Summary of new Italian electoral system

If you have been unclear on what the new Italian electoral system–to be used the first time this March–really is, there is a good summary.

Broadly, it is mixed-member majoritarian (MMM, and definitely not MMP, contrary to a few claims I have seen). But with only 3/8 of the seats elected from single-seat districts, it stretches the definition at least a little bit. Anyway, the components (nominal-district and list-PR) are allocated in parallel.

There are some complicated provisions regarding the relations of votes for district candidate and lists, having to do with parties running in alliances, but there is no way to split across alliances. There is no partial compensation mechanism as there was in the MMM system (which had a balance tilted more in favor of the nominal tier) that Italy used between 1994 and 2001.

39 thoughts on “Summary of new Italian electoral system

    • It will complicate things. M5S is in first right now, and while they don’t seem to have much of a platform as to how they will govern, at least under the old system a government would be formed. Now you’ll have a protest part that despises both the center-left and the Berlusconistas as the largest part and potentially needing to form a coalition to govern.

      • The issue with the old system was that the majority bonus handed to the largest party in the Chamber of Deputies couldn’t be reflected in the Senate (given that Senate seats can only be apportioned based on region-wide results)-the old system did limit the number of parties that could lead a government (basically to 1), but it’s not as if it didn’t require broad and unwieldy coalitions. Indeed, the incumbent government, as I understand it, started as a grand-ish coalition between the Democrats, the Monti coalition and Berlusconi’s party.

        While it’s true that M5S will be the largest *party*, it will probably be in second or third place after the Democratic Party joins up with minor left and centre groups and Forza Italia and Lega Nord run jointly.

  1. Dear Matthew Shugart,

    we disagree with your assessment and classification. In the new Italian electoral system seat allocation is not linked (the two tiers run almost in parallel) but the vote linkage is absolutely present. Given its ballot structure (no vote-splitting and no individual vote), there is an automatic transfer of votes from the nominal tier to the list tier (and vice versa). Such a system of full/absolute vote linkage makes the new electoral system work as a MMP, prioritizing the list-PR tier, and accordingly the principle behind majoritarian systems (that is, giving an advantage to a large party) entirely disappears.

    All the best,
    Gianfranco Pasquino (University of Bologna), Marta Regalia (Luissi Guido Carli-Roma), Marco Valbruzzi (University of Bologna)

    • A well-informed friend comments “The Italian professors are wrong and Shugart is right. Yes there is a transfer of votes – including those from successful candidates. MMP deducts those votes either directly or indirectly (by deducting seats). This being a one-ballot system doesn’t make it MMP.

      Granted, in practice in may be more proportional than AMS (Scottish MMP) given the high number of PR seats, but that’s not really relevant to categorizing the system. How proportional a system is depends on circumstance, whereas the system itself is categorized by how it behaves across all circumstances.”

      I defer to those who can read Italian.

      • He adds “I should add that this is a massive improvement over the super-majoritarian system previously used in Italy though. 29% was good enough for a majority last time around. They’ve harmonized things more with their senate too which should reduce their propensity for deadlock.”

  2. how could anyone define a system that allocates about 66 per cent of the seats according to a PR formula less than Mixed Member Proportional is a mystery that we will allow the allegedly “well-informed”, though, apparently not knowledgeable, friend to solve. The previous electoral law was PR plus a majority bonus. If the coalition obtained 40 per cent of the votes the majority bonus was 14 per cent additional seats. In parliamentary democracy rarely do the deadlocks depend from the electoral law. Mostly they are the product of the inability/unwillingness of party leaders to reach agreements.

    • The system is not more proportional. But because 100% of the seats are not aligned proportionally, give or take, it is not MMP. MMP does not describe a system where some members are seated based on proportional voting, it is a system where every member sits in a caucus sized in proportion to its overall vote, but with some members selected locally.

      • Whoop! Meant to say that the new system is more proportional than the old system

  3. If Google Translate and I have correctly interpreted the Gazetta Ufficiale record of the law of 3 November 2017, the multi-member districts are to have between 3 and 8 seats for the Chamber of Deputies, between 2 and 8 seats for the Senate. [Presumably, they are coterminous with groups of between 2 and 4 single-member districts, in such a way as to achieve the overall 5-to-3 ratio between MMDs and SMD.]

    It is not obvious that the large fraction of PR seats will result in good PR, given the small district magnitudes (compared to Scottish MMP regions, say). A bug or a feature?

    • Proportional seats are FIRST distributed nationally and THEN allocated to local districts. So, no reductive effect of small districts.

  4. I have a feeling people are arguing from different perspectives. To the Anglophones, the lack of seat (or vote) deduction is decisive in making Rosatellum bis not proportional; while to the Italians (I think), given Italy’s experience with scorporo, disallowing vote splitting and having 66% proportional tier seats makes it proportional, or at least more proportional than say New Zealand, whose very proportional elections are more results of people acting irrationally (i.e. not setting up decoy lists) rather than institutional design.

    • The Italians can call it whatever they want to, but the new system does not fit any definition of “mixed-member proportional” as that term is conventionally used in English language political science. If 3/8 of the seats are awarded on a plurality basis, and then 5/8 of the seats are awarded proportionally not counting the plurality seats, that is pretty much a textbook example of a “mixed-member majoritarian” system.

      Parties in New Zealand don’t run decoy lists because the political culture is vastly different than Italy and such a move would likely result in the party losing seats. I’d also say it’s far from “rational” to expect the average voter to understand an electoral system well enough to vote in order to game that system. It’s absurd to suggest that because of a minute theoretical possibility of using a decoy list that the Italian system, which has a built-in non-proportional component, is somehow more proportional than the New Zealand system.

      If you’re going to assume that the New Zealand system acts in a highly improbable manner, you’d also have to presume that the Italian system is going to deliver an unorthodox result, like M5S winning all of the constituencies on 20% of the national vote. If that were the case, M5S would win 309 seats in the Chamber of Deputies (49% of seats) and 155 seats in the Senate (49% as well) without vote splitting or decoy lists. That is hardly a “proportional” system by any definition. And that scenario is hardly an outlandish prediction for the Italian system.

      • And if New Zealand is the “outlier,” what explains Germany, Scotland, Wales, and London where, with real differences in sitting list members, none of them have decoy lists or other ways of gaming the system.

  5. It would seem to me that while Italy’s new electoral system is clearly mixed-member and mostly proportional, it is not mixed-member proportional (MMP), given the absence of a top-up mechanism to partially or completely compensate for the distortions introduced by the FPTP component.

    At the same time, I can see that it may appear counter-intuitive to describe Italy’s new and predominantly proportional electoral system as mixed-member majoritarian; nevertheless, in the same manner as typical MMM systems – namely those featuring a large FPTP component – with Rosatellum bis “the list-PR allocation is unlikely to prevent any party that can emerge from the nominal tier over-represented (perhaps substantially) from retaining over-representation. […] any party in a position to be over-represented in the nominal tier will also obtain a large (and approximately proportional) share of the list-tier seats. It thus will retain some degree of over-representation far and away beyond what it would have with any compensatory mixed-member system (MMP), even one with a relatively small PR tier and/or small magnitudes in that tier.”

    To illustrate the impact of Rosatellum bis compared to MMP, I re-ran last September’s Bundestag election in Germany under the new Italian system. Excluding the special cases of Valle d’Aosta and the expatriate seats, the remaining 617 Italian Chamber of Deputies seats have a PR-to-SMC ratio of 386:231; since Germany has 299 single-member constituencies, there would have to be 500 party list seats to attain the same ratio, for a total of 799 seats. (This figure may seem large compared to the minimum Bundestag size of 598 seats, or even its current size of 709 after last year’s election, but simulations of 1990-2009 Bundestag election results under the fully proportional system introduced in 2013 indicated that an 800-seat Bundestag was well within the realm of possibility.)

    Now, Rosatellum bis does not allow split-ticket voting, so the second vote (Zweitstimme) was used to allocate both single-member and party-list seats. This would have resulted in an even more lopsided CDU/CSU victory in the single-member constituencies, with the Union parties capturing 256 of 299 direct mandates (210 CDU and 46 CSU), while SPD would have been reduced to just 29 seats; meanwhile, both The Left and AfD would have won seven seats each. (Compared to the actual first vote outcome, CDU would have lost four seats to AfD, SPD would have lost 29 seats to CDU and one to The Left, which would have also captured the sole seat won otherwise by Alliance 90/The Greens.) Following the nationwide distribution of 500 list seats by the largest remainder method of PR, the composition of the Bundestag under Rosatellum bis would have been as follows (with numbers in parentheses showing the percentage of mandates and the difference with respect to the actual MMP outcome):

    CDU – 351 (43.9%; +151)
    SPD – 137 (17.1%; -16)
    CSU – 78 (9.8%; +32)
    AfD – 73 (9.1%; -21)
    F.D.P. – 57 (7.1%; -23)
    The Left – 56 (7.0%; -13)
    Alliance 90/The Greens – 47 (5.9%; -20)

    Incidentally, it would have made no difference for CDU and CSU to run separately or in coalition for the allocation of list seats, as the Union parties would have won the same number of mandates under either scenario. More importantly, with 429 of 799 seats (53.7%) Frau Doktor Merkel would have secured an absolute majority of 59 seats with just 32.9% of the vote, which would have allowed her to form a CDU/CSU government right after the election. While this is a purely notional outcome – there is no way of telling how voters would have cast ballots under a different electoral system – the difference with respect to the actual, inconclusive outcome of last September’s election is quite dramatic.

    Finally, it should be noted that while Rosatellum bis did under-represent minor parties in the notional scenario, they would have still fared better than under nominally proportional systems such as that of Spain. In fact, the levels of representation attained by the smaller parties would have been roughly comparable to those achievable under bonus-adjusted PR systems such as Italy’s now-repealed Porcellum.

  6. Interesting thread; thank you to all who have commented. To be clear, I am using the definitions of MMM and MMP that Wattenberg and I introduced in our 2001 book. What places a system in one or the other category (and, yes, intermediate cases are possible) is not whether there are two votes (allowing ticket-splitting) or what the percentage of seats is in each component is. It is all about the process by which the list-PR seats are allocated: Do nominal (plurality or majority) wins count towards the seats won by PR, or are the seats won in the two components simply added to each other?

    It is my understanding that the new Italian system does the latter, which would be parallel allocation. Therefore not MMP. If my understanding is wrong, then I would need to revise my conclusion.

    I already addressed in the initial post the question of whether “MMM” was a good name for a system where the percentage of PR seat is so high. Perhaps not. But I believe for reasons of terminological clarity, we should reserve the term “MMP” for compensatory mixed-member systems. And, yes, just as “PR” systems can have low magnitude districts and thereby be relatively non-proportional in effect (as in the Spanish case Manuel references), so can “MMP” have small compensation regions (e.g. Scotland). By the same token, perhaps it is not too much of a stretch to allow “MMM” to have a high percentage of PR seats, as long as the mechanism is parallel and thus a large party can win an above-proportional share of seats via winning many plurality races, for which the other parties are not compensated.

    • District magnitude is not always the main contributor to disproportionality in MMP elections to the Scottish parliament. In 2016 the SNP won 64 seats, while its 41.7% share of the regional vote would have justified a total of 54 seats. The SNP won 60 constituency seats, of which 6 seats came from overhangs in 5 of the 8 regions. Going to 1 region of magnitude 129 have seen just 4 SNP regional seats redistributed among the Greens, Lib Dems and UKIP.

      In 2011 the situation was reversed. The SNP gained 69 seats, would have been entitled to 57 by the popular vote and had 5 overhang seats. In this case the region magnitude (16) was more important, eliminating 5 parties that each had more than 1/129 of the popular vote and giving a “D’Hondt bonus” to larger parties.

      Overhang compensation is especially important in MMP if one party has >40% of the vote and 3 or 4 other parties are scrabbling for the remaining <60%, as in Scotland in 2016.

      • The described phenomenon in Scotland is one of “compensation magnitude”–the size of the regions (nominal seats plus compensation list seats) over which compensation is carried out. Of course, the formula used will matter, too, and all the more so when compensation magnitude is low.

        I looked into some of these issues in more detail, as reported in one of the on-line appendices to Votes from Seats, linked at the book’s webpage (click on “Resources”).

  7. The electoral system introduced by the latest Italian electoral law (n. 165/2017) is clearly a mixed-member system. However, the world of mixed-member system is vast and eterogenous. The so-called “Rosato law” refers to the subtype of mixed-member majoritarian (MMM). In fact, in those latter cases the two tiers (proportional and majoritarian) are not linked (see Shugart and Wattenberg 2001). Therefore, the majoritarian boost received by a list/party in the nominal tier is not wasted/weakend by the proportional allocation of seats. This latter aspect is confirmed by the fact that votes conferred to candidates in the plurality part only are proportionally allocated to the lists of the coalition, so that the lists are not affected in their final score in the PR part. And, therefore, it is confirming the “division” between the two tiers and the MMM nature of 2017 Italian electoral system.

    • If I am not mistaken, this means that the Shugart and Wattenberg definitions from the 2001 book do not regard the electoral system of Baden-Württemberg as MMP, since it does not allow for split-ticket voting. I would find that most peculiar.

      • As MSS states above: “What places a system in one or the other category (and, yes, intermediate cases are possible) is not whether there are two votes (allowing ticket-splitting) or what the percentage of seats is in each component is. It is all about the process by which the list-PR seats are allocated: Do nominal (plurality or majority) wins count towards the seats won by PR, or are the seats won in the two components’ simply added to each other?”

      • Baden-Wurttemberg is the one that was called “Dual Member Proportional” in a recent Canadian poll and where a runner-up is selected from each district to balance the FPTP winners is it not?

        If it is not MMP, what is it then?

  8. Right you are. I must’ve misread something earlier! By those criteria the B-W system would rightly not be classed as MMM.

    • What is the B-W system called? Here’s a radical idea: let’s ask them what they call it? Answer: the Personalized No-List Proportional System. Germans call MMP the Personalized Proportional System, or sometimes the Improved Personalized Proportional System. So the distinctive element of the B-W system is “no list” (ohne listen). Some call it a “best losers” system, or a “best runners-up” system, but not the Germans.

      • Mark,
        Dual Member Proportional differs a bit from B-W. (1) the losers in a region are ranked in order of the percentage of votes in their district contest while B-W ranks by number of votes, (2) there is exactly one top-up candidate drawn from each district while B-W may select 0, 1, 2, 3…from a given district

      • I would submit that this very thread offers ample evidence for why we should not classify and name systems based on what those in the jurisdiction call it. We should instead apply clear principles from electoral-system science.

        The B-W system is MMP. It is an example of MMP in which the list is comprised solely of candidates who run (but do not win seats in) the nominal tier and ranked by their relative performance there. It is still a list, even if it is not constructed separately by parties prior to the election. We actually addressed this question in Shugart and Wattenberg (2001). The Italian Senate’s former MMM system had something similar. Japan’s MMM has a twist on the concept as well.

        I don’t have a name for it, and I am not crazy about inventing new names for every combination of rules someone has invented. “Best losers” is used in the literature (and I think I have used it in published work), but I do not really like it because it puts too much emphasis on the single-seat district component (as does “Additional Member System” or the dreadful concept of “zombie“).

        It could fit under a category of “District-Ordered List“, although that names has its issues, too. (“District” is not synonymous with single-seat district.)

        Anyway, I just call systems with such list mechanisms MMP or MMM, using the usual classification scheme. Then we can go on into details of implementation in a given jurisdiction.

  9. How does the national (chamber) / regional (senate) allocation ‘rain down’ to the electoral districts (“collegi plurinominali” of max. 8 seats) ?

  10. Another system that challenges definition boundaries is Fair Majority Voting. As in FPTP, candidates compete for votes in a single-member district, one of those candidates will be elected and, as a rule, the elected candidate is the one preferred by a plurality of voters in the district. However, the exceptions to the rule arise from considering votes for each party’s candidates throughout a multi-district region.

    FMV could be described as best-losers MMP in the limit where the top-up fraction goes to 100% of the seats, with the one-per-district allocation rule of Dual Member Proportional. This is no longer a mixed system. Presumably FMV goes in the Proportional category, despite the possibility that an election could produce the same winners from the same SMDs as an FPTP system.

    • Isn’t the Romanian system form 2008-2012 FMV with possibility of being elected directly from your district if you won over 50%?

    • I’m not familiar with the details of your proposed system, but I assume that all seats would be allocated based on party totals in the multi-district region. In such a case, that would seem to resemble a list-PR system with a somewhat novel mechanism for allocating seats to candidates. The difference between your proposal and MMP/MMM/plurality is that under those systems, winning the most votes in a constituency *guarantees* your election to the legislature, wheras in your case it would not always be enough. It’s true that you could have the same results as FPTP, but that would seem unlikely under normal circumstances, and is anyway irrelevant to a generalised classification.

  11. I don’t claim Fair Majority Voting as my system — it was proposed by Michel Balinski:
    http://www.mathaware.org/mam/08/EliminateGerrymandering.pdf
    He achieves regional PR from SMDs by explicit re-weighting of votes for candidates of one or more parties in all districts, or of votes for candidates of all parties in one or more districts.

    I described a procedure where the re-weighting is less visible: sequential allocation of seats to candidates down a ranked list, subject to a one-seat-per-district constraint. I believe it gives the same result as Balinski’s, but have no proof.

    Yes, FMV is different from MMP/MMM in that local plurality is not a guarantee of a seat. And it could be described as list-PR with post facto list formation and constrained allocation of seats from the list. In terms of options for voters and incentives for candidates, FMV is much more like best-losers (“no list”) MMP.

  12. Pingback: MMM not MMP | Fruits and Votes

  13. Pingback: Italy, 2018 | Fruits and Votes

  14. Very belated comment here. The debate between whether this is MMP (mixed-member proportional) or MMM (parallel) seems to miss the point. This is what I call a split (or mixed) single-vote system. As Matthew notes, it is clearly not proportional due to the presence of majoritarian districts and the lack of compensation on the PR side. But the dominance of PR districts, as the Italian professors such as Gianfranco Pasquinto point out, tends to limit the strategic behavior of voters in casting ballots due to the presence of plurality districts. As a result, my sense is you get voters participating mostly as if in a PR system but with allocation according to a mixed system.

    The closest parallel I can think of is to an old South Korean system (before they split the votes for each part) but the crucial difference is that plurality seats composed the vast majority of seats so they dominated over the PR seats.

    Another parallel, though true MMP, is Lesotho. Like in Italy but more so, parties figured out how to completely subvert the corrective aspects of the MMP system (a real MMP system in Lesotho unlike the complex scorporo arrangements in Italy but same idea). As a result, Lesotho switched to MMP but with one vote system in order to avoid this manipulation in future. However, there are not enough PR seats to assure full correction if one party does really well. See my discussion at:
    http://www.electionpassport.com/electoral-systems/lesotho/.

  15. Question if anyone knows and is still reading: My understanding is that seats in the Chamber are distributed on a national basis first and then devolved down to specific constituencies. However, there also appear to be some special provisions. I know that Valle d’Aosta just has its single seat elected by plurality. Some more questions:

    (1) Are there special provisions for either Trentino Alto-Adige or Molise for allocation of the Chamber seats as in the Senate, where my understanding is that they use the old scorporo system (please correct if wrong)?

    (2) How are seats allocated down to constituencies?

    • Concerning (2), I discuss that very subject in my comment here.

      As for (1), Molise never had scorporo for Senate elections: during the 1993-2005 “Mattarellum” era the region had no PR compensation seats; instead, its two senators were chosen in single-member colleges. On the other hand, under “Porcellum” Molise had no majority bonus, and its Senate seats were allocated by the largest remainder method of PR. Now, under the current “Rosatellum bis” system, both Molise and Trentino-Alto Adige/Südtirol have the same Chamber and Senate electoral systems as the rest of Italy (minus Valle d’Aosta), albeit with their own peculiarities in each case.

      Regarding Trentino-Alto Adige, the region’s six single-member colleges are identical for both houses of Parliament. This is so because the 232 Chamber of Deputies single-member colleges are based on the 232 Senate single-member seats that were in place under “Mattarellum;” in turn, the number of Senate seats in Trentino-Alto Adige is fixed by a 1991 law, part of a package of measures favorable to the population of Alto Adige, which led to the settlement of the South Tyrol issue with Austria in 1992. Because of this, Trentino-Alto Adige has only five PR list seats in the Chamber and one in the Senate; scorporo is no longer in place for Senate elections in the region.

      Meanwhile, Molise now has two SMC seats and one PR list seat in the Chamber, while in the Senate the entire region is one single-member college, with the remaining seat chosen by PR.

      These arrangements are peculiar because in most regions the proportion of PR list seats to SMC seats is approximately (or exactly) five-to-three for both chambers; moreover, in most regions the proportion of SMC Chamber seats to SMC Senate is approximately (or exactly) two-to-one.

  16. Pingback: Italy 2018: Interim government, early elections | Fruits and Votes

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 )

Google+ photo

You are commenting using your Google+ 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 )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.