Poland 2015-Some observations on threshold

Poland recently held elections to both houses of its parliament: the Senate, which is the upper, newer and less powerful house, and the Sejm, which is the lower, older and more powerful house. The Senate is virtually powerless, in fact. A majority vote of the Sejm can override a Senate rejection of any bill. This rarely matters, though, because the Senate is elected at the same time as the Sejm, and uses a less proportional electoral system.

The impact of the threshold

Poland uses D'Hondt open party-list proportional representation in districts with an average magnitude of 11. The system includes a threshold of 5% on parties and 8% on coalitions of parties, which is implemented on the national level. An exemption is made for lists that represent ethnic minorities.

District allocation and a nationwide threshold both have a negative impact on proportionality. However, which one has the most significant impact says a lot about a nation's party system. If a nation has a party system where parties have consistent support across the country, then the use of district allocation will have a more substantial impact on proportionality. But if the party system is geographically diverse, the use of a nationwide threshold will have a more dramatic impact.

Let's look at Poland's party system. At this election, there were two major parties; Law and Justice and Civic Platform. Law and Justice is a traditionalist conservative party, while Civic Platform is a more liberal party on both social and economic issues. Civic Platform was in government coming up to this election. Musician Pawel Kukiz also formed his own party, after winning 20.8% in the first round of the presidential elections earlier this year. He is an anti-establishment figure, and his party's main issue is the introduction of the single-member plurality system in Poland, which he claims will remove party domination of Poland's politics.

The Polish left has not been especially strong in recent years. It's divided between post-communists, modern greens and socialists, and liberals. At the last election, the Democratic Left Alliance, which is the major left-wing party, won 8.2% of the vote and 27 out of 460 seats in the Sejm, while Palikot's Movement, a left-liberal party, won 10% of the vote and 40 seats. However, Palikot's Movement has dramatically lost support over the past few years. At this election, it (now called Your Movement) formed a coalition with Democratic Left Alliance and a number of other left-wing groups. The coalition was called United Left. Another more left-wing party was also formed, called Razem (Together).

A number of liberal parties also contested the election. The largest was Modern, which is a pro-EU centrist party, in a similar vein to the Liberal Democrats in the United Kingdom and the Free Democrats in Germany. It was only formed this year. The second is KORWiN, which is a more radical libertarian, Eurosceptic party. Its leader, Janusz Korwin (see what he did there?)-Mikke, is something of a controversial character. He has made statements about women that have gotten him in trouble, and his views are somewhat polarising. His former party, the Congress of the New Right, won four seats in the European Parliament at the 2014 elections, and Korwin-Mikke himself is a member of the European Parliament.

The Polish People's Party is the other substantial political presence in Polish politics. This party is one of the oldest, and was around before the Communists took control and in some form before World War 2. It is focused on rural issues, and is economically leftist and socially conservative.

It's worth noting that none of these parties are focused on local issues. Poland is also fairly ethnically homogeneous. As a result, I would expect that national instead of regional allocation would have the most substantial impact in terms of improving proportionality.

The graph above shows the results. As you can see, regional application of the threshold would lead to United Left and  However, this does not give us an indication of proportionality. To do this, the Gallagher index is used. This is a standard index of disproportionality that goes from 0 (most proportional) to 100 (least proportional). On this scale, the actual result scored 12.55. If the threshold was applied regionally, the index would be 9.86. However, if allocation was national but between the parties past the threshold, the index would be 9.41.

The reason that applying the threshold regionally would lead to a smaller increase in proportionality is that Poland has a geographically consistent party system, with none of the parties below the threshold receiving especially strong support in a geographic area.

A good country to compare Poland to is Turkey. The nations have similar electoral systems, but very different political environments. Turkey uses the D'Hondt method applied in small districts (average magnitude of 6.5) with a 10% national threshold. This has led to some very disproportional elections. For example, at the 2002 election, 45.3% of the vote was cast for parties that fell below the threshold. This led to Recep Tayyip Erdogan's Justice and Development Party winning two-thirds of the seats off 34% of the vote.

Now, if Turkey had gone with the approach of applying the threshold at a regional level, the results would be substantially different. The below graph shows the results, using the three methods described above.

As you can see, Turkey is quite different from Poland, mostly due to the large number of parties that fell below the threshold at this election. National allocation would, in theory, go some way to relieving this disproportionality. However, it removes Turkey's only alternative for candidates to be elected outside their parties; running as independents. Independents in Turkey are exempt from the national threshold, and are counted as single-candidate parties in district allocations. The Kurdish ethnic minority, following their failure to receive 10% nationwide, runs independent candidates in order to get some representation in the parliament.

The lack of independents means that a national allocation would actually make this election more disproportional. National allocation would score 27.47 on the Gallagher index, which is worse than the 27.41 the actual result got*. 27.41 is massive on the Gallagher index. Even in the UK, no result has not even been close to that.

Regional allocation would be a more dramatic change. As you can see, it would allow the entry into the parliament of other parties, although they would be very weak compared to their vote. The True Path Party, a centre-right secular party, would win 6.73% of the seats for 9.54% of the vote, and the centrist Young Party would win 2.36% of the seats and 7.25% of the seats.

The only party below 10% to receive more seats than their vote share is the pro-Kurdish Democratic People's Party, which would get 8.18% of the seats for 6.22% of the vote. This is because the Kurds are an ethnic group that mostly live in a small number of provinces in the nation's east. As a result, they can win these seats, whereas the other small nationwide parties have low vote shares everywhere, and cannot win many seats. This pattern has been repeated in recent elections, too; at this year's most recent election, the Kurdish-based People's Democratic Party won 59 seats with 10.76% of the vote, while the national-based Nationalist Movement Party won 40 seats with 11.9%.

Both results, however, show something important. Even under proportional representation systems, vote distribution is important. The lower the magnitude, the more important it is. Parties with uneven vote distribution will win more seats, but this assumes they get above the threshold. For parties like this (for example, the pro-Kurdish parties in Turkey), applying the threshold on the regional level would have the most positive effect. However, for parties with geographically consistent support, like Poland's Modern party, national allocation would have the most substantial impact.

*Note: Data is not consistent on the result of the election. However, this is likely to be close enough, although do not use this data for anything where accuracy is very important.