A central theme in the committee’s inquiry into the system for the election of the members of the House of Commons is the problem of proportional representation of political parties. Hitherto, members of Parliament have been elected in single seat constituencies. However, Parliament’s daily work relies on party affiliation, not on regional provenance. This mismatch has been the cause of many debates and initiatives for electoral reform. Similar electoral issues have been encountered in German history. Based on the German experience, we would like to sketch some ideas that may possibly aid in identifying feasible solutions for the Canadian problems.
In section 2, we give a short review of the essentials of the electoral system for the German Bundestag. In section 3, we turn to the Canadian House of Commons and sketch a hypothetical electoral system, tagged as “SMP and PRP”, that we view to be a natural enhancement of the current single-member plurality system.
I will begin with section 2.
I'm very pleased to appear here as a witness and to introduce to you the main features of the German electoral system. The double vote system for the election of the members of the Bundestag has become something like a democratic export hit. In contrast to other exports, like German diesel cars, it has caused no considerable mischief. Quite the contrary, it is held in high regard. Without doubt, this esteem is not undeserved, but there are at least some precautions you should take when implementing the German system.
To explain the system as a whole would be a very demanding task because the current German system is one of the most complicated systems in the world—but this is valid only for the way it works, not for the way it is executed. Furthermore, the complex intricacies of the German system are mostly due to its federal structure and the specific way the German system is adapted to that.
Although Canada is also a federal state, due to constitutional constraints, especially the fixed numbers of seats for every province, a one-to-one transfer of the German electoral system to the context of Canada is probably not possible. The only solution we could imagine, therefore, consists in the application of the system within each province separately. This is also the way it was used in the first German federal election in 1949. The separate application within each province makes things much easier. Therefore, I will concentrate on the main features of the German electoral system only.
In the literature, the German system is often referred to as a mixed member proportional system. The key point is its combination of two ballots that are used at two different tiers: the direct and personal election of candidates in single-member districts, and the voting for party lists in an upper tier, which is big enough to ensure that the proportional distribution of seats between the parties can be guaranteed. The intention was to maintain the proportional system of the Weimar Republic, but to complement it with the advantages of directly elected local representatives to which the citizens could establish a special personal relationship.
In the German electoral system each citizen has two votes. With the first vote—the Erststimme—the voter selects one of the candidates in his constituency. With the second vote—the Zweitstimme—the voter votes for a so-called Landesliste, which is a party list for one of the 16 Länder of the Federal Republic of Germany. For a better understanding, I will refer to the second vote as the “party vote”.
One-half of the seats in the German Bundestag are constituency seats and are attributed to the candidates who have won the plurality of first votes in their constituency. The second half are list seats. Only those parties that have won more than 5% of all valid second votes, or have won at least three constituency seats, are entitled to participate in the proportional distribution of seats according to the Sainte-Laguë procedure.
We can skip, for reasons of simplicity, the complex distribution of seats among the Länder. What is important is the fact that in the end each party is entitled to a certain number of seats according to its share of party votes. From that number, the number of constituency seats that the party gained in that Land is subtracted. The remaining seats are distributed according to the ranks in the Landesliste.
Persons on the list who have already won a constituency seat are not considered, so it is possible that the number of constituency seats achieved in one Land is higher than the number to which the party list is entitled. Such seats are called “surplus seats” or “overhang seats”. As long as there are no overhang seats, the distribution of seats among the parties is more or less proportional, depending on the rounding effects and the effective threshold.
The linkage between the single-member district tier and the upper tier of the Land guarantees that the whole number of seats of a party, including the constituency seats it has won, is covered by its second votes. Thus, usually the first votes are important only in regard to the personal occupation of seats. They are, with the exception of the emergence of surplus seats, irrelevant for the number of seats a party gains.
The whole purpose of this linkage between the two tiers is to correct for violations of proportionality that occur in the course of the distribution of seats in the single-member districts.
As mentioned, the situation is more complicated when overhang seats emerge. These are seats for which the appropriate [Inaudible--Editor] and party votes are not raised, so proportionality is violated.
One possible solution to restore proportionality could be to enhance Parliament until the overhang seats are covered by the proportionally distributed seats for the parties. This solution is applied in all electoral laws of the German Bundesländer, which also have mixed member proportional systems, and since 2012 it has also been valid in the federal electoral law.
This solution is obviously not available if the size of Parliament is fixed. In this case, restoring proportionality isn't possible as long as the gain of a constituency seat is guaranteed. It would certainly violate fundamental considerations of fairness to deny the winner of a plurality of first votes his constituency seat, so some deviations from perfect proportionality may be the necessary price for preserving the principle of direct representation by personally elected MPs in the single-member districts. At least, this price should be no higher than is required by the wish to satisfy our taste for fairness.
The best response to that problem consists in the attempt to prevent the creation of overhang seats whenever possible. Sometimes overhang seats are the consequence of the structure of the party system. Since this is a materialization of voter preferences, it cannot be manipulated. However, some overhang seats are created by strategic voting, especially by ticket splitting, which occurs if a first vote is given to a candidate who is not a candidate of the party to which the second vote is given. These overhang seats can simply be avoided by abolishing the two-vote system. Then, the voters have only one vote, which they cast for candidates in their constituency. The party votes, which are the base for the proportional distribution of seats, are calculated by summing up all personal votes for constituency candidates.
This system was also used in 1949 for the first German federal election, and it is still used in Baden-Württemberg. This would also have the nice advantage that parties have an especially strong incentive to nominate attractive candidates.
Mr. Pukelsheim will now continue with the presentation.
Since Confederation in 1867, members of the House of Commons have been elected using the single-member plurality system, which has its focus on constituency representation.
On the other hand, when Parliament convenes, it is not local representation that is dominant, but party affiliation. However, the number of seats a party holds in the House is visibly at odds with the support a party enjoys in the electorate. With our German background, we propose to rectify this representational mismatch by enhancing the current provisions in the direction of a system implementing single-member plurality combined with proportional representation of parties.
The Canadian constitution includes detailed rules to determine how many members of the House of Commons are assigned to each province and territory. In electoral jargon, these seat guarantees are referred to as district magnitudes. In order to meet the constitutionally mandated district magnitudes, our hypothetical system allots seats separately per province and territory. Hence, our model calls for 13 separate seat apportionments. The 13 apportionment calculations are split into two categories. The first category assembles the districts with a district magnitude too small for proportionality to take effect. The second category comprises the other districts.
In first-category districts we propose to maintain the single-member plurality system as is. The three territories belong to the first category because they command just one seat each. Evidently, a single seat is insufficient to achieve any degree of proportionality whatsoever. As a matter of fact, for proportionality to function properly, theoretical investigations recommend that the district magnitude should meet or exceed twice the number of participating parties. Therefore, we also place Prince Edward Island, with four seats, in the first category, and Newfoundland and Labrador, with seven seats. Whether to do so or not is a political decision. If so, the already large constituencies do not have to be enlarged yet further, which is good. On the other hand, the votes that are not cast for constituency winners are wasted, which is bad.
In summary, first-category districts enjoy the same electoral system as in the past. For the eight provinces in the second category, the old single-member plurality system is enhanced by proportional representation of parties. In order to set some seats aside for the system's proportionality component, we propose to reduce the number of constituencies. Of course, it is a genuinely political decision to fix the number of constituencies per province. For our hypothetical model we choose to roughly halve the number of constituencies—for instance, by merging two into one. Then, about half of the seats are filled by way of single-member constituencies, and the other half from party lists.
Specifically in the last two elections, our model apportions the district magnitude among parties using the Sainte-Laguë method, as it is known in New Zealand and continental Europe, or Webster method in the Anglo-Saxon world, or from a more systematic point, the divisor method with standard rounding, because it functions in the following way: a party's vote count is divided by an electoral key—that's the divisor—and the ensuing quotient is rounded to yield the party's seat number. For example, in 2015 in British Columbia, every 56,000 votes justified roughly one seat, and “roughly” reminds us that the quotients have to be rounded. All resulting seat numbers for the parties happen to meet or exceed the number of constituencies won by a party in a province in the last two elections.
In every instance there are enough seats for constituency winners. Any additional seats beyond constituency seats are filled from closed party lists. Closed lists encourage parties to promote social cohesion and to include under-represented groups. Generally, the legal provisions would codify not the ordinary divisor method with standard rounding, but it's direct-seat restricted variant. The variant inhibits the occurrence of overhang seats and thereby ensures adherence to the constitutionally mandated district magnitudes.
The direct-seat variant imposes minimum restrictions, which the ordinary method neglects. A party is allotted at least as many seats as are needed for its constituency winners. In cases where the minimum restriction becomes active, the required seats are transferred from the competing parties to those parties that feature an excess number of constituency winners. That is, proportional representation is compromised in favour of constituency representation.
Finally, we address the question of which vote pattern to use. The answer is as simple as can be: nothing changes. Voters are issued ballot sheets with the old design they are accustomed to. Every voter casts a single vote that is a composite appreciation of eligibility of a person and preference for a party. Our proposal only changes the law's scheme of evaluating the information supplied. In second-category provinces, where the old system is enhanced by proportional representation of parties, every vote is tallied twice: once for the candidate, towards constituency plurality; and once for the party, towards district-wide proportionality. The essential novelty that people need to understand is that now their votes are more carefully evaluated by the law by lending particular weight to party affiliation. At this juncture, every vote counts.
In our brief we illustrate our hypothetically proposed system by applying it to the last two general elections. Our proposal is seen to achieve more proportionality among parties than the status quo. Due to its hybrid character, it does not coincide with pure proportionality, but it preserves much of the charm of past traditions.
Thank you very much for your attention. Back to the new world.
Mr. Chair, I'm going to do the whole presentation. Thank you for inviting us to give evidence to your committee.
My purpose is to provide a brief presentation, which we hope will stimulate some questions afterwards. I'll seek to very briefly explain who we are and what we do. I'll then seek to explain Scotland's electoral systems and the four voting methods that we use, explain how we manage to finish up with four different electoral systems, and then raise some issues that might be of interest to you in your deliberations as we've sought to administer various elections in Scotland.
First, the EMB is the Electoral Management Board for Scotland and we have an electoral commission. I don't intend to spend too much time on this. Slides 1 to 8 of our presentation are really there for information, but I would probably point out to you the following salient points because we administer elections slightly differently than you do in Canada. Neither the EMB nor the electoral commission run elections in Scotland. That's important to note. That's the job of the individual 32 returning officers, of which Mary is one.
The Electoral Management Board seeks to coordinate returning officers during their administration of elections. Mary, as convener of the Electoral Management Board, has a power of direction for local government elections and recommends courses of action in other elections, such as a common issue date for postal boards across the whole of Scotland.
We, the commission, provide advice to governments on electoral law and guidance to returning officers and electoral registration officers. We register the political parties who campaign. We seek to ensure that candidates, agents, and parties, understand the financial rules within which they campaign. We administer the observer scheme for observers at elections and we also seek, through public awareness activity, to ensure that voters know how to register and how to vote, which is important because we have four different electoral systems. Both our organizations seek to put the interests of the voters first and what we do by thinking about our actions, how they will impact on the voters, and ensure that they impact in a positive way.
Turning to slide 10, which briefly lays out Scotland's four main elections, which we use a different electoral system at each time. The Scottish Parliament, established in 1999, has 129 members. It uses the additional member system—“MMP” to everyone else in the world apart from the Welsh, but we call that AMS. It's a combination of first past the post and a closed party list using the modified D’Hondt system for calculation. For our 32 councils, we use a single transferable vote to elect our 1,223 counsellors in 353 multi-member wards. The U.K. House of Commons uses first past the post; and for the European Parliament, we elect 73 members through a closed list in the U.K., with Scotland electing six members.
Turning to slide 11, we have 4.2 million electors in Scotland. We estimate that 86% of the eligible electorate is currently registered. A recent innovation in Scotland is that 16 and 17-year-olds can now vote in the Scottish Parliament and local government elections.
Turning to slide 12, I don't intend to talk about the U.K. Parliament electoral system, because you understand that is first past the post. For the European Parliament election, the six MEPs elected in Scotland are all elected for the whole of Scotland. Scotland is an electoral region for that election. We use a single X on a closed list and we use pure D’Hondt to allocate seats.
Turning to the Scottish Parliament elections, which were first held in 1999, we have 129 members. Of these, 73 constituency seats are elected by first past the post—that's 57% of the seats—and there are 56 regional seats, with 7 members elected in each of the 8 regions that Scotland is divided into. We use an X on two ballot papers, with one X on each ballot paper, and there is a maximum number of 12 names on a regional list.
If you turn to slide 15, there is a representation of the two ballot papers. The lilac ballot paper is the constituency ballot paper. You use an X for whichever candidate you have. On Scottish ballot papers you would see the candidate's name, the party, and a description underneath. It's not on this one that I show here, but there is also an emblem or a logo next to the box in which you put your X.
On the region, you can stand as an independent or individual on the regional list. Individuals appear at the bottom, alphabetically. Parties appear above the independents, and they also appear alphabetically. We use two separate ballot papers, unlike, I believe, in Germany—correct me if I'm wrong—and New Zealand. We did try a joint ballot paper in 2007. I think it's fair to say that poor design issues led to an increase in the number of rejected votes on that occasion. Politicians decided that we wouldn't use the combined ballot paper again, so we went back to two individual sheets of paper in 2011.
On slide 16 we've shown you a calculation for the regional list. How the seats are allocated under modified D'Hondt—essentially, the connection between the constituency and the regional votes— is that the number of seats a party wins in that electoral region is added to one to create the divisor. For instance, in the representation, party 1 has won two constituency seats, so its divisor is three, and that carries through. You then do simple division, and the one who has the most seats and most votes is allocated the first seat, until you've allocated all seven. As you gain a seat, your divisor goes up by one.
Turning to slide 17, for Scottish Council elections we have used the single transferrable vote since 2007 to elect the 32 councils. Again, we have 353 multi-member wards. The number of seats elected per ward varies. It was a very contentious issue during the passage of the bill in the Scottish Parliament. You either have three- or four-member wards, and I think that people who would argue that STV is the best electoral system would probably argue that you need more seats per multi-member ward, but politicians decided it would put three or four in Scotland.
If you turn to slide 18, there's a representation of the Scottish ballot paper. You vote your preferences—1, 2, 3, 4, and so on. You don't have to vote for all candidates; you can stop your preferences whenever you want. We tested that ballot paper with electors before it was used, and we saw that the instructions were very important.
It is the only electoral system we use in Scotland where numbers are used. When we first brought it in, it was held on the same day as an EMS election, and the concern was that we were using Xs as well as numbers, so we spent a lot of time making sure electors understood the difference.
To count STV, we use the weighted inclusive Gregory method, which we count electronically. It is thought that it would take us possibly two days to manually count a weighted inclusive Gregory STV election. Our colleagues in Northern Ireland manually count elections in which they use STV and they regularly go into the second day. However, the Scottish government agreed to fund the e-Counting Project. The system is common throughout all of Scotland. There are 32 count centres. The system isn't connected; they're individually set up, but it's the same system throughout Scotland to ensure consistency.
Slides 20 and 21 are an explanation of how STV works. I didn't intend to go into it, although my colleagues who administer the elections are happy to answer questions. I thought I'd move swiftly on to slide 22 and try to explain why Scotland has four different electoral systems.
Essentially it's the decision of politicians and government. Mary, Chris, and my job is to advise on the workability of the electoral systems that we employ. It's not our decision to say which one is best; in fact, we don't have any views.
For the four systems, the U.K. Parliament has always used first past the post. It's an act of the U.K. Parliament. For the European elections, member states agreed in the seventies that it had to be a proportional system, and the U.K. Parliament in the seventies passed an act that requires the use of a closed list. I think at the time that was seen as very important, because it used an X, the same as first past the post, and they simply voted once.
Turning to the Scottish Parliament, why do we use the additional member system? It's widely accepted in Scotland that the desire amongst politicians was to maintain the constituency link with voters, so they were very keen to ensure that constituency seats still existed. It was also thought that the use of Xs was important because it was familiar to the electorate.
We finished up with that because the Scottish Constitutional Convention, which was a cross-party civic society organization that existed in the 1980s and 1990s, suggested the use of AMS, which led to the document “Scotland's Parliament. Scotland's Right”, which was published in 1992, ahead of a U.K. parliamentary election at that time. When the Labour government of 1997 came in, they passed the Scotland Act 1998, which brought in AMS, which we first used in 1999. Finally, we finished up with STV for council elections, which is a departure from Xs.
Essentially the story behind that is the McIntosh Commission on Local Government and the Scottish Parliament proposed the use of proportional representation in 1999. The Working Group on Renewing Local Democracy, known as the Kerley committee, then suggested in 2000 that we consider single transferable vote. There then occurred in 2003 a Scottish Parliament election in which negotiators between the coalition partners of the Labour Party and the Liberal Democrats agreed to introduce single transferable vote. I think it was considered to be a deal breaker for the Liberal Democrats, who were very pro-STV at that time, which led to the Local Government (Scotland) Act 2004 and our first use of STV in 2007.
We have highlighted some issues that you might want to consider as part of your deliberations. Firstly, we would say that whatever you do, change needs time, planning, resources, and testing to be successful. You would expect that. We're a bunch of bureaucrats after all. However, we did have a period of major change, I think it's fair to say, in the period 2005 through to 2007. We then suffered from some local difficulties in our elections. I think it's commonly accepted that too much change was brought in in too short a period of time, which led to higher rejection rates than we normally expect and the e-Counting system didn't work quite as we expected. It essentially broke down on the night in certain places.
We would argue that testing voter-facing materials, such as, of course, the ballot paper, is very important. Make sure that what you're giving the voter works, and have good guidelines, something that the electoral commission produced post the 2007 event. We produced “Making your mark” guidelines to explain how we thought a good ballot paper should be designed and then tested.
I think one thing which also came out of 2007 was not to be driven by process. You have to put the voter first. The needs of the machinery for e-counting are secondary to the needs of the electorate. In two of our electoral regions in 2007 at the very last moment, because of the size of paper which the counting machines needed to use, the instructions to voters at the top of the ballot paper were reduced. This led in those two regions to an increase of 2% over the Scottish average for rejected votes. Certainly, that's something we learned. In 2011, it didn't happen.
We would also suggest that you consider the size of your constituencies. Certainly the six MEPs who represent Scotland are widely thought to find it difficult to connect with their electorate. They literally represent the whole of Scotland. Whilst it is a small country, it's a relatively big place to try to connect with 4 million electors.
Under single transferrable vote, as an elector, you get three or four councillors to represent you, and that can be seen as a bonus, as a positive. It also means that in remote rural areas, councillors also have to represent very large areas, areas of several islands, which generally but not always have ferry connections. You have to think about the size of the wards.
The Kerley committee suggested two member wards in certain remote areas. Lord Steel, way back in the sixties, recommended that if you were going to introduce STV, you could consider single-member wards in very remote rural areas. It's fair to say that neither of those proposals have been brought into legislation yet, but that's an option for trying to get away from the problem of remoteness from your electorate.
There are also some other issues which surround single transferrable vote. I should say that these aren't the views and opinions of the U.K. electoral commission. This is taken from a paper by Professor John Curtice of the University of Strathclyde, a famous commentator and psephologist in Scotland and the U.K. There are issues that we have noticed of alphabetic bias, identified by academics in 2007. We know this, of course, because we have e-counting, so there is data is available in an anonymous form after the event which people can then analyze.
In the first STV election in 2007, around 60% of all voters gave their first preference to a candidate higher on the ballot paper than the candidates for whom they gave their second preference. Also, it was found that if a party nominated more than one candidate in a ward, then the voter was more likely to vote for the party candidate higher up the ballot paper than the second party candidate. That happened 80% of the time in the 2012 Scottish local government elections. However, of course, that doesn't always happen. If they were a well-known or an incumbent candidate in 2012, they'd go against that trend.
Politicians in Scotland are aware of this and are considering solutions. The solutions talked about include rotating—Robson's rotation—the candidates on the ballot paper so they appear equally in all positions. They are also considering bracketing party candidates together, and perhaps also rotating them within the brackets so that all party candidates appear in different places.
The commission doesn't have an opinion on that. What we've said is that they have to consider two things. Voters are used to seeing ballot papers with candidates laid out alphabetically, so there is a need to see the effect of changing that.
In general, you should test any ballot paper before giving it to the electorate, so we recommend that you test the the voter-facing product. We also think you might want to consider the following: do voters, while filling in the ballot paper, fully understand an electoral system? For instance, under EMS in Scotland—