National Assembly for Wales

Good morning, everyone. Can I welcome you all to the Committee on Assembly Electoral Reform? We've not received any apologies for this morning's meeting. The National Assembly operates through the medium of Welsh and English languages, and there are headsets for simultaneous translation on channel 1 and for sound application on channel 0. As this is a formal meeting, Members do not need to operate the microphones themselves. In the event of an emergency, an alarm will sound and ushers will direct everyone to the nearest safe exit.

So, we move on to item 2 on electoral systems and boundaries. Can I welcome Professor Roger Awan-Scully and Jess Blair, who are here this morning to help inform our inquiry on electoral systems and boundaries? I think probably the best thing for us to do is to move—. We've had your papers, so thank you very much for that. I think probably the best thing for us to do is to move straight into questions to maximise the time available to us. Can I start with a few questions, please, and start by asking you whether you agree with the expert panel's assessment of which principles should be used to evaluate electoral systems?

Thank you. Shall I go first? Okay. Well, thank you for that question. I think drawing up a definitive list of the principles that should underpin a good electoral system is a very difficult task. It's quite difficult to say, 'These are the principles, these only, and no more.' I think the expert panel's report was a very good piece of work in total. I think this is a pretty good go at developing a very comprehensive set of principles. But, I think, in most circumstances, even principles as good as the list of 10 that they gave are often going to be in some tension.

We should also note that, depending on the context, principles that are very important in some electoral contests may be much less important or even wholly irrelevant in others. For instance, people have often talked about the ability to form a stable Government as being one important principle with electoral systems. Well, in things like, say, electing police and crime commissioners or electing Members of the European Parliament, which we've done in the past in the UK—where Government formation is not at stake, therefore—that principle is not really relevant in those contexts. Whereas, in other contexts, it may well be judged to be very important, possibly the most important principle. So, we do need to be very much aware that the different sort of contexts we're talking in can make some principles more or less relevant.

Absolutely. We broadly support the expert panel's principles, which they used. What we would say is that there is, obviously, a balance between them and some are clearly more important than others. For example, we would back proportionality, voter choice and an equal mandate as being quite fundamental there; also, obviously, measures to increase diversity.

You did talk—I'm trying to think whose paper it was, but there was this balance between promoting proportionality and avoiding hyper-proportionality. Did you both comment on that? I can't remember whether those were your views, Roger.

I haven't submitted any written evidence to this panel, but I think there is a balance that can be drawn reasonably satisfactorily there. Yes, there are certain types of very proportional voting systems that do tend to lead to very fragmented party systems; that in Israel is often cited as an example. But, there has been some very good work in political science, for instance, by John Carey and Simon Hix who have talked about there being an electoral sweet spot you can find, where you have relatively low district magnitude in the number of representatives per area, while still having an overall strongly proportional formula that can lead to good levels of proportionality between the major parties, but without fragmenting representation into dozens and dozens of tiny parties. So, it is possible to arrive at a reasonably good balance between these principles. It's not just a strict trade-off, you can have a bit more of one or a bit more of the other. You can, potentially, I think, find a fairly good place that balances the two pretty effectively. 

This is where the importance of setting a threshold comes in, really, doesn't it?

Yes. Whether that is a formal threshold, say of 5 per cent or whatever, as they have in Germany, or, if you have relatively low district magnitudes, relatively small numbers of representatives per area, then you have in practice an effective threshold that may be of a similar magnitude or slightly bigger. 

Okay. One of the panel's principles was about Government accountability and effectiveness. How would you define effective Government? What does that mean to you in terms of the context of voting systems?

So, I would say there are a few different ideas of Government effectiveness, and one of those is obviously that Government effectively represents what voters have chosen. Obviously, we saw in the 2019 general election the Conservative Government having a majority of 80 with only a 1.3 per cent increase in vote share from 2017, so there's an element, I think, of consideration about how voters are properly represented by their Government and making sure that voters aren't frustrated, and building that kind of trust there. But I think there's also that argument that Roger mentioned earlier, about not having fragmented systems where you promote lots of different parties. You want to try and build as much consensus as possible, while not necessarily delivering a majority Government all of the time. 

I would just add to that that I think there's been a tradition in the UK of thinking that effective Government means almost by definition clear single-party Government with a strong parliamentary majority. Now, I think some of the systematic evidence on that is far from clear. For instance, the groundbreaking multinational research that was done some years ago by the Dutch political scientist Arend Lijphart suggested that, if anything, political systems that have more proportional voting systems, whilst they certainly do not have obviously worse policy outcomes—whether you look at economic growth rates, delivery of major public services like health and education, or whatever, it's not clear that necessarily single-party Government is more effective. Single-party Government may well be more effective in helping that particular party deliver its electoral manifesto, but of course that would be implementing both the good ideas and also the bad ones. Overall, it's far from clear, trying to look at the international evidence in the round, that single-party Governments with a majority actually produce better government. They may well make it easier for those holding the reins of power to get what they want done, but getting what they want done is not necessarily the same thing as governing well. 

Sure. And we've seen examples, haven't we, of—. I take Italy as an example of where we have constantly changing Governments. So, we want to avoid that kind of situation, because that doesn't lead to stable Government, does it? 

No. Well, of course, the Italian situation—I think our traditional view of Italy is really based more on what happened in Italy in the 1950s, 1960s, 1970s, and, actually, in the last 20, 25 years, while one can certainly find things to criticise about particular Italian Governments, it hasn't been a case of a Cabinet reshuffle every month, or a new Government formation every month. And part of that is due to the fact that they changed the electoral system in Italy in the 1990s. My own view is that, again, in all these things there needs to be a balance, but you can find electoral systems that deliver a reasonable amount of proportionality that don't lead to a highly fragmented party system that might well often produce minority Governments or coalition Governments, but don't necessarily need to do so by aggregating 12, 15 parties or something into a Government. That's the sort of level—if you get into that sort of situation, with a very fragmented party system, that's I think when you have real obstacles to any sort of effective or coherent Government. 

Yes. Anything you want to—

No, I fully agree with that. 

The next point I wanted to explore with you was about the complexity of some of the systems, and how easy it is to explain some of these systems to the electorate. Is it more important that the electoral system is fair, or is it more important that it's simple and people understand what they're doing?

I think you can do both, actually. So, obviously, we support single transferrable vote as our preferred electoral system of choice for Senedd elections, and we've seen it used across the UK really successfully, and seen voters interact with it really well. I think probably the best example is how quickly Scottish voters got to use STV and got used to using STV when it was introduced for Scottish local elections. So, we—in our report, which was written by Professor John Curtice, he said that, in 2012, the second STV election, you saw an 8 per cent increase in voters using second preferences under STV. So, you're seeing, actually, voters really very quickly adapting to different systems, using them properly and understanding how they work. I think you can deliver that alongside fairness and proportionality. 

Yes. Clearly, it is important that the voters should have some broad understanding of how they should cast their votes and also, ideally, it would be certainly preferable that voters have a fairly clear understanding of how their vote leads to a political outcome down the line.

I have to say I think there are two levels of complexity we need to be aware of, though: the complexity of an individual voting system on its own terms, but also then within the broader political context in which that voting system might be applied or introduced. I do think we actually have problems with that already at the moment in Wales—well, actually, in both regards. With regard to the additional Member system, for instance, it seems very clear from all the different ways that scholars have tried to look at it that lots of voters really don't understand particularly what the regional list vote is about. You could see that, just—. A very good PhD student of mine, Jac Larner, has tried every possible way you can cut the data to find whether there's clear evidence of voters using their regional votes strategically, and it's very difficult to find this. You could take this more simply just by the fact that, for instance, in the three south Wales regions last time, lots of people voted Labour on the regional list vote and, frankly, they might as well have just set fire to their ballot papers, because the chances of any of those votes leading to Labour delivering a regional list seat were absolutely infinitesimal. So, I think there's a problem with the current system at the moment.

But there are also, I think, problems at the moment where we have, obviously, these different electoral systems being used. And now one unintended consequence of the wholly justified decision to delay this year's local and police and crime commissioner elections is that, next year, we're going to have, for the second successive time, on the same day as the Assembly election, where we have these two votes and voters don't seem to wholly understand what at least one of those is really for, another simultaneous election also using a sort of two-vote system, but a different one. Again, this will be the only election for which people use that particular two-vote system. So, people are going to be asked to cast, in effect, four votes on the same day, and it's very unlikely that many people will actually really understand the implications of what all of those four votes are. So, I think we already have significant problems in this regard. 

In 2016, when this happened in 2016, certainly in my constituency—I can't speak for others; I don't know whether you'll know—we found a lot more spoiled ballot papers in 2016 than we'd seen previously. Was that generally what we picked up?

I don't recall and I don't have in front of me the systematic figures across Wales for the spoiled ballot papers. People have sometimes criticised STV saying that, back in 2007, when it was introduced, there were a lot of spoiled ballot papers, but actually the spoiled ballot papers were on the AMS vote for the Scottish Parliament—an issue of really bad ballot paper design. STV—I think there's an issue of complexity for parties, which you may want to get to in a minute. In terms of the voters, it's not clear that it's really any more complex than the current system. As long as you can basically count one, two, three, four, five, six, you can engage with STV as a voter. I think there are issues of complexity in terms of the counting and in terms of party strategy, but, for the voters, I think it's far from obvious that it's any more complex than what we have now. 

Okay. And, just in terms of the preferred systems, Roger, you haven't necessarily identified a preferred system, but, Jess, you have. Is part of that that you believe the connection with the constituency is important—the local connection, the constituency link. Is that what led to your conclusions, partially, or—?

Partially. I think one of the main reasons we would endorse STV in this circumstance is because of voter choice. Obviously, voters get a lot more choice on the ballot paper than they might under any other system, realistically. But, yes, STV has shown, particularly in Ireland, to deliver that local link still and retain that constituency element. Also, obviously, we really endorse the expert panel's recommendations of an integrated gender quota too.

Okay. I've just got one final question, before I ask Dai Lloyd to come in. Have you seen any evidence, or can you point us to any evidence, that shows that changes in voting systems actually change voter behaviour?

Okay. There is an awful lot of, obviously, international research trying to look at the impacts of electoral systems, and that does show fairly clearly, for instance, that more proportional voting systems, particularly those that allow for representation of smaller parties, do tend to lead to somewhat more fragmented party systems. People are more likely to vote for smaller parties in a context where they think that vote might count, whereas, say, under first-past-the-post, those parties may well get squeezed, certainly in general elections. 

It's more difficult to look at the impact of electoral system change, because electoral system change in a context of otherwise political stability is actually a relatively rare thing. But we can see, for instance, very direct evidence ourselves within the UK. If we look at our history of European Parliament elections, we can see some very clear evidence there, and the first four European Parliament elections that were held in the UK were held under first-past-the-post, except in Northern Ireland. Then, from 1999, we moved on to the use of the regional lists, and you see a very clear acceleration of the fragmentation of the party system there, and a decline in the dominance of the Conservatives and Labour, reaching to the point last May where, between them, they won something like 22 per cent of the vote. So, there's some very clear evidence there. 

Obviously, we have had the recent experience that Jess knows plenty about of the introduction of STV, the local government elections in Scotland. That's another example. One of the main changes—. Well, I think there have been a couple of main changes there, but that's got rid of the one-party fiefdoms that you tended to get in some places. It's also meant that, in Scotland, unlike in Wales, you've tended to have all of the seats being competitive. So, you can certainly see from some examples of electoral system change—. But electoral system change, unless there's a broader political system change, is still actually relatively unusual in established democracies, for the fundamental reason that people who are elected under the current system are almost by definition beneficiaries of it, and to get approval for reform you generally have to get those people to then vote to change the thing that they are by definition beneficiaries of. 

Huw. 

Do you have any examples of where parties that have prospered under the first-past-the-post system previously have adapted, have modified, their approach to prosper also under a proportional system? Because all the things you've mentioned so far flag up danger signals for, if you like, dominant parties. 

I would say, actually, moving to a more proportional system opens up constituencies where a dominant party might not have been able to play before. So, if you move to a system, say, where one party gets 50 per cent of the vote and you move to four seats in a multimember constituency, what that actually means is that you can compete in other areas. So, for a dominant party—and I'm assuming they're not dominant across the whole of Wales—it opens up those constituencies that might not before have been in play. And I think that's a benefit for larger parties and also existing smaller parties to compete with. It just means that there's more of an equal playing field across the whole of the country, rather than in specific areas. 

That's really helpful on a conceptual basis of what may happen, but I'm really interested in, historically, what has happened as well. 

Yes. As I said, it's actually quite difficult to find particular examples of this, because electoral system reform is often part of a much broader tumult in the political system, and it's often been the downfall of existing dominant parties for other reasons that then prompts the electoral system reform. Obviously, very close to home, I suppose we can see an example here just in Wales, where the Labour Party won the last 27 general elections here in Wales. We then introduced AMS with a degree of proportionality for the National Assembly, but Labour have still been the dominant party here in the Assembly as well, and the fact that there was that proportional element introduced has not fundamentally undermined Labour dominance. 

We have, of course, also seen in some established and reasonably stable democracies in recent times a substantial decline of some of the traditionally dominant parties—places like Spain, places like Germany, Belgium as well—without electoral system reform as well. As I said, I think change to the electoral system is often more of a symptom than the cause of fundamental changes in the party system, which is exactly, for instance, what we saw in Italy in the early 1990s.

Thank you for that. I wonder if I could ask something that may be equally as difficult to look upon from a historical concept, but, again, it's to do with the way that politics and political parties change under a more proportional system. Do we have any evidence, from what has happened before, that, as well as that fragmentation, and the introduction of new parties and new actors within a political space, there is any fluidity amongst the mainstream parties?

We look back to under a first-past-the-post system, where there was a famous example of a Plaid Cymru/Green Member of Parliament, Cynog Dafis, in Ceredigion. Is there any evidence—? Conceptually, I can see that this might be something that might happen, but where, let's say, there is a Conservative/something candidate, a Labour/something or an SNP/something, it clearly happens within the mix there, because when you form a Government you have to do it, but is there any evidence of either individuals standing who say, 'Well, I'm actually a Labour/whatever', or, alternatively, standing on a combined platform?  

Again, I think this is relatively unusual, although if we go back quite a long way in British electoral history, of course, the emergence of Labour as a serious parliamentary party was facilitated by the Lib/Labs—that was part of the whole story of the emergence of Labour as a serious representative political party.

Most commonly, what happens is that if the electoral system really places very strong incentives for there to be a very small number of parties, then you tend to get, most often, obviously, broad churches, and the arguments get contained within a different political party. If the electoral system allows for more parties to have chances of winning representation, then those differences will tend to lead, you know, to rather more brands being out there and on the ballot paper.

Of course, we do have the very famous example in the UK of parties working together of the Liberal/SDP alliance in the 1980s. If there'd been a more forgiving electoral system then, who knows, maybe the SDP would've continued to be a stand-alone organisation in the long term, but that was, clearly not, politically, going to be a runner.

I just wanted to—. Michael Gallagher, an academic at the University of Dublin, has done a lot of research on elections in Ireland, and I just want to quote something that he said, which was that

'rivalry within constituencies expresses itself not in candidates' staking out distinct policy positions but in their seeking to earn reputations...as local promoters.'

So, it's less about individual policy and more of that real embeddedness in that constituency.

Yes, okay, thank you for that. We'll move on to the next set of questions from Dai Lloyd, please, Dai.

So, first-past-the-post, you know, we use, in its classic variant, for Commons elections in the UK. It's relatively unusual, actually, by international standards, but it is used for general elections in Canada. It's used in pretty much all sorts of things like elections to the House of Representatives and many legislatures in the United States and in provincial elections in Canada as well.

It is clearly a very simple system. There's simplicity for voters, there's simplicity for the parties and the electoral administrators. It’s a very simple electoral formula. Everyone knows who their representative is. And I was just checking some evidence that we gathered in the 2016 Welsh elections study—a large representative sample of voters in Wales—and we asked people a simple question and gave them a list of five plausible-sounding but false names, and then the one correct name: ‘Can you pick out your constituency MP here?’. We gave them only 30 seconds to answer the question so that they couldn’t Google it and we found that—excuse me while I just double-check the stats here—we think it was 70 per cent, I think, of people picked out their constituency MP and 53 per cent also correctly picked out their constituency AM. So, there is quite a level of recognition of those individuals. Name recognition for regional list AMs in their region or MEPs—we still had them at the time, obviously—was much, much lower.

So, I think there is quite a strong attachment to this idea of the constituency representative. Of course, there are potentially different ways of still taking that, as we have for the Assembly: we have the additional member system, which has a constituency component, but also the regional list component. And, as Jess was mentioning earlier, I think one of the internationally recognised strengths of a system like single transferrable vote is the way in which it facilitates a pretty good level of proportionality, but also very strong local representation. That is a somewhat different form of local representation and you don’t have the one person who’s identified with a specific geographic area—you have them in a slightly larger area; a small number of individuals who all represent that particular locality.  

In terms of the strengths of first-past-the-post, we judge them to be rather limited, especially when you consider the consequences. So, as Roger mentioned, the simplicity of first-past-the-post is often praised, however, it’s actually led to voters not just understanding the system, but understanding how to ‘game’ the system and tactically vote, and all of those things that we think are quite negative components of the first-past-the-post system.

The other thing, obviously, is that it drives a huge lack of voter choice. So, 52.2 per cent of voters in Wales in the 2019 general election didn’t vote for their MP. That’s under first-past-the-post, so I just think that the downsides massively outweigh the strengths.

Well, of course, the 2011 referendum was essentially a choice, for Westminster, between first-past-the-post and one specific alternative. It wasn't like processes in the past, where, for instance, in New Zealand in the 1990s, they asked people, first of all, 'Are you broadly, in principle, in favour of reform?' and then go to, 'Well, are you then in favour of a specific reform?' This was, 'Do you want first-past-the-post, or this very specific system?' which is actually not very much different from first-past-the-post; it's retaining single-member district systems, but instead of the plurality based on a categorical vote, it's an ordinal vote. 

And I think there are two fundamental reasons why that lost: one is that it was such a limited reform, it was very difficult to get many people that enthusiastic about going out and campaigning for it. And the second reason that it was lost is Nick Clegg, to put it quite bluntly. It was very effectively associated with him and by that stage, for reasons I think we're all familiar with, he was a deeply unpopular political figure. And one of the common findings from international research on referenda is that particularly if it's an issue that people don't have long-standing, deeply rooted feelings about, they often take cues from which political elites are supporting or opposing a thing. And I'm afraid, by that stage of his career, Nick Clegg was pretty toxic. If there had been a referendum on nicer weather, and he'd been backing it, then it might have been lost, I think, simply because it was associated with Nick Clegg.

'We don't agree with Nick.' [Laughter.]

Exactly. Sadly, politics, as you will all know, is a brutal game. Twelve months previously, his association with AV might well have worked well, but by that stage, it was a disaster for the advocates of that cause.

We would echo much of that. I think referendums are rather blunt tools and can often be hijacked by other issues. The British Columbia citizens' assembly in 2004, when it was actually given a lot more time to consider different electoral systems, they came down and recommend STV. So, that kind of tool, that kind of deliberative model would work much better in this kind of case.

As I mentioned earlier, there are always inherent difficulties in campaigning for electoral reform. Obviously, it depends on the exact nature of the political process, but you normally, at some stage, will have to get endorsement of that reform by people within the Chamber who have been elected under the current system. And there's the 'turkeys voting for Christmas' sort of thing—people who are, by definition, pretty much beneficiaries of the current system, have to vote to change it. And there's always something of a leap into the unknown. But I think, with relatively new institutions, maybe in places where you already have experience of using some different electoral systems, it may be more possible to get a mood in favour of change.

Taking this out to the public, if you were going to think about having a referendum on changing the electoral system, what we should maybe learn from the 2011 AV referendum—and possibly from other more recent experiences of referendums as well—is that you do need to try and explain to people, try and have some sort of process by which people can receive clear understandings of, 'Okay, if we vote one way, this happens; if we vote the other way, this happens.' Whatever way people end up voting, they ought to be clear as to what are the consequences of their vote and have the alternatives explained clearly to them, and lots of opportunity to engage, not just with the campaigns of the two sides, but also with neutral, non-spun information that tries to explain the technicalities to them of what it is about.

To be quite blunt, to be honest, I think the challenges in Wales of getting this reform through are party structures and people in positions of power in those parties who are nervous about implications for their own parties. But I do think we have encouraging signs. I think there's leadership in a lot of parties currently sitting in this Senedd actually coming around to these conversations. The fact that this committee has been established, I think, is a really promising sign there.

I think one of the other benefits for this conversation happening in Wales is it's not taking place in a vacuum; it's being considered in the round with an increase in the size of the Assembly, in the size of the Senedd, and also measures to promote diversity. So, I think that's a really powerful set of reform packages there, and it's something that really should be pushed through by people in those positions of power.

Thank you.

Thank you. Huw.

Thank you. If we can move on to some of the recommendations of the expert panel around different types of proportional systems, one of the recommendations was around flexible list systems as one of the possible ways forward. Could you just take us through what you see as the pros and cons of flexible list systems?

Flexible list systems could be implemented on pretty much the same constituency boundaries as STV, in that sense that systems are similar. You also, under a flexible list, as you have under STV, have the ability for voters to choose both between parties and between individual candidates. You enhance the degree of voter choice. Under closed-list party systems—also, by the way, first-past-the-post is a closed-list party system, with the district magnitude of one—you, essentially, if you're a supporter of the party, have to take the candidates or candidate that you are given. Under open list, as well as STV, you have some ability to choose both individuals as well as parties. The clear difference, of course, is that, under open list, you’re still basically coming down with a single, categorical vote for one person and/or party—so, giving all your vote to them and none of your vote to everybody else. Whereas STV, of course, the whole nature of the voting system is that it is a gradation of preferences, and different people, obviously, could have different views about whether it is better in a voting system just to have an all-or-nothing decision or whether it is better reflecting the nuances of feelings and whatever to have a gradation of preferences as the way you cast your vote. That is the central difference between the two.

In terms of proportionality, in terms of delivering—

In terms of proportionality, obviously a great deal will depend on—I’m sure the committee will be familiar by now; there is an amazing amount of arcane detail I can get into. So, the precise mathematical formula you choose—is it, say, D'Hondt, or Sainte-Laguë or modified Sainte-Laguë or whatever, the size of the constituencies, the district magnitude—a lot of that can affect things crucially.

We at Cardiff University have, both for the McAllister panel report and, actually, even previously in some work we did with the Electoral Reform Society, tried to model this as best one can—based on the behaviour of voters who voted under one system, what would have happened under others—and we’ve taken the best information that we can get and applied appropriate health warnings.

We do believe that it is likely that STV would be, more often than not, a little bit more favourable to, broadly speaking, as I say, parties of the centre left than open list. And the fundamental reason for that is, and I’ll try and explain very simply, that a big proportion of those who would, say, vote Labour in an Assembly election, their second preference would be Plaid Cymru. The converse is also true. And so that would likely mean that, under STV, quite a lot of lower preferences being transferred from, say, Labour to Plaid Cymru. There is also—. We found, say, in 2016, that, quite a lot of people who were then voting for UKIP, the Conservatives were their second option. So, there would be some more on the right-of-centre parties, but this effect seemed to work more powerfully for the benefit of parties that are broadly left of centre. So, they would, under most scenarios, do just a little bit better overall in STV than open list. You’re not talking about large numbers of seats, but, obviously, in what, even in an expansion, would still be a relatively small Chamber, one or two seats here and there could be important.

Sorry, can I just bring David in a minute here? 

I'll just put a counter argument to that, if I can. As you say, the centre left would be represented by a number of different parties, so someone, if they're centre left, would vote for maybe one of four parties, whereas centre right might only have two parties at the centre right—say, for instance, my previous party and the Conservatives. So, someone voting who was centre right would only have one other alternative party to vote for, whereas someone centre left would have four parties as alternatives to vote for. Wouldn't that counteract the effect?

I'm really glad you made that point, David, thanks, because what we were not really able to seriously account for in the modelling is how the parties would deal with these systems, in particular STV, as Jess could tell you a lot more about. STV raises some very important questions for the parties about party strategy and how many candidates you stand in each area. If you stand too few, you miss out on winning some seats, as, for instance, Sinn Fein did in Ireland recently; if you stand too many, you fragment your vote too much, and if voters don't all transfer votes properly, you can end up with lots of people getting respectable votes but almost none of them getting elected. So, parties would need to—. I think there'd be very high fees being charged by Irish political consultants. Parties would need to learn how to play the system quite quickly and make some important strategic judgments, but the issue that you've raised there, David, I think, is precisely why STV would tend to be a bit more favourable than open list, because, under open list, you might have, say, perhaps three or four parties centre left, you vote for one or you vote for the other, and the vote could be fragmented in several ways. Under STV, you do, at least, have the possibility to have your lower preferences going to those other parties broadly on the centre left, so those votes can be cycled around those different parties and aggregate towards eventually getting somebody elected, provided the parties have got their candidate strategies in the right sort of ballpark.

Can I just—? Before I bring Huw back in, can I just ask, on that—? If we had—. Under STV, of course, you don't have to use all your preferences. So, that, in a sense, can also skew results and can also see people voting tactically. So, would you advocate, if we were going to use STV, that we should have a compulsory voting system, where people have to use all their preferences, to get that proportional result?

We wouldn't advocate for any kind of compulsory element to the vote. What we have seen in Scotland, as I mentioned earlier, is that over time people learn how to use the system and use more preferences. So, in Scotland, in 2012, at the local elections, in seats where you had four representatives, 88 per cent were using their second and third preferences. So, you start to see that kind of increase as time goes along. What I would suspect is that, the first election under STV, you're likely to get a much lower use of preferences than in subsequent elections, and it will build over time. 

Yes, and I personally would not favour the idea of forcing people to transfer their votes all the way down the ballot because, I think, certainly, particularly in the first one or two applications of the system, that might lead to a very substantial number of spoiled ballot papers. So, I think it would be preferable to avoid that.

This would be, I suppose, another element of party strategy that would need to be thought about quite intensively if we were to introduce STV—what attitudes do the parties take to each other. So, under an open-list system or closed-list system or under first-past-the-post, essentially parties have a purely competitive relationship for the votes: 'You either vote for us, or you vote for somebody else, if you vote at all'. Under STV, there is the possibility to send out somewhat more nuanced messages, that, 'Okay, look, if you really can't vote for us on the first or second preference, but you prefer us to those other people, maybe give us your third preference or your fourth or something like that to overall ensure that parties in our broad area of the political spectrum do better and it isn't that those people that we really dislike end up getting the seats.'

Again, that would be one of the elements where, I think, a lot of the big questions are not so much how this impacts directly on the voters, but how the parties would, say, take a political system and seek to apply it and the sorts of messaging that they would need to do, and that may well need to be of a rather different type to what we've traditionally seen in the UK.

Sure, okay. Huw. 

Thanks. I just wanted to, Jess, take your thoughts on the list system, open systems, closed systems, flexible list systems and what not, but, before I do, I agree with your—well, it's more than an assumption with you—analysis that, based on historic and current voting patterns, if you were to move to STV, there would broadly be a centre left, not a bias, but it's a bias within the population, and, with the current options on the table, they would—. But that doesn't take into account, of course, that things can change quite rapidly over time. Who's to say that, in a decade—? And I'm saying this—I'm not a disinterested observer in this, I'm a participant, but you could have a centre-right bias within—. So, I'm just interested in your response to that, because I utterly agree with you that, under historic voting patterns, that's where we would be, but, actually, it could change.

Yes, and this is, I suppose, maybe one specific manifestation of a more general point that electoral systems, the way in which they function, can change over time. So, we've seen in the UK that a traditional argument for first-past-the-post was that it delivered clear, single-party majorities, and then we saw for two out of three general elections, oddly, it didn't. We saw, from the 1970s onwards, first-past-the-post suddenly started—because there was a rise in people voting for other than Labour or Conservatives—to be much more disproportional in the outcomes it was producing. Electoral systems can change in the consequences they have.

I think, yes, the best we can really do as researchers is try and get the most accurate information we can at the time, and there are several times we've gone back and tried to do this for STV. That's all been, obviously, from about 2011 onwards. We've come to broadly the same conclusion, that it tends to be a bit more favourable to parties of the centre left.

If there were a fundamental reorientation of the political preferences of the Welsh people in some way, then, of course, it could end up working in a fundamentally different way. It's impossible as researchers to gather systematic evidence on that, of course.

Yes. The reason I mention that is, regardless of what electoral system we have, whether it's traditional first-past-the-post or a form of proportionality, it is still there for political parties to make their case, and, as you say, they need to play the game of whatever electoral system there is, but they need to make the case to either continue to win the arguments for centre left or harder left, or the centre right or harder right, regardless of what the system is, and bring the public with them. Sorry, it's an interesting one, because sometimes we get into this mindset that, if you shift to a certain system, it's going to have a prejudice towards a certain outcome only based on historic voting patterns, which are a good indicator—it's the only indicator we have—but it could change entirely.

So, Jess, your take on list systems.

Yes. So, in particular, the flexible list system, we judge it to have a lot of merit, particularly in countries that it's widely used in, which—it is one of the most common systems used in the world. It has delivered quite a high level of proportionality. You can bring in thresholds and change district magnitudes to ensure that there isn't this element of over or high proportionality, where really small parties gain a lot of representation.

We also find that flexible lists generally return representatives in proportion of what people have actually voted for. So, that's obviously a really big positive. There's also that element of voter choice, in that voters can choose whether to elect a party or an individual from that party, which I think is a very positive thing. But, again, as Roger said, it's very much a case of you get less voter choice than a preferential system, which we would prefer, because it does give a lot more nuance and a lot more representation of what voters actually want.

The nuance, for you, is important.

I think so. And I think elections where people only have one vote are quite blunt elements and don't really represent actually how people are actually probably feeling about their local area or their constituency or who should represent them, and giving voters preferences allows a lot more nuance to come through.

Yes. And I suppose I would just add to that. It's not so much to disagree with Jess, but I'd maybe put it in a slightly different way. Jess said STV gives you more choice, open list gives you somewhat less, and I suppose I'd put it: open list, it's a fundamentally different type of choice that you make. STV is this sort of gradation of preferences: 'I'm a bit more favourable to this, a bit less to that.' Open list, there is still this fundamental thing that you ultimately choose to back this, and therefore not everything else.

Okay, thanks. Can I turn to what this feels like, what this looks like, to voters, if we were to make a change to either a flexible list or to STV? We've touched on the issue of how voters can and do adapt to these systems; we've seen that, historically, it can be done. I'd be interested in your thoughts on how that process could be managed in terms of public awareness, information, et cetera, to lead up to that in transition to it.

But, also, what do we know from the evidence about what this does? The big thing that electoral reformers always say about voter engagement and voter turnout: do we have hard evidence that this does actually improve voter turnout? In an age of—. We keep on hearing about people who are not just apathetic towards politics, but genuinely hostile to it. Does this turn things around?

Just in terms of some of the first bits of what you're asking about there, Huw—so, I think if you move to open list or if you move to STV, then I think, clearly, you would need to try and make sure there was some sort of voter information campaign. But I don't think—. It's not that big an ask of the voters. Actually, a large number of voters would have some experience of preferential voting if they've been, for instance, members of trade unions or members of political parties that have used preferential voting in the past, or if they've been students. A pretty large proportion of people, actually, would have some experience of preferential voting, and you'd need to reiterate that to them, and have it explained to them that they were casting their votes in order of preference.

With open lists, there would probably be a smaller jump still for the voters. You're still casting your vote as a single x or whatever, you would just need to have—. I think for open lists particularly you'd need to think quite seriously about the ballot paper design, just to make sure that that was constructed in a way that was very user friendly for the voters, and they could understand the nature of the choice that they were being asked to make there. So, I think, on that, it's evidently doable.

In terms of actually getting people to engage more, the evidence very broadly internationally is that more proportional systems, even once you've accounted for other things, have a modest upward impact on voter turnout in elections. But the dominant factor in this is still only a fairly minor effect. I think, Jess, you probably know more about some of the recent research that's been done, for instance, particularly on the introduction of single transferable vote and things like that. You might want to talk about that.

Yes. So, one of things we would say about any kind of voting system reform is that it should be seen in the round about how the public engages and understands devolution and politics, firstly. We know that that is rather limited in Wales. So, I actually think a good communication campaign around reform and changes to the voting system would be quite helpful in also communicating what the role of the Senedd is to the public. So, that's one positive there. We would obviously say that any change to the voting system requires a very effective communication campaign, and it requires the public to be properly informed about that change.

In terms of evidence, particularly in Scotland where STV was obviously introduced, our stats are slightly skewed because the introduction of the first STV election in Scotland was held on the same day as the Scottish parliamentary election. So, it's quite hard to read into exactly the impact that you had under STV. I think we would agree with Roger that what you tend to see is a very, very small increase in voter turnout, and much more engagement as years progress in terms of how people engage with those ballot papers. 

I'm conscious of time, but I just want to ask if there's anything you want to add about the impact of either flexible lists or STV for political parties. Now, there is one significant party that isn't represented here, which is the Conservative Party, and we'd love to see them here, but they aren't, so I'll speak for them, in a sense. [Laughter]. What would the impact be for political parties in terms of the way they approach it, which you've touched on, and in terms of the impact on them, including the main parties and also emerging minority parties as well? 

I think, if we were to move to STV or a flexible list system, then you're looking at somewhat differently constructed boundaries. So, whichever of those we move to, parties would have to be thinking about candidate selection within different geographical units than what they've maybe used in the past. So, there's an element there you'd need to think about.

There's then a more dynamic element of how parties, when the system is up and running, campaign and message around it. In open list, I think the challenges are a little bit less for the parties. You're still talking about, 'Vote for us rather than anybody else', and there aren't quite the same issues of how many candidates you stand and the voter management side of STV. So, I think for the parties, open list is probably a little bit less demanding. There would though still be this dynamic of voters potentially having the option to choose both between parties and between individual candidates. The question that parties will all have to think about is: 'How do we manage that? Do we go for a clearly identified No. 1 candidate, or do we just try and have some sort of balanced ticket and use that to hoover up as many possible votes from different sub-groups in the electorate as possible?'

I think STV, on top of all of that, imposes a somewhat further degree of demands for political parties. They have, as we were talking about earlier, this whole thing about how many candidates do you stand and the optimal strategies. They have things like, if you've decided to stand two candidates in a multimember constituency, how do you play that? Do you, as they do in Ireland, for instance, sometimes, roughly geographically split things? So, in half of it you say, 'This is our number one candidate', and in the other half you say, 'This is our number one candidate', and try and get them both fairly high up the preference order. Do you identify one clearly as the lead candidate? There might, at times, be difficult issues for parties in managing the internal dynamics of rivalries between people within the party and in the same area.

And then finally there's the issue, as was just talked about a few minutes ago, of how parties handle the lower-down-the-ballot-paper preferences and their relationships with other parties. Because under STV, as I said, you're not just a competitor purely with other parties—yes, you are that—but you also want to have a nuanced message to say to people who are strong supporters of some other party in broadly the same area of the ideological space: 'Okay, we might not be your number one choice, but perhaps we'll be your number two choice.' That would be a rather different type of messaging that parties would have to send in the UK. They would need to learn how to do that.

Yes. Anything you want to add, Jess?

I think I want to echo Roger's comments about boundaries, and I would say that is something that hasn't really been touched on that much today. I'm quite interested to see what the committee recommends in terms of appropriate boundaries for Wales. I think there's a question about coterminousity between either Westminster boundaries or local government boundaries in Wales.

Our view is that a boundary review should be established to properly look at these issues. In an ideal world, you would have consistent boundaries between different electoral arrangements and different systems in different elections. But I think there's a question here about Wales following the lead of another institution that might not necessarily have those same boundaries in five, 10 years. I think that's something that absolutely has to be considered.

Yes, thanks for that.

Can I just ask you both, because I'm conscious of time, are you both okay for a few more minutes after 11:00? Yes, you're fine. Thank you.

Just one final question, and it's to do with if we don't take this big leap but if we actually look to continue with the mixed-member system that we currently have—but, of course, there are different types of mixed-member systems: Germany, New Zealand and others—if we were to continue with it, are there suggestions you would have about how we might modify and make some improvements to what we've got?

Obviously, this is partly in the context of the related issue of increasing the size of the Assembly. If there were a decision to do that, I mean, we'd obviously have to do something to the electoral system. Now, there are ways, and the McAllister report talks about how you could adapt the additional member system with roughly the same level of proportionality as you have at present to a larger Assembly.

That brings in issues of coterminousity of boundaries, of course, yet my own personal preference would be for a somewhat greater level of proportionality. Obviously, you can do that very easily with the current boundaries when everyone just adds in more list members, but that has very clear advantages and is good for some parties, but much less good for others. All parties can be expected to defend their own individual interests there.

Yes, you can adapt the current system to a larger Assembly. That is doable; I think the McAllister report said that can be done. But on balance it is probably less preferable than looking at these two other alternatives. I think they came to a very nuanced judgment on that, and it's one that, on balance, I would overall agree with.

Absolutely. One of the fundamental things that we are advocating for is an increase in the size of the Senedd, and I really do feel that that has to be fundamentally addressed. Obviously, with that, you need to then address the electoral system, as Roger said.

One of our concerns with continuing the use of AMS is that there would be a limit on the number of representatives you could have under a larger Senedd without a boundary review, and I think that that would get rather complicated. Given there's a clear lack of understanding about how the list system works anyway in Senedd elections, it's clearly not that popular or embedded, at least with the public, I do think there's merit for looking at a new, fresh system that can probably deliver that larger Senedd.

Okay. Thank you, both. David.

Jess, you'll be glad to know that some of my questions are about the boundary review, but before that, if I can seek the Chair's indulgence, I want just to talk about the apportionment of votes after they've been cast. We know all about the D'Hondt system, but we also had in the papers an example of the Sainte-Laguë—whichever way you say that—system, and Professor Denis Mollison says that there's something called a weighted inclusive Gregory vote that's used in the Scottish Parliament. I won't ask you to explain any of these. Simply, I'm going to ask—. But he says his preference is for the Meek system of apportionment. Now, I haven't had time to look at either of those two, but do you have a preference for the nuances of the apportionment? Which one may be the best?

I haven't myself been able to look at Denis's evidence there, so it's probably best I don't comment on the detail of it. Depending on the system you have, there can then be slightly different nuanced mathematical formulas. So, we have a mathematical formula, D'Hondt, that is used right now, for instance, for allocating the regional list seats within each region. There are alternatives, like the Sainte-Laguë formula or a halfway house modified Sainte-Laguë.

Of those, D'Hondt is slightly more favourable to larger parties, Sainte-Laguë is somewhat more favourable to relatively smaller parties. So, if you were thinking about applying this to an open list system—. We did some modelling on this: D'Hondt brings out a more favourable result to parties that are first, second and third in terms of vote share at the time; Sainte-Laguë is more preferable and gives a few more seats to the parties that might be fourth or fifth most popular in Wales at the period. We've produced a whole bunch of simulations for the McAllister panel.

It doesn't make a huge difference, but if we increased all the way to 90 Members, it would still be a relatively small Chamber. Two or three seats here and there could actually, post election, be important in terms of what Governments are possible. The choice of particular mathematical formulae inter-relates with the size of the district magnitude and, overall, you have to try and come to some sort of balance. But of the available mathematical formula under PR, D'Hondt is actually a relatively simple one, so at least we can have some degree of simplicity on those matters, and that is probably better.

We tend to back D'Hondt. It's very much a case of whatever suits the specific environment and specific requirements, especially district magnitude.

So, in our submission, should we be recommending a system of apportionment?

I think there are maybe first order decisions to be made in terms of the basic system: should it be STV, open list, AMS, or whatever? Then, the McAllister panel got into looking at different types of boundaries and whatever. 

There are, sometimes, ways in which you can trade these things off. So, if you have smaller constituency sizes the Sainte-Laguë formula might work out as equally favourable to smaller parties as larger boundaries with more members but D'Hondt. So, these two things can balance out a little bit if we overall go for larger constituencies of more members. That tends to lead to more parties winning seats. But then, if you can balance that by using the D'Hondt formula, that may lead to a similar outcome as if you have smaller constituencies but with the Sainte-Laguë formula.

Ultimately, decisions will need to be made on this, and I think what we tried to do for the McAllister panel was model as many plausible scenarios as possible to see, on the best evidence we had, how much difference does it make. It doesn't make a huge difference, but one or two seats could well be important, post election. So, it's certainly worth trying to get as much evidence as you can on this.

Okay, fine. Thank you very much. If I can move on, then, to the boundary situation? What, if any, impact will the 2018 Westminster boundary review have on electoral reform in Wales? Do you think it will have an impact?

Obviously, we have in the Assembly thus far maintained coterminousity of Westminster and Assembly constituency boundaries, which of course they stopped doing in Scotland. If we do move to a system whereby there's equal representation—equal constituency size across the different nations of the UK as well as within it—then inevitably that reduces significantly the number of Welsh MPs. If we want to retain coterminousity, then we need to think about doing something fundamentally different for the Assembly electoral system.

I think—and Jess has already talked about this—while, other things being equal, it is probably better to have one set of constituency boundaries because that's simpler for parties and voters, in practice that is almost certainly going to mean the Assembly electoral system being held hostage, if that's not too pejorative a word, to decisions that are made at Westminster for elections to the House of Commons, which will probably be made without any regard to how this Assembly should be elected. So, I think it's therefore only right and proper that the McAllister panel looked at other ways of trying to develop constituency boundaries, and one was local authorities. Again, in terms of simplicity, if you combine Assembly boundaries with local authority boundaries, that would be good, but of course local government boundaries are not necessarily set in stone either. So it may just be that we have to move to a completely different set of boundaries for the Assembly, which means a little bit more complexity. But, overall, that's not a major deal. It would be good if it could be avoided, but maybe it can't, and as we've seen in Scotland, that is perfectly workable and doable.

A lot of these decisions, we feel, should be made by a boundary review, a full, proper one before 2026 if a new voting system was going to be implemented. What we would say is that the expert panel's suggestion of almost pairing the current 40 constituencies, and developing multimember wards from those 20, is probably a nice place for a boundary review to start, because it does move away from that dependency on either Westminster or local government models, and it does drive a much larger Senedd as well.

Thank you. How much variation is appropriate in the number of electors per Member in each multimember constituency?

I know the Electoral Reform Society have thought a lot about this issue. My own personal preference is that you have the minimal possible difference in terms of constituency size and number of Members per constituency while respecting as much as possible natural communities, and things like that. We have provision at the moment in the UK in other places for recognition of particular circumstances. So, in Scotland, for instance, Shetland and Orkney having their own individual seats for the Scottish Parliament or whatever, although that's not really justified in terms of the population. And I think, as much as possible, if we could try and formulate constituencies around reasonable natural boundaries, like Swansea, for instance, or Newport, or something like that—what seem like relatively natural communities should be the locus of representation. If that means a little bit of movement up or down in terms of the number of Members, then I think it's probably preferable to accept that than cut lines on a clearly very artificial basis of just having lines on a map. I think most people would rather have one more or one fewer representative in what felt like a genuine community than have an absolutely rigid application of a mathematical formula with complete disregard for what feel like natural areas to be represented. 

Of course, in south Wales we've got the Valleys as the communities, et cetera. 

I think that's absolutely right. Our official position is that between three and seven delivers the right kind of proportionality on a general basis. However, we need to look at a specific Welsh example here, and I think it is right to go alongside communities where possible. Broadly, you would want to see an equivalent number of electors per Member elected, and I think it's also important to consider that that might be more complicated in rural areas, which actually might need looking at more specifically in the Welsh context. One of the concerns in Scotland under STV is how large some of those rural wards are for local elections, so I think that's something that we'd want to avoid in the Welsh context.

So, moving on just a little bit, with regard to do you support incorporating factors such as geographical areas of deprivation or distance from legislation in calculating how many Members should represent each multimember constituency? We are now really twisting the system.

I think geographical areas, definitely, as we referred to in terms of communities. I think it would be quite difficult to try and measure in distance from the Senedd as one of the indicators, and obviously, I really do feel that population size, in terms of number of electors per Member, and natural communities are the right two main principles to go by.

There are countries, I think, a few examples of places that have tried to implement this. I understand Norway have a quite complex mathematical formula where they have thrown several of these things into an equation and that generates particular results and consequences for representation. I think it would be quite a complex argument to have, but I don't see any fundamental reasons in principle why you couldn't think about maybe modest overrepresentation for areas that are further away and may, either geographically or socially, feel marginalised and isolated. I think we just need to make sure that, as with some of these higher level decisions on voting system reform, these are done on the basis of a broad consensus because, obviously, within the technicalities of a formula like that, there's a potential for individual parties to pursue their own personal self-interest. So, you'd try to make sure that it was done in a fairly open and transparent and consensual manner. 

Okay. My fourth question was about coterminousity, which we've talked about. I'll just make one comment: coterminousity with the parliamentary constituencies electing 40 Members under our system now was much easier to explain to people than going on to the regional system and trying to explain to people how the regional AMs are elected. It did make it much easier for me to explain it to people.

Yes, and as I said, straight after the election in 2016, we looked at this, the Welsh Election Study, and we found that lots of people knew who their constituency AMs were and knew who their constituency MPs were, but when you get down to the regional level, voter knowledge drops off quite substantially. If you just look at the way that people cast their regional votes, lots of them really don't understand to any degree of sophistication the way in which that whole process works. 

Yes, I would say that, absolutely, in an ideal world, we should have similar boundaries between the Assembly and Westminster; however, I just think, realistically, it might lead to a situation where the Senedd chooses boundaries just on that basis that are, perhaps, less good for itself, and then Westminster might reform their own boundaries. So it's very much a game of balance and pragmatism, and judging what is the most likely scenario to happen. 

Huw. 

Thanks, Chair. Sorry, I'm being cheeky as we head to the end of the session. So, away from the technicalities of boundaries and types of electoral reform and so on, you mentioned during the course of this discussion that one of the biggest difficulties is actually taking these reforms, if there was agreement on them, through an institution that, in some ways, is small 'c' conservative in terms of these reforms, or appears to be at the moment. There isn't an obvious consensus.

So, let me just give you the opportunity, if you want to take it, to—. Plaid Cymru are on board with these reforms and my own party has accepted the need for additional Members, but there is a difference between that and taking a final step towards that. Now we've been talking about wider electoral reform of the systems here today as well. How would you argue to the leaders of the Conservative Party and my own Labour Party in particular that this was in their interests, as opposed to being conceptually ideologically sound in the interests of the voter, in the interests of voter engagement, that it was in their interest to take these reforms forward?

I think there are two main areas that we would focus on and one of them is around information. We know that a lot of people are quite nervous about reform because they don't have all of the information about what that might mean for their party. So, one of the things we've been trying to do is really engage with different political parties and give them a better idea of what it would actually mean for them in practice. I think that there has been a change in mood in the Labour Party in particular, and there's certainly been an increase in the number of Members thinking about it at least or starting to back reform. So we can see that shift happening. 

The other thing is obviously modelling what it would actually mean in practice in electoral terms. It's a risky business, because you don't know whether that's necessarily going to be a good thing for any party or not, but I think it is worth having the information to hand, and that is something that we are going to do later this year. 

I have long been convinced that there is an absolutely compelling case for increasing the size and, particularly, the legislative and scrutiny capacity of this institution. If you've decided that on principle then, as I mentioned earlier, you have to do something to the electoral system. We're looking at this and it's a good opportunity to review all the various possibilities. I think the McAllister panel has delivered a really excellent report in terms of what are the major possibilities here. It's given a limited menu of options for the political parties to engage with.

What I said earlier about it's always been difficult to get people to vote for reform, that was not particularly a point or a dig at this institution; I think that's the case with all representative institutions. Getting people who've been voted in under one system to agree to change it is always a difficult ask, and I don't think there's anything particularly deficient in Members of this institution in some of them finding that that's a process fraught with uncertainties. But if we actually want this place to be a serious law-making Parliament, then we have to do something about the size of the Assembly. If we have to do something about the size of the Assembly, we have to do something about the electoral system, and we now have, I think, at least a couple of very good proposals for how we could implement that over a period of time and have it in place for the 2026 Assembly elections. I think there are enough serious-minded people in this institution that they ought to be willing to engage with that, and find some sort of consensual way forward. 

Thank you. I think we've come to the end of that evidence session. Can I thank you both very much for coming in? It's been very, very helpful to our deliberations. You'll be sent a transcript of the proceedings to check for factual accuracy in due course. Thank you very much. 

Okay, if we can move on now to item 3, papers to note. There are several papers to note; I won't go through them all individually. The responses to the committee's consultation on electoral systems and boundaries as listed on the agenda, and included in the papers. Is everybody happy to note those papers?

Yes.

Okay, if we move on to item 4, then, which is a motion under Standing Order 17.22 to elect a temporary Chair for the meeting on 20 April. We will not be dealing with that item this morning. 

If I can move on now to item 5, a motion under Standing Order 17.42 to resolve to exclude the public for the remainder of the meeting. Is everybody content that we do that?