The Additional Member System (AMS), otherwise known as Mixed Member Proportional (MMP), is one of the leading contenders as a Proportional Representation system for UK General Elections. However, as this piece describes, it can turn alarmingly disproportional when the number of parties in contention increases to the levels we are seeing today: five parties in England, and six in Scotland, are now polling at over 10%.
AMS/MMP has been used for the Scottish and Welsh parliaments, and for the London Assembly, since devolution came in in 1999 – though the Welsh Parliament (Senedd) has decided to use a closed List system from next year’s election onwards.
Recent polls for next year’s Scottish Parliament election were widely reported as suggesting that the SNP are predicted to win 62 of the 129 seats (48%).
What has been less reported is that this is despite their predicted proportional entitlement being only 43 seats (33.3%). How can this happen in a supposedly proportional system?
The answer lies in what are known as ‘overhangs’, where a party wins more of the constituency seats, elected using FPTP, than their proportion of the list vote justifies.
For example, in a Scottish electoral region with 9 constituencies and 7 list seats, if a party gets 33% of the list vote, its proportional entitlement will usually work out at 6 seats (out of 16). So if it wins 8 of the 9 constituencies, which is quite likely if there are many other parties splitting the votes, it has 2 more seats than its proportional share. The ‘top-up’ component of the system will give it no list seats, but it already has too many for proportionality. What is or can be done about this?
In Scotland, as in the London Assembly and formerly in Wales, a party is allowed to keep its overhang seats, and other parties’ shares are reduced. According to analysis by Ballotbox Scotland current polling data suggest that in next May’s Scottish Parliament election the SNP would have 19 overhang seats; these would come at the expense of Labour (19 instead of 25), Reform (17 instead of 23), Conservatives (11 instead of 14), Liberal Democrats (10 instead of 13) and the Greens (10 instead of 11). Note that one implication is that pro-independence parties (SNP/Greens) would have a majority in the parliament of 72 seats (56%), against their proportional entitlement of 54 (42%).
Germany, which was the first country to adopt an AMS/MMP system, has had a similar problem of overhangs in recent years. But it has dealt with it differently, prioritising proportionality above constituency entitlements. In both 2017 and 2021 it allowed parties to keep overhang seats, but added extra list seats to maintain strict proportionality at national level. This required adding 111 and 133 seats respectively to the German parliament in those two elections. [And that was with a modification that allowed numbers of seats to vary disproportionally between regions (länder); to avoid that would have required many more additional seats.]
Following those results, Germany has now adopted a different way of dealing with overhangs, so as to keep the Parliament at a fixed size (630): if a party wins too many constituency seats in a region, some are disallowed, using the percentage achieved in their constituency vote as criterion. In this year’s election 23 constituency winners were disallowed, with those constituencies left without an FPTP elected representative. The alternative under the previous arrangement would have again required adding well over 100 seats to the parliament.
If we look for the broad reason why these systems are not working as intended, it lies mainly in the increased number of parties in contention. This is because the more parties there are in contention, the lower the vote a party needs to win all or nearly all the FPTP constituencies in a region.
There is a standard statistic for the ‘effective number of parties’, introduced by Laakso and Taagepera in 1979, np = 1/sum(p_i^2), where p_i is the proportional vote for party i. If, for instance, you have 5 parties in contention with equal votes, so that each p_i=1/5, then the formula gives np=5 as you would expect; if their votes are widely different, np will be significantly lower.
In both Scotland and Germany until recently, support among the main parties has varied sufficiently that the effective number of parties has been roughly in the range of 3 to 4 and there have been few overhangs. The exception to this was Scotland’s ‘rainbow’ election of 2003, when np was 4.7 and there were a significant number of overhangs: 7 for Labour in central belt regions, and 2 for the Liberal Democrats – one of them at the expense of Labour – in the Highlands and Islands region.
The effective numbers of parties in the German elections of 2017 and 2021 were similar: 4.6 and 4.8.
In Germany’s 2025 election np rose to 5.5, while in the opinion polls behind the latest seat predictions for Scotland np=5.4. In last year’s Westminster election, the effective number of parties rose sharply, from 2.7 in 2019 to 4.0. Current opinion polls for England, with five parties very much in contention, give an estimate of 4.6, not quite as high as in Scotland, but close to the levels that upset Germany’s system from 2017 onwards. These polls suggest that it is Reform that would benefit from the disproportionality under AMS, similarly to the prediction for the SNP in Scotland, at the expense of Labour and other parties. It could lead to a right-wing majority on a minority of votes.
Proportional Representation is clearly the answer to the democratically unacceptable disproportional results of FPTP elections – especially recent ones – but if the number of effective parties remains high, anyone advocating for AMS/MMP as a replacement needs to address its overhang problem. Otherwise, AMS/MMP seems likely to fail on the most basic PR criterion, the proportionality of seats won to a party’s votes.
* Denis Mollison is a former Chair of Liberal Democrats for Electoral Reform, and has been a member of the party since joining the SDP in 1981.
Follow @libdemvoice.org on Bluesky
Like us on Facebook
Subscribe to our feed
Sign-up for our daily email digest
Why non-fiction books need their indexes
Changes in status: Local Elections wrap up
What should the Liberal Democrats stand for?
Ben Wood
Ed Sanderson
Jack
Matt (Bristol)
John Grout
32 Comments
Totally agree. NZ also uses MMP and they have a good system for eliminating the overhang. Simply create additional MPs to ensure proportionality. NZers like their electoral system and have voted several times to keep it. You also need party vote to be entire country, not areas/districts. The way MMP is used in this country is a joke. MMP, when used properly, is the most proportionality, easily understood and explained (which is very important) and protects the local MP.
The problem Germany had with adding seats to deal with overhang is that they insisted on adding new seats to every region and did so just by expanding the size of every region’s representation and awarding the seats like normal until the overhang was gone, and no overhang was allowed at either regional or national level meaning an insane number of new seats were needed.
You could do it with allot fewer extra seats if you just awarded the minimum number of seats needed to bring the overhang down to no more than one seat and assign the extra seats to the most underrepresented parties nationally, and choose whichever candidates of theirs came closest to getting elected to award the seats to.
The problem with the Scottish form of AMS is that the system used was a compromise agreed by, and designed to meet the needs of, Labour and the Liberal Democrats, in preparation for the re-establishment of a Scottish Parliament in 1999. Labour wanted to keep its then dominance in Scotland, that FPTP gave it, by agreeing to a form of AMS where 73 of the 129 MSPs would be elected in constituencies by FPTP – Labour expected it would therefore always be in government either by itself or as a senior coalition partner. The Liberal Democrats wanted to elect a large enough number of MSPs via the regional lists that would enable it plus Labour to be able to form a majority government. Having 7 MSPs to be elected per region by corrective PR meant that the Liberal Democrats expected to be able to win 16-24 MSPs (2 or 3 per region, either by winning individual constituencies or gain seats via the list).
Now this cosy arrangement worked in 1999 and 2003, but then the SNP began to rise and soon replaced Labour as the dominant party, winning most constituency MSPs.
The solution to the overhang problem – if we must stick with AMS – is to alter the proportions so that fewer MSPs are elected via individual, and more elected via regional lists. The lower the proportion of constituency MSPs, the smaller the overhang problem will be.
Jenny Smith mentions the establishment of P.R. for Holyrood in Scotland. It’s worth noting there was also a re-organisation of local government.
It all produced a hybrid scheme of directly elected Holyrood constituencies plus huge Holyrood regional seats elected by a form of PR and fairly typical Council Wards (over twenty miles long in the Borders).
In my bit of of southern Scotland I have a Westminster M.P., three ‘local’ Councillors elected under PR, a directly elected MSP and seven regional list MSP’s. elected by a form of PR in an area stretching over 160 miles from Stranraer to Dunbar.
Whether this plethora of representatives produces better government and decisions is a nice question, but they all get salaries and index linked pensions – which must cost a bit.
I do wonder why in all these PR discussions no-one mentions Roy Jenkins’ AV+, which had the merits of:
1) retaining the constituency link;
2) requiring anyone elected in a constituency to receive over 50% of the votes, not just first preferences;
3) including an Additional Member system to provide proportionality.
We should not get hung up on achieving exact proportionality; in elections other than for Parliament we would be quite happy with STV in seats returning ~5-7 members and we know the level of disproportionality which that size of constituency implies.
Which is why we should campaign for STV which is more proportional, makes nearly every vote count and avoids having two categories of elected members
TBH the simple answer to this in Scotland is for the Scottish Parliament to adopt STV for its elections. Technically, there remains a majority in Parliament for this – it’s still, I think, SNP policy – so all it needs is for someone to propose it.
The question is, though, would the SNP dare to be the turkeys who voted for Christmas?
The problem with all electoral systems is that they all have problems STV is arguably the best over all but still has problems. I think it would be better if PR supporters recognised this
@Russell. MMP does not protect the local MP. As others have said, you need most of the MPs to be non-constituency and those that are will have to represent bigger areas. If the constituencies are three times the size, they are viable (if not ideal) STV seats.
The current local government reorganisation means that most of England will be covered with ‘Local’ authorities of a sensible size for STV seats. They may not be real communities in many cases but there is an advantage to MPs representing an area coterminus with an authority.
“if a party wins too many constituency seats in a region, some are disallowed, using the percentage achieved in their constituency vote as criterion.” Surely this makes the constituency element meaningless. Voters in the affected constituencies would feel cheated if the person they thought they’d elected was “disallowed” solely on the basis of numbers and proportionality. It’s bad enough that someone the voters thought they’d kicked out returns to Parliament via the back door of the party list. Forfeiting the seats of legitimate winners is even worse.
I don’t know what planet @Andrew Tampion is on; supporters of electoral reform are often very well versed on the flaws of specific electoral systems.
@Peter. NZ has 60% constituency MPs. The NZ system, how it was implemented and received is worth a look if anyone is interested in proportional representation. The UK experience of AMS/MMP has been poisoned by the ridiculous systems they’ve ended up with. MMP beat STV by a huge margin in the public vote.
The NZ electoral reform referendum process is also worth looking at, given brexit process. The public had several votes. 1 Do you want to change? 2 if change, which system? 3 status quo v preferred system. 4 after next ge, still happy? 5 after next ge, still happy?
As everyone else has said, 60% constituencies seldom gives proportionality. New Zealand has just been lucky.
Of the many things referenda are bad at, determining whether a system is fair to minorities ranks pretty high.
The main problem with any electoral system which includes constituencies electing a single member (even if this is then made more ‘proportional’ with other members elected to create proportionality, such as an AMS system, thereby creating a two-tier pattern of representatives) is that this removes choice by the electorate BETWEEN elections. It is a mirage to suggest merit in a system where electors choose a single ‘local’ representative, as though this somehow helps the average elector, because the person elected CANNOT represent the views of all the voters in that constituency; and in an election with 5 or 6 parties contesting with roughly equal levels of support (as is now becoming the case), that means the ‘local’ representative would share the views of only 20% of the voters (or even a smaller proportion). The voter seeking help with a problem would generally have to rely for that help from a ‘local’ representative whose views and beliefs were diametrically opposed to solving that problem. This is already the main problem with the current FPTP system. There is no value in such a system, and the voters’ views would thus be better met by having several (at least 5 or 6) members elected with equal status in a multi-member constituency.
Alex Macfie — I think Andrew means ‘I think it would be better if a pro-PR campaign was more loudly honest upfront about this, and the tradeoffs involved’. Which I agree with but I think I (and Andrew, if that’s what he means) am being naieve in terms of how political communication works these days.
Overall I agree with Dennis from the point of pure proportionality. AMS is a hybrid system, and the joints show.
I’m not sure I agree with the solution Dennis proposes, as it attacks the very compromise which is at the heart of AMS. Indeed its whole point.
And there are many who would say the degree of disproportionality in AMS compared to Westminster FPTP still looks like the sunlit uplands of utopian political representation.
Very few of our political systems in the UK were designed for this number of viable parties and this degree of split vote (even the STV ones, with relatively small constituencies – possible exception for 5-member wards in Scottish Councils). I’m sure many people are simply (also naievely?) hoping it goes away and is a passing phase.
The reality in uk general elections is that although you’re really electing a PM it is via an MP. That’s why MMP works. It recognises the historic connection with local constituency AND provides a fair and proportional result. The BIG advantage of MMP over STV is that it is easily understood (and therefore explained). AV is quite straightforward a system but explaining that in the referendum was difficult for people who wouldn’t give 5 minutes of their time. NZ was a cut and paste of the UK. Looking at what NZ did is instructive. NZ didn’t “get lucky”.
AMS as used in the UK is nothing like MMP as used in NZ
For proof that STV is best, look at the Republic of Ireland. Elections to the Dail used to produce party strengths to within 2% of the share of first-preference votes This also happened in 2020. In 2025, the largest three parties got proportionately slightly more seats than they should. As in FPTP, the Greens with their votes thinly spread lost out. Generally, however, smaller parties’ seats can still align with votes under STV. And of course voters can seek help from a choice of representatives in STV multi-member constituencies
MMP doesn’t have margin of error. If you get 5% of votes you get 5% of seats
That would be 32.5 MPs. There is always a margin of error.
More seriously, At the last election, on your 60-40 split, Labour would probably have won 248 constituency seats. Their share of the vote would only have allowed them 219. That is a significant overhang.
The perspective I’ve moved to – entirely personally – is that I would rather a moderately disproportional multi-member system where candidates are as elected as individuals, there are no lists, voters can mix-and-match candidates from different parties and there are not so many candidates and choices that voters can’t meaningfully compare their options.
AMS – with its mix of different types of candidate, its closed list system etc – just doesn’t cut it for me, viewed ideally.
However proportional it is, I’m not convinced 5-to-7-member STV would genuinely encourage the same degree of meaningful participation from the average voter, particularly not those immediately transitioning from FPTP at its crudest, safe-seat-est (where we might get, even now, no more than 3 meaningful, actively campaigning non-‘paper’ candidates).
The old (forgotten) Swedish system of sequential proportional approval voting, or something similar, would work for me, in uniform three-member constituencies. I know Denis would not approve. Personally I would prefer it over STV, at least as an entry into (semi-) proportionality. This is a niche and obscure opinion and I expect no one else here to hold it.
People were told AV was movement towards PR. People didn’t believe that. It has to be full PR (MMP obviously) or nothing.
@matt – I agree, the approach where voters can mix-and-match is more democratic than party lists, and better fits local needs. It also means each member of “the team” has some level of democratic accountability in that people have had to explicitly vote for them.
I get that may mean capable candidates, who don’t present well on the doorstep not getting elected, but given in the last few elections in my area we hardly saw anything whether in person or leaflet of many candidates, I doubt this really is a hindrance to the major parties.
One big difference between the electoral reform referendums in NZ and the AV referendum in the UK is that NZ had fact-checking. So there weren’t any spurious claims like the need for “expensive voting machines” under AV (ironic when many electoral reform supporters oppose e-voting), or the heartstring-tugging “He needs [X] not an alternative voting system” stuff based on over-inflated costings and the idea that spending can simply be transferred that way (the £250M supposed cost of changing the system was hypothetically spent multiple times; only one of those could ever have been done). At least one claim made by opponents of MMP in NZ during the confirmation referendum campaign, that the system would prevent any reduction in size of Parliament, was disallowed after its first outing and was never mentioned again.
If we are to have referendums on anything at all (whether electoral reform or, say, EU re-entry) ever again in this country, then we should follow the example of NZ and fact-check all claims made by campaigners.
Sorry, I think it was £250k not £250k. But my point still stands. No-one has ever pointed to anything like the things mentioned in the No2AV ads that was done because we didn’t switch to AV.
Not £250M!
@ Alex Macfie The Liberal Party could have had PR in the 1918 Representation of the People Act, but Asquith /Lloyd George et al chose not to do it. It’s all in Hansard if anyone cares to read it. Asquith thought the new PR constituencies ‘would be too big’.
As for Clegg & Co back in the recent Coalition, did they ever seriously believe they would get it in the way it was done ?
@Alex Macfie
“If we are to have referendums on anything at all (whether electoral reform or, say, EU re-entry) ever again in this country, then we should follow the example of NZ and fact-check all claims made by campaigners.”
Who would do the fact checking and who carries the cost?
@Nonconformistradical: Fact-checking could be done by a beefed-up (and re-liberated) Electoral Commission or a new public body. As for who pays, that should probably be part of the cost of running the vote, although it also seems reasonable that any campaign organisation that makes a claim ruled to be false should pay the costs of investigating it. The facts themselves are usually available (the government said there were no plans to start using voting machines if the AV referendum passed, and for Brexit the £350M claim on the side of the bus was widely debunked, including by the ONS), but there was no-one with the power to nuke the false claims, so they remained part of the conversation (the more they were debunked, the more airtime they got). There is already a law against making false statements about opposing candidates, so it shouldn’t be too difficult to extend it to outlawing other claims that are known to be factually incorrect.
@David Raw: While I don’t disagree with your assessment about Nick Clegg, I doubt that the campaigns in any future referendum on electoral system will refer back to the one that he presided over. Most voters hardly remember it. And they certainly won’t care about political machinations from a century ago.