Electoral Reform Society submission to the Richard Commission

Summary

In the 2003 Assembly elections, FPTP might have resulted in Labour winning 75% of the seats with only 40% of the vote, and even that 40% might overstate true support for Labour. Such a result would have deprived the Assembly of legitimacy and of support from the Welsh electorate.

Using the Additional Member System (AMS), a broadly proportional electoral system, ensured an Assembly which is very much more representative of the Welsh electorate than would have been the case under FPTP.

Our submission is about how the electoral system could be improved further, and not just in terms of party proportionality.

In spite of the benefits gained by using the Additional Member System (AMS), the system has a number of drawbacks.

Two general problems associated with AMS are:

• it creates two types of elected members – the potential for tension between the two is heightened by all Labour AMs being constituency members and most of the opposition being list members;

• although creates a unique, but perhaps weak, link between constituency members and their electorates, it also creates members who are only very loosely connected to the electorate.

However, the Assembly elections on 1st May of this year well illustrated a number of other problems with AMS including:

In this submission we therefore consider a number of ways in which the particular version of AMS could be revised by:

• increasing the number of list members, thereby improving proportionality;
• using a national list, thereby increasing opportunities for smaller parties;
• increasing voter choice through the use of the Alternative Vote in constituencies and open lists for the election of list Ams.

We conclude that none of these changes would fully overcome the problems referred to above.

We have considered list systems (closed, open, and semi-open) as an alternative to AMS. While list systems offer good proportionality and flexibility in the choice of electoral regions and numbers to be elected in each region, anything short of fully open lists denies voters choice and risks reducing the accountability of AMs to the electorate.

STV would offer the flexibility of list systems, but would go much further than any other system in overcoming the problems encountered with AMS (as listed in (3) above).

In considering the ‘representativeness’ of the Assembly, the Commission’s consultation paper notes issues other than party proportionality.

• With regard to "the close identification between an individual elected Member and the area represented"; we conclude that AMS does not produce strong links for all members, and not even for constituency members; closed list systems seriously weaken links, but STV in appropriately-sized electoral areas could improve links;

• With regard to the extent to which the Assembly might represent the diversity of Welsh society, including minority interests, we conclude that STV and open list systems are best able to allow the electorate to choose not just what party they want, but what type of representatives;

• We recognise the important role of parties’ candidate selection procedures in producing a representative Assembly: systems which require parties to select groups of candidates simultaneously are more likely to lead to greater diversity, while STV and open lists score over AMS and closed lists in that they require parties to present a team of candidates from which their voters make choices, rather than a ranked list.

STV in constituencies each electing 4 to 6 AMs is therefore the system most likely to overcome the existing problems of AMS and to improve the representativeness of the Assembly.

We have considered what might be the simplest approach to the introduction of STV – the pairing of Westminster constituencies and making each pair a 4-member Assembly constituency. While it is not possible to say how the Assembly might have looked if such a system had been in place on 1st May 2003, it is reasonable to assume that the result would have been slightly more proportional, but with Labour still holding close to half of the seats. Other problems associated with AMS would not have arisen.

We have also considered the ‘Hain proposal’ for STV in 2-member seats. We accept that this would overcome some of the problems of AMS, but would do next to nothing to make the Assembly more representative.

Labour is, at least at present, the party in Wales with the strongest support. Any electoral system which did not give Labour a pre-eminent position in the Welsh Assembly would not be an option worth considering. However, although Labour is strong, a majority of Welsh voters support parties opposed to Labour. It would therefore have been wrong to use an electoral system which allowed Labour to dictate policy in Wales against the wishes of parties which collectively have more support than Labour.

That, however, is what would have happened in Wales if FPTP had been used for the Assembly elections, as can be seen from the results from the constituency contests (of the AMS system) on 1st May.

%age constituency votes

No. constit. seats

%age seats

40.0%

30

75.0%

21.2%

5

12.5%

20.0%

1

2.5%

14.1%

3

7.5%

4.8%

1

2.5%

%age votes

Number seats

%age seats

48.4%

34

85%

14.2%

4

10%

20.8%

0

0%

13.7%

2

5%

2.9%

0

0%

Thus FPTP, however introduced, would have denied any effective influence to parties representing a majority of Welsh voters. That would have resulted in the Assembly no longer being a legitimate collective voice of people in Wales.

AMS, as a broadly proportional system, provided all major parties in Wales with representation, as the results from 1st May demonstrate:

%age party votes

Number seats

%age seats

36.6%

30

50%

19.7%

12

20%

19.2%

11

18.3%

12.7%

6

10.0%

11.8%

1

1.7%

The Liberal Democrats, Plaid Cymru and the Conservatives would have won nearly their proportional share of the seats, whereas with FPTP the Conservatives, with nearly one vote out of every five, would have won only one seat.

Constituency members are viewed as having greater legitimacy as representatives of issues in their constituencies while list members are not regarded as local representatives in the same way.

Tensions between constituency and list representatives do not appear to be as strong as in Scotland, but this may be because the Scottish Parliament has more executive power and constituency MSPs are therefore likely to have more casework referred to them. There is also concern that list members can "cherry pick" issues, deciding to focus their activities on those issues most likely to raise their profile or create problems for their constituency opponents. There have been accusations that list members have concentrated their energies in constituencies in their regions where there are future prospects of winning constituency seats. If the Welsh Assembly were to acquire greater powers, it must be anticipated that the tensions between list and constituency AMs will increase.

The problem is exacerbated by the split of constituency and list seats between the parties. All Labour AMs are constituency AMs, but only one third of opposition AMs were elected from constituencies.

That FPTP creates a unique link is clear, but that does not necessarily mean that it creates a strong link. As has been noted above, the constituency contests of the Assembly elections are generally fought by candidates of 4 major parties, and a consequence of using FPTP is that many votes are wasted. Winning candidates will not necessarily be the most popular candidates and they should not expect those who opposed them (in most constituencies the majority) to feel a strong link. Later (7 below) we will argue that stronger links can be created with STV in multi-member constituencies.

A second problem is that AMS retains the unique link of FPTP for only some members, but at the expense of having other members only loosely connected to the electorate.

As noted above, this over-representation of Labour arises from those electoral regions (North, South Wales Central and and South Wales West) in which Labour was able to win more than its proportional share of seats from the constituency seats alone.

In the contest for list seats, it was clear that at least in 4 of the 5 electoral regions Labour would have next to no chance of winning list seats (unless they performed quite disastrously in the constituency contests).

Those candidates at the top of some lists (e.g. Conservative in South Central and South East, and Plaid Cymru in South Central and South West) were almost certain to be elected, while candidates at the bottom of lists had negligible chances of election.

Safe seats and predictable outcomes are neither good for democracy nor for participation in elections.

Votes which ensured victory of the winning candidate  26.4%

Votes which contributed to unnecessary majorities  20.8%

Votes cast for losing candidates                                   52.8%

Thus 73.6% of votes were 'wasted' in this sense.

In the list contests, votes were also wasted in that they were cast for parties which failed to win seats.

It is also likely that there was tactical voting for party lists. In the electoral regions in which Labour was strong, for example, some Labour voters may have voted for other parties knowing that Labour would have little chance of winning list seats.

An electoral system which requires voters to vote other than for their preferred candidates or parties in order to make their votes effective is far from desirable.

However, even where the voting system was understood, it is possible that many voters did not appreciate how to make their second votes effective. That more than a quarter of a million list votes (nearly a third of all list votes) were cast for Labour in regions where Labour had no chance of winning list seats might suggest that voters simply wanted to demonstrate the extent of support for Labour, but it is also likely that many did not appreciate that votes for their preferred party would be ineffective.

At present there are 4 list seats in each of the 5 electoral regions. The table below examines what would have happened if there had been 6 or 8 list seats in each region.

%age of party vote

%age total seats (at present)

%age total seats if top up seats in each region increased by:

0

2

4

36.6%

50.0%

45.7%

43.8%

19.7%

20.0%

21.4%

21.3%

19.2%

18.3%

20.0%

20.0%

12.7%

10.0%

11.4%

13.8%

11.8%

1.7%

1.4%

1.3%

This change would do nothing to correct Labour's over-representation. It would, however, produce a more proportional result amongst the remaining parties. In particular, it would allow the Green Party [check on others] to gain representation.

An all-Wales list would, however, make other options more possible. With 5 regions, options for the size of the Assembly are 60 (40 + 20), 65 (40 + 25), 70 (40 + 30) etc., but with a national list the number of list seats need not be restricted to multiples of 5. This, however, might not be regarded as a great advantage.

The advantages of this change would be to improve electoral competition and remove the need for tactical voting in the constituency contests. All constituency winners would need at least 50% support (even if not all from first preference votes). In the 25 constituency seats in which the winner did not receive at least 50% of the vote in 2003, the winner would have needed to attract (lower preference) votes from supporters of other parties to win. In the 2003 election, only 47% of votes were cast for winning candidates, but with AV+ this percentage would increase (although we cannot predict by how much without knowing how voters would rank their preferences).

This would overcome the criticism that list AMs owe their positions to their ranking by party selectorates, rather than by the electorate. All AMs would then have the satisfaction of knowing that they were chosen by the voters and not just by party activists.

Although this would greatly improve the democratic nature of the system, it would also further complicate it.

The Jenkins Commission proposed that voters should have the option of voting for a particular list candidate or simply for the party, thereby accepting the party's ranking of its candidates (i.e. making the lists 'semi-open'). This would create yet a further level of complication in the system. Moreover, experience in countries in which semi-open lists are used suggests that voter preferences do not often alter the ranking of candidates made by the parties: they can also, depending on the particular form of semi-open list used, create anomalies whereby a candidate can be elected with fewer individual votes than an unsuccessful candidate.

List systems also offer considerable flexibility. For example:

• We could retain the existing 5 regions and elect 12 AMs in each, but instead of 12 we could have any number of AMs elected in each region. Indeed, we could vary the number in each region to reflect the sizes of the electorates.

• Instead of using the existing regions (generally of 8 Westminster constituencies), we could have 10 regions of 4 Westminster constituencies each electing 6 AMs (or any other number of AMs) while still preserving broad proportionality.

• Regions could be of different sizes to reflect natural regions of Wales, with the number of AMs elected in each reflecting the electorates of the regions.

If a list system had been used in the 2003 elections using the same electoral regions and electing the same number of AMs in each region as at present, the outcome would have been:

However, list systems have their disadvantages. To consider these we must differentiate between open and closed lists.

With a closed list system, the voter votes only for the party, not for an individual candidate. A candidate's chances of election depends on where their party places them on the list than on the their popularity with the voters. Candidates at the top of their parties' lists (at least for the major parties) could expect to be elected, even if the voters considered them to be the worst candidates they had ever encountered, while candidates at the bottom lists would have no realistic chance of election, whatever their individual merits. When candidates' chances of election depend more on their party selectorates than on the electorate, there is a danger that their accountability will shift from the electorate to their party selectorates. That cannot be good for democracy: it could seriously weaken the link between AMs and those who elected them, and by weakening the influence of the voters over those who are elected we could expect interest in elections, and therefore turnouts, to fall further.

Open lists go a long way towards putting power into the hands of the voters. If a party won, say, 3 seats in an electoral region, then it would be those who voted for that party who would decide which 3 of that party's candidates filled the 3 seats.

However, there is a drawback. Suppose a party fielded a list of 6 candidates (A, B, C, D, E and F) but only won 2 seats. If A and B had the most votes, then A and B would be elected. But some voters might have voted for party only because they wanted to support C (or D or E or F) and might be opposed to the election of A or B. Nevertheless, their votes would have contributed to the election of A and B while their preferred candidates were unsuccessful. Voters can therefore find that they have supported outcomes which they might not have wanted.

With open lists, parties can put well-known and popular candidates at the top of their lists: these candidates attract individual votes, which are also votes for the list, thereby securing the election of others in whom voters might have little interest.

Semi-open lists present a compromise between open and closed. Parties rank their candidates and in voting for a list, voters can, if they choose, vote for a particular candidate on the list. If a candidate receives a significant number of individual votes (different formulae are used in different countries where this system is used), the candidate can move up the list improving his or her chances of election.

However, semi-open lists have been criticised for creating an illusion of voter choice. Practice elsewhere suggests that most voters are content to vote for the party without indicating a preference for a particular candidate, and individual votes therefore do not often alter the ranking proposed by the parties.

Moreover, anomalies can occur. Candidate A might be place high on a party's list but received few individual votes. Candidate B might have a significant number of individual votes, but not enough to achieve a higher ranking than A. As a result, A might be elected, but not B who might appear to have more votes. While such outcomes are understandable in terms of the system, they are not likely to generate voter confidence in the system.

Moreover, a major advantage of STV (and indeed list systems) over AMS is that all members are elected with the same democratic mandate. STV does not produce two categories of member – constituency and list members – with the conflicts over roles and legitimacy which can arise.

Below we examine how STV can tackle each of the problems encountered with AMS.

STV takes account not just of first preferences, but also of how voters rank candidates. For example, a party which had no first preference votes but was the second choice of all voters should, in conventional proportionality, receive no seats, but STV would take account of its broad support.

With STV voters vote for candidates rather than for parties (although most will select candidates on the basis of their party affiliations). As a result, STV can produce outcomes which are proportional in terms of any criterion which significant numbers of voters consider important. For example, if most voters consider gender to be important, STV will produce a result which reflects the voters’ gender preferences.

The Commission’s consultation paper notes a number of possible dimensions of ‘representativeness’ other than measures of party proportionality. Here we consider how representativeness might be affected by changes in the electoral system.

Firstly, we would point out that while AMS creates a unique link between constituency members and their electors, list members under AMS (one third of all AMs in the case of the Assembly) have only very weak links with the electorate. If the number of list AMs were to be increased, we would have even more loosely connected AMs.

While the link of accountability of a member to the electorate is clearly important, we do not accept that it depends on single-member constituencies. In local government multi-member wards are common, and our research demonstrates that councillors in multi-member wards regard their links with ward members are being at least as good as those of councillors in single-member wards. The link does not depend on there being a unique link between members and those who elected them.

Indeed, the link between a member in a single-member constituency and his or her electors might not be as strong as some argue. It is understandable that those with a monopoly of representation are keen to avoid competition, but from the perspective of the elector things might look very different. A single-member might ‘represent’ all constituents in the sense that they are prepared to attend to routine casework, but there is no way in which a single member can represent the diversity of views found in any constituency. On many issues electors are likely to prefer being able to choose from a number of representatives the one which they feel is most likely to be sympathetic to their case. Generally an elector is much more likely to feel a link with a member of his or her party and with a person they have helped elect: in the case of AMS many electors do not have such a member, but they would in a multi-member STV constituency.

We also note the experience of the Republic of Ireland. On two occasions Irish politicians called for a referendum in hope of changing away from STV, not because it weakened constituency links but because, in the view of some TDs, it made the link too strong. On both occasions, however, the proposals for change were defeated by the electorate which felt that the member-constituency under STV was not something they wanted to lose.

With a system which requires parties to field a list or team of candidates, parties have an incentive to field a broadly representative team in order to attract wider support. List systems and STV are therefore more likely to lead to greater diversity amongst candidates, and therefore AMs.

STV and open lists have a further advantage. Voters can choose from the teams presented by the parties: those who want older or younger AMs can choose on the basis of age, just as others will make choices influenced by gender and ethnicity, occupational and/or social background, and even policy positions which lie outside the scope of party manifestos. This contrasts not just with FPTP, but with AMS where voters must accept the choices made by their parties – or refrain from voting for their favoured parties.

With electoral systems in which parties need to choose more than one candidate in each constituency/electoral area, it should be easier for parties to promote candidates from under-represented groups while preserving opportunities for good candidates from more traditional backgrounds. It should be possible to achieve an acceptable gender balance without needing to resort to women-only shortlists and other devices which some regard as controversial.

With closed lists (and AMS, part of which is a closed list) parties must not just select but also rank their candidates. This can be a difficult and divisive task as a small selectorate might need to decide which candidates will almost certainly be elected and which will only be on the ballot paper for show. With STV and open lists, parties do not need to rank their teams: it is the voters of a party who decide which of the party’s candidates will be successive (effectively a primary and an election rolled into one). Not only is this more democratic, but it should help avoid some of the internal party tensions which result from selection decisions.

• provide better proportionality;
• overcome tensions arising from two categories of AM;
• strengthen links between members and the electorate;
• overcome the many problems, listed in 3.2 above, which arose with AMS in the 2003 Assembly election;
• allow flexibility in the sizes of constituencies and number of AMs in the Assembly;
• increase voter choice;
• lead to an Assembly which better reflects the diversity of Welsh society.

We recommend the use of constituencies which elect between 3 and 6 members. As with all electoral systems, there is a trade-off between the size of the electoral areas and the degree of proportionality which will be achieved.

• In 3-member constituencies the ‘quota’ which parties must reach to gain representation would be 25% of the total vote, which could make it difficult for some parties to gain seats. in some areas (although the quota could be reached with transferred votes as well as first preference votes). We would not therefore recommend 3-member seats other than where the geographic size of constituencies could create problems.

• In 6-member constituencies the quota would be 14.3% (which again could be reached by transfers). In some circumstances constituencies of more than 6 members could be considered, but as well as arguments about the geographic size of constituencies, there is a danger that the list of candidates can reach a length at which voters might be less discriminating in their choices.

There are, however, two ways in which STV could very easily be introduced without the need for new boundaries to be drawn:

• we cannot assume that voters would use their first-preference votes in the same way as their FPTP constituency votes (and we only have details of list votes at regional level) because (a) some may have voted tactically and (b) there would probably be a wider range of candidates from which to choose;

• we do not know how voters would use their second and subsequent preferences.

If STV were to be introduced in 4-member constituencies (each being a pair of Westminster constituencies), and if voters had used their first-preference votes in the same way as their FPTP votes, it appears that the proportion of seats won by each party would have been roughly the same as at present (Labour would have had at least 40% of the seats, but probably more than that and a chance of getting near to 50%). However, as a result of the involvement of smaller parties and less tactical voting, we would anticipate STV on this model producing a slightly more proportional outcome.

Peter Hain has proposed STV in two member constituencies (each Westminster seat being a constituency). Peter Hain calls this ‘2-member AV’, but we do not want to get involved in a debate about terminology.

We accept that 2-member STV could overcome some of the problems associated with AMS, but it would be barely proportional. In any constituency a party would need to gain 33% of the votes (including transferred votes) to be sure of election, compared with 20% under 4-member constituency STV. As a result some of the major parties would not be represented in many areas of Wales.

The result would only be moderately proportional. If people had used their first preferences in the same way as their FPTP votes (see our warnings about making such assumptions above), Labour would have been almost certain to win about 45% of the seats, and would have had a realistic change of winning about 55%.

This proposal does not therefore appear to have a great deal to recommend it.

We therefore recommend the use STV as a much better system than AMS for Assembly elections. We recommend the use of 3- to 6-member constituencies, but advise against the use of constituencies of only 2 members.

Appendix: What the Voters Think

This poll was conducted for the coalition which became ‘Make Votes Count’. A full summary is provided as an appendix to this submission and detailed notes from focus group discussions can be provided if needed. Here we quote only the opening paragraph of the summary:

"This poll … suggests broad support among the British public for moving to a new electoral system based on proportional representation (PR). A substantial bloc of the public – 72 per cent – say they will vote "yes" to replace the current "first-past-the-post" system with a PR system; nearly 60 per cent strongly support the change. That suggests a general political environment favourable to reform: nearly three of every five members of the public say the current system for governing Britain needs to be improved "quite a lot" or a "good deal". Indeed, overwhelming majorities seem upset with a political system full of bickering and politicians that are unaccountable. The public wants to move to a system that is fairer and that, above all, puts more pressure on politicians to address the country’s major problems. A public that has supported constitutional reform in Scotland, Wales and Northern Ireland seems prepared to take the reform process one step further."

This poll was conducted by System Three for the Electoral Reform Society. Responses to the three questions which were asked are summarised below:

Agree

Disagree

70%

12%

28%

50%

75%

10%

The researchers found strong support for the voting system used for the Scottish Parliament and Welsh Assembly and for extending electoral reform to local government. On whether "the UK should introduce PR so that the number of MPs each party gets … matches more closely the number of votes each party gets", they found:

"Some people think we should change the voting system for general elections to the House of Commons to allow smaller parties to get a fairer share of MPs. Others say that we should keep the voting system as it is to produce effective government. Which comes closer to your own view?"

This question produced a majority in favour of the second option which the authors summarise as "keep it as it is". However, the question appears to be worded in a way which would lead those (the majority) who want "effective government" towards the second option, when we would dispute the implication that a reformed voting system would lessen the effectiveness of government. The argument given for electoral reform ("to allow smaller parties to get a fair share of MPs") is a very narrow one, only likely to appeal to the members and supporters of smaller parties.

This poll, conducted by ICM Research, asked whether people were in favour of a referendum on changing the way we elect MPs. It found

For a referendum being held        69%
Against a referendum being held 17%

This poll was conducted by ICM Research for the Joseph Rowntree Reform Trust. Its relevant findings are summarised below:

This poll was conducted in Wales by Beaufort Research for the Electoral Reform Society. It was commission to test electors’ knowledge of the Sunderland Commission report and asked for views on that Commission’s recommendation of electoral reform for local government in Wales. Results were:

This poll was commissioned by the Electoral Reform Society and conducted by NFO WorldGroup (formerly System Three). It examined public attitudes to the Scottish Executive’s proposal to change the voting system for Scottish council elections.

To the question: "How fair do you think the new system of proportional representation would make local council elections compared to the current system?" responses were:

On being asked: "How do you think proportional representation would affect local government?" responses were:

Total

Cons

Lab

Lib Dem

SNP

Other

25%

21%

20%

33%

28%

34%

28%

36%

32%

26%

30%

30%

21%

18%

22%

21%

21%

14%

9%

14%

10%

8%

10%

10%

5%

8%

7%

5%

4%

1%