Rural-Urban Proportional

Rural-Urban Proportional Representation (RU-PR) is an approach to blending the strengths of STV and MMP that was inspired by a suggestion made by former Chief Electoral Officer Jean-Pierre Kingsley, who proposed that we use multimember ridings in the more urban parts of the country and retain single member ridings in the more remote parts.

What is the Rural-Urban Proportional Model?

The Rural-Urban PR model is a Made-for-Canada voting system that offers tremendous flexibility to adapt to the wide range of situations found across Canada.  As suggested by Kingsley, it uses multimember ridings in the more heavily populated parts of the country, and allows single member ridings where deemed necessary.  To ensure a high degree of overall proportionality, RU-PR also has a small number of seats distributed in top-up regions on a compensatory basis to any parties that are under-represented in the single and multi-member ridings, much as is done with MMP.

An Example From BC – How Rural-Urban PR Might Work on Vancouver Island

RU-PR for Vancouver Island
Rural-Urban PR Example for Vancouver Island.  The single member riding is shown in yellow, multimember ridings in green, and the topup region in red.  The number of MPs in each region are indicated by the ‘head and shoulders’  icons.

The figure above shows how RU-PR might work on Vancouver Island.  There are currently 7 single-member ridings on the island.  In 2015, one party took 6 of these seats with only 33% of the vote.  These 7 seats could be replaced by a four-member riding in the more densely populated southern region (shown in green), a two-seat riding in the mid-Island region (also in green), and we could keep a single member riding in the north Island (shown in yellow).  Since the two multi-member ridings would be much more proportional than single member ridings, we could add a single top-up seat for the entire island (shown in red).  Alternatively, if we wished to keep the total number of seats unchanged, we could decide to have only six seats in the first tier and one seat in the top-up tier.

What Might Have Happened in 2015 With Rural-Urban PR:  In the 2015 election, the vote shares of the various parties were NDP 33%, GP 24%, CP 21% and LP 21%.  The NDP won 6 of the 7 seats, while the Green Party took the 7th seat.  Under RU-PR, the NDP would likely have won one seat in all three local ridings, the Conservatives would likely have won one seat in the two southern ridings, and the LP and GP would have won the other two seats in the southern riding.  The GP would likely have been awarded the top-up seat.  Overall, this result (NDP 3, GP 2, CP 2 and LP 1) would have much more closely matched each party’s vote share.

Design Choices with Rural-Urban PR

Fewer Top-Up Seats Needed:  RU-PR requires far fewer top-up seats than are required with Mixed Member Proportional (MMP) because the results in multimember ridings are so much more proportional than those in single-member ridings.  If more multimember ridings are used (even if there are relatively few MPs elected in each), we can generally achieve excellent proportionality with roughly 15% topup seats, compared with ~40% in most MMP models.  In the example above, we added one seat to seven existing seats.  Doing so would allow us to keep existing riding boundaries intact and to form multimember ridings by merging adjacent ridings.

Small Increase in Single-Member Riding Size if No Change in Number of MPs:  However, if we wanted to keep the overall number of MPs unchanged, we would need to ask an Electoral Boundaries Commission to set aside 15% of the seats and allocate the remaining 85% of the seats within new boundaries.  In the Vancouver Island example above, this would likely mean leaving six seats in the single and multimember ridings and allocating one seat to the top-up region.  In comparison with MMP, this means that in the more remote parts of the province where we might particularly wish to use single member ridings, the population in these ridings could be less than 120,000 people instead of the ~170,000 people that would be needed with an MMP model with 40% of seats set aside as top-up seats;  the single member riding size under RU-PR would therefore be much closer to the ~100,000 population we have with our current Single Member Plurality system, even if we kept the total number of seats unchanged.

Fewer Top-Up Seats Needed With More Multi-Member Ridings:  The number of top-up seats we would need to get strongly proportional results would depend on the fraction and size of multimember ridings we use – the more multimember ridings and the more MPs per multimember riding we have, the more proportional the result will be in these ridings and the fewer top-up MPs we might need.  RU-PR is extremely flexible – we could opt for mostly multimember ridings as in the example above, or we could use more single-member and 2-to-3-member ridings, as shown in the example below of how Rural-Urban PR might work in the Interior of BC:

Rural-Urban PR Model for BC Interior
Rural-Urban PR Model for BC Interior

In the Interior, we might want to have single member ridings in the most remote parts of the province – particularly the northwest and the southeast.  Near Prince George and along Highway 1 from Kamloops eastwards we might want to create two two-member ridings, and perhaps a three-member riding in the Okanagan area, with perhaps one top-up seat for the entire Interior.  In 2015, the CP won 37% of the vote in the Interior, while the LP won 30% and the NDP 28%.  In that election, the CP won 5 of the 9 seats, the NDP 3 and the LP 1.  Under RU-PR, we would predict that the NDP would have won the Skeena seat, the CP and the LP would have won near Prince George, the CP and LP in the Kamloops/North Okanagan, the NDP in Kootenay-Columbia, and perhaps one seat for each of the CP, LP and NDP in the Okanagan, for a total of 3 seats each for the CP, NDP and LP.  The CP would have won the top-up seat, which would make the regional result essentially perfect:  4 seats for the CP (37% of the vote), and 3 seats for each of the LP (30%) and NDP (28%).  This demonstrates how much improvement in proportionality comes from even small multi-member ridings (2 or 3 seats).

Larger Urban and Suburban Areas:  Rural-Urban PR also works exceptionally well in more urbanized areas where no single-member ridings would be needed.  Consider the following two examples from the Lower Mainland (Greater Vancouver and the Fraser Valley):

Rural-Urban PR Example for Greater Vancouver
Rural-Urban PR Example for Greater Vancouver


Rural-Urban PR Example for Fraser Valley
Rural-Urban PR Example for Fraser Valley

The Greater Vancouver example shows a 2-seat riding on the North Shore, a 6-seat riding in the city of Vancouver, and two 3-seat ridings in Richmond/Delta and Burnaby/New Westminster.  This 14-seat region might have two regional top-up seats.  The 12-seat Fraser Valley example shows four 3-seat ridings – one North of the Fraser, one in Northwest Surrey, one in Southeast Surrey/Cloverdale/Langley/Aldergrove, and one in Abbotsford/Chilliwack/Hope;  this region also might have two top-up seats.

We estimate that in Greater Vancouver, the LP would have won 7 of the 14 first tier seats, the CP 4 seats, and the NDP 3 seats, with the NDP and GP each picking up one of the top-up seats.  This means that the LP would have won 44% of the seats on 44% of the vote, the CP and the NDP 25% of the seats each on 26% and 23% of the vote, respectively, and the GP would have won 6% of the seats on 5% of the vote – essentially a perfectly proportional result for this region.

In the Fraser Valley, the CP and LP would each have won 5 of the 12 first tier seats, with the NDP picking up the other two, and the NDP and GP each picking up one of the top-up seats.  Overall, the LP and CP would each have won 36% of the seats on 39% and 35% of the vote, respectively, the NDP would have won 21.4% of the seats on 21.6% of the vote, and the GP would have won 7% of the seats on 4% of the vote – again, an essentially perfectly proportional result for the region.

Flexibility of Rural-Urban PR:  RU-PR can be custom-tailored to the needs and desires of each part of the country.  In the examples given here, we see that it can work well with about one top-up seat for every 7-9 local seats and that we can have multimember ridings that range in size from 2 to 6 (or more, if desired).  We saw excellent results in the BC Interior with 9 local seats and a single top-up seat, though generally the more small and single-member ridings we use, the larger the top-up region we should have and the more top-up seats we should use.  For example, Fair Vote Canada has shown a model for Alberta with 11 of their 34 seats remaining in single-member ridings, with two top-up regions with three top-up MPs in each;  even with almost a third of the seats remaining as single-member districts, 15% top-up seats were enough to provide excellent overall proportionality.

RU-PR Similar to Scandinavian Voting Systems:  We note that a number of the Scandinavian countries currently use a model very similar to RU-PR. In particular, Sweden uses an open list system which includes 29 multimember ridings ranging from 2 to 36 seats, totalling 310 seats, plus 39 ‘adjustment’ seats allocated at the national level (11% of the total number of seats).

Why Another Model?

Given that most of the public discussion to date has revolved around two well-known voting systems (STV and MMP), why do we feel that we should be discussing yet another option?

While it’s true that both STV and MMP would be excellent choices for Canada, with key strengths that address important values Canadians share regarding our electoral processes (such as effective representation of all voters, fundamental fairness, representation of our society’s diversity, the importance of clear connections between voters and MPs, etc), various aspects of these two systems are sometimes perceived as potentially problematic, both by critics of reform in general and by politicians who are being asked to endorse a new voting system.  If we could defuse the concerns of these people, it could smooth the path to reform.

With both systems, concerns are sometimes raised about the increase in constituency size needed to ensure representation of diverse political perspectives amongst voters. Since all proportional voting systems in some way depend on having multiple MPs representing voters in each geographic area, if a PR system is to be used, then a key question is what the optimal way to do this might be.

With STV, adjacent ridings are merged and the merged riding sends more than one MP to Ottawa. In BC, the Citizens’ Assembly recognized the challenges of serving more remote areas and so recommended that the more remote parts of the province have smaller merged ridings – e.g., some districts of two and three members were proposed for parts of the BC Interior – while larger ridings of up to about 7 representatives could be imagined in the bigger cities. Nonetheless, even with two-member districts, this would imply a minimum rural riding size at the federal level of roughly 200,000 people (as compared with the roughly 100,000 in an average current single-member riding – see figure below). In addition, if a two-member riding were used in these areas, then voters who did not support one of the top two candidates would not have a representative of their choice in Ottawa; assuming a typical vote distribution of 40-30-20-10, 70% of voters would be happy with one of their riding’s two MPs, but 30% might not be.

With MMP, single-member ridings are expanded, and regional seats are added. Assuming a 60-40 split between single-member and regional seats, the single-member ridings would contain approximately 167,000 people (compared with our current 100,000), and the regions could contain anywhere from 800k to 1.4M people (assuming regions of 8-14 seats were used).

Sizes of the smallest ridings under our current FPTP voting system (red) and three proportional voting systems (blue). Note that with STV the smallest riding will have two MPs, not just one as with the other systems.

Because both STV and MMP would require an increase in size of the current single member ridings, some have proposed leaving the rural ridings as they are and only using PR in the cities. Indeed, such a scheme has been used before at the provincial level in both Alberta and Manitoba. Predictably, voters in the cities were well-represented by a diverse slate of MLAs, but representation in the rural parts of the provinces was heavily skewed towards one party or the other, so the overall results were not particularly proportional and denied the benefits of alternative representation to roughly half of the rural voters.

Figuring out a way to bring the benefits of proportionality to all voters in a province is the driving force behind the proposal for this Rural-Urban Proportional model.

The Rural-Urban PR Ballot

Finally, we note that the Rural-Urban PR ballot would be very simple – in the single member ridings, it could look exactly the same as it does today, or we might opt to use a preferential ballot. In the multi-member ridings, it would look very much like an STV ballot (i.e., with names of multiple candidates from the same party on the ballot). Voters could either be asked to rank their top few choices, or simply to mark a single ‘X’. Either way, the top few vote-getters in each multi-member riding would be elected. The top-up seats would then go to the strongest remaining candidates of any parties that deserve additional seats.  Note that a version of RU-PR in which we use mainly multimember ridings and a preferential ballot could simply be considered an extension to STV – in our description of STV, we call this enhanced model STV+.

The ballot for a multi-member Rural-Urban PR riding would look the same as an STV ballot. Voters could either rank multiple candidates or simply place an 'X' beside their preferred candidate.
The ballot for a multi-member Rural-Urban PR riding would look the same as an STV ballot. Voters could either rank multiple candidates or simply place an ‘X’ beside their preferred candidate.  The top-up MPs could be the candidates from the most under-represented parties in the region who won the most votes in their local races, so there is no need for a second regional ballot as is normally used with MMP.

National Performance of Rural-Urban PR

We simulated how Rural-Urban PR might have performed in the 2015 election. Essentially, it would perform as well as any other PR method with equivalent region sizes. That is, if we used typical region sizes of about 15, with a few smaller regions where required (e.g., in some of the Atlantic provinces, or more distinct regions such as the BC Interior or northwestern Ontario), we would expect to see performance equivalent to the 14-seat MMP model or the STV+ model (which itself could simply be considered a particular form of RU-PR). We therefore feel that RU-PR would be an excellent choice for Canada to adopt.