The Additional Member System (AMS, or Mixed-Member Proportional (MMP) as it is known in much of the world) has been used since 1999 in elections for the devolved Scottish Parliament and National Assembly for Wales. But AMS is not used in identical form in the two nations. The division between constituency and list members in the two institutions is as follows:
Scottish Parliament: 73 constituency MSPs, 56 regional list MSPs
National Assembly for Wales: 40 constituency AMs, 20 regional list AMs
Not only is the total number of elected representatives much greater in Scotland than Wales; so also is the proportion elected from the regional list (43.4% against 33.4%). This difference is more than just an arithmetical detail. With a smaller explicitly proportional element the Welsh variant of AMS is far more likely to produce disproportional results. That indeed has transpired. The chart below shows Gallagher Index of Disproportionality figures for the four sets of devolved elections thus far; on each occasion, the Welsh figure exceeds the Scottish one, indicating less proportional results in Wales.
The levels of disproportionality seen in Wales are similar to those normally experienced, for instance, in Australian lower house elections, conducted using the non-proportional Alternative Vote system. They are also notably higher than those typically experienced for any EU member state that uses an explicitly proportional voting system in parliamentary elections. This shows that AMS, as applied to Welsh Assembly elections, cannot seriously be described as proportional representation. The system deployed in Wales can more accurately be described as semi-proportional representation.
What would be needed for similar levels of proportionality in Wales as in Scotland? A 70-member Assembly, with 10 additional regional AMs, would produce a situation where List members comprised almost the same proportion (42.8% of the total) as they do in Scotland. In the 2011 Assembly election, 10 additional list members would have produced the following overall outcome:
Labour: 28 Constituency + 5 List = 33
Conservative: 6 Constituency + 12 List = 18
Plaid Cymru: 5 Constituency + 8 List = 13
Lib-Dems: 1 Constituency + 5 List = 6
Wales’ Gallagher Index score would then have been 8.97 – compared to 7.45 in Scotland, and 10.47 in the actual 2011 Welsh Assembly election. An 80-seat Assembly, with members split equally between Constituency and Regional List, would enhance the proportionality of AMS even further. In 2011, this would have produced the following result:
Labour: 28 Constituency + 8 List = 36
Conservative: 6 Constituency + 14 List = 20
Plaid Cymru: 5 Constituency + 10 List = 15
Lib-Dems: 1 Constituency + 5 List = 6
UKIP: 2 List = 2
Green: 1 List = 1
This yields a Gallagher Index score of 6.53 – much closer to the norm in properly proportional voting systems.
One other difference in the implementation of AMS between Scotland and Wales has emerged since 1999. A ban on ‘dual candidacy’ (i.e. the same person being a candidate for both a constituency and on the party list) was instituted by the UK Labour government for the 2007 Assembly election, but has not been introduced in Scotland. During the first two sets of devolved elections dual candidacy was permitted, as occurs in most countries that use AMS, and prominent Labour figures (notably Rhodri Morgan and Sue Essex) were both constituency and list candidates in 1999.
The dual candidacy ban was introduced in a rather partisan manner by the Labour party, apparently seeking to make life difficult for its opponents. It prompted acrimonious arguments, and the main justification offered for the change (that dual candidacy ‘both devalues the integrity of the electoral system in the eyes of the public and acts as a disincentive to vote’) was never supported by any substantial or credible body of evidence.
A detailed investigation of AMS was recently conducted in New Zealand, where the system was introduced to replace First Past the Post in the 1990s. The final report concluded that:
“Candidates should continue to be able to stand both for an electorate [constituency] seat and be on a party list” (p.37).
This conclusion was reached for several reasons: one has particular pertinence to Wales:
“It is both proper and desirable under MMP that political parties can protect good candidates contesting marginal or unwinnable electorates by positioning them high enough on their list to be elected.” (p.38)
The Assembly has only 60 members – as many have argued, an inadequate number for the Assembly to do all aspects of its job properly. Given the paucity of AMs, it is particularly important that the quality of those elected be the highest possible. Yet banning dual candidacy actively inhibits parties giving the best chance of election to their best people. Removing the ban, as has been suggested by the current UK government, would be wholly welcome.
As with dual candidacy, partisanship almost certainly accounts for the low level of proportionality in the Welsh variant of AMS: Labour offered the minimum proportional element necessary to win Plaid Cymru and Liberal Democrat support for its devolution proposals in the 1990s. Many within Labour, though, appear to resent even this modest concession to proportionality: Labour have dominated the majority of constituency seats at every Assembly election and without the list element would therefore have won clear majorities every time.
To some extent, inter-party disputes about electoral systems are inevitable. Politicians and political parties are unavoidably ‘interested parties’ with regards to all aspects of electoral arrangements. That is why non-partisan Boundary Commissions exist to review constituency boundaries – to avoid attempted gerrymandering by those currently in power. The distinctly flawed Welsh experience of reforms to AMS suggests that this model might perhaps usefully be broadened in remit, so that decisions on other aspects of the electoral system also involved avowedly non-partisan bodies.
Were powers over electoral arrangements for devolved elections ever given to the National Assembly, it would surely also be a good thing if super-majorities (i.e. two-thirds or three-quarters approval) were needed for changes to be introduced, to ensure that a broad, cross-party consensus supported any change. Any future reforms to the electoral system for the National Assembly should certainly be aimed at enhancing politics as a whole, not merely at helping the electoral prospects of one political party.