Mixed member proportional (MMP) is an electoral system used with some variations in a number of countries, including Germany and New Zealand.

The essential idea is that you have two votes: one for your representative, and one for a party. The idea is to combine the benefits of proportional representation (PR: parties with uniform support across the country but less than 50% in any given area still can win representation) and constituency or electorate MPs (a representative accountable to a community).

The problem with MMP is that it does not eliminate one of the drawbacks of a pure PR system: parties have a disproportionate say in who gets elected, since a party generally gets to draw up and rank its party list. In South Africa, for example, the power of the party to decide who gets to be a candidate and how high up the list they get to be has had a significant effect on the independence of MPs.

Representative Member Proportional Voting

I propose a variant on this system, which I call Representative Member Proportional Voting (RMPV), with several benefits:

• counting is easier
• there is no need to make complicated preference deals

In this model, the first round of counting results in a member of parliament being elected on a first-past-the-post basis. In the second round, all votes countrywide are tallied, to determine the composition of parliament on a PR basis, and MPs who did not make it the first round are elected to correct party representation up to their proportion of the vote. The MPs elected in the second round are on the basis of a party list system, in which candidates are ranked on the basis of their popular vote in the first round. The party list is therefore decided by voters, not by the party. The parties do still get to decide who to put up as candidates, but the power of the party is significantly reduced compared to the traditional party list system.

If the MPs are split as equal numbers in both rounds of counting, it is impossible for one party to win an outright majority. Consider the extreme case where a party wins every seat on the first round. They will not have any MPs left to compete in the second round, and will therefore end up with exactly 50% of the seats in parliament. Choosing the size of the second-round vote therefore constrains the maximum majority the winning party can attain. For purposes of an example, let us fix the first round at 100 MPs, and the second at 80 MPs. This means that a majority of 20 is possible to attain.

Example 

Let's consider a specific scenario to see how this could work (using Australian parties for purpose of example but with made-up numbers; 120 seats in the first round, 80 in he second). We assume (with some relation to reality but, again, the example is made up) Liberal and National have geographically limited concentrations of votes, and Labor and Greens more uniformly distributed. Family First has no geographic  area of strength big enough to win a seat.

First, round 1, the first past the post election of electorate MPs could look like this:

• Liberal: 30%, 40 seats; 20% used up electing MPs
  
• Labor: 47%: 60 seats; 30% used up electing MPs
  
• National: 8%, 20 seats; all votes used up electing MPs
  
• Green: 10%, 0 seats; 0% used up electing MPs
  
• Family First: 5%, 0 seats; 0% used up electing MPs

The Nationals result could happen for a party that happened to only run in seats they won, and that they won on an average of 40% of the vote, which could happen in a first past the post system. In round 2, they would already be over-quota and would get no more seats. The Liberals and Labors results allow them to transfer votes over the 50%+1 needed to win each member to the second round; the two smaller parties transfer their entire vote to the second round.

In round 2 then, we have 80 seats to distribute, based on the remaining unallocated 42% votes:

• Liberal: 10%, 19 more MPs, total 59 (30% of parliament)
  
• Labor: 17%, 32 more MPs, total 92 (46% of parliament)
• National: 0 more MPs, total 20 (already at 11% of parliament)
• Green: 10%, 19 MPs, total 19 (10% of parliament)
• Family First: 5%, 10 MPs, total 10 (5% of parliament)
  

The end result is not too far off a PR vote. The Liberals are spot on, Labor within 1%, the Nationals a bit over and the minor parties right. The Nationals example is not terribly likely; I took an extreme case to see if anything very unfair could result. I am happy if an outcome like this is extremely unlikely and can at best favour a small party by a few points, which is still a lot better than most non-PR systems (including Australia's preferential system). Add to this the gain of having a constituency or electorate MP and avoiding the problem with pure PR of putting too much power in the hands of parties, and this idea looks good to me.

The MPs in the second round are those who had the highest vote out of the candidates who were not elected, ranked within their respective parties. Independents who are not part of a formal alliance could be treated as a single party list in round 2.

Reflection on Example

We can end up with something slightly off PR because we have allowed a theoretical majority of up to 20 for a party that won all the first round seats, and (in this case) a party with very concentrated support in a small part of the country can win slightly more than its share). However it is much closer to PR than you would have in the current electoral system in Australia that uses transferable preferences. Another advantage of RMPV is that, unlike in a pure PR system, you get not only a representative but also potentially multiple alternative representatives. If you are in an area where one party is very strong, the "alternate" may be quite far away -- for example, in a large rural electorate of a type the Nationals may dominate, you may not have a Greens alternate anywhere close enough to be practical to call on. However, in an inner-city area where several parties may have at least 5% of the vote, you may have several MPs to call on. On the other hand, in an inner city area where the Labor vote is high, you may have to look further afield to find representatives of another party. Provided the Labor member is doing her or his job well, this is not necessarily a problem.

The RMPV result is not a bad outcome. If a community is fairly homogeneous in its composition, it makes sense to have a smaller variety of MPs representing their interests in parliament. There is however no dilution of their representation: all their votes have contributed (give or take the effect of fewer of their preferred party's MPs on the second round than on the first).

Further evaluation of the idea would be useful: working up a range of scenarios to make sure that bizarre or pathological outcomes are not possible.