Germany electoral system change

Note: This is a revised version of an older post (originally 15 Jan. this year). The electoral system changes (with some modifications) have now been passed. Rather than make a new planting I am just re-upping this older one, as the comment thread has continued to grow with useful information. I particularly recommend a new comment by Thomas D for good detail. In addition, see the post by Verfassungblog for background and Twitter thread by Heinz Brandenburg which has both background and excellent detail on the final version. (The Verfassungblog post refers to an earlier version of the draft; in the finally passed version the 3-districts alternative threshold is indeed being abolished.) The law is sure to be challenged before the Constitutional Court, and some or all if it may fail the constitutional test. As Thomas notes, the law could be a mortal threat to the CSU as we know it, as well as to the Linke. And it certainly means no more “personal constituency mandate” and thus is not a mixed-member system.


[original post, edited] The parties in Germany’s current governing coalition have agreed a package of reforms to the electoral system. The proposal keeps the size of the Bundestag at 598 and eliminates Überhangmandate and “balance” seats.

For details, see first comment. This is a big deal. I’d understand this proposal as Germany abolishing MMP (and Verfassungblog and Brandenburg–linked above–agree). It is not a mixed-member system if it’s possible for a candidate with the most votes in a nominal district not to win the seat.

77 thoughts on “Germany electoral system change

  1. From the piece (my translation):

    “In the past twenty years, the Bundestag has become larger and larger. This was due to overhang and compensatory mandates. Overhang mandates occur when a party wins more direct mandates in a federal state than it is entitled to according to the results of the second votes. To ensure that the other parties are not disadvantaged as a result, there have also been compensatory mandates since the 2013 Bundestag elections. The draft bill of the traffic light parliamentary groups now provides for a radical cut. In future, there will be no more overhang and compensatory mandates. As a result, the Bundestag will always have its standard size of 598 MPs. However, this may mean that a candidate who gets the most votes in a constituency will still not be allowed to enter the Bundestag. Example: If the CSU in Bavaria were to win all 46 direct mandates, but according to its second vote result would only be entitled to 41 seats in the Bundestag, the five constituency winners with the lowest first vote result would not get a Bundestag mandate. The number of constituencies in Germany – currently 299 – is not to be changed. This will avoid a complicated redrawing of constituencies. The so-called basic mandate clause is also to be retained. It provides that a party that fails to clear the five-percent hurdle but wins at least three direct mandates may still enter the Bundestag according to its second-vote results. The Left Party benefited from this rule in the last Bundestag election. Even if no one in the traffic light coalition says it openly: by maintaining this exemption, the traffic light coalition wants to secure the Left’s consent to the new electoral law. (…) In order to make the significance of the respective votes clearer, the coalition now also intends to change names. In future, the first vote will be called “constituency vote” – and the second vote “main vote”.”


  2. Interesting <a href=”|>article on a not dissimilar topic from the Malay Mail.

    Malaysia desperately needs redistricting reform but that’s apparently seen as a separate issue. The Pakatan Harapan minority government headed by Ibrahim Anwar is committed to redistricting reform but hasn’t announced any details yet.


  3. RCV is an ambiguous and infelicitous term. It theoretically includes non-STV systems like Condorcet, Borda and the Condorcet-Borda hybrid that Ned Foley supports, but is actually used in the US almost exclusively to mean STV where M=1. Tom is correct in a previous thread that we need greater precision in these discussions, but I’ve got no proposal for terms for:

    STV where M=1
    Non-STV but preferential systems
    Tom’s own contribution to ambiguous terminology of arguing that minimum preferential systems and optional preferential systems are the same.


    • In Bavaria, Free Voters – which didn’t clear the nationwide 5% threshold – reached second place with 16.7% in Rottal-Inn, one of the 11 Wahlkreisen where CSU would have been forced to give up a single-member mandate in 2021 had the proposed system been in place at the time. Of course, in that case an alternative would have been to award the seat to the party in third place – which in that case would have been AfD with 12.7% – CSU won the seat with 35.1% – but I doubt that would have gone down well.


      • Wahlrecht
        Ampel will Bundestag deutlich verkleinern
        15. Januar 2023, 16:48 UhrLesezeit: 2 min

        Die Koalition einigt sich auf einen Gesetzentwurf: Statt bisher 736 Abgeordneter soll es künftig nur noch 598 geben.

        Von Robert Roßmann, Berlin

        Die Ampelkoalition hat sich auf eine Reform des Wahlrechts verständigt, mit der der Bundestag auf 598 Abgeordnete verkleinert werden soll. Derzeit gibt es 736 Abgeordnete. Die Fraktionen von SPD, Grünen und FDP wollen an diesem Montag einen entsprechenden Gesetzentwurf vorstellen. Er liegt der Süddeutschen Zeitung bereits vor.

        In den vergangenen zwanzig Jahren ist der Bundestag immer größer geworden. Das lag an den Überhang- und Ausgleichsmandaten. Überhangmandate entstehen, wenn eine Partei in einem Bundesland mehr Direktmandate gewinnt als ihr Sitze nach dem Zweitstimmenergebnis zustehen. Damit die anderen Parteien dadurch nicht benachteiligt werden, gibt es seit der Bundestagswahl 2013 auch Ausgleichsmandate.

        Der Gesetzentwurf der Ampelfraktionen sieht nun einen radikalen Schnitt vor. Es soll künftig keine Überhang- und Ausgleichsmandate mehr geben. Dadurch wird der Bundestag immer seine Normgröße von 598 Abgeordneten haben. Dadurch kann es aber passieren, dass ein Kandidat, der die meisten Stimmen in einem Wahlkreis erzielt, trotzdem nicht in den Bundestag einziehen darf.

        Ein Beispiel: Wenn die CSU in Bayern alle 46 Direktmandate gewinnen würde, nach ihrem Zweitstimmenergebnis aber lediglich Anspruch auf 41 Sitze im Bundestag hätte, würden die fünf Wahlkreissieger mit dem niedrigsten Erststimmenergebnis kein Bundestagsmandat bekommen.

        Die Zahl der Bundestagswahlkreise in Deutschland – es sind derzeit 299 – soll dagegen nicht verändert werden. Dadurch wird ein komplizierter Neuzuschnitt der Wahlkreise vermieden. Auch die sogenannte Grundmandatsklausel soll erhalten bleiben. Sie sieht vor, dass eine Partei, die an der Fünf-Prozent-Hürde scheitert, aber mindestens drei Direktmandate gewinnt, trotzdem entsprechend ihrem Zweitstimmenergebnis in den Bundestag einziehen darf. Von dieser Regelung hat bei der vergangenen Bundestagswahl die Linke profitiert. Auch wenn es in der Ampelkoalition niemand offen ausspricht: Durch den Erhalt dieser Ausnahmeregelung will sich die Ampelkoalition die Zustimmung der Linken zu dem neuen Wahlrecht sichern.

        Die Vorsitzenden der drei Ampelfraktionen bemühen sich aber auch um eine Verständigung mit CDU und CSU. Am Sonntag schickten sie deshalb ihren Gesetzentwurf an Unionsfraktionschef Friedrich Merz und boten Gespräche an. Vor allem die CSU hat sich bisher vehement gegen eine Regelung gewehrt, wie sie jetzt von der Ampel angestrebt wird. CSU-Landesgruppenchef Alexander Dobrindt nannte sie sogar verfassungswidrig.

        Allerdings gab es eine vergleichbare Regelung eine Zeit lang auch im bayerischen Landtagswahlrecht. Außerdem werden inzwischen viele Wahlkreise mit niedrigen Ergebnissen gewonnen. Bei der vergangenen Bundestagswahl gab es mehr als 80 Wahlkreise, in denen der Sieger nicht einmal auf 30 Prozent kam. Ein Wahlkreis wurde sogar mit 18,6 Prozent gewonnen.

        Um die Bedeutung der jeweiligen Stimmen klarer zu machen, beabsichtigt die Koalition jetzt auch Namensänderungen. Die Erststimme soll künftig “Wahlkreisstimme” heißen – und die Zweitstimme “Hauptstimme”.


  4. right to vote
    Traffic light wants to significantly reduce the Bundestag
    Jan 15, 2023 at 4:48 p.m Reading time: 2 mins

    Conversion work in the German Bundestag

    The coalition agrees on a draft law: Instead of the previous 736 MPs, there should only be 598 in the future.

    By Robert Rossmann , Berlin

    The traffic light coalition has agreed on a reform of the electoral law, with which the Bundestag is to be reduced to 598 MPs. There are currently 736 MPs. The factions of the SPD, Greens and FDP want to present a corresponding draft law this Monday. It is already available to the Süddeutsche Zeitung .

    In the past twenty years, the Bundestag has grown in size. This was due to the overhang and compensation mandates. Overhang mandates arise when a party wins more direct mandates in a federal state than it is entitled to according to the result of the second vote. To ensure that the other parties are not disadvantaged as a result, there have also been compensatory mandates since the 2013 federal election .

    The draft law of the traffic light factions now provides for a radical cut. There should be no more overhang and compensation mandates in the future. As a result, the Bundestag will always have its standard size of 598 MPs. As a result, however, it can happen that a candidate who achieves the most votes in a constituency is still not allowed to enter the Bundestag.

    An example: If the CSU in Bavaria won all 46 direct mandates, but was only entitled to 41 seats in the Bundestag based on its second vote result, the five constituency winners with the lowest first vote result would not get a Bundestag mandate.

    The number of Bundestag constituencies in Germany – there are currently 299 – should not be changed. This avoids a complicated reshaping of the constituencies. The so-called basic mandate clause should also be retained. It stipulates that a party that fails at the five percent hurdle but wins at least three direct mandates can still enter the Bundestag according to its second vote result. The left benefited from this regulation in the last federal election. Even if no one in the traffic light coalition says it openly: by maintaining this exemption, the traffic light coalition wants to secure the approval of the left for the new electoral law .

    The chairmen of the three traffic light groups are also trying to reach an understanding with the CDU and CSU. On Sunday, they therefore sent their draft law to Union faction leader Friedrich Merz and offered talks. The CSU in particular has so far vehemently opposed a regulation such as that now being sought by the traffic light. CSU regional group chief Alexander Dobrindt even called it unconstitutional.

    However, there was a comparable regulation for a while in the Bavarian state electoral law. In addition, many constituencies are now being won with low results. In the last federal election, there were more than 80 constituencies in which the winner did not even get 30 percent. One constituency was even won with 18.6 percent.

    In order to make the meaning of the respective votes clearer, the coalition now also intends to change names. In the future, the first vote will be called the “constituency vote” – and the second vote will be called the “main vote”.

    Liked by 1 person

    • The cited example doesn’t fully convey the potential impact in Bavaria of the proposed electoral system, under which CSU would have been assigned just 34 of 598 seats in 2021, even though it won in 45 of 46 single-member districts in Bavaria. Therefore, it would have had to give up 11 single-member mandates where it won with the lowest percentages, among them three in Munich (out of a total five seats), both Nuremberg seats and one of two Augsburg seats. In other words, Bavaria’s major cities would suffer a disproportionate impact under the proposed system, and for that as well as other reasons I suspect Karlsruhe won’t be amused.


  5. Thomas Däubler on Twitter:

    A variant that had been considered but didn’t make it into the proposal: everyone can indicate a second preference in the SMD tier, and if your first one turns out to be for one of the worst winners, the vote would transfer.


  6. My suspicion that a big reason a lot of polisci nerds prefer MMP is that they get to see frequent debates and changes to any such system in the real world is strengthening by the day.

    Liked by 1 person

  7. Pingback: The Problem With MMP, and How Germany Is Fixing It – Koji's Posts

    • They do not want a substantial reduction in the number of SMD because they want to avoid a complete redrawing of the map. Vested interests of those incumbents elected in the SMD tier.


      • Gotcha. Good to see the “turkeys don’t want an early Thanksgiving” rule applies outside of Canada.


        • Outside of North America that phrase doesn’t really have much meaning, but I take your point.


        • I’ve heard British commentators use “Turkeys don’t vote for Christmas” as a saying in this context. Perhaps at least somewhat more broadly applicable than the Thanksgiving reference.


  8. So does this mean if a party wins the most votes in one or two districts – along with few to no votes elsewhere – they will not enter parliament, whereas previously they would have won an overhang seat or two?

    And will another effect of this be to increase the effort put into local campaigns by the larger parties, and within-party relationships? Manuel Alvarez-Rivera mentions that the CSU would have lost seats in the large Bavarian cities because of this. So now the CSU incumbents would be more inclined to campaign for local voters (over and above the amount needed to come first), so that they can rise up above CSU candidates elsewhere in the state. This might mean making concessions to local voters on issues that are very important to voters in other parts of the state/country. On the other hand, this is to the benefit of the CSU as a whole because more constituency votes in Munich or Nuremberg might lead to more main votes overall – and more seats. So is this kind of change associated with an increase in factionalism?


    • A party that wins 1 or 2 single-member district seats but gets under 5% of the list vote keeps them and gets no further seats. That is the rule now and remains intact under the latest proposals.

      A party that wins at least 3 constituency seats is entitled to compensatory seats on the list no matter how they do on the list vote, but not to ‘overhang’ ones. This is also not proposed to change.

      So no, there will be no lesser incentive than before for hyper-local campaigns. If you’re wondering about the SSW, they currently have 1 list seat because they’re exempt from the threshold. They did not win a FPTP seat in 2021.


      • I take it that you mean the party that wins the 2 districts get to keep them only if they also got enough PR votes to qualify for at least 2 seats out of 598 (but less than 5%)? Or would the 2 district winners be treated as if they were independents?
        By the way, running district candidates as independents would be what I would do if I were the CSU should their constitutional challenge fail. Granted, they could already have done that under current rules and they would still have gained advantage, but the effect of distortion will be greater if the Bundestag size is fixed.


        • Fair question. At the moment, the two district winners are treated as if they were independents (so it’s worth noting that in 2002 the PDS, forerunner of Die Linke, won just 2 seats despite gaining enough second votes to be entitled to around 25 seats because they fell below the threshold). I’m not sure if the proposed change to the electoral law will change the formula to make sure that the FPTP winners are not from some micro-party that didn’t run candidates anywhere else (or got 0.001% if they did).

          That would be an interesting tactic by the CSU if they were to try it. It certainly feels like it’d be open to a legal challenge.


        • Perhaps I am wrong (and my German is not good enough to confirm it 100%), but if an actual independent wins a direct mandate, aren’t the second votes cast on the winner’s ballots thrown out to prevent exactly this sort of thing?


        • Indeed, see Bundeswahlgesetz [federal electoral act], § 6, (2), second sentence: “Nicht berücksichtigt werden dabei die Zweitstimmen derjenigen Wähler, die ihre Erststimme für einen im Wahlkreis erfolgreichen Bewerber abgegeben haben, der gemäß § 20 Absatz 3 [=independent candidate] oder von einer Partei vorgeschlagen ist, die nach Absatz 3 bei der Sitzverteilung nicht berücksichtigt wird [=party not clearing the threshold] oder für die in dem betreffenden Land keine Landesliste zugelassen ist.”

          Liked by 1 person

  9. Two quick replies:
    1. I don’t think the change is that much of a big deal in one sense. Since the introduction of “Ausgleichsmandate”/compensatory mandates in the 2013 elections, that compensated parties for the Überhangmandate of other parties, the system had become de facto PR as winning local districts above and beyond the proportional vote share didn’t result in more seats vis-a-vis most other parties.
    2. It’s far from clear that this proposal, even if it becomes law, would actually be used in the next election. The three governing parties explicitly want to pass the law this April, so that the Constitutional Court could rule on the law. A challenge by the CDU/CSU is widely expected. The CDU/CSU also introduced their own proposal for a reform that would reduce the number of directly elected seats today. From the FAZ today (Translated with (free version):

    “In the dispute over reducing the size of the Bundestag, the head of the CDU/CSU parliamentary group is proposing to the traffic light coalition to reduce the number of constituencies from 299 to 270, according to media reports. This is reported by the media house Table.Media and the “Süddeutsche Zeitung”. The newspaper refers to a “flash briefing” of the parliamentary group leadership to its delegates, the media house to coalition circles.

    A reform of the electoral law has been disputed for years. The standard size is 598 mandates, by overhang and compensatory mandates the parliament had grown in recent years always further – last on the record size of 736 delegates. In a draft bill, the traffic light parliamentary groups have now proposed that in future there should be no more overhang and compensatory mandates in order to return to the standard size. This could mean that directly elected representatives in a constituency would not receive a seat in the Bundestag. The CDU/CSU rejects this.

    Amendment of the so-called basic mandate clause
    According to both media reports, the leadership of the CDU/CSU is proposing to reduce the number of constituencies to 270. This would also significantly reduce the number of overhang and compensatory mandates. It is also “conceivable” that “up to 15 overhang mandates could remain unbalanced. These arise when a party wins more direct mandates in a federal state than the number of seats it is entitled to in the Bundestag according to its second-vote results. To ensure that the other parties are not disadvantaged as a result, there have been compensatory mandates for them since the 2013 federal election.

    According to the Süddeutsche Zeitung, the CDU/CSU parliamentary group is also proposing a change to the so-called basic mandate clause. This stipulates that a party that fails to meet the five-percent clause can still send representatives to the Bundestag in accordance with its second-vote results if it has won at least three direct mandates. The Left Party benefited from this in the last Bundestag election. According to the report, the CDU/CSU is now proposing to increase the number to five constituencies to be won.”

    I am not sure how serious the proposal is, given that the Union parties dragged their feet on this for two legislative periods when they were in power. The proposal might even appeal to some Social Democrats, given how it might lead to the expulsion of the Left Party from power and that the SPD is the other main beneficiary of overhang mandates. At least, there are now a number of alternatives on the table.

    Liked by 1 person

  10. I expect the Constitutional Court to strike parts of this down, but with any luck it won’t derail the whole thing, because there are good parts such as the change in terminology on the ballot paper.

    I think a lot of ordinary German voters who aren’t political anoraks are still under the impression that the ‘Erststimme’ is the more important one.

    I am intrigued by the final sentence in the post: “It is not a mixed-member system if it’s possible for a candidate with the most votes in a nominal district not to win the seat.”

    This raises the question of how the present Bavarian ‘Gesamtstimme’ system would be classified, if not as MMP? The 5% hurdle is considered absolute there. Just because it’s never happened in Bavarian to my knowledge, doesn’t mean it hasn’t been possible for however many previous Landtag elections.

    As we can see from the example in Manuel’s earlier comment, there is nothing explicit in the law which states that a district seat won by a party which is not eligible to receive further seats is automatically awarded to the SECOND-PLACED candidate, although for simplicity’s sake it’ll probably be reported as such.


    • Regarding that final sentence, this is the standard definition of mixed-member: There are “nominal” seats that are won by candidates based on votes cast for them (by name, not party), and there are “list” seats won by candidates ranked on party lists–obviously in a compensatory manner if it is MMP.

      So it’s not MMP if someone who gets the most votes in a single-seat district is unable to take the seat due to their party not being entitled to it. Is that already the case in Bavaria? I’ve long had questions about whether the Bavarian system “counted” as MMP or not. I don’t think I ever came to a conclusion on that.

      Note: that it might not be MMP does not mean it is a “bad” idea. It just is something different, and as a result it may have different political consequences.

      Earlier, kazuario (18/01/2023 at 6:55 am) suggested some plausible impact that such a provision could have. Whether or not that is precisely the effect we would expect (I am not sure, need to think on it more), I think we (and Germans discussing these reforms) should at least be aware that the change could be substantial enough to have political consequences. Nils-Christian Bormann (20/01/2023 at 12:00 am) suggested the changes already in 2013 had made the districts less politically significant. That may be true, although I am not sure I agree it has had a substantive effect. For one thing, the basic tier even in 2021 still produces outcomes we’d expect if the districts matter to the outcome, even if arguably they no longer do. (On this latter point, see my summary of the 2021 election, in particular the third paragraph after the second table.)


      • Yes, as far as I am aware, that is the letter of the law in Bavaria at present. It’s just that with the level of CSU dominance there, I don’t believe any party has ever been in danger of having a FPTP winner actively deprived of their seat in practice.

        There are a handful of districts in the two largest cities, Munich and Nuremberg, that can be won on a good day by the SPD and latterly the Greens, but they are both not in any real danger of falling below the 5% threshold.

        In the days when the FW (Free Voters) list was still below 5%, they may have finished a distant second to the CSU in some ultra-safe rural districts, but I think that’s as close as we’ve ever come to having that part of the electoral law actually applied.


      • The name changes to “Main vote” and “Constituency vote” are central to the proposal.

        In MMP the first step is the filling of all the SMD seats with the FPTP winners according to the “first votes”. The next step is one-by-one distribution of top-up seats according to the party (“second”) votes. In countries with a previous history of SMD-only elections, there may be little (NZ) or no (Scotland) overhang compensation. This gives results that are “Proportional if possible, but possibly not proportional.”

        Conceptually, the proposed changes in Germany begin by allocating all seats according to the party (“Main”) votes, provisionally filling them with list candidates. The next step is one-by-one awarding of SMD seats, according to the candidates’ shares of “constituency votes”. Each successful SMD candidate replaces one of their party’s provisional list candidates. An FPTP winner will not be seated is there is no longer a list candidate to replace: “FPTP if possible, but possibly not FPTP.”

        Perhaps the system could be called PMM – Proportional Mixed Member — to emphasize the reversal of MMP priorities!

        Liked by 1 person

  11. Can MMP be combined with 2 member districts using the SNTV with party lists, would this resolve Germany’s problem with overhangs?


    • Something that was actually proposed in the Netherlands. I was among a team of advisors asked to give our opinion. We told them they were nuts (not in those exact words).


        • This was in 2005 or thereabouts. Memory fuzzy. But mostly the usual objections to SNTV, plus the fact that putting things together that haven’t been tried together before is always risky when talking about electoral reform. Basically, even if the effects were to be benign, I’d struggle to see the point of such a composite. And I recall that is was pretty evident they had not really thought it through when presenting it for comment.


      • Did the Dutch want an MMP system using two members plus party lists? Would this system be disproportionate? Or too much like Chile’s old binomial system. Would the 2nd placed party get too many seats causing more overhangs? Something I never thought of could be problematic.


        • Surely, if the overall composition is determined by the list vote using PR, the overall problems of SNTV should only appear in a very mild form, if at all (maybe akin to open list PR).


        • I am not sure about that. It is certainly plausible, but I would not count on it, and would not feel professionally safe saying we could predict how it would work.

          Anyway, as I said, it was a long time ago, and Rob brought it up I had not thought about it in years!

          (Also, down-thread, see the comment by Steven Verbank, who offers a link to the proposal itself.)


  12. If a party wins “too many” district seats, what solutions are there to choose from?
    I see the problem and its possible solutions as if in the ideal situation the result is “zero-zero” : no deviation from proportionality, no extra seats.

    A method with Überhangmandate chooses “one-one” : the total number of seats is raised with one and a party has one seat above its proportional share.
    As an example: a target of 10 seats with 5 districts won by party A and list vote result: A 35%, B 27%, C 24%, D 11%, E 3%.
    This will give A 5 district seats (but only list votes for 4) and B gets 3 list seats, C 2 and D 1. (11 seats in total)

    The Scottish top-up method chooses “zero-two” : the total number of seats cannot change (“zero”) but a party gets more than its proportional share, even more than with Überhangmandate (because the other parties do not get an extra list seat) (“two”).
    In my example: A wins 5 district seats and B gets 2 list seats, C 2 and D 1.

    A method with Ausgleichmandate chooses “two-zero” : the total number of seats is raised even more than with Überhangmandate only (“two”) so that proportionality is restored (“zero”)
    In my example: A wins 5 district seats and B gets 3 list seats, C 3 and D 1. (12 seats in total)

    The German proposal want to get back to “zero-zero” (no deviation from proportionality, no extra seats), but it then has to give in on something else. It chooses to make exceptions to the simple “plurality wins” method in the district tier.
    In my example: A wins only 4 district seats (one district remains empty) and B 3, C 2 and D 1 list seat.

    Liked by 1 person

    • Project management involves the iron triangle of time, cost and value. You want fast, cheap and good but sometimes the reality is “on time, on budget or on spec – pick any two”.

      Design of an electoral system with an SMD tier and a compensatory tier has a similar tension in its design. You want: each party’s seats in proportion to its votes; only FPTP winners filling the SMD seats; predetermined, fixed size of assembly. Sometimes the outcome of an election is such that all three objectives can be achieved, but some election outcomes require at least one of them to be relaxed in order to maintain the other two criteria.

      MMP in Scotland is willing to relax proportionality; current MMP in Germany allows assembly size to vary; the proposed change in Germany would allow some SMD seats to be filled by FPTP also-rans.

      Liked by 1 person

  13. Using highest average instead of largest remainder + overhang + compensation would eliminate the problem, but it could reduce proportionality slightly.

    As it is, I believe, Germany calculates the total number of mandates per party using largest remainder. So, for example (I’m going to use vote % rather than whole votes and quotas, just for simplicity), hypothetical result:

    Lander with 5 SMD / 10 total
    Party A: wins 5 SMD, 40% of party vote–entitled to 4 total seats – 5 SMD = overhang mandate = 5 seats
    Party B: wins 0 SMD, 36% of party vote–entitled to 4 total seats – 0 SMD = 4 list seats = 4 seats
    Party C: wins 0 SMD, 24% of party vote–entitled to 2 total seats – 0 SMD = 2 list seats = 2 seats

    And from there you have to work out the compensation for B, C for the over-representation of Party A.

    If, however, you use highest average (d’Hondt for simplicity) to determine overall seats, you can avoid the problem (again, it can impact proportionality, but it gets rid of overhang mandates). Same numbers (again using % for simplicity):

    Lander with 5 SMD / 10 total
    Party A: wins 5 SMD, 40% of party vote
    Party B: wins 0 SMD, 36% of party vote
    Party C: wins 0 SMD, 24% of party vote

    Five seats determined, use highest average to award remaining five [votes / (seats + 1)]:

    6th seat
    A 40% / (5+1) = 06.7
    B 36% / (0+1) = 36.0 — 1 seat
    C 24% / (0+1) = 24.0

    7th seat
    A 40% / 6 = 6.7
    B 36% / 2 = 18.0
    C 24% / 1 = 24.0 — 1 seat

    8th seat
    A 40% / 6 = 6.7
    B 36% / 2 = 18.0 — 1 seat
    C 24% / 2 = 12.0

    9th seat
    A 40% / 6 = 6.7
    B 36% / 3 = 12.0
    C 24% / 2 = 12.0 — 1 seat (let’s assume for sake of argument C has a few more total votes)

    10th seat
    A 40% / 6 = 6.7
    B 36% / 3 = 12.0 –1 seat
    C 24% / 3 = 8.0

    A 40% 5 constituency + 0 list = 5 total
    B 36% 0 constituency + 3 list = 3 total
    C 24% 0 constituency + 2 list = 2 total

    Might not be constitutional under court’s ruling that led to the current compensation scheme, but solves the problem mathematically, and without having to take away SMD mandates.

    Liked by 1 person

    • Actually, I got something wrong: Germany is already using St. Lague. But when they calculate (s +1), s = 0 for the first seat assigned. If on the first pass s = constituencies already won, then it achieves what I suggested.

      Liked by 1 person

  14. The main points of the reform are:

    Fixed number of 630 seats. Number of SMDs stays at 299.
    The allocation of seats between parties (and across states) is based on the PR vote.
    There are no overhang or compensation seats any more. When there is an overhang in principle, the district winners with the lowest vote shares in the respective party and state do not receive a seat.
    Parties (with the exception of those representing national minorities) need to win 5% of the national vote. The rule that 3+ districts won are also sufficient is abolished. District winners from parties under 5% are not elected, but district winners who ran independently are.

    The Left (4.9% of the national PR vote in 2021) and the Bavarian CSU (5.2%) are furious and will bring it before the Constitutional Court.


    • Thank you for this information. So the CSU actually could be threatened with extinction as a distinct party under the new threshold rule? When you refer to “district winners who ran independently” do you mean as non-party candidates, or assuming they run with a party that does not present a party list?

      It is quite a significant reform, and as I said above, I would not call this MMP anymore if it goes into effect. A mixed-member system, by standard definition, allows a party to hold any seats where it won under the district-level formula, regardless of the performance of the associated list.


      • Yes, both the CSU and The Left are under threat now, since winning 3+ districts no longer qualifies a party for the PR-based allocation. The CSU won almost 32% of the PR vote in Bavaria, and The Left gained around 10% in former East Germany (more than 8% in any of the five East German states).
        I agree that some form of a sub-national threshold looks like a good alternative. Some legal experts suggest that the 3+ districts rule could have been deemed unconstitutional in combination with the overall strengthening of the PR principle in the new law, while others disagree. Hence, many perceive that the parties in government (SPD, Greens and Liberals) dropped the 3+ districts rule out of self-interest.

        The story with non-party SMD candidates is complex. The German electoral law distinguishes between party-nominated SMD candidates and others. The new law also adds a rule that SMD candidates who are not nominated by a party cannot be included on a party list. This is meant to disencourage parties from running their folks as non-party candidates while also providing them with a good list position as a safety net. However, a party (think of the CSU) could still run non-party SMD candidates in either all districts, or in safe districts in combination with party-nominated ones (with a list back-up) in others, and after the election all of them could join the parliamentary party group. (Combining party-based SMD nominations without a list would not make sense, since you don’t win any PR votes.) I have seen comments, though, that the maneuver with non-party SMD candidates may not be accepted by the election authorities. There may also be reputation costs for a party if they are seen as gaming the system.

        The point that non-party SMD winners are awarded a mandate but SMD winners from parties under 5% are not is also one the Constitutional Court can be expected to look at.

        Liked by 1 person

  15. A 5% threshold would be more appropriately applied at the state level, rather than nationally, in a federal or devolved country. Even (especially?) separatists should feel they are given a fair shake by the electoral system.


    • The original NZ MMP recommendation included no threshold for Maori parties, with no practical suggestion as to how you define them!


      • Yes, I remember that. The idea was to have a provision like Germany’s whereby “ethnic” parties are not subject to the threshold, which would have allowed for the removal of the set-aside seats for Maori. The definition problem could be thorny, however.


  16. The end of an era. With the demise of MMP in its place of birth surely by now even the most ardent supporters of MMP must admit it is at least a very problematic and overrated system and possibly not worth the trouble.


        • Ireland, Malta, and Australia are very blessed to have STV, and interestingly enough that electoral system is used for the indirect elections for the Senate of Pakistan and Rajya Sabha of India.


        • Yes, STV has a sparse application around the world, mostly in former British imperial jurisdictions, in addition to relatively more recent adoption for various offices in the UK itself (Scottish local elections, Northern Ireland’s devolved assembly, etc.), and a few local councils in the US. The only non-Anglo case I am aware of is Estonia, for one election (1990). It just is a very rare system, for better or worse.

          It is not often appreciated just how differently it can be expected if there is, or is not, a pre-existing developed party system. In Malta, STV probably barely matters because party lines so rigidly structure voting behavior. In Ireland, the parties are pretty “strong” too, but there is much more cross-party ranking. And I believe twice the country has had governments that wanted to abolish STV, but in the required referendums, the voters opted to keep it.

          (This is supposed to be a thread on Germany, so I guess I am kind of breaking my own rules by engaging in a discussion of STV here. But you asked a good question, Rob! There are many other threads here about STV…)


      • I’d also add to this puzzle the observation that it makes little sense to blame MMP, per se, for the decision in Germany to prioritize full proportionality and avoid overhangs and assembly-size increase from election results. If that is what those in power want, then MMP is not the right system for them. That does not prove MMP is itself problematic. All systems have tradeoffs. Upon completing the research on the effects of MMP on parties’ committee assignments in both Germany and New Zealand, for the Party Personnel Strategies book, I became more convinced than ever that MMP does indeed offer the “best of both worlds” in its ability to balance national-partisan and local-constituency representation. The current German government has decided (in effect) that it no longer wants that balance. I should add that the Constitutional Court also made that decision (again, in effect) with the ruling that exacerbated the assembly-size problem in the first place.


        • If “MMP” doesn’t apply when the election of SMD plurality winners is conditional rather than unconditional, is there an alternative acronym that applies to the new German system?


        • I say we are too harsh to deny it to be called MMP because the allocation of say 1% of the seats will deviate from the ideal MMP model. That’s just a matter of political science definitions. On the legal front however, I wouldn’t be surprised when the Constitutional Court uses “equality” as an argument against some districts being denied its own district MP.


        • I think this is fair. MMP in Germany has been subject to so many constraints that it’s hard to pretend that it’s somehow a meaningful archetype. It could be said that the increasing fragmentation of the system made the trade-off between proportionality and local representation harder, but that’s not really unique to MMP.

          I do think that the “best of both worlds” dynamic won’t end entirely as a result of these changes. For an individual candidate, it still seems like it would be advantageous to seek votes in the single member districts even though election isn’t guaranteed if you win. The mechanical effect of compensating overhangs was that the number of members being elected in SMDs was consistently being diluted, even though every candidate elected in a SMD was elected.


        • If it is possible under the rules for a candidate to “win” a district and yet not be awarded the seat, it is NOT a mixed-member system, as this is absolutely core to the definition.

          Whether it will have a substantial effect on the behavior that leads us to conclude MMP offers the best of both worlds is uncertain.Any such change in behavior may take a while, but I’d expect it to happen.

          This is now–if implemented–some form of PR with nominating districts or “district-ordered lists.” Really, a hybrid of that and closed-list PR. I would not be willing to call it even a little bit of MMP anymore.


      • On the “best of both worlds” point, do you feel it has advantages even over the low-magnitude PR systems that Carey/Hix propose?


        • Yes. In Party Personnel Strategies, we include Portugal (which has a system pretty much in their “sweet spot”) and I would interpret our results as favorable to MMP over moderate-magnitude districted PR. But note that our outcome is committee assignments, and by extension policy responsiveness, which is not the same as the outcomes Carey & Hix are interested in.


        • Yes, Steven. You are right. I should not have suggested that Portugal was in the Carey-Hix “sweet spot.” The district magnitude may be a little too high even on average, but the wide magnitude variance across districts certainly means it is not what they had in mind.

          In fact, in the Party Personnel analysis, we exploit this variance in the Portuguese case, and it certainly makes a difference. So our results do not speak to how a districted PR system of more uniform magnitude would work, except perhaps by extended inference.

          Also, that Monroe and Rose article is one of my favorite articles on electoral systems of the past 20 years or so. (More than 20 now!)

          As a somewhat related aside: in Votes from Seats, we note that we see no difference due to magnitude variance in the fit of the effective number of parties (nationwide) to the seat product model prediction (based on mean M). However, such variance certainly affects which parties any given voter has as viable to win a seat in their own district, according to their local magnitude (a theme of Ch. 10 of Votes from Seats). And it affects the intra-party dimension, including the harnessing of constituency connections and expertise of members (key themes of Party Personnel Strategies).


        • What I found most interesting about Carey and Hix’s work was around voter-legislator congruence, and the observation that low-magnitude PR had better congruence both SMDs and high-magnitude PR.

          Click to access Carey-Hix-Public-Choice-2013.pdf

          Their work and Hagemann’s – – have shaped a lot of my thinking on what the “ideal” should be. I guess I still fail to see what advantages if any MMP has over such proposals, and if the added complexity of MMP or the arbitrary nature of SMDs are really worth it.


  17. Courts have been known to reach strange conclusions. But it is hard to see how the voters of Munchen-Nord would have been denied equal representation if the CSU candidate with 25.7% of votes had been set aside in favour of the Green Candidate with 24.2% of votes (or even the SPD candidate with 21.9%).


  18. While we are discussing Carey and Hix’s “electoral sweet spot” of low-magnitude (4 to 8 seats per district) proportional representation, I will just note that Canadian electoral reformers included it as an option. In Fair Vote Canada’s 2016 submission to the Electoral Reform Committee we said of MMP “Parliament might decide that the average region should have 14 MPs, eight MPs, or some other number.” We said that, if we seek some uniformity in the size of MMP regions across the country including the smaller provinces, that could be an average of 11 or 12 MPs per region. And then we gave them a specific example of how MMP regions could be configured with an average size of eight MPs per region.

    Liked by 1 person

Leave a Reply

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

You are commenting using your 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.