoral-reform.org.uk/ers-cymru/



In 2013, the UK’s Changing Union project, in collaboration with the Electoral Reform Society Cymru, produced a report, Size Matters: Making the National Assembly More Effective, which made a reasoned case for increasing the number of Assembly Members (AMs). To bring it into line with the capacity of other comparable legislatures, the report suggested, the National Assembly for Wales should have around 100 members.

Since 2013, there has been further devolution of taxation powers with the Wales Act (2014); yet more change is on the horizon with the forthcoming Wales Bill, and the implications of Brexit on devolution. These developments strengthen further the core argument of Size Matters: as more powers accrue to Wales it is necessary to ensure that there are enough Assembly Members to scrutinise a powerful Welsh Executive.

But one question which our previous report left entirely unaddressed was how a
larger Assembly might be elected. Increasing the number of AMs would require some adjustment in current electoral arrangements – but how, and on what basis, should this be done? This is the subject of our new report.

We do not, in Reshaping the Senedd, seek to promote one specific pet scheme for electing a larger Assembly. Rather, we seek to promote debate about the following:

  • The principles on which any reform should be based;
  • The key political realities with which any proposals for reform will have to deal; and
  • The extent to which alternative potential electoral reforms might or might not be consistent with those principles and realities

We hope that Reshaping the Senedd can make an important contribution to public debate about these matters.

November 2016

Wales Governance Centre at Cardiff University

Professor Roger Scully
Professor Richard Wyn Jones

ERS Cymru

Dr Owain Llyr ap Gareth



In a previous report, Size Matters, we argued strongly that the membership of the National Assembly for Wales should be substantially increased – preferably to around 100 AMs.1

One thing we did not do in that report was make recommendations as to how this larger Assembly should be elected. Increasing the chamber’s size would require some change to the electoral system. But change to electoral arrangements in Wales is
now very likely anyway. Implementation of the ‘reduce and equalize’ legislation for
the House of Commons, due to occur before the 2020 general election, means that Wales’ Westminster representation will likely be reduced from 40 MPs to 29. And with the number of Westminster constituencies going down, some change to how we elect members to the Assembly would be inevitable – even if it was only the minimal change of abandoning the ‘coterminosity’ of Westminster and Assembly constituencies (that is, having identical constituencies for both institutions’ elections).

The Wales Bill intends to transfer powers over the electoral arrangements for devolved elections to the Assembly. However, the Bill also sets a high ‘supermajority’ threshold for agreement, requiring the support of two-thirds of members. This means that there will be an opportunity for politicians in Wales to shape how our National Assembly is elected. But there is also an obvious question. Will it be possible to achieve suficient cross-party agreement on a method of electing a larger Assembly?

This is the question that Reshaping the Senedd seeks to address by offering a framework based around rst principles about what an electoral system seeks to do. We do not
set out our own pet scheme for electoral reform. Rather, in an area where party self- interest can produce considerable heat but precious little light, we attempt to introduce principles and evidence. We do so in a way that seeks to take account of the political realities and the particular context in which the National Assembly for Wales operates. While one of the partners producing this report, ERS Cymru advocates the Single Transferable Vote, our aim here is to provide policy makers with options and a framework within which to come to a judgement – to provide evidence and promote reasoned debate, rather than to be overly prescriptive. In providing the basis from which to work towards agreement, we hope to prompt a process of negotiation and consensus-building towards an effective, workable and sustainable solution.

Speci cally, in Reshaping the Senedd we do three things:

  • First, we set out several basic principles that we think should guide debate on electoral reform in Wales.
  • Second, we show that several major electoral systems clearly cannot satisfy these principles, and therefore do not offer a satisfactory way forward.
  • Finally, we outline three potential routes to electing a larger Assembly, and present estimates as to the electoral outcomes that these systems might have produced if they had been used in the 2016 Assembly election.

Key Principles for a Good Electoral System


Debates around the rules of the political game should be different in character to everyday partisan politics.

Though political parties inevitably and unavoidably have a deep interest in how an electoral system is constructed, members of all parties and of none have a common interest in developing a system based on principles that all can recognise and respect. That is the process we are seeking to advance here.

Our analysis is based around seven key principles. These principles, we believe, provide the basis for a stable and lasting electoral system for the National Assembly for Wales. Prospective voting systems will be tested against these principles:

  1. PROPORTIONALITY: A new electoral system should be likely to produce outcomes no less proportional than those produced by the present system, and ideally more proportional than the current system.
  2. SIMPLICITY AND COTERMINOSITY: Electoral boundaries for the National Assembly should, as far as possible, be coterminous with others (such as mirroring Westminster boundaries). This makes things simpler for both voters and parties.
  3. SUSTAINABILITY AND STABILITY: Any reformed electoral system should be sustainable. It should not need to change again fundamentally in the near future. This may require that it is also flexible and adaptable to minor changes as necessary.
  4. BROAD-BASED CONSENSUS: Decisions on the electoral ‘rules of the game’ should always be based on as broad a consensus as possible; there should be checks against individual parties being able to change electoral systems for partisan gain.
  5. STRONG AND EQUAL MANDATES: All Assembly Members should have clear and equal mandates; if mandates are different there should be no sense of some representatives being ‘second-class’ AMs.
  6. REPRESENTATIVENESS: Insofar as possible, the electoral system should produce a body of representatives who reflect the electorate, in terms of race, gender, disability, religion, age, social class and diversity of opinions.
  7. SUBSTANTIAL SUPPORT: Election to the National Assembly should require a substantial level of public support; the effective threshold for election should reflect this.

We acknowledge that balancing these key principles is precisely that – a question of balance. There are overlaps and tensions, and different and legitimate viewpoints will put more weight on some principles than others. Nevertheless, it is our contention that these principles constitute the basis around which future electoral arrangements of the National Assembly should be constructed.

We will now examine each of these principles in more detail.


A new electoral system should be likely to produce outcomes no less proportional than those produced by the present system, and ideally more proportional than the current system.

The more a party’s share of seats reflects its share of the vote, the more proportional an electoral system is. The system currently used in Assembly elections is generally termed the Additional Member System (AMS). AMS is a hybrid voting system: it combines elements of First Past the Post (FPTP) where voters choose a candidate to represent their constituency, and party-list Proportional Representation. List seats are allocated to parties in a way that partially compensates for the disproportionality associated with First Past the Post elections.

How AMS works in Wales is different from Scotland. In Wales 40 AMs are elected through the constituency seats, and 20 on the list. In Scotland, 73 are elected through the constituency seats, and 56 from the party regional list. Having only one-third of members allocated proportionally via the list is a relatively low percentage and means that the Welsh version of AMS is inherently less proportional than that used in Scotland or in most other countries and regions that use the AMS system. At least half of all members of the German Bundestag are normally elected via the list,2 and 42 per cent of New Zealand MPs. However, AMS as used in Wales is substantially more proportional than the results generally produced by FPTP for general elections in Wales. The system might thus most appropriately be described as semi-proportional representation.

Given that the current AMS system was an important part of the devolution settlement endorsed by the Welsh people in the 1997 referendum, any new system should be likely to produce outcomes no less proportional than the current one; ideally, we believe that it should be significantly more proportional, although we recognise that different political parties will have a direct interest in the system being either more or less proportional.

Simplicity and Coterminosity

Electoral boundaries for the NAW should, as far as possible, be coterminous with others (such as mirroring Westminster boundaries). This makes things simpler for both voters and parties.

Any voting system should be as simple as possible for voters to use and not generate any unnecessary confusion.

Electoral boundaries that mirror others (or, ‘coterminosity’ as it is termed) are preferable as they make things simpler for both voters and parties. Therefore, using either Westminster boundaries or local government units as the basic geographic unit of representation is preferable to having differently-sized units for local, devolved and Westminster elections (as now happens in Scotland).

Losing coterminosity may also matter in terms of public service delivery, as having different boundaries adds to complexity there as well. The Williams Commission on Public Service Delivery noted that Wales’ public services map is very complex, and this has the effect of making citizens’ rights and entitlements unclear, ‘requiring them to understand and navigate complex overlapping responsibilities to access the services they need.’ With such a complex public services map, AMs and MPs often serve as the first point of contact for their electorate and as a guide and advocate through the system. Separating the Westminster and Assembly boundaries from each other would likely lead to yet greater difficulty and less clarity in this regard, particularly as the public often have very limited knowledge of what is and is not devolved.

We would not expect decoupling the link in UK and Welsh Assembly constituencies to have as dramatic an effect as the advent of a new devolved institution and polity. However, it will have an effect on parties’ organisation, how party members’ organise local electoral campaigning, and raise issues about which level of governance should have primacy within the party organisation.

Currently, whether Westminster boundaries will change remains somewhat uncertain. Boundary changes are expected to be implemented before the 2020 election (in line with the Parliamentary Voting System and Constituencies Act 2011); this would reduce the number of Westminster MPs from 650 to 600. Wales has long been over-represented (due to generally having constituencies with smaller electorates than the average) and will therefore lose proportionately more MPs than the rest of the UK. The number of Welsh MPs is therefore scheduled to be reduced from 40 to 29.

National Assembly constituencies are at present identical to Westminster ones, so either the two levels will have to be decoupled or there will have to be some change to the Assembly electoral system anyway. Moreover, equalising Westminster constituencies through a strict formula based on the electoral roll for all future boundary reviews

would make the Assembly’s electoral arrangements vulnerable to instability caused by demographic changes – and not necessarily even changes only in Wales. Demographic changes, or large increase in the electoral roll in one part of the UK will impact on
the constituency boundaries of other parts of the UK. As well as volatile levels of representation, the new boundary rules are also likely to promote volatile boundaries themselves, with constituencies liable to radical change at each successive election.

With Westminster parliamentary constituencies likely to change, it would be prudent to have an electoral system exible enough to adapt to such changes.

Stability & Sustainability

Any reformed electoral system should be sustainable. It should not need to change again fundamentally in the near future.

Whatever electoral system is adopted, it should be sustainable for the long term. There should be sufficient Assembly Members to adequately scrutinise the Welsh Government, even if the Assembly and Government’s powers are widened or deepened by further developments in the Welsh devolution settlement, beyond those anticipated in the current Wales Bill. An electoral system should also be immune to, or able to adapt to, other plausible changes (such as changes in the number of Westminster constituencies over time).

Broad-Based Consensus

Decisions on the electoral ‘rules of the game’ should always be based on as broad a consensus as possible; there should be checks against individual parties being able to change electoral systems for partisan gain.

As stated earlier, debates around the rules of the political game should be different in character to everyday partisan politics. To sustain public confidence in the integrity of democratic politics it is vital that the electoral system should be protected from the possibility – or even the perception – of manipulation for partisan gain.

Both the St David’s Day Command Paper and the Wales Bill provide for a substantial ‘supermajority requirement’: while powers over electoral arrangements for the Assembly in Wales are to be devolved, changes must gain the support of “at least two-thirds of the total number of Assembly seats”. We strongly support this safeguard. But we also note its implications for our recommendations here. Any recommended electoral system for the National Assembly must be able to surmount the two-thirds threshold within the Assembly. This will make a broad-based consensus not merely desirable but actually essential. We believe that adopting the principles outlined here as building blocks to consensus will make this process easier and more attainable.

Strong and Equal Mandates

All Assembly Members should have clear and equal mandates; if mandates are different there should be no sense of some representatives being ‘second-class’ AMs.

AMs’ mandates should be clear to the electorate: it should be readily apparent on what basis someone is elected, and for what geographical area. All AMs should be seen as equal in terms of mandate. There should be no sense that some members are ‘second- class AMs’, and all should have broadly equal expectations in terms of responsibilities in their role as AMs. If members’ mandates are different, it is important that they are still of equal status.


Insofar as possible, the electoral system should produce a body of representatives who reflect the electorate, in terms of race, gender, disability, religion, age, social class and diversity of opinions.

Proportionality of representation is usually evaluated in terms of fairness between political parties. But an electoral system should also be concerned with other aspects of the representation of a society. An electoral system should facilitate, or at a minimum not inhibit, the election of a body of representatives who reflect the broader electorate in all its diversity of characteristics and opinions. The system should not be an additional barrier to members of traditionally under-represented groups seeking election.

Some types of electoral system are more amenable to parties producing a ‘balanced ticket’ than others; it is certainly easier when more than one candidate from a party stands in the same geographic area. This opens up space and potential for parties to take action to increase representation of under-represented groups, such as by age, gender, disability, and race.

Substantial Support

Election to the NAW should require a substantial level of public support; the effective threshold for election should reflect this.

While a strong democracy requires the inclusion of a diversity of voices, there are also potential negative consequences for a political system when elected representation ‘fractionalises’ into a large number of small parties. Among these consequences can
be that extremist parties, with limited public support, gain the legitimacy of an elected platform; that such parties may sometimes be in a strong bargaining position to influence government formation and policy; and that effective governments become more dif cult to form and sustain across the multitude of parties.

We therefore maintain that individual representatives, and political parties, should require the support of a substantial section of the electorate in order to gain representation to the National Assembly. In practice, some form of electoral ‘threshold’ should operate to ensure this is the case. This can be achieved in two ways:

  • Electoral Law: An electoral system can include in its design a specific level of support that a party must reach to win representation. For example, in Germany there is a national threshold of 5% in order for a party to gain representation in the Bundestag. Under some electoral systems, it would be possible to put in place such a threshold;
  • Effective Threshold: This is not a threshold set out in electoral law, but rather reflects that under some electoral systems a certain minimum level of support would be required to stand a chance of election. The fewer the number of members elected from the same geographic area, the higher the effective threshold generally will be.

Any new Assembly electoral system should balance a diversity of opinions and political standpoints with the avoidance of fractionalisation.

The principles outlined here offer a sound basis through which to evaluate systems that might be used in National Assembly elections. The following section will review some of the main available electoral systems, and identify several which do not plausibly satisfy all, or even most, of these seven criteria.

FAIL! Systems that do not pass the test


The seven principles outlined in Section 1 can be used to assess different electoral systems that might be used to elect an enlarged National Assembly for Wales. Here we outline several systems that clearly fail to pass this test.

Single member constituency voting systems


  • First Past the Post (FPTP)
  • Alternative Vote (AV) (which also encompasses Supplementary Vote (SV) in this analysis)

These systems all elect a single representative from a constituency. Under First Past the Post (FPTP) voters cast a single vote for their chosen candidate; the candidate with most votes wins. Under Alternative Vote (AV), voters rank candidates in order of preference. Candidates are elected outright if they gain more than half the votes as first preferences. If not, the bottom candidate is eliminated and their votes re-allocated to the second preference marked on the ballot papers. This process continues until one candidate has more half of the votes and is elected. Supplementary Vote (SV) is an abbreviated variant of AV. Under SV, there are two columns on the ballot paper – one for voters’ first choice and another for their second choice. The count follows a similar procedure to AV. For our purposes here SV and AV can be treated as similar.

What do these systems do well?


The main bene t of FPTP is simplicity. Voters are used to it, voting is easy, and voters know who represents them. AV is also fairly simple, with voters placing their candidates in order of preference.

Under these systems each member has a strong and equal mandate: each member is elected on the same basis to represent a distinct geographic area.

These systems also require a high threshold of public support for a candidate to be elected, with AV requiring a still higher threshold.

Where do they fail?

Proportionality: FPTP tends to produce distributions of seats that bear little relationship to the proportion of votes won by parties. In the 2016 Assembly election, the Labour party won 67.5 per cent of the constituency seats from 34.7 per cent of the vote. If FPTP had been used in devolved elections, Labour would have won strong to overwhelming majorities in every Assembly election, despite never approaching a majority of the vote.

A primary aim of increasing the number of AMs is to increase the National Assembly’s capacity to scrutinise the Welsh Government. But this aim would be neutered if that increase were accompanied by a change in electoral system that entrenched single-party dominance.

Using AV rather than FPTP would likely make little difference. Indeed, AV can sometimes produce even less proportional outcomes than FPTP.

Stability and Coterminosity: If, as is likely, the number of Welsh Westminster constituencies are to be subject to frequent change in future, single-member systems are problematic in terms of establishing a stable number of AMs. For sustainability and stability Assembly constituencies would probably have to be decoupled from Westminster ones. As noted in Section 1, this presents its own difficulties for voters and parties.

Broad-based consensus: While the two largest UK parties – Labour and Conservatives – have generally supported FPTP for Westminster, Conservative AMs have taken a pragmatic approach and supported PR for the National Assembly. It is unlikely that they would provide Labour with the two-thirds super-majority to change the system if that change would entrench Labour dominance to all other parties’ detriment. Both Plaid Cymru and Liberal Democrats support other systems, as do the Greens and UKIP. It is thus dif cult to envisage consensus on a single member system.

Representativeness: International experience has shown that single member systems provide less of an incentive for gender balance and diversity than multi member systems.9 With one candidate to choose there is more likelihood of selecting what is perceived as a ‘safe’ candidate; more than one candidate encourages a more ‘balanced ticket’. Despite this, Labour have achieved better gender-balance than other parties in the Assembly, even with its AMs coming largely from FPTP constituencies. The ways to achieve this were sometimes necessarily prescriptive (Twinning and All-Women Shortlists10) and prompted some internal party conflict. Other parties in the Assembly have to a greater or lesser extent used mechanisms regarding regional lists to help achieve greater equality.

Assessment: FPTP and AV are both unsuitable for electing a larger National Assembly. 

Multi-Member Majoritarian System


  • Two member First Past the Post
  • Multi member FPTP
  • Two member STV (often referred to as 2-member AV)

Two member and multi member FPTP work similarly to single member FPTP, but with voters putting an X next to more than one candidate, and more than one member elected to each constituency. Under two member STV, voters rank the candidates in order of preference. Candidates are elected outright if they gain a threshold of first preferences. If not, the bottom candidate is eliminated and their votes move to the second preference marked on the ballot papers. When one candidate is elected his/her surplus votes are also transferred. This process continues until two candidates reach the necessary threshold and are elected.

What do these systems do well?

As with single member systems, these systems are relatively simple for voters, and would provide a strong and generally equal mandate for each AM.

There is more scope to provide gender balance and diversity when parties select two or more candidates to a constituency. It provides an incentive for a balanced ticket (of a man and a woman for example). This incentive can be reinforced by party rules (that they must select a man and a woman) or even by legal quotas for parties.

These systems would require substantial support for candidates to be elected.

Where do these systems fail?


As with single member systems, these systems are relatively simple for voters, and would provide a strong and generally equal mandate for each AM.

There is more scope to provide gender balance and diversity when parties select two or more candidates to a constituency. It provides an incentive for a balanced ticket (of a man and a woman for example). This incentive can be reinforced by party rules (that they must select a man and a woman) or even by legal quotas for parties.12

These systems would require substantial support for candidates to be elected.

Where do these systems fail?

Proportionality: As with single member systems, these systems would likely be substantially less proportional than the current AMS system. In local government, it is
far from unknown for one party to win 100% of the representation from a multi-member ward on a minority (sometimes quite a narrow minority) of the vote. Repeated across different wards, this can make particular councils grossly imbalanced. There is little reason to expect a different outcome for the Assembly. If the 2016 Assembly election had been conducted on two member FPTP, with the current forty constituencies, our best estimate is that Labour would have won 54 of the 80 seats. If it had used three member FPTP, on the revised boundaries envisaged for 29 constituencies in Wales, then Labour would have won an estimated 57 of the 87 seats.

Under STV, introducing only one extra member per constituency provides so little proportionality it would make very little difference outside very marginal constituencies.

In terms of proportionality these systems are therefore, at best, every bit as unacceptable as single member systems.

Coterminosity and Stability: If coterminosity with Westminster boundaries is maintained, then two member representation would mean an Assembly with 58 members. This provides for no progress towards the goal of a larger Assembly – quite the reverse.

Simplicity: If one merit of single member system is that electors know their single representative, then two member systems clearly undermine that simplicity. The ‘sacred bond’ of the single member constituency link would be foregone without the balancing bene ts of more proportional multi-member systems.

Broad-based consensus: These systems are unlikely to enjoy broad cross party support.

Assessment: These systems are all unsuitable for electing a larger National Assembly.

Mixed Member Majoritarian (MMM)

Mixed Member Majoritarian (MMM) systems are similar in many respects to the current AMS/MMP system: members are elected both via FPTP in constituencies and party lists. The big difference is that, unlike with AMS, constituency results do not affect the allocation of list seats – they are essentially two parallel elections to the same chamber. List seats thus allow some representation for smaller parties but do not compensate for disproportional constituency results.

What does this system do well?

MMM systems retain some of the strengths of the current system. It allows smaller parties that cannot win constituency seats some opportunity to secure representation. However this is less so than under the current system.

The list allows parties to ensure gender equality and diversity in ways probably less contentious than in FPTP systems alone.

It would require parties to gain a substantial level of support to gain entry into the Chamber.

Where does this system fail?

Proportionality: Because constituency results are ignored in allocating list members, MMM systems are inherently much less proportional than AMS. For example, if the 2016 Assembly election had been conducted on the existing boundaries but under MMM rules – so that list seats were allocated without regard for the constituency results – then Labour would have won 9 list seats, on top of its 27 constituency seats. That would have given Labour nearly a three- fths majority in the Assembly, on an overall vote-share below one-third.

In order to retain similar levels of proportionality to the present, the ratio of list to constituency members would need to be adjusted substantially. Constituency members would have to be greatly outnumbered by list members in the chamber. For example, under the revised boundaries with 29 constituency seats, we estimate that between 60- 70 list AMs would be required to achieve the same level of proportionality as actually obtained in the 2016 Assembly election. Given some of the controversies on list members over the course of devolution, as well as the British tradition of constituency work, this seems an unlikely scenario.

Simplicity and Coterminosity: In terms of simplicity, MMM offers no clear gain over the current system. In terms of exibility to allow for linking with boundaries at the UK or local level, it would retain a similar level of exibility as the AMS system, and we could add list members to t with the total number of AMs required (although there would be debate about the desirability of different ratios).

Strong and Equal Mandates: Mandates would be more or less the same as in the current system. The different mandates might be less controversial, as the largest party would be likely to have more list AMs, so the differences in mandate would lose some of their partisan avour. However, this would come either at the cost of lower proportionality, or (as described above) one would require many more list than constituency members.

Broad-based consensus: It seems unlikely that this system would achieve widespread consensus.

Assessment: MMM systems would retain the main problems identi ed with AMS, while reducing its bene ts. This is a less desirable system for National Assembly elections than the current AMS/MMP system.

National list as part of any AMS/MMP System

The five regional lists currently used in Assembly elections would be replaced with a single national list for the whole of Wales. Within an enlarged National Assembly, more than the current twenty list AMs would likely be elected. Moving from regional lists to a national list could be part of a broader package of reforms to how we elect the Assembly.

What does this system do well?

This system would increase the number of parties likely to gain election into the Assembly, as the ‘effective threshold’ for election would be a lower percentage of the vote. Indeed, to adhere to the principle of ‘substantial support’, and avoid excessive fragmentation of the political system, it would likely be necessary to create a legal threshold, where a party would need to gain, say, 5% of the vote in order to be eligible for election into the legislature.

A National list would retain some of the strengths of the current system, in allowing positive action for gender equality and diversity on the list, for example. Indeed, this would be likely to increase, since academic research suggests that a larger ‘district magnitude’ (a larger constituency with more members elected) will often lead to a more diverse group.13

Where does this system fail?

Strong and Equal Mandates: This system might exacerbate tensions between constituency and list AMs, with list AMs losing even the broad regional link they currently have, and dozens of AMs all representing Wales as a whole (or choosing to target particular areas). Losing AMs’ link to particular regions which feel less connected to the Assembly in Cardiff (such as North Wales) could exacerbate these problems. It may also encourage parties to choose representatives from the areas in which they have the most strength and membership, so there is a danger that parties may be less diverse in terms of their geographical representativeness. List AMs would also be encouraged to try to play to the areas with the greater population for votes (such as Cardiff).

Simplicity: While the way the system operates would be similar to the present system, so there are few gains in simplicity. Indeed, bringing ve regions into one electoral list could make things more complicated for voters. Notably, the list would likely be very long.

Broad-based consensus: Given parties’ stated preference for some form of local (or at least regional) representation, this option is unlikely to be preferred to a variant of the current AMS/MMP system, if one favours a mixed system for elections to a larger National Assembly.

Assessment: A national list appears to offer few advantages for National Assembly elections.

Table 1: Summary Evaluation of How Different Electoral Systems Perform on Seven Key Principles


Adapting the Current System


An obvious starting point for designing a future electoral system for the National Assembly should be the current system. Could it be adjusted, and made to work for the future?

Perhaps, but only with some dif culty.

Implementation of the ‘reduce and equalize’ legislation, and the ensuing decline in
the number of Welsh Westminster constituencies to 29, would inevitably lead to some changes in how the National Assembly is elected. This is one instance in which no change really would not be an option. If Wales were to opt to maintain coterminosity
of Westminster and Assembly constituencies, and also to keep the same proportion
of constituency and list AMs as at present, then we would still see a large change: that change would be one of substantially reducing the size of the National Assembly – to 29 constituency AMs and 14 or 15 list members! The Assembly is already inadequately small at 60 members; one of 43/44 AMs would be disastrous.

To avoid that fate, while keeping the AMS electoral system, we would need to do one
of two things. The rst option is to abandon coterminosity. If that were done, then signi cant exibility is created. The current size of the Assembly could be maintained, with the 40/20 split between constituency and list members. Or constituency boundaries could be re-drawn to increase further the number of constituencies; the number of list members could also be increased to maintain the existing proportion of constituency and list AMs. Abandoning coterminosity is clearly possible – it has already been done in Scotland. But for reasons we discussed earlier, requiring two completely different sets of constituency boundaries is not ideal either for voters or for parties.

The second option would be to maintain the same constituency boundaries as for Westminster, but to elect two AMs for each constituency. (This would give a total of 58 constituency AMs; the number of list AMs might then be increased to 29). This could, in turn, be done in one of two ways: either splitting each constituency into two geographic halves; or by choosing constituency members via two member FPTP, as described above.

However, both of these options also come with some dif culties. Splitting Westminster constituencies in two requires yet a further set of boundaries to be drawn, a task which will create scope for additional boundary disputes, and which will have to be repeated every ve years once new Westminster boundaries are set. Electing two members

per constituency, as discussed earlier, undermines the whole point of having a single constituency representative. It would also require that voters in Assembly elections be given three votes in total: two for both of their Assembly constituency AMs, and a further one for the party list members. This would complicate further an electoral system that some voters already struggle to fully understand.

Furthermore, whichever of these two approaches was taken, there would be further potential problems. A minor one is that the ve-yearly boundary reviews can produce, and currently has produced, numbers of constituencies for Wales that do not translate very neatly into an AMS system with one-third list members. Allocating 29 list seats across ve regions would presumably mean that four regions would get six list members and one would have only ve.


A more serious objection about using AMS to achieve a larger Assembly, and one
that would apply pretty much however it was done, is that increasing the numbers of list members for each region would also increase the chances of small, and extreme, parties gaining seats in the Assembly. By having more list seats per region, the effective threshold (the share of the list vote you need to win representation) is lowered. We note that the BNP would have won the sixth regional list seat in North Wales in 2007.

None of these objections to AMS strike us as necessarily a ‘knock-out blow’. Using AMS and 29/30 constituencies as the basis for a larger Assembly clearly could be done. But it could not be done very elegantly. Moreover, all of the existing problems with AMS – most obviously the fact that the existence of the two types of members has created tensions, such as the arguments over ‘dual candidacy’ – would remain if the system were to be adjusted to try to provide for a larger Assembly within an overall total of 29 constituencies. And in at least some respects, those existing problems of AMS might be added to, such as the added complications of two members per Westminster constituency.

All this suggests to us that, while AMS should certainly remain under consideration as a route towards achieving a larger National Assembly, some alternative potential solutions should be explored.

Systems that are Workable, Sustainable and Pass the Principles Test

Passing the Principles Test

Which electoral systems could provide for a larger National Assembly for Wales while also according with the principles outline in section 1 (no reduction in proportionality, strong and equal mandates etc.)? Section 2 outlined a number of electoral systems that, to our mind, are clearly not suitable. The previous section showed how AMS could be adapted for a larger Assembly, though only with some dif culties. Here, we explore two more promising options.

In recent years there has been increasing academic interest in the use of Proportional Representation systems in fairly small districts. Such systems have the bene ts of keeping representation fairly close to the people with all representatives having a similar mandate. They also has the advantage of generally avoid giving seats to small, extreme parties, through having a high effective threshold for representation. At the same time, they can produce results that are fairly proportional between the major parties. Thus, it has been suggested, PR in fairly small districts hits an ‘electoral sweet spot’.14 This work suggests to us two alternative ways in which the National Assembly for Wales’s electoral system could be re-designed. Both of them are based around the new, 29-seat, constituency map.

Single Transferable Vote (STV)

One approach would be to use the STV system that has long been advocated by the Electoral Reform Society. This system is used for nearly all elections in Ireland (north and south), in Malta, in Australia (Senate) and also in Scotland (local government); STV was also advocated for use in National Assembly elections by the Richard Commission in 2004.

How would it work? As is fairly well known, STV uses smallish multi-member constituencies. Parties can (and often do) stand more than one candidate in a constituency. Voters cast their choices as a numbered order of preferences, rather than a categorical all-or-nothing ballot.15

There are several ways in which STV could be used for National Assembly elections.16 But if Wales moves towards 29 constituencies for Westminster elections, then a simple way of applying STV would be to use those same constituencies as the electoral boundaries for elections to the Assembly. If we elected three AMs for each constituency, one would arrive at a chamber of 87 members. This would provide for a signi cant increase in the size of the Assembly – albeit not to quite the extent advocated in Size Matters.

Used in this way, the STV system would not normally be a particularly proportional system: the more representatives are elected per district under STV, the higher the level of proportionality. The system might not also be wholly stable: the number of AMs would change if the number of Westminster constituencies were to be altered in the future.

However, such a form of STV would achieve at least the following goals:

  • It would allow for the maintenance of coterminosity between Westminster and Assembly constituency boundaries;
  • It would avoid dif culties in re-calculating and allocating respective numbers of constituency and list seats if future boundary reviews changed the number of Westminster constituencies;
  • It would put an end to all current arguments about dual candidacy under AMS, as under STV all AMs would be elected on the same basis; and
  • As best we can judge, such a system would achieve a very similar overall level of proportionality as now, while avoiding lowering the thresholds for representation to parties with little substantial public support.

Table 2 presents outline results from our best attempt to simulate the use of this form of STV in the 2016 Assembly election, alongside estimates for other alternative systems. We see that, compared to the actual election outcome, the overall level of disproportionality (as indicated by the Gallagher Index gures) is almost identical.

Table 2: Estimated Results of 2016 Election Under Different Electoral Systems in a 87- Seat National Assembly


We should make clear that such a form of STV would not be our ideal electoral system. All the authors of this report would personally prefer a more proportional form of STV. It might in time be possible to achieve the necessary consensus within the Assembly
for such a system – one could transition fairly smoothly from a system based on three members per constituency to four, which would make the system distinctly more proportional. But a three-member version of STV as outlined here it would be a perfectly workable means of producing a significantly larger NAW based around 29 Westminster constituencies, while maintaining the current degree of proportionality in how the Assembly is elected.


Open List

An alternative method to STV would be some form of party list system. Under this type of system, each constituency could again elect three members; thus, with 29 constituencies one would have an 87-member NAW.

There are different types of party list system that could be used. One could potentially use a ‘Closed List’ system, such as is currently deployed for European Parliament elections in mainland Britain. Here, voters choose between party lists, and seats are allocated proportionally to the parties; if a party wins two seats in a region, those seats go to the top two candidates on that party’s list.

However, this Closed List system has substantial drawbacks for the voter. Closed party lists are impersonal, weakening the link between the representative and the constituency. The system offers very little in the way of voter choice: all the power, save that of choosing a party for government, resides with party leaders and the party machines. Parties can stifle independent and minority opinion within their ranks. As nearly all the power over who gets seats lies with the party machine, so too does the power to voice opinions. As candidates are selected by the party networks, they may also be more likely to put ‘safe’ candidates near the top of the list, at the expense of traditionally under- represented groups.

Among party list systems, our very strong preference would therefore be for an Open List system. Under this form of electoral system, voters normally cast a single, categorical vote, just as they do in a Closed List system.18 The main difference with Open List is
that voters can choose to cast their votes for an individual candidate. Seats would then be allocated proportionally to parties according to the total number of votes their candidates received. But which individuals are actually elected depends not on their place on a party list, but instead on the number of individual votes they receive.

The key bene t of Open List is the greater degree of voter choice it provides compared to Closed List, while maintaining overall proportionality between the parties. As with STV, although in a slightly different way – voters cast one, categorical vote rather than an ordered series of preferences – voters are able to choose not only between parties but also between candidates (if parties stand more than one candidate), and so all elected representatives would have a clear personal mandate.

If Open List were to be used for National Assembly elections, one would also need to give careful consideration to the Electoral Formula used. Our gures (see Table 2) suggest that using a 3-member system with the current D’Hondt formula would substantially favour the largest party: if it had been used in the 2016 Assembly election, our estimates suggest that the system would have produced a clear majority for the Labour party, and increased the overall disproportionality of the system considerably. But use of the Modi ed Sainte-Laguë formula – which is currently, for example, in Norway and Sweden, and which is slightly more helpful to smaller parties – would produce levels of disproportionality only slightly greater than STV, or than the actual election outcome in 2016.



Following on from the publication of Size Matters in October 2013, this report has sought to answer the one unanswered question of that document: how do you elect a larger National Assembly for Wales? An increase in the number of AMs in Wales will require some change in the electoral system. The key question is how this might be done.

In Reshaping the Senedd, as in our previous report, we have followed a principle- and evidence-based approach. We have drawn on international examples of different electoral systems to identify all plausible options, and provided a framework for judgement based on the balancing of clear but essential principles. We have also used the best evidence available on the likely electoral implications of different systems to evaluate them in an open and rigorous manner.

Following the principles we outlined in section 1, some electoral systems immediately fail the tests we set out (as we discuss in section 2). These systems, we believe, do not come close to meeting the principles that should underpin any electoral system used to elect the National Assembly for Wales. But we believe that there are three more viable options – electoral systems that might or do meet these principles.

One of these systems is the current one, Additional Member System. But we observe some potential problems with using AMS to elect a larger Assembly. We believe that two other systems, the Single Transferable Vote and Open List, may offer more attractive routes towards achieving a larger National Assembly. Indeed, we believe that both the latter systems could readily be implemented alongside, and in tandem with, the move to 29 constituencies in Wales.

Prominent figures from nearly all the main political parties in Wales have agreed that there needs to be an increase in the number of Assembly Members. But the Wales Bill requirement of a supermajority for change ensures that consensus building will be needed. We believe that is it now time for political leaders in Wales to move forward on this issue on that basis.

We offer this report to them, and to the people of Wales, in the hope that it can advance the debate.


Assumptions Made for Estimating STV and Open List Elections

The following assumptions are the basis for estimates of the result that the 2016 National Assembly for Wales election might have produced if conducted under STV and Open List, applied to 29 constituency seats.

The 29 constituency seats are assumed to be those outlined by the Boundary Commission in September 2016 for Wales in line with the proposals under the 2011
Act to ‘reduce and equalize’ representation in the House of Commons. The actual constituency results from the 40 constituencies used in the 2016 National Assembly election have been translated into ‘notional’ estimated results for these 29 seats. These national estimates were produced by Anthony Wells, Director of Polling at YouGov UK. (We are grateful to Anthony for supplying these figures; he bears no responsibility for our interpretation and use of them).

For Open List, each party is assumed to gain the same number of votes as in the 2016 notional constituency vote results. Seats are then allocated to the parties via either the D’Hondt or the Modi ed Sainte-Laguë formulae.

For STV, matters are a little more complicated. The following assumptions are made:

1. In each constituency, every party is assumed to get the same number of rst preference votes under STV as the total number of votes received on the constituency ballot in May 2016.

2. Each party is assumed to stand two candidates per seat. The party’s aggregate total of votes is initially divided between these two candidates on the following proportions: 0.6 and 0.4. Surplus votes for elected candidates, and votes from eliminated candidates are assumed to transfer entirely to the other candidate of the same party in the first instance.

3. Once both candidates for a party have been either elected or eliminated, votes to be transferred are then re-allocated to the leading candidates from each of the other parties in proportions based on data drawn from the 2016 Welsh Election Study. This asked all survey respondents the following question:

“Please indicate how you would have voted in the National Assembly for Wales election if you had been asked to rank the parties in your order of preference. Put 1 for your most preferred party, then 2 for your second best party, 3 for your third choice etc. You may rank as many or as few choices as you wish.”

Votes are therefore transferred to candidates of other parties according to the proportion of respondents selecting a party as their first preference who told the 2016 Wales Election Study that they considered another party to be their second preference. For example, of all those Welsh Election Study respondents who gave Labour as their first preference, 43.5% indicated that they would have chosen Plaid Cymru as their second preference. Therefore, 43.5% of any surplus Labour votes would be transferred to the lead Plaid Cymru candidate.

4. The detailed figures given for second preferences in the 2016 Welsh Election Study were:

Labour: Of those indicating that their first preference would have been Labour, the following second preferences were chosen:

  • 43.5% of those choosing Labour as their first preference listed Plaid Cymru as second
  • 15.0% nominated the Greens
  • 13.8% chose the Liberal Democrats
  • 6.1% chose UKIP
  • 3.4% selected the Conservatives
  • 8.0% chose ‘Other’ parties

The remaining 10.2% of those choosing Labour as their first preference did not indicate any second preference. That proportion of Labour votes was therefore not transferred.

Conservatives: of those indicating the Conservatives as their first preference, the following second preferences were indicated:


  • 32.5% UKIP
  • 21.0% Plaid Cymru
  • 19.9% Liberal Democrats
  • 5.7% Labour
  • 3.3% Greens
  • 4.7% ‘Others’

The remaining 12.9% of those choosing the Conservatives as their first preference did not select a second preference party. That proportion of Conservative votes was therefore not transferred.

Plaid Cymru: For those nominating Plaid Cymru as their first preference, the following pattern of second preferences were indicated:

  • 35.6% Labour
  • 20.4% Greens
  • 11.7% Liberal Democrats
  • 8.6% Conservatives
  • 8.6% UKIP
  • 5.9% ‘Others’

The remaining 9.2% of Plaid supporters did not nominate a second preference party. That proportion of Plaid Cymru votes was therefore not transferred.

Liberal Democrats:

  • 29.4% Plaid Cymru
  • 25.7% Labour
  • 17.3% Greens
  • 15.9% Conservatives
  • 2.3% UKIP
  • 4.2% ‘Others’

The remaining 5.2% of Lib-Dem supporters did not nominate a second preference party. That proportion of Lib-Dem votes was therefore not transferred.


  • 34.3% Conservatives
  • 18.0% Plaid Cymru
  • 8.9% Labour
  • 6.9% Liberal Democrats
  • 3.2% Greens
  • 15.4% ‘Others’


  • 37.4% Labour
  • 34.0% Plaid
  • 11.6% Liberal Democrats
  • 3.4% Conservatives
  • 2.7% UKIP
  • 5.4% ‘Others’

The remaining 5.5% of Green supporters did not nominate a second preference party. That proportion of UKIP votes was therefore not transferred.

Simulated 2016 Election Results for 87-Seat NAW based around 29 Constituencies

Read more publications...