Three Ontario Mixed Member Proportional models With local and regional MPPs who all faced the voters, all accountable

Personal viewpoint (Wilfred Day) Principles

1. I’m making this submission to the Ontario Citizens’ 3. Your nine principles for electoral systems are all Assembly on Electoral Reform as a former school important. You can design a voting system to meet them trustee who served on the Northumberland - Clarington all. Our present system doesn’t. Seventy-five countries, Board of Education. I was elected four times as one of including most major democracies, use proportional a pair of trustees representing Port Hope and the Town- representation systems. The goal is to maximize the ship of Hope. That school board used to be the largest by number of citizens who can help elect representatives they enrollment between Oshawa and Ottawa, and is now want: make every vote count. Polls show between 63% larger still: in 1998 it became part of the Kawartha Pine and 71% of Canadians support proportional represen- Ridge District School Board, based in Peterborough. As tation, so you shouldn’t worry too much about the a lawyer and active local citizen I’ve been involved in 60% threshold. standing up for our community over 37 years with govern- ments of all three parties, making me a bit non-partisan. 4. Most of us agree with your principles. But then we’ll ask worriedly “but how would party lists work?” Winner-take-all? While supporting the principles, most Ontarians are wary of any change which will replace single-district representa- tives with legislators elected from lists. 2. With the current system, only one MPP is elected from each riding. The only way we can gain political 5. So in my opinion a mixed member proportional representation is if we live in a riding where we happen to (MMP) system would be best. MMP is designed to allay support the most popular party. This is even worse than a these concerns. It uses lists like other PR systems, but at local council with single-member wards, because they’re the same time it retains single-member districts. Every non-partisan, while our MPP may be someone we cam- voter has both a local representative and regional paigned against. When I was a trustee for a two-mem- representatives. ber ward, voters had a choice which trustee to go to: two competing trustees.

Three of a range of MMP models 6. Because the MMP system is flexible, a range of Table of Contents: good MMP models are possible. You already have three Three models and suggested improvements p. 3 professional submissions on MMP models: Women and minorities p. 6 1. One from Sean Geobey, a Ph. D. student at Thresholds and other issues p. 7 Wilfrid Laurier University in Waterloo. It’s on Two simulations of Ontario’s 2003 election p. 8 your website as submission number 1039. Technical options p. 12 Open list methods p. 13 2. One from Greg Morrow, a Ph. D. student in the Options, alternatives and variations p. 15 School of Public Affairs, University of California, Los Angeles, raised in rural Eastern Ontario, submission number 1122, amplified by submis- sions number 1138 and 1218. 8. As you may know already, the Mixed Member 3. One from Brian Doody, a Ph. D. Student and Proportional (MMP) system Lecturer in Political Science at University of - was recommended by the Law Commission of Western Ontario, submission number 1087. Canada in 2004. This submission expands on a presentation I made at your - British authorities in West Germany created the MMP hearing in Peterborough Dec. 5. It will say why I like system in 1946. It merged British direct elections with these three models, and show a couple of examples of European proportional systems: “personalised PR.” how you might blend them and make the result even - it balances personal accountability with party ac- better. I also have a suggestion on open list methods. countability; “the best of both worlds.” 7. With any proportional system every vote counts - It caught on across Germany, then in New Zealand, equally. With these Mixed Member Proportional , Wales and elsewhere. Quebec has been models, your Ontario Legislature can accurately working on an MMP model. reflect the popular vote, and the voter can have more - Ontario’s Students’ Assembly on Electoral Reform accountable MPPs. chose MMP in November. 2

A Voice at Queen’s Park for Ontario’s You have two votes Regions: 13. On voting day you have two votes. You cast a local 9. With a regional Mixed Member Proportional vote for your local member, as we do today, and you cast system like any of these three you can go to your local a regional vote for your choice of party and, with open list, for your regional MPP. None of these models rule out MPP or one of your regional MPPs. The core concept open list. of MMP is to respect the voter’s first choice.

10. Some MMP models in New Zealand and some MPPs will now have competition German provinces don’t tie the proportional MPPs to a region. They don’t make them personally accountable to 14. If a party wins lots of local seats, it gets few regional the electorate in natural regions. These three models MPPs. As a result, being on the regional list is no look more like Scotland’s. But with open list they’ll be guarantee of election, so that regional MPPs hope to even better: this makes Regional MPPs accountable to run locally as well as regionally next time. So, as Prof. voters, providing both effective representation and Louis Massicotte has shown, they have their own con- accountable government. stituency offices and compete with local MPPs to serve constituents. Bad for the politicians, perhaps, but Accountable Regional MPPs good for the voters. Competing MPPs: good for voters 11. Regional MPPs will make regional issues like economic development and regional health services heard at Queen’s Park. If possible, the regions should be No more wrong-winner elections consistent with District School Board boundaries. 15. Quebecers have been talking about proportional Stable government representation for decades. Action started after the 1998 election. That’s when Quebec’s Liberals won more votes 12. The German provinces all use similar models. Do than the PQ, but the PQ government won more seats and they suffer from unstable coalitions or minority govern- was re-elected. BC started action after the 1996 provincial ments? Since 1947, their total 152 parliaments have election. That’s when BC Liberals won more votes than lasted an average of 4.03 years each. MMP has produced the NDP, but the NDP government won more seats and current true majorities in Bavaria, Hesse, Thuringia, and was re-elected. In Ontario’s 1985 election the Liberals (in previous elections) many other German provinces, as won more votes than the PCs but the PCs won more well as in other jusidictions. seats. With a good proportional representation model, this cannot happen.

Demographic representation and ef- But as I looked at what has happened in other jurisdic- tions and started to consider how we can actually, all of fective parties us, in all three parties, stop talking about wanting more women and actually try to produce more women from 16. Janet Ecker was Ontario’s Finance Minister, a our respective nominations and various processes, the Progressive Conservative. Here’s what she told Law Commission's recommendation about a portion of Ontario’s Select Committee on Electoral Reform, as your seats I started to find very attractive. one of the three spokeswomen for the multi-partisan group Equal Voice: “As the result of your nomination process and your election process, a political party, indeed a govern- “When I started with the group, primarily because I agree ment, may find itself with a lack of representation in that we need to have more women in our elected pro- some area, whether it's geographic, whether it's cess at all levels, I was opposed to proportional repre- gender, whether it's whatever: urban, rural, you sentation because I believe very strongly in that link to a name it. For caucus, cabinet and party discussions constituency . . . I still think there are a lot of strengths to to adequately assess an issue, I think you need as that system and I don't think we should lose that. much diversity in that room as you can get. Propor- tional representation provides a political party with an opportunity to round out the slate, if you will, of what Why Janet Ecker likes MMP the nomination process may well have produced for them. . . That is one of the reasons why I have been convinced that a portion of the seats be proportional representation.” 3

Sean Geobey’s model -- He suggests the local Riding Seats be kept much the same in the North, and almost the same in south- 17. Here’s the basics of Sean Geobey’s model: ern Ontario, where every six ridings will become five.

-- He would keep 85 local MPPs and have 45 regional -- He suggests 11 regions: two smaller regions in the MPPs, bring the size of the legislature back to the 130 North, one for Toronto with 27 MPPs, and eight others we had before 1999, with 35% regional seats. in the South, typically with 12 MPPs each. Each would have two local MPPs for every regional MPP: 33% -- He has flexible options for regions. He suggests the regional MPPs. Any disproportional excess seat in local Riding Seats can be kept the same in non- any region would be adjusted by deducting a regional metropolitan regions, but “in densely populated urban seat of that party from the region in which that party areas (Toronto, Ottawa, Hamilton, parts of the 905- earned the lowest fraction of a seat. belt) ridings could be enlarged” to add more regional MPPs. In large-urban regions proportional representa- tion is likely a higher priority than geographic repre- sentation, and consequently more individual ridings Brian Doody’s model can be placed in one region. 19. Here’s the basics of Brian Doody’s model: Note: Different ratios of regional MPPs in different areas is not just Sean Geobey’s idea, nor is it a novel idea. In -- He would keep the 107 local MPPs and have only 27 September 2005 Fair Vote Ontario Chair Dr. Joseph regional MPPs (or maybe 33), for a legislature of 134 Murray suggested to Ontario’s Select Committee on (or maybe 140): only 20% or 24% regional MPPs. Electoral Reform that the Citizens Assembly be allowed to recommend expansion to 139 MPPs, reflecting -- He notes that, if voters voted as they did in 2003, population growth, and “that a northern Ontario region one party would still have gotten 55.4% of the seats have a higher proportion of riding seats to list seats with 48.5 per cent of the vote, so a further nine MPPs to ensure smaller ridings. . . We think it's important to would have to have been added in 2003 to make the recognize some of these concerns around the north result proportional; but he does not prefer this option. and rural areas” but still “allow for province-wide proportionality.” Wayne Smith added “As we men- tioned before, if you set up a northern region, maybe it's more important to have good geographic coverage than Further variations it is to have representation of diversity in that area, so you have 20% list seats. In downtown Toronto, maybe 20. You can work with these models to provide propor- it's more important to represent diversity than it is to tional representation, accountability, regional representa- worry about geographic coverage, so maybe you have tion, demographic representation (women and minorities), 50% list seats." (Note: this submission’s views are my voter choice, and your other goals. own, not Fair Vote Ontario's.) 21. There are five main design decisions with an -- He suggests a minimum of five local riding seats per MMP model for Ontario. region, and at least one regional MPP, but generally he aims for over 10 total seats per region. He prefers a. First, proportion of regional seats to local seats. over 20 seats per region, because Geobey calculates the regional seats only at each separate regional level, b. Second, the number of regions, and their size. so larger regions are needed for good proportionality. c. Third, how big is the legislature?

d. Fourth, do you have province-wide proportionality or Greg Morrow’s model unlinked regions? e. Fifth, do we have closed lists or open lists? 18. Here’s the basics of Greg Morrow’s model:

-- He would keep 90 local MPPs and have 45 regional Five design decisions MPPs, adding the 27 MPPs we had before 1999 to the 107 we’ll have in the coming election, plus one more. 22. You independent citizens are in a good position to decide these issues from the local voter’s viewpoint. 4

1. We need 40% Regional MPPs 29. Morrow suggests 11 regions, Doody 10. Geobey 23. Can’t we keep 107 local ridings unchanged, and isn’t as specific, but his assumptions lead to about 10 just add 32 to the House for the regional MPPs? regions, which I’ve described in Appendix 1 (page 8.) Wouldn’t 23% Regional MPPs be enough? No. In 2003 the Liberals got 72 of 103 seats on 47.1% of the four-party vote. To compensate would take 153 MPPs, so that the 3. Restoring MPPs to the Legislature Liberals would have 72 of the 153, or 47.1%. That means 50 more MPPs than in 2003, which is 32.7% of the new 30. The one defect of MMP is when the local ridings are 153 MPPs. In 1987 the required figure was 33.7%. In 1990 larger than today’s. In a metropolitan region where many it was 31.6%. It was lower in 1995 and 1999. voters don’t even know what riding they live in, regional MPPs would be as accountable as local ones. But for If you keep today’s ridings and add those Ontarians who do not live in larger urban areas, would some communities lose their MPP? This would be only 23% more MPPs -- “MMP-lite” unacceptable in the sparsely populated north, where many -- you lose proportionality. ridings are already vast.

31. The Citizens’ Assembly’s model can restore the 24. In Germany it’s 50%, in New Zealand 44%, in 27 MPP positions lost from the legislature in 1999. Scotland 43%. No MMP model anywhere but Wales has Norm Sterling M.P.P. authored the Act that cut the 27 less than 40%. When more voters vote their first choice MPPs. His latest private member’s bill would add some and have more choices, we’ll need more proportional back. seats than in past elections. 32. This would stop big jumps in local riding size. 25. Geobey proposes 35% or more, while Morrow wants With the coming 107, 27 makes 134, but what about 33%. But 40% gives the winning party regional MPPs population growth? Should it be 135, 139, 140, 147, for regions where it won few or no local seats. This 172, 179? (See page 15.) Geobey’s model has 130, gives these regions a voice in the government. As Morrow’s 135, but with 139 both models work even well, this helps women and minorities get elected in better, and Doody says you could go to 140. Geobey’s the winning party. And 40% gives us a reasonable proposal of 85 local seats would, with a House of 139, chance of proportionality even with future voting patterns. have 39% regional MPPs.

2. How many regions, and how large? More MPPs would stop big jumps in local riding size 26. These three models all have 10 or 11 regions, because regions can be too small or too large.

27. At one extreme are the 27 tiny unlinked five-seat 4. Province-wide proportionality. or three-seat regions in the Quebec government’s first “large-party” model. This was rejected by its Citizens’ 33. Doody’s model has province-wide proportionality, Committee, because a party would need at least 17% of and Morrow’s has a big element of it. Geobey’s doesn’t. the vote in the region to win a seat. And with such little regions, wouldn’t all the lists be too short to allow for 34. Scotland’s eight regions show you the problem. diversity? Lists shorter than five are little help to women. What if one party gets 44% of the vote in a 16-MPP region in Scotland, wins all nine local seats, but deserves only 28. Regions can also be too large. Scotland’s 16-MP seven of the 16 MPPs? The other seven "top-up" MPPs regions are a bit large for some parts of Ontario, as are aren't enough for proportionality, but the other regions Geobey’s preferred 20-seat regions. Some regions of aren’t adjusted to correct this, because each region is Ontario are too small to rate 15 or 20 MPPs. Would we self-contained, not linked to the others. This happened want a region running from Pickering to Brockville? In in four of Scotland’s eight regions in both elections. Last large regions, some voters will ask “is our regional MPP time, Labour got 39% of the seats with only 29% of the really accountable to a real region?” As your geography votes: a big “winner’s bonus” more typical of “win- working group said, the real attachment is to informal ner-take-all” systems. If you want accountable regional regions and communities. You want natural regions. In MPPs from smaller more natural regions than Scotland fact, Scotland’s Arbuthnott Commission recently has today, you have an even bigger problem with small reached the same conclusion, and recommended (at page unlinked regions: they give a bonus to larger parties, 26 of their Report) revised regions running from 11 and exclude small parties. members up to 16. 5

39. Doody’s model is expressly based on the Quebec Small unlinked regions give Citizen’s Report, although they had 40% from regional a bonus to larger parties lists, and the other two models resemble it in many ways. and exclude smaller parties. I’d prefer something like the Quebec Citizens’ model. But Ontario needs a larger House, and personal accountability through open lists.

35. The solution is to borrow the Scandinavians’ 40. The Quebec Citizens’ Committee said the total method of linking regions to the overall vote. Sweden number of regional MPPs should be calculated province- and Norway don’t use MMP, but their list systems show wide, that is, counting the party’s total votes. Then, these how province-wide proportionality helps. They have seats would be distributed among the regions by the smallish locally-accountable regions. Sweden’s have following formula. The first regional seat for Party A would mostly only eight or 12 MPs each. Then they add "top-up" go to the region where it was the most under-represented, MPs with 5% (Norway) or 11% (Sweden) of the MPs, to if that region had regional seats still available; and so on create nation-wide proportionality. But they don’t use for the other seats. Or to use the math term, the regional national lists. The "top-up" MPs for a given party are the seats go to regions across the province using the votes in candidates next on the list in the regions where that party each region, in order of highest remainders until the party was closest to winning another seat. or region runs out of seats. Doody’s model does some- thing comparable. 36. As Jonathan Rose told you, with MMP the regional MPPs can be counted both province-wide 41. As Geobey notes “The key strength behind the MMP and regionally, just as they do in Sweden: no system outlined here is that it can achieve a strong province-wide lists, natural regions, and regional balance between proportional and local representation MPPs calculated both regionally and provincially . while providing extensive flexibility in handling the repre- sentative needs of Ontario’s very diverse regions. . . . Regional MPPs calculated both each region in Ontario can be served by the local-propor- regionally and provincially. tional mix most appropriate to its needs.” But ten unlinked regions would risk too much fragmentation. Even eight unlinked regions give a bonus to larger parties in 37. That’s just what we see in the two examples of Scotland. In Ontario with only 12 seats they would MMP models made in Canada by a Citizens’ group. exclude small parties. The logical extension of Geobey’s Quebec’s Citizens Committee reported in April 2006. The model is province-wide proportionality. This would also BC Citizens Assembly designed an MMP model in 2004. end the need for such vast non-metropolitan regions. Both MMP models have the same simple basics: - 60% local MPPs, 40% regional MPPs. Every voter is 42. Morrow’s model goes part-way: if a party wins too represented both by a local MPP, and by regional MPPs many local seats in a region (as Labour keeps doing in from regional lists of candidates. half of Scotland’s regions), it loses the same number of - province-wide PR calculation, giving full proportion- regional seats in other regions, namely, those regions in ality. Every vote counts equally. which they earned the lowest fraction of a seat. Morrow’s - a province-wide threshold for parties to get seats: 5% model corrects the bonus to larger parties, but not neces- (Quebec) or 3% (BC.) sarily the exclusion of small parties. Morrow doesn’t guarantee the Greens that, if they get 5% of the vote province-wide, their voters will elect 5% of the MPPs. Other citizen-designed MMP models. 43. Province-wide proportionality adjusts dispropor- tional results, in the North or elsewhere, automatically, 38. This process is like round two of discussions in the since the number of MPPs from each party is calculated BC Citizens Assembly on October 16 & 17, 2004, and province-wide, and the regions are linked. But if the North similar to discussions at Quebec’s hearings. The BC CA has more local MPPs, does this give Northern voters more decided to design two systems, and then vote which one power than others? No. With province-wide proportionality, they preferred. For their MMP model they decided on each party’s number of MPPs is set by the total provincial some kind of open lists. They ran out of time before vote. Every vote counts equally, north or south. deciding the number of regions. They also ran out of time before they decided what they meant by open list, and voted for STV instead. You can take up the task where Every vote counts equally, they left off. north or south. 6

5. Open List gives more voter choice The party’s candidate with the most 44. You have two votes. You cast a local vote for your votes on the regional ballot local MPP, as we do today. But with open list, you cast gets its first regional seat. a regional vote for your regional MPP, which is also for your choice of party to govern Ontario. On the regional ballot are the party’s candidates for that 47. Parties will likely want to run a full list: as many region. That’s the party’s face as it competes for govern- as the total number of MPPs from the region. In Germany ment. and Scotland most regional MPs were also local candi- dates. In case most of the names ranked highest on the 45. You choose whom you like from the “open list.” list get elected locally already, you need more regional Morrow prefers open list. Geobey suggests closed or open candidates. but prefers closed. Doody doesn’t specify. 48. And this lets parties add women and minorities 46. For Regional MPPs, the party’s candidate who who didn’t get local nominations, as Janet Ecker won the most votes on the regional ballot gets its first suggested. They will, voters will expect, be nominated regional seat, and so on. democratically in the region. These choices affect women and minorities

49. In the last election only 21% of the MPPs elected 53. All over the world, PR systems lead to more were women. In a recent poll 90% of Canadians want to women being nominated. Similarly, it allows for fair see more women elected; 86% of men do, and 94% of representation of minorities. With an “open list” system, women. Yet when party members hold nominations voters still choose who’s elected, but if we have women to one at a time, 80% of the time they nominate “the vote for, we can elect them. strongest man,” typically a middle-aged professional with 2.5 children. This is the major bar to women. 54. The longer the list of regional candidates, the more chance women and minorities have to get nominated. So Ontario can’t afford not to use you either need big regions or more MPPs. But not too all its human resources. big: with open lists, in a region running from Toronto to Ottawa what hope would a woman in Gananoque or Ajax have? They would be lost in such a long list. Who would 50. In a competitive world, Ontario can’t afford not to win in such a vast region? Big names, big men, incum- use all its human resources. bents, TV stars, sports stars? As the Quebec Citizens’ Committee warned, if voters have to learn about too many 51. With these models, almost 40% of the MPPs will be candidates this “would probably cause them to vote for the elected regionally. When party members in the region only candidates they have heard of.” nominate a group of candidates, they can nominate whom they like -- proven winners, new faces, candidates 55. But even a five-seater with three local MPPs and who are also running locally, or additional candidates. two regional MPPs would have a list of five, so if the three Parties will be able to respect the diversity of the local ridings happened to nominate all men, party mem- electorate. That’s one problem with Doody’s model: If bers voting in the regional nomination might well nominate 76% or 80% of MPPs are still nominated by our present two women to balance the list. A list of five will help. system, this won't advance women's representation as much as it should. Fewer male incumbents; good for women. Lists help women and minorities

52. Take an urban region with 13 MPPs, 8 local and 56. Scotland and Wales have excellent numbers of 5 regional. With open list, it can include the 8 candi- women elected, partly because they are fairly new legisla- dates nominated locally and 5 more nominated region- tures (no male incumbents blocking women's way) and ally. These lists of 13 are long enough to give voters partly because of their MMP system. If you recommend a choice from a good diversity of ethno-cultural restoring the 27 MPPs cut from Ontario's House in minorities, and good representation of women. 1999 and adding a few more, the new seats with no male incumbents will help kick-start improvement in women's representation. 7 Other issues Thresholds for parties to win seats 63. When Germany decided to keep a high (5%) threshold, they changed to the “highest remainder” method as a consolation prize for small parties. D’Hondt, 57. Parties have a conflict of interest in excluding on the other hand, penalizes small parties. If your new competition. A Citizens’ Assembly is the perfect group to system is already fair to small parties, Ste-Lague is more decide what size of party qualifies for party seats. With neutral. provincial calculation the threshold issue can't be finessed by using the size of unlinked regions as a proxy for it, as Scotland did. Independents

58. Doody suggests a 5% threshold. Morrow suggests 64. Geobey and Doody do not mention independents, 3%. Geobey states no threshold because his regions are but of course they can still run for local MPP. Your party unlinked: an unlinked region of 13 has a natural threshold ballot is for the party you want to govern Ontario, so most of 7% or so. MMP models don’t let independents run for the top-up regional seats on the party ballot. 59. Should a party with only three per cent of the vote qualify for seats? One argument is, every vote should 65. However, Morrow also would let a single independ- count equally, as far as possible. High thresholds keep ent candidate win a regional seat if can win enough votes newer views out of the Legislature. Lower thresholds on the regional ballot by him or herself. Scotland lets favour free expression of minority opinion. Why should any independents run for regional seats too, but they need at group be excluded? Voters should decide for themselves least 6 percent of the vote across a 16-MP region, not how many parties they want. Federal Election Financing easy for ordinary people. See the note on page 12. extends to any party getting more than 2%. 66. In Scotland's last election, an MP who left her party 60. The other argument is: a threshold should be high won re-election as an independent Regional MP. How enough for the Legislature to be workable, not unduly easy do we want to make it for independents to be elect- subject to extremists or “vanity parties.” We don’t want so ed? That’s the tradeoff between effective local representa- many mini-parties that it’s chaos, with only one or two men tion and responsible government. (and it’s usually men) each. Even Israel has a 2% thresh- hold. Hitler took advantage of Weimar Germany’s instabil- Just local MPPs, by preferential ballot? ity. So 2% may be too low. 67. To elect a single post like Mayor, the preferential The threshold must be shown to be a ballot is good: drop candidates with the fewest votes until the winner has a “true majority.” Could we use that reasonable limit: 2%, 3%, 4% or 5%? provincially?

68. Not to elect a government. The result would not be 61. If challenged in court under the Charter of Rights, proportional to the popular vote at all. Most Ontario the threshold must be shown to be a reasonable limit. If it voters are already "represented" by their second or was set by an independent Citizens’ Assembly, the courts third choice - that's the problem, not the solution. might show more deference than if politicians had set it. Some voters have a sincere second preference, but others The threshold should be 3%, 4% or 5%. would only try to stop the party they dislike. If you assume NDP voters would have made the Liberals their second Rounding method preference, the 2003 election would have resulted in only 7 PCs and 7 NDP MPPs. With 8 needed for party status, 62. When parties win a fraction of a seat, a mathemati- that's a one-party Legislature. This would be worse cal formula must be used to round the fraction up or down. than our present system. Germany uses “highest remainder” as does Morrow. By using the “highest remainder” method, as Morrow explains MMP is one of the PR family in his Interpretation Bulletin #5, he compensates somwhat for the lack of a guarantee of a fair share of seats across 69. It’s important to note that, although these models are the province for small parties like the Greens. New MMP models, there are a variety of proportional voting Zealand uses the Ste-Lague formula, as does Geobey. models: other MMP models, List PR, and STV. Each Scotland uses “highest average,” also known as D’Hondt. meets the OCA’s principles better than either the current Doody uses either Ste-Lague or a modified D’Hondt system or the Alternative Vote (preferential ballot). formula. 8 Appendix 1 How would it work? sample regions, and simulated recalculations of Ontario’s 2003 election results

70. This simulation assumes voters voted as they did 72. Many people want a threshold of 3%, 4% or 5%. A in 2003. Of course, unlike in 2003, more voters would good compromise is 4%, but on the 2003 votes, this would have voted who saw no point wasting their votes in safe exclude the Green Party with 2.85%. With a 4% threshold, ridings. Also, more voters would have felt free to vote for would PR have empowered Green Party supporters their first choice, not against their last choice. enough to bring them up to 4%? Quite likely.

73. In these simulated recalculations of the 4,497,244 More would have voted. votes cast in 2003, only 1.35 percent are “wasted”: the More would been free to vote 60,681 votes for candidates of parties who got less than for their first choice, 2%. In the actual 2003 election 48.3% of the votes were “wasted” in that they did not count toward the out- not against their last choice. come.

71. For these calculations we assume a threshold of Of 4,497,244 votes cast in 2003, 2.5% for parties to qualify for seats. The model works only 1.35 percent are “wasted” either way. The lower threshold shows how the model works with four parties. I used the “highest remainder” rounding calculation because it’s easier. Sample Regions Central-East (everything between the City of Toronto and Ottawa, including Durham Region) 15: 11 local, 4 regional, 74. Morrow gives you a useful map of his 11 sample compared to the present 12. regions. Doody lists the ridings in his 10-region model, but never explains why the regions are all the same size Three northern regions with existing ridings: instead of following natural regions. Geobey’s assump- tions lead to about 10 regions, such as: Northeast (Parry Sound to Cochrane) present 4, plus 1.

Four large-urban regions with larger ridings: Near North (Sudbury - Sault) present 4, plus 1. Toronto 27 MPPs: 15 local, 12 regional, compared to the Northwest present 3, plus 1. present 22 MPPs. York-Peel 19: 11 local, 8 regional, compared to the Plan A: 11 middle-size regions: present 15. Hamilton-Niagara-Halton 14: 8 local, 6 regional, compared 75. This variation on Geobey’s model uses province- to the present 11.5. wide proportionality. Therefore, it’s free to be a bit more flexible, so it’s more like Morrow’s map. You’ll see Ottawa-Cornwall 10: 6 local, 4 regional, compared to the that using 139 MPPs helps too, which I compare with the present 9. 107 ridings to be used this year: Three large southern non-metropolitan regions with ridings almost today's size: Seven regions (62 local MPPs, for- merly 82 local ridings). London-Oxford-Brant-Haldimand-Sarnia-Chatham-Wind- sor 16: 12 local, 4 regional, compared to the present 13. Toronto 29: 16 local, 13 regional (now 22.4 local) Waterloo-Wellington-Dufferin-Simcoe-Grey-Bruce-Perth- Huron 15: 11 local, 4 regional, compared to the present Peel 12: 7 local, 5 regional (now 8.5) 12. York-Durham 14: 8 local, 6 regional (not counting Brock and the part of York Region in York-Simcoe, now 10.6) 9

Hamilton-Niagara-Grand Erie 13: 8 local, 5 regional (now 10, including Brant and Haldimand-Norfolk.) Plan B: Fifteen Smaller Regions Here’s a version with 15 smaller regions, and only 135 Waterloo - Halton - Upper Grand 12: 7 local, 5 regional MPPs, again using province-wide proportionality. (including Wellington and Dufferin, but not counting the 3 north Wellington Townships in Perth-Wellington, now 9.5.) Nine large-urban regions (97 MPPs): Ottawa Valley - Cornwall 13: 8 local, 5 regional (includ- ing Prescott & Russell, Renfrew and Mississippi Mills, Toronto has 28 MPPs: 15 local, 13 regional now 10.) Peel - Dufferin - Halton 16: 8 local, 8 regional (Miss- Southwest 14: 8 local, 6 regional (now 11 including issauga, Brampton, Caledon, Dufferin, Halton) London, Windsor, Chatham-Kent, Sarnia, Oxford, & Lambton-Kent-Middlesex.) York Region 9: 5 local, 4 regional. (Including the new York-Simcoe riding). Four Northern and Rural Regions (25 Durham Region 6: 3 local, 3 regional. ridings unchanged, now with 32 MPPs) Hamilton - Niagara 10: 5 local, 5 regional.

As discussed in the OCA recently, these have “regional list Ottawa - Cornwall 11: 6 local, 5 regional (Ottawa, Pres- members specifically mandated to represent northern or cott & Russell, Stormont-Dundas & Glengarry.) rural interests.” Based on the 107 ridings to be used this year, these 25 local ridings are unchanged. Waterloo - Guelph 7: 4 local, 3 regional.

Do the 15 southern ridings really need to be unchanged? London - St. Thomas 5: 3 local, 2 regional. (London, Would five for every six (Morrow’s model) , or 11 for every Elgin, Thames Centre.) 12 (my interpretation of Geobey’s model), be close enough? You decide. Windsor - Chatham 5: 3 local, 2 regional (Not including Lambton-Kent-Middlesex) North-East/Le nord-est 9: existing 7 local plus 2 regional (including Sudbury, Nickel Belt, Algoma - Manitoulin, Sault Six other regions (30 local ridings Ste. Marie, Nipissing, Timiskaming - Cochrane, Timmins - James Bay.) unchanged, now with 38 MPPs):

North-West 4: existing 3 local plus 1 regional Again these 30 local ridings are unchanged.

Georgian Bay - Huron-Perth 10: existing 8 local plus 2 Mid-West Ontario 9: 7 local, 2 regional (Haldimand- regional (including Barrie, Simcoe North, Parry Sound - Norfolk, Brant, Oxford, Perth-Wellington, Huron-Bruce, Muskoka, York-Simcoe, Simcoe-Grey, Bruce-Grey-Owen Lambton-Kent-Middlesex, Sarnia.) Sound, Perth-Wellington, Huron-Bruce.) East Central & Mid-East Ontario 10: 8 local, 2 regional Limestone-Kawartha 9: existing 7 local plus 2 regional (Peterborough, Haliburton-Kawartha Lakes-Brock, Prince (including Kingston, Lanark-Frontenac-Lennox & Adding- Edward-Hastings, Northumberland-Quinte West, Kingston, ton, Leeds-Grenville, Prince Edward-Hastings, Lanark-Frontenac-Lennox & Addington, Leeds-Grenville, Northumberland-Quinte West, Peterborough, and Renfrew.) Haliburton-Kawartha Lakes-Brock.) Georgian Bay 5: 4 local, 1 regional (Barrie, Simcoe North, Simcoe-Grey, Bruce-Grey-Owen Sound.) (Note: for simplicity, in this example I have put Parry Sound-Muskoka in the Georgian Bay region, because North Bay-Timmins 5: 4 local, 1 regional (Parry Sound - Muskoka is generally part of that region. However, the Muskoka, Nipissing, Timiskaming - Cochrane, Timmins - Boundaries Commission would have to put Parry Sound James Bay) into the Northeast region, not only because it is generally part of Northern Ontario, but also because that would Sudbury-Sault 5: 4 local, 1 regional (Sudbury, Nickel bring the Northeast’s 9 seats within the criterion of a 25% Belt, Algoma - Manitoulin, Sault Ste. Marie) variation in the average population per MPP.) Northwest 4: 3 local, 1 regional (Thunder Bay, Kenora - Rainy River) 10 Simulation of Ontario’s 2003 election results

76. Each region has the following MPPs. Note that Hamilton - Niagara - Grand Erie 13: the PCs have 4 MPPs comparisons are to the 2003 results transposed onto the (2 local, 2 regional), not just 2; the NDP has 3 MPPs (1 2007 boundaries. The seat calculation rules are on page local, 2 regional), not just 1; and the Liberals have 6 MPPs 12. (5 local, 1 regional), not 7.

Waterloo - Halton - Upper Grand 12: the Liberals have 4 Plan A MPPs (2 local, 2 regional), not just 3; the NDP and Greens each have a regional MPP, not be shut out; the PCs have 77. Start with a provincial calculation for the 139 MPPs. 5 local MPPs, not 6.5. The Liberals are entitled to 65.48 MPPs, the PCs to 48.85, the NDP to 20.70, and the Greens to 3.97. Rounding by Southwest 14: the PCs have 4 MPPs (1 local, 3 regional), the “largest remainders” rule, Liberal voters elect 65 not just 1; the NDP has 3 regional MPPs, not be shut out; MPPs, PC voters 49 MPPs, NDP voters 21 MPPs, and and the Liberals have 7 local MPPs, not 10. Green voters 4 MPPs. In this calculation, every vote counts equally, including Northern Ontario votes. The Ottawa area’s 45,706 NDP voters Toronto 29: the PCs have 7 regional MPPs, not be shut get a regional MPP, not be shut out. out; the NDP has 6 MPPs (3 local, 3 regional), not just 3; the Greens elect a regional MPP; the Liberals have 15 MPPs (13 local, 2 regional), not 19. Georgian Bay - Huron-Perth 10: The Liberals have 3 MPPs (2 regional, 1 local), not just 1 local (note: the 2007 redistribution costs them Perth); the PCs still have 7 local Toronto’s 223,942 PC voters get 7 MPPs. The NDP is still shut out, the only region where this regional MPPs, not be shut out. happens, due to the PC near-sweep of the local seats. Limestone - Kawartha 9: The PCs get 3 MPPs (2 local, 1 Peel 12: the PCs have 5 MPPs (1 local, 4 regional) not regional), not just 2; the NDP gets a regional MPP, not be just 2; the NDP has a regional MPP, not be shut out; and shut out; the Liberals still have 5 local MPPs. the Liberals still get 6 local MPPs. Northeast/Le nord-est 9: the PCs get 2 regional MPPs, not York - Durham 14: the PCs have 6 MPPs (4 local, 2 be shut out; the Liberals still have 5 local MPPs; and the regional), not just 5; the Liberals still have 6 MPPs (4 local, NDP still has 2 local MPPs. 2 regional); and the NDP and Greens each have a re- gional MPP, not be shut out. (Note: this Region gets more MPPs in 2007 with population growth.) The Northwest’s PC voters get a regional MPP, not be shut out. Ottawa Valley - Cornwall 13: the PCs have 5 MPPs (2 local, 3 regional), not just 2; the NDP and Greens each get a regional MPP, not be shut out; and the Liberals have 6 Northwest 4: the PCs get a regional MPP, not be shut out; local MPPs, not 7. the Liberals still have 2 local MPPs, and the NDP still has 1 local MPP.

Plan B Large-Urban Regions:

78. Start with a provincial calculation for the 135 MPPs. Toronto 28: the PCs have 7 regional MPPs, not be shut The Liberals are entitled to 63.6 MPPs, the PCs to 47.4, out; the NDP has 5 MPPs (2 local, 3 regional), not just 3; the NDP to 20.1, and the Greens to 3.9. Rounding by the the Greens elect a regional MPP; the Liberals have 15 “largest remainders” rule, Liberal voters elect 64 MPPs, MPPs (13 local, 2 regional), not 19. PC voters 47 MPPs, NDP voters 20 MPPs, and Green voters 4 MPPs. Again, every vote counts equally, including York Region 9: the PCs have 4 MPPs (2 local, 2 regional), Northern Ontario votes. not just 2; the Liberals have 4 MPPs (3 local, 1 regional), not just 3; the NDP has a regional MPP, not be shut out. (Note: York Region gets more MPPs with population growth.) 11

Peel - Dufferin - Halton 16: the PCs have 6 MPPs (2 local, Other regions: 4 regional) not just 3; the NDP has 2 regional MPPs and Greens 1 regional MPP, not both be shut out; and the Georgian Bay 5: The Liberals have a regional MPP, not be Liberals get 7 MPPs (6 local, 1 regional), not 8. shut out; the PCs still have 4 local MPPs. Ottawa - Cornwall 11: the PCs have 4 MPPs (2 local, 2 Georgian Bay’s Liberal voters get a regional), not just 2; the NDP and Greens each get a regional MPP, not be shut out; and the Liberals have 5 regional MPP, not be shut out. MPPs (4 local, 1 regional), not 7. East Central & Mid-East Ontario (Peterborough-Kingston) Hamilton - Niagara 10: the PCs have 2 MPPs (1 local, 1 10: The PCs get 4 MPPs (3 local, 1 regional), not just 3; regional), not just 1; the NDP has 2 regional MPPs, not the NDP gets a regional MPP, not be shut out; the Liberals just 1; the Greens have a regional MPP; and the Liberals still have 5 local MPPs. have 5 MPPs (4 local, 1 regional), not 7. Mid-West Ontario (Haldimand to Sarnia) 9: The NDP gets Waterloo - Guelph 7: the Liberals have 3 MPPs (2 local, 1 a regional MPP, not be shut out; the PCs have 4 MPPs (3 regional), not just 2; the NDP has a regional MPP, not be local, 1 regional); the Liberals have 4 local MPPs. (Note: shut out; the PCs still have 3 MPPs (2 local, 1 regional). the 2007 redistribution costs them Perth.)

Durham 6: The NDP gets a regional MPP, not be shut out; North Bay - Timmins 5: the PCs get 2 MPPs (1 local, 1 the Liberals have 2 MPPs (1 local, 1 regional), not just 1; regional), not just 1, the Liberals still have 2 local MPPs; and the PCs still have 3 MPPs (2 local, 1 regional). and the NDP still has 1 local MPP.

London - St. Thomas 5: the PCs and the NDP each have Sudbury - Sault 5: the PCs get a regional MPP, not be a regional MPP, not be shut out; and the Liberals have 3 shut out; the Liberals still have 3 local MPPs, and the NDP local MPPs, not 4. still has 1 local MPP.

Windsor - Chatham 5: the NDP and PCs each have a Northwest 4: the PCs get a regional MPP, not be shut out; regional MPP, not be shut out; and the Liberals have 3 the Liberals still have 2 local MPPs, and the NDP still has local MPPs, not 4. 1 local MPP.

Trade-offs 80. You have more effective local representation: you 79. The basic trade-off with MMP is simple. For exam- can go to your local MPP or one of your regional MPPs. ple, take an area like London - St. Thomas in Plan B. The result across Ontario matches the popular vote. In Instead of four local ridings that elected four Liberals, you return, local urban ridings are, on average across large- get five MPPs: three from slightly larger local ridings (still urban Ontario, 32% larger than today (Plan A) or 48% all Liberals, of course), and two regional MPPs (one PC, larger than today (Plan B with 135 MPPs). one NDP.)

More effective local representation: you can go to your local MPP or one of your regional MPPs.

Note:

Spreadsheets and calculations for these simulations available upon request. 12 Appendix 2: Technical options The regional calculation concept 87. That would happen once in Plan B. The PCs would run out of seats before the Northwest regional seat is assigned, and the Greens would wind up with an MPP 81. Let’s take another look at regional calculation. Let from the Northwest for only 1,949 voters. me cast the problem in concrete rather than abstract terms. In a region of 200,000 voters with five MPPs, say 88. To prevent this, first, each region gets up to two 80,000 (40 per cent) vote Liberal: they elect two MPPs. regional MPPs for the parties most under-represented Say 54,000 (27 per cent) vote NDP: they deserve 1.35 in that region. Then, as recommended by the Quebec MPPs, and get at least one. Say 50,000 (25 per cent) vote Citizens’ Committee, the remaining seats are distrib- PC: they deserve 1.25 MPPs, and get at least one. And uted among the regions in order of highest remain- say the other 16,000 (8 per cent) vote for the Green: they deserve 0.4 MPPs. Who gets the fifth MPP? ders until the party runs out of seats. Adjustments from the provincial results would be made only in the larger 82. The Green says, “We’re the largest minority. By the regions. But for Ontario as a whole, the regional seats highest remainder rule we do.” The New Democrat says, would compensate for any imbalances in the local results. “Hold on – with two MPPs we would have 27,000 voters for each MPP. You Greens have only 16,000 votes, so we 89. In 11 of Plan B’s 16 regions, the regional MPPs are do.” An STV fan says, “Neither of you has a good claim. elected by voters in that region only. In the others, we’re Let’s decide by seeing who is the second choice of the focussed first on what parties the region’s voters 10,000 surplus Conservative voters.” “No way,” answer support, not on where the party wants its MPPs to be. the Greens and the NDP in unison. “Why should we let Then, similar to Morrow’s model, we make adjustments in Conservatives decide between two candidates of the left?” regions where a party has earned a fraction of a seat, to compensate for any disproportional regions. 83. There is another option. We can say, “None of you has a clean claim on this seat. Let’s see which party is We focus first on what parties the underrepresented in the province as a whole and use the seat to correct the imbalance.” region’s voters support

Let’s give it to the party 90. So, with linked regions and province-wide propor- tionality, an NDP voter in East Central helps elect an NDP that’s under-represented elsewhere. MPP in Peel-Dufferin-Halton, and Green voters in 12 regions help elect Green MPPs in three other regions. That’s just what province-wide proportionality and Regional calculation in practice linked regions has to mean. This model automatically corrects local anomalies by adjusting elsewhere. The 84. You’ll know the election result on election night, as result is the four Green MPPs are in their best three to the number of MPPs from each party, the local winners, regions, plus one from Hamilton-Niagara. and most of the regional winners. The final regional calculation -- which parties have regional MPPs in your 91. Why would an NDP or Green candidate bother trying region -- may not be known precisely until the final count. to win in Georgian Bay? Well, next time the PCs may not sweep it. You never know who’ll win the last seat. 85. Of the 53 regional MPPs, in Plan B 29 are their party's only regional MPP from the region. Only one 92. Plan A, with fewer regions, looks cleaner, but the person fits the title "The PC MPP for London - St. Thom- same rule is needed to try to prevent “orphan seats.” as." The other 24 regional MPPs would no doubt divide the region between them for constituency service pur- poses. They could choose a name for their sub-region. Independents and regional parties 93. A winning independent would remove that seat Regional calculation rules, and “or- from the provincial calculation and regional assignment of phan seats” party seats. That region would have one less MPP to be divided among the parties. But Scotland has no nation- 86. Some regions would have only one or two regional wide threshold like 4%; a party can win one regional seat seats. Could regions with unfilled seats end up with MPPs if it gets 6% in a single region, just like an independent. If for whom few in the region voted: “orphan seats?” Ontario has a province-wide threshold like 4%, should an independent need fewer votes than a regional party? 13 Open list methods: Who gets the seat?

94. Lists help reflect Ontario’s diversity: parties want 99. To “break the slate,” an individual must get personal to nominate balanced slates. Some MMP models have votes equalling at least 8 per cent of the party vote in the “open lists” which offer more extended voter choice. constituency. Any list candidate who exceeds this thresh- More MMP models have “closed lists,” but could parties fill old is ranked by number of votes received, ahead of their lists with obedient unknowns? party-ranked candidates. In the Commission’s example, the “Red Party” wins four list seats in a constituency. Personal votes for the fifth candidate on the Reds’ list Open lists: all MPPs face the voters. equal 10 per cent of the Reds’ party vote, while personal votes for the seventh candidate equal 8 per cent. There- fore, both of these candidates break the slate, and they 95. When we know which parties have seats in which get two of the seats. The other two go to the party’s two region, we see whether the voters have “broken the slate” highest-ranked candidates. (changed the regional list ranking.) If a candidate on the list also ran locally and won a local seat, we pass on to the 100. This helped a couple of women break the slate in next name. the 2002 election. Here, 90% of Canadians want more women in elected office. Once women get past the Party lists: Open, closed, or flexible gatekeepers of the single-seat ridings, won't Canadian voters vote for them? 96. The biggest challenge for the Ontario Citizens’ Assembly may be designing an open list MMP method. 101. One of the difficulties in designing a flexible list The BC CA needed expert help to design their MMP system of this kind comes in deciding what weight each model, but didn’t get enough. Pure list jurisdictions have personal vote should be given in relation to party votes. a variety of open list methods which can be adapted to The 8 per cent rule makes sense in Sweden, where the 29 MMP. These models would work with any of them. Which regions have, on average, 12 MPs each (8 per cent being shall we use? roughly one twelfth). Sweden’s new model helped a couple of women break the slate in 2002.

102. However, the proposed models for Ontario use The Citizens’ Assembly’s biggest “natural regions” of sizes varying from four MPPs to 27. challenge may be designing This variation makes the model more responsive to local an open list MMP method. conditions: voters can choose known names from the list. However, it also means that the 8 per cent threshold does not fit. 97. The only place in Germany with open lists is the state of Bavaria. New Zealand looked at it, but hasn't 103. It’s normal in MMP systems that local candidates copied it, for good reason. In Bavaria, all party votes are also run for regional MPP. A candidate who lost locally cast by voting for one candidate of the party. The party's may win a regional seat. Does this bother anyone? nomination process ranks candidates. Voters with no preference tick the one at the top of the list. Most votes 104. In Scotland, their Arbuthnott Commission has just are wasted on the top man. Lower rankings can be quite reviewed their system. It said “There is no survey evi- random, since the list order has no weight in the counting. dence to suggest that dual candidacy is an issue for voters. Candidates coming in second or third place who 98. At the federal level, the Law Commission of Can- are then elected through the regional list are only “losers” ada’s 2004 study recommended adopting Sweden’s new in the context of a first past the post, “winner takes all”, “flexible” list system, where voters have the option of . This logic does not sit well within a proportional system. . . preventing dual candidacy would voting for the party list or casting a “personal vote” be undemocratic and would place “an unnecessary for an individual on the list. restriction on the democratic rights of . . . local electors to have as unrestricted a choice as possible in an election. . . Open lists present a better way forward.” 14

105. To weight single preferences fairly in relation to list 112. So, what weight should the “personal vote” get? If votes, my version of the Swedish system is to give each a party wins 5 list seats, does a vote for the list count as single preference as many points as the number of list a vote for each of the five, while a personal vote counts as MPPs the party has won in that region. Thus, if a party only one vote? Some European models do this, but wins five list seats in a region, each of that party’s voters candidates rarely break those slates, when many party in effect has five preferences. A vote for the slate then voters choose to vote the slate. counts as one point for each of the top five candidates. A single preference for a particular candidate amounts to five points for that candidate. I’ll explain this in more detail. What weight should the “personal vote” get? Will PR encourage parties to splinter? 113. Belgium recently decided list votes will count as half a vote for each list candidate, to even the playing field 106. This ties into the threshold debate. Critics say PR slightly. Still, a faction that took all its supporters with it systems encourage parties to splinter. Indeed, one merit would do better as a separate party. At the other extreme of PR is that unnatural big-tent parties, with nothing in is Brazil, where a vote for the list has no weight at all common but power, are free to split into natural parts. But within it. About 90% of voters cast personal votes. The will this create a pizza Parliament? One answer is a high proverbial “strong man” still wins too often. (5%) threshold. But Canada gives public funding to parties getting 2% of the vote. 114. Voters can cast more than one personal vote. In Peru, you can vote for two. The popular slogan "Make 107. Consider party factions. With closed lists, they one of your votes count for a woman" seems to have must choose between bargaining for a good list position, worked a bit: they have 18.3% women (high for South like the SPD’s left wing in Germany, or splintering off, like America), while Brazil, where you can vote for one, has Germany’s Left Party. only 8.6% women. A personal vote for two candidates would be counted at the polls as a ½ personal vote for 108. Very low thresholds hold the power of the central each, but then in the seat calculation it would be multiplied party apparatus somewhat in check. When Israel had a by the number of list seats the party won in the region. 0.8% threshold, if a particular faction within a party—or even if a single high-profile politician in that party—was 115. The Quebec Citizens chose a closed-list model so sufficiently frustrated with the party and its direction, they the party nomination process can put minorities in a good could leave, form a new party and frequently did get list spot, and voters can’t “break the slate.” elected. 116. I propose what political scientists might call an intra- 109. But with open lists, a party faction can appeal party d’Hondt quotient. To give the single preferences as directly to the voters for support. Even with a threshold much weight as a list vote, give them as many points as as low as 2%, such a system need not create incentives the number of list MPPs the party has won in the region. for parties to splinter, if factions can stay and fight. Like Belgium, but more so. If a party wins five list seats in a region, in effect, every voter has five preferences, but a single preference amounts to a cumulative five points. A Factions can stay and fight. vote for the slate then counts as one point for each of the top five. Two preferences each get 2.5 points.

110. How, under the Charter, does one demonstrate that a 5% threshold is a justified limit on political freedom? Voters always think Many countries are moving toward a consensus on “plumping” will help. Now, it will. thresholds.

111. Scandinavia, Austria and many others use 4%. 117. Voters always think “plumping” should help; now, it Many use 3%. Germany has had a problem with neo-Nazi will. For example, women voters could plump for parties getting close to 5%. Israel has just come up to 2%; women. Palestine uses this too. Few other countries but Israel have had problems with micro-parties in Parliament, 118. With this model, there’s no incentive for factions to except Brazil and Columbia which have open lists gone leave a party. If the Blue Party gets 30% of the vote, and berserk, and party-hopping on a world-class scale. Yet 6% of voters support the party’s Purple Caucus, they can South Africa does fine with no threshold at all. elect one-fifth of its MPs, and would do no better if they formed the Purple Party. 15 Appendix 3: Options, alternatives and variations:

Of course OCA members can consider alternative MMP models, and more variations to these models.

1. Region size: larger or smaller regions

Lists give women more chance of an equal voice, and In fact, with overhangs, you also need "balance seats" for minorities of election, especially in larger regions with full proportionality, to balance the overhangs; several longer lists. In Plan B 6 of the 53 regional MPPs -- 11% German states do this, but Doody omits this option. Just of them -- are the only one from a little five-seat region, but as well: that would make a House of 189. All of which that’s a trade-off with northern and rural geography. We shows why MMP-lite is not the first choice of most elec- could have even longer lists than Plan A. York, Durham toral reformers, and AV would make it worse. and Peel could all go in with Toronto. Hamilton - Niagara - Halton - Waterloo could be a single urban region with 13 MMP-lite is not the first choice local MPPs and 9 regional. Yet with open lists, will a woman in Guelph or Fort Erie have a hope? of electoral reformers, and AV would make it worse On the other hand, smaller regions have better ac- countability. We want both fair representation for women and minorities, and accountability. Should This feature is not necessary. Combined with open lists for Hamilton and Niagara be separate regions? As long as the regional seats, the number of choices would be too you have province-wide proportionality -- not the large. large-party model first discussed in Quebec, with self- contained regions -- the result is still fully propor- Some voters would decline to state any second prefer- tional. ence, so the winner might still not have 50%. Some voters would have a sincere second preference. Others would Should Hamilton and Niagara be only try to stop a party they dislike: more negative cam- paigns. This would add complexity illogically: the “true separate regions? majority” would be a false majority based partly on nega- tive voting. Scotland’s unlinked regions have 16 Members of Scottish No MMP jurisdiction does this, but Ontario could do it if the Parliament. For a party to win a seat with only 6% of the OCA wishes to. Still, the core concept of MMP is to vote, unlinked regions must be around 16 MPPs. The Law respect the voter’s first choice. Commission of Canada's demonstration model used this method, but with huge self-contained regions of 36 or so MPs, for low thresholds. The big problem with this model 3. A larger or smaller Legislature? is vast regions, which reflect no community of interest or accountability. At the other extreme, Quebec is If Ontario had Quebec’s ratio of voters to MPPs, we’d studying a Scottish-style model with only five MPPs per have 179 MPPs. region. You’d need 17% of the vote to win a regional seat. After the 1981 census, we went from 125 MPPs to 130 2. Preferential ballot for local MPPs MPPs. From 1981 to 2001 Ontario’s population grew by 32.3%. To keep pace, we’d have 172 MPPs. In that case we could keep all 107 local ridings unchanged and add 67 Doody proposes that, within MMP, local MPPs be elected regional MPPs, 38.5% of the total. Or we could have 73 of by a preferential ballot or Alternative Vote (AV), so the 180, 40.6%. local MPP has a “true majority.” But if one party was everyone’s second choice, it could easily win so many From 1991 to 2001 Ontario’s population grew by 13.1%. local seats that MMP could not compensate. To keep pace we’d go from 130 to 147. At the Select Committee hearings Dr. Hiebert said we could go to 140. If AV had been used in the 2003 Ontario election, it's quite feasible that the largest party would have won 89 of the Conversely, if we don’t expand the Ontario House beyond 103 local seats, with 47% of the vote. For full proportional- 107 MPPs, we can’t keep the present northern ridings, let ity (Doody’s "hypothesis B," with overhang seats) there alone the rural ones. So should all parts of Ontario have would have been 28 overhang seats, and a House of 158. regional MPPs? If not, the three northwest ridings and But overhangs alone do NOT make full proportionality, as Timmins-James Bay could be unchanged isolated ridings. you can see: 89 is 56% of 158. With the total number of seats being calculated by province-wide proportionality, this would do no harm. 16

4. France's parity law

France is the first country in the world where its “Parity Law” says that for the majority of elections there must be If voters rank the top two regional as many women candidates as men. Since 2000, for candidates as a man and a woman, elections in France using a list system (municipal elections the man wins a local seat, and for at least 3,500 inhabitants, regional and European one person wins a list seat, elections, and elections to the Senate), 50% of the list candidates must be women. The objective is an equal that’s the woman. voice for women.

For European elections and elections to the Senate, the Therefore, what matters is -- are there at least two MPPs lists must be "zippered," that is, follow the principle of of each party elected from the region, including at alternating parity. For regional and municipal elections, least one regional MPP? Even in Plan B, this happened the parity law requires each list to be divided, from top to with 37 PCs, 79% of their MPPs. It happened with the bottom, into blocks of six candidates, of whom three must Liberals in regions with 41 (64%) of their 64 MPPs be- be women. (The "lumpy zipper.") cause, since they win so many local seats, they rate only 10 regional MPPs. Since the NDP rates only 20 MPPs, the No MMP jurisdiction does this yet. Each party could NDP does this only in regions with 45% of the MPPs. nominate twice as many regional candidates as the number of MPPs from the region: half men, half women. Three MPPs from a region, at least one being on the A Parity Law could say that, if three men won three local regional list, is even better, since some parties may find seats and the party got two regional seats, the top women 1/3 women enough. Even in Plan B’s simulation, 31 of the from the regional list would be regional MPPs. If men won 47 Conservatives (66%) were from such a region. two of the three local seats, again the first two regional MPPs would be the top women on the party’s list. German experience shows how this works. In the German SPD, only 16.1% of its MPs elected in 1987 were women. In 1988 they adopted their new policy of putting women in 5. Unofficial Parity 40% of the list seats. In 1990 they elected 27.2% women.

But do we need a law? MMP works for women even with A woman elected on a regional list last time will, if she no Parity Law. Parties don’t dare present biased lists. We expects her party to do better this time, make sure she can expect most parties to nominate a gender-balanced gets a good nomination in a winnable local seat, since her regional list, either naturally or by internal party arrange- regional list seat may evaporate as her party does better ment. Some parties will choose to “zipper” their lists, nationally. So as women's nominations spread into the alternating women and men. Others won’t. local ridings, the number grew: 33.7% in 1994, 35.2% in 1998, and 37.8% in 2002, exactly as one would expect. But 90% of voters want more women in politics, and we have an “open list” system where voters can rank women on top. As women's nominations spread into local ridings, the number grew Quebec’s Citizens Committee recommended that, if the number of women is not significantly higher after two elections, zippered lists be considered. That would work Why did the SPD not take this step before 1988, when for Ontario too. German SPD women must have been seeking election since the 1960s? First, it shows the entrenched strength Take the worst case, in a little region with three local of the SPD male oligarchy, which makes the subsequent MPPs and two regional MPPs. Say three men win all three progress in the local constituencies more noteworthy. local nominations. Say the party members add two women to the regional list of five. If a party wins two seats, and if Second, German Greens would no doubt point out the voters rank a man and a woman as the top two names on value of competition from more parties: they entered the regional list, and the man wins a local seat, so one the German Parliament only in 1983 with only 5.6% of the person is elected from the regional list, that would be the vote, just over the threshold. Their policy of equal numbers woman, even if she was ranked second, because the of men and women MPs required action from the SPD man was already elected locally. after the 1987 election when the Greens stayed in Parlia- ment with 8.3% of the vote. 17

6. "Two classes of MPPs?" In Scotland's 2003 elections, 88% of successful regional MSPs had fought, and lost, a constituency seat, but the Does the MMP system produce two classes of MPPs? Commission finds "There is no survey evidence to suggest This has been exhaustively studied by Louis Massicotte that dual candidacy is an issue for voters, or a disincentive (see p. 61, Chapter 8, "Two Classes of Elected Mem- to their participation in the political process. Few of our bers?") who concludes "The above data strongly support consultation responses raised dual candidacy as an issue, the prevailing consensus in the literature: the existence of nor was it raised spontaneously in our focus groups. . . two types of parliamentary mandates within the same Dual candidacy is a common and accepted feature of parliament does not produce two unequal castes, and the mixed member proportional systems across the world . . distinctions based on this criterion are far from having the . We suggest that dual candidacy only seems problematic crucial importance that some give it. As we will see in the to some people here because of the legacy of constitu- following chapters, this reality has a lot to do with the way ency representation within British political culture and the the compensatory seats are filled." hegemony which this has secured for some parties." So it’s normal in MMP systems that candidates run for 7. “List-free MMP”: best runners-up? both local and regional MPP. But what about leaders of small parties that don’t know where their strongest regions Some voters just don't like party lists, even when they're will be? Should they be able to stand on more than one locally nominated in natural regions, and even with "open regional list? Some countries let the party’s top three or lists." They prefer the "best runner-up" model. It works just five candidates do this. like these MMP models, but the "top-up" MPPs are the party's "best runners-up" for local seats in the region, not from regional lists. This is used in the conservative 9. STV (the )? southern German state of Baden-Wurttemberg, where they wanted to keep politics local, with no central party In September 2005 Ontario’s Select Committee on control. The small regions in this model already meet the Electoral Reform was studying electoral systems. They objective of keeping politics local, with no central party had been to Ireland to learn about the Irish STV system, control. but they could find no academic expert here who recom- mended it. The only witness who liked STV was Tasha With no lists, there is less chance for party members Kheiriddin, provincial director of the Ontario division of the to decide to promote women candidates. Baden-Wurt- Canadian Taxpayers Federation. She recommended STV temberg has fewer women elected than most German for the two-thirds of Ontario’s ridings that are in multi-riding states, because local riding members are “gate-keepers” urban areas. just as in Ontario, nominating one man at a time. Still, some best runners-up would be women. Many ridings get However, for the other one-third she said “the problem two MPPs, a few get three, one even has four, some have with STV in rural ridings is that to have a multi-mem- only one. Quebec considered using this option, but ber district you would have territories that are so huge decided against it. it would be impossible, unfeasible, to even campaign in them. . . in the north, it would just be enormous.” You could take the runner-up as the candidate with the For Ontario, MMP is the answer. Also, there’s no consen- highest percent. Baden-Wurttemberg defines it as the one sus how to count STV ballots: Gregory Method, IGM, with the highest vote total. This gives an incentive for high WIGM, Dail Eireann, Cambridge (MA.), or Meek? turnouts, and preference to fast-growing ridings. 10. Solving the Twin Party problem: 8. Dual candidacy; Multiple races Those politicians who want to manipulate the MMP The Arbuthnott Commission Report in Scotland is an system may try to run “twin parties” or a “dummy party:” excellent review of the design issues with MMP, from one party label for their local candidates, and a second for pages 29 to 67. It considers the Dual Candidacy issue at their regional list candidates. This means that the list page 44. winners, rather than compensating for local disproportion- alities, are added to the local winners. This gives large parties a big bonus. It has happened already, and is a reason why Quebec considered a single-ballot model. 18

The best solution is an Elections Act that gives the In Northrhine-Westphalia they start with only 29% top- Chief Electoral Officer strong powers to rule that ups, not 50%, but in the 2000 election the “overhangs” and “Twin Parties” are in reality a single party. However, “balance seats” added 15% more seats. parties could still launch campaigns directed at only regional seats without any corresponding local campaigns, In Ontario we could keep the 107 local MPPs we’ll have which would undermine the proportionality function of the after redistribution, but initially have only 27 regional regional seats. MPPs -- only 20% of 134, which would have been enough in 1999 -- but if “overhangs” and “balance seats” require Another solution clearly within your mandate is to cap the more MPPs for proportionality, we would automatically party vote: a party's Final Total Vote shall be not more add more MPPs. If we need 26 more, that would bring us than twice the sum of votes received by its local up to 160 MPPs; 33% of them would be regional MPPs. candidates. For example, this would not penalize a party However, to maintain regional balance, we’d have to add getting 12% of party votes unless it had less than 6% of the additional “balance seats” in regions other than the local votes across Ontario. ones with “overhang” seats, so that each region still had its proper share of the seats. This makes regional candidates somewhat dependent on the success of their local counterparts. This encour- In Northrhine-Westphalia, by the way, German's largest ages cooperation between local and regional candi- province shows once again it's a fallacy that PR always dates of a party (if they are not already dual candi- produces coalitions. In 1980, 1985 and 1990 it elected dates), who can otherwise have an incentive in MMP majority SPD (Social Democrat) governments (how's that systems to hinder each other because of the compen- for stability?). In 1995 and 2000 it was an SPD-Green sating mechanism between regional and local seats: the coalition, defeated in 2005 by the current Christian fewer local MPPs the party elects, the more regional Democrat - Liberal coalition. MPPs it gets. Author’s note: Wilfred Day has been a lawyer in Port 11. Aboriginal seats? Hope, Ontario for 36 years, and served on the Ontario Bar Association’s Council and its Familly Law Execu- New Zealand has had, for a long time, a separate aborigi- tive. He was elected as a school trustee from 1982 to nal voters list. Seven MPs are elected by those Maoris 1994, and served on the Provincial Policy Committee who choose to vote on the Maori list, from seven Maori of the Ontario Public School Boards Association. An ridings, a parallel set of local ridings located across the active electoral reformer, he has prepared these whole country, with their “party votes” part of the national comment as an individual. No political philosopher, he total. With at least 188,310 people with aboriginal identity has been called an electoral mechanic, which he takes as a compliment. in the last census in Ontario, they would get two or three aboriginal seats. These sample models do not include aboriginal seats, since Ontario’s First Nations have not Comments to: [email protected] decided if they want it, but we could easily add them.

12. “Top-ups” percent, “Overhangs” References: Massicotte Report “In Search of A Compensatory Mixed Here’s a Canadian-sounding option: an expanded Electoral System for Québec” legislature if necessary, but not necessarily expan- http://www.institutions-democratiques.gouv.qc.ca/public sion. It’s the Dusseldorf solution: the biggest German ations/mode_scrutin_rapport_en.pdf state, Northrhine-Westphalia, does this. Arbuthnott Report In most of Germany, half the MPs are from party lists. In http://www.arbuthnottcommission.gov.uk/docs/Final%20 New Zealand it’s 44% "top-up." In Scotland it's 43%. Report/Low%20Res%20Arbuthnott%20Print.pdf We've used 40%, often said to be the minimum required. But what if 40% isn’t enough? Equal Voice presentation and Fair Vote Ontario presenta- tion to Ontario’s Select Committee on Electoral Reform: In Scotland, if a big party wins more riding seats than it http://www.ontla.on.ca/hansard/committee_debates/38_ deserves from its share of the party votes, the House size parl/session1/ElectoralReform/ER003.htm#P473_151284 and region size are fixed: the other parties lose out. In Germany and New Zealand, the big party keeps the extra Quebec Citizens’ Committee Report (In French) seats (called "overhangs"), and other parties don’t lose. In http://www.assnat.qc.ca/fra/37legislature2/commissions/ some German states the other parties also gain “balance csle/rapport_comite_csle.pdf seats” to balance the overhangs. The size of the House goes up by the number of overhangs and balance seats For ongoing discussion of these and other models, see: until the next election. http://citizen2citizen.ca/model_rating_program