Paul's Internet Landfill/ lj/ MMP 102: Effects of the 3% Threshold

MMP 102: Effects of the 3% Threshold

Most list-based proportional representation systems have some kind of threshold that parties must hurdle in order to win list seats. These thresholds are intended to keep tiny parties out of power and to act as a disincentive for parties to splinter too much. Although this practice offends extremists and PR purists, most people agree that it is better to have systems that are slightly disproportional than to suffer from the coalitions of coalitions that Italy has to deal with.

The hope is that most people can find a party that matches their views pretty closely because PR allows several parties to be viable. Compare this to FPTP systems, which feature two or three big-tent parties that are distant from a lot of voters.

The Ontario Citizens' Assembly chose 3% as its threshold: in order to qualify for the 39 list seats, parties must earn 3% of the party vote (tallied provincewide). Based on the 2003 Ontario turnout of 4.5 million voters, that translates to about 135 000 votes.

Scotland and Wales don't have formal thresholds. They shut out parties by splitting their jurisdiction into small regions, and tallying the party votes in each region independently. For example, Wales uses five regions to split up 20 list seats. As a result, parties need quite a large share of the party vote to qualify for anything -- in the Wales 2007 election, no party received list seats with less than 11.7% of the vote, and three parties reached 3% without getting any seats (the highest of which was the British National Party, with 4.3% of the vote). Scotland did better in 2007: an independent won a list seat, and the Scottish Greens got two seats while earning 4.0% of the list vote spread across all regions.

Germany and New Zealand both use a 5% party threshold, which according to Massicotte (Massicotte Summary 2004, p. 9) is pretty standard. However, both Germany and New Zealand have an "electorate threshold" that can also be used to qualify for list seats. In New Zealand parties that get less than 5% of the vote can receive list seats if they win one riding. In Germany parties can get list seats if they win three ridings. Ontario has no such rule.

Thresholds and Wasted Votes

As many critics point out, electoral reformers tend to be obsessed with the idea of wasted votes. Fair Vote Canada defines a vote as wasted if it does not contribute to helping get somebody elected. For first-past-the-post systems, all votes cast for people who don't win their ridings are wasted, which is exactly why FPTP results in huge distortions between the popular vote and the percentages of seats each party wins. In MMP, my understanding is that people usually count the percentage of party vote that does not help get anybody elected. This is a bit sneaky, since it ignores the local riding vote. The idea is that even though a lot of the riding vote will continue to be wasted, most electors will continue to have their ballots count for something through their party vote. If nothing else, counting wasted votes in this way gives us a rough sense of the degree to which our voting systems disenfranchise voters.

It's pretty easy to see that thresholds will increase the number of wasted votes that are cast -- after all, one of the objectives of thresholds is to deny hardcore extremists their voice in legislature. The question in my mind is how thresholds affect the number of wasted votes that are cast.

It appears that MMP systems tend to waste about 5% of the vote: according to Fair Vote dogma (which I don't have sources for) in 2003 Scotland wasted about 6% of the vote while in 2005 Germany wasted about 4%. New Zealand had a much lower wasted vote total at 1% in 2005, but I will argue below that this is freaky and undesireable. Wales does much worse than other Anglophone MMP systems: according to my back-of-the-calculator calculations, in 2007 Wales wasted at least 15.7% of the vote (and in fact wasted more because I did not count the vote within regions). That's pretty hideous, but it's nothing compared to what we deal with in Canada: in the 2006 federal election 51% of us cast wasted votes, and according to my calculations in the Ontario 2003 election we cast 49%.

It is pretty clear that we would do a lot better under MMP than FPTP in terms of wasted votes, but it is not at all clear how many wasted votes we can expect, or even whether we should expect fewer or greater wasted votes than places that use a 5% threshold. One letter to the editor I read in The Record cavalierly predicted 10% wasted votes (Breithaupt 2007) but I have a feeling that the writer picked this number out of the air. Certainly I have not yet found good evidence to support this conclusion.

Massicotte (Massicotte 2004 Long, p. 41) has a table that shows the average wasted votes in the German Ler (provinces). Tallying the averages, we get the following ranges:

Range of wasted votes on average (x) Number of provinces
x <= 5.0 4
5.0 < x <= 7.5 4
7.5 < x <= 10.0 2
10.0 < x 2

The higher wasted votes came uniformly from East German provinces, which had fewer elections (four at the time of writing) and had recently transitioned from communism. I suppose it is conceivable that we could waste around 10% of our vote under MMP, but I would not bet on it and I strongly believe that the average will go down in the long run.

How do 3% thresholds compare against 5% ones? One possibility is that a 3% threshold will waste fewer votes than a 5% threshold because more parties will be able to reach the threshold. The other option is that the smaller threshold will waste more votes, because more parties will attempt to hit the lower target and fall short. In Germany and New Zealand there is a pretty big jump between parties that meet the threshold and parties that don't; most parties that miss the cut get 1% of the vote or less. It is not clear to me how things will play out in Ontario.

Effects of the electorate threshold

Ontario does not have an electorate threshold, where parties can dodge the 3% requirement by winning some ridings. wilf_day expressed relief at this, because New Zealand has been exploiting this threshold like nobody's business. In 2005 eight New Zealand parties received seats in parliament; of these four parties earned less than 5% of the party vote, and received seats because they had won some ridings. It appears that the two big New Zealand parties work to ensure that their favoured coalition partners win ridings. One way big parties do this is to avoid contesting certain ridings, and then to tell their supporters to cast their riding vote for certain small parties instead (Vowles Banducci Karp 2006). As a result, there have been quite a few calls to have the single seat electorate threshold abolished in New Zealand (Vowles Banducci Karp 2006) (NZ MMP Review 2001, p. 50)

On the other hand, the electorate threshold gives smaller parties some incentive to contest ridings. Currently, my understanding is that small parties need not contest any ridings in order to appear on the party vote portion of the ballot. The Scottish Greens did exactly this in the 2007 elections. I suppose this is a rational strategy, but I don't like it one bit. I prefer deciding how to vote via in-person situations like all-candidates meetings; this gives me a sense of the quality of candidate each party recruits, as well as how different party policies play off each other. I also worry that parties that cannot scrape enough volunteers to run candidates in most ridings are parties that are not serious enough to consider for office. For these reasons, I far prefer the policy of the New Zealand Greens, who encourage their list candidates to run in ridings as well. I would like to believe that parties who do not run many local candidates will have a harder time passing their thresholds, but there is no way to be sure.


As usual, I leaned upon Wikipedia to get election results, which is going to get me in real trouble one day:

(Breithaupt 2007) James Breithaupt. "Proposed election reform is wrong approach" (Letter of the Day). The Record (Kitchener), June 20 2007.

(Massicotte Long 2004) Louis Massicotte. In Search of a Compensatory Mixed Electoral System for Quc, Gouvernment du Quebec, 2004. ISBN 2-550-43379-3. Available from <http://www.institutions->

(Massicotte Summary 2004) Louis Massicotte. In Search of a Compensatory Mixed Electoral System for Quebec: Summary, Gouvernment du Quebec, 2004. ISBN 2-550-43380-7. Available from <http://www .institutions->

(NZ MMP Review 2001) Rt. Hon Jonathan Hunt, chair. Inquiry into the Review of MMP: Report of the MMP Review Committee, New Zealand House of Representatives, August 2001. Available from:

(Vowles Banducci Karp 2006) Jack Vowles, Susan A. Banducci, Jefferey A. Karp. "Forecasting and Evaluating the Consequences of Electoral Change in New Zealand", _Acta Politica, vol 41, 2006, pp. 267-284. Available from NZES website:

Livejournal URL:

Mood: distraught