Nationwide PR in a big country

Ukraine and the Russian Federation have represented, at various times, the only two examples I know of using a single-nationwide district with a magnitude greater than the 150 used in the Netherlands* and Slovakia. (Israel’s single district has M=120, Namibia’s M=72.) [But see JD’s comment for an intermediate example.]

As it happens, both Ukraine and Russia have used the same magnitude, 450, with closed lists, when they have had the single-national district. For Ukraine, such a system was used in 2006 and 2007; for Russia, 2007 and 2011. By contrast, in 1998, 2002, and 2012, Ukraine used a mixed-member majoritarian system (225 M=1 districts, and a nationwide non-compensatory M=225 district), as did the Russian Federation in post-Soviet elections before 2007.

Nationwide closed lists could have the effect of biasing representation towards the capital and other major cities, given the (potential) control of the lists by the central party leadership, and the absence of institutional imperative to offer regional or personalized representation. On the other hand, they could encourage parties to present candidates from even those regions where they are not strong, because a vote anywhere counts towards the party’s overall seat total, and because even in closed lists the presence of candidates from a region might signal to voters in the region that the party is responsive to their needs. In the only study I know of in the political science literature to address such questions, Latner and McGann find some bias towards the most important cities, but also an over-representation of peripheral regions in Israel and the Netherlands.

What about Ukraine? The pattern could be different in a much larger country, with clear regional divides in its politics. A blog post by Erik Herron, Univ. of Kansas, and one of my “Party Personnel” collaborators, offers interesting data on candidate and MP residency in the 2007 election.

Key point regarding 2007 winners:

Kyiv residency is dominant, accounting for more than half of all elected deputies. The Party of Regions is better represented through the reported residency of its elected deputies in some eastern areas (e.g., Donetsk) and the opposition is better represented in western areas (e.g., BYuT in Galicia). But, parties can also claim elected deputies who report residency in “enemy” territory.

Meanwhile, Russian Federation president Vladimir Putin today signed into law a return of his country’s electoral system to the mixed-member system. While the article is not explicit about the relation of the two tiers, I assume it will again be MMM (non-compensatory). Given the decline in the standing of the ruling United Russia, it makes sense that Putin would prefer a move towards a system that is both disproportional and favorable to “independents” who have local bases of support that exceed the popularity of the ruling party’s label. In this respect, it would be identical to the change in Ukraine prior to the 2012 election. That change worked strongly in favor of the Putinist forces of that country, buying them time to acquire the finest in home furnishings.

Now that Russia is moving back to MMM, and Ukraine is moving on from the Yanukovych/Party of Regions era, maybe Ukraine will go back to the pure PR system. If they ask me, I certainly would not recommend the single national district, however. Either districted PR, without too much variation in magnitude, or MMP would be my advice.

___________
* In a very technical sense, the Netherlands has districts for nomination purposes. But for all practical purposes, it is a single district. It also allows preference voting for candidates on the list (though list ranks are more important), as does Slovakia, and as Israel does not. Russian and Ukrainian lists have always been closed, as are Namibia’s, to the best of my knowledge.

14 thoughts on “Nationwide PR in a big country

  1. Serbia has a nationwide district for its 250-member National Assembly.

    Doesn’t Russia have regional lists, with only overall seat allocation calculated on the national level?

  2. As I wrote that line, I did so with trepidation, because I knew there had to be cases I was not aware of. Thanks, JD! I had no idea about Serbia. At one time it was MMM, I think, but that may have been some time ago. It is very interesting that several Balkan countries adopted MMM and then moved to pure PR systems, though in other cases either districted or two-tier (or both). I believe this can be said of Albania (which was MMP at some point as well), Croatia, and Macedonia. Bulgaria, too, though there was only one MMM election (1990).

    I should also have noted Kosova, which uses M=100, one district, if I recall correctly (not counting the special minority seats).

    On the Russian lists, now that you mention it, that may be so. I think even the MMM system used regional lists somehow, although I recall that it was more like the Netherlands case. That is, candidate lists are not the same everywhere, but it still really is one district. However, for exactly the question raised in the post (and in Latner and McGann), it may be that a case that has any regional provisions on its list should not be conflated with cases of a single district (and thus a single national list for each party).

    • Bulgaria also had a second stint with MMM, in 2009 with 31 out of 240 seats elected by SSP, although, as you’ve asked before, such a small nominal tier probably doesn’t justify the first ‘M’…

      • Right. I manage to forget the partial return of nominal “SSP” election in Bulgaria. So is that still the system on the books, or has it been changed again?

      • It was changed right back to the previous system, which might be seen as a mistake, considering the deadlock that the 2013 election has caused.

      • Had the old system been retained in 2013, the results would have been GERB 100, BSP 89, MDF 36 and Attack 20, giving the BSP-MDF coalition that resulted a majority instead of the exact half of the assembly which they received in reality.

  3. Moldova is another case of single national district (though certainly not a big country). M=101. It is possible that it is also another case that was once MMM, though I am not sure. It has been a PR system since at least 1998.

  4. I am surprised that small island countries in the Caribbean do not use PR for the whole country, but most were colonized by the English and thus use FPTP/SMP. The largest size multi-member district should be no larger than 25 members, and the natural threshold would be 4%.

    • Yes, the formerly British Caribbean countries use M=1 plurality because, well, they always have. Except when they have used M>1 plurality (as a few have in the past).

      They also all have undersized parliaments, relative to population, which only exacerbates the disproportionality. And, other than Trinidad and Tobago (and maybe Belize), they seem to have fairly uniform swings, which suggests not much regionalism.

    • Well, it’s actually less than 4% (depending on the system). I don’t remember where but I came by a formula saying the effective threshold for D’hondt is 3/4 of a hagen-bischoff quota (votes/(seats+1)). In my own models, I’ve come by cases where is actually drops to as low as 2/3 of such a quota. I don’t know, but suspect that it has to do with the amount of fragmentation. In any case, in M=25, it’s likely to be more like 3%.

      • Searching for a suitable Webster-quota (I only use D’Hondt divisor tables to explain the system, not for quick real life calculations), I always start with votes/(M+(P/2)), with P being the number of parties that are big enough to gain a seat – this is a circular reasoning, I know, but it’s rather useful in practice. In a 2-party contest, this formula collapses to votes/(M+1) = the Droop quota.

  5. Just to be extra pedantic ( I hope you don’t mind), Droop is actually votes/(M+1)+1. v/(M+1) is hagen-bischoff.

    • To compound the pedantry, it is “Hagenbach-Bischoff”, but I always understood the H-B quota to be the same as Droop. And I just checked the Gallagher and Mitchell glossary, and it agrees. However, Wikipedia agrees with JD. Sorry, JD, but I’m going with G&M, who actually say, “The Hagenbach-Bischoff quota equals the Droop quota; there is no disagreement about this”!

      Droop (and so apparently H-B) is just a direct generalization of majority: a fractional share of [1/(M+1)]+1 is 50%+1 when M=1. Note that 1/(M+1) would allow two candidates to obtain the quota at M=1.

      The G&M glossary notes that there is disagreement about what the term, Hagenbach-Bischoff method, as opposed to quota, refers to, but suggests in practice it refers to different ways of implementing the D’Hondt divisors.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s