Ukraine honeymoon election today

Ukrainians are voting today in an assembly election. It is a relatively extreme “honeymoon” election, as the new president, Volodomyr Zelensky, was just elected in March-April of this year (two rounds). There was already an assembly election scheduled for October of this year, which certainly would have qualified as a honeymoon election. But in his inauguration, Zelensky announced he would dissolve the Verkhovna Rada and call an election even earlier.

And why not? Based on much experience in presidential and semi-presidential systems, we know that there is a strong tendency for the party of a newly elected president to gain a large boost in votes the earlier it is held following the presidential election. This topic of the impact of election timing has been a theme of my research ever since my dissertation (1988), an early APSR article of mine (1995), and most recently in a whole chapter of Votes from Seats (2017).

At the time Zelensky was elected, various news commentary had the all-too-typical concern that the new president would be weak, because he is an “outsider” with no established political party. We got similar useless punditry when Emannuel Macron was elected in France in 2017. And we know how that turned out–his formed-on-the-fly party did slightly better than the 29% of votes I projected, based on an equation in Votes from Seats, prior to Macron’s own runoff win. (The electoral system helped turn that into a strong majority in the assembly.)

In May of this year, I projected that Zelensky’s Servant of the People party could get around 34.5% of the votes in an election held on 28 July. (One week earlier obviously does not change anything of substance.)

Early polling had him short of this (not even 25% just before the presidential first round), but predictably, SoP has been rising in the polls ever since Zelensky took office. The party almost certainly will beat this projection, and may even have an electoral majority. If short of 50% of votes, the party still looks likely to win a parliamentary majority, given the electoral system (discussed below).

A bigger boost than average (where the average across systems with nonconcurrent elections is what my projections are based on) is to be expected in a context like Ukraine, in which the party system is so weak. That is, poorly institutionalized party systems would tend to exaggerate the normal electoral cycle effect. The effect will be only further enhanced by low turnout, as opponents of the new president have little left in the way of viable political parties to rally behind. Thus a performance in the range of the mid-40s to over 50% of the vote would not be a surprise.

As for the electoral system and election itself, Ukraine is using again (for now, at least) its mixed-member majoritarian (MMM) system. It consists of 225 single-seat districts, decided by plurality, and 225 closed-list proportional representation seats, in a single nationwide district. The two components are in “parallel”, meaning seats won by any given party in districts and seats won from party lists are simply summed; there is no compensatory process (as with MMP). There is a 5% threshold on the list component; quite a few small opposition parties may waste votes below this bar. Due to parts of the country being under Russian occupation, only 199 single-seat contests will take place.

In some past MMM elections in Ukraine, a large share of the single-seat districts have been won by independents or minor parties, whereas the national parties (such as they are) have, obviously, dominated the nationwide list seats. It is probably quite likely that this rather extreme honeymoon election will result in most of the seats in both components being won by “Servants.”

On that theme, a tweet by Bermet Talant makes the following points (and also has some nice polling-place photos) based on conversations with voters in Kyiv:

• Ppl vote for leaders. Few know other candidates on party lists, even top5

• Servant of the People = Zelensky. Bscly, ppl vote for him again

• In single-member districts, ppl vote for a party too, not candidate

This is, of course, as expected. It is a completely new party. Many voters will be wanting to support the new president who created the party. The identity of candidates will not matter, either on party lists (where at least the top ones might be known in a more conventional party) or in the districts (where the vote is cast for a candidate). The single-seat districts themselves are referred to as the “twilight zone” of Ukrainian elections in a fascinating overview of the candidates and contests in the district component published in the Kyiv Post. These contests attract “shady candidates” many of whom are “largely unknown”. If a given election lacks a strong national focal point, it would tend to favor independents and local notables. In an election with an exceptionally strong focal point–as in a honeymoon election, more or less by definition–that will benefit whoever has the “Servant of the People” endorsement.

The timing of the election, and the likely dominance of an entirely new pro-Zelenskyy party, really is presidentialization at its very “finest”.

I am just going to quote myself, in the final paragraph of an earlier post about Macron’s honeymoon election, as it totally applies here, too: “All of the above should serve as a reminder of two things: (1) the purpose of the upcoming election is to ratify the new executive’s direction, not to be a second chance for an alternative vision; (2) the honeymoon electoral cycle matters.”

Expect the new Verkhovna Rada to be Servants of Zelenskyy.

Following the Indian campaign

The first of several phases of the 2019 Indian general election is underway. I just found a really neat feature at the Times of India that gives summary of the two major party leaders’ activities each day.

Also, on India:

“180 turmeric farmers refused to withdraw their candidature from Nizamabad on the last day of nomination… no choice but use ballot papers” instead of electronic machines.

That’s a lot of candidates, and also a pretty expensive protest (apparently around $400 per nomination), as well as a headache for election administrators.

Pakistan, 2018

Pakistan elections were today.

Something to watch is how well new religious parties do. One of them, Tehreek-e-Labbaik Pakistan, has a campaign poster that features a woman candidate, although you will need a little imagination to see her.

One of the party’s male candidates explains, “The party has nominated a few women… because… it is mandatory under the election law.”

I understand (via experts on Twitter) that there could be many by-elections in the weeks to come. One even spoke of a “wave” of them. Some of these will be mandated by a provision that invalidates any constituency result in which at least 10% of women on the voter roll did not participate. Others will be necessary because candidates can run in multiple constituencies and, if they win more than one, they have to step aside in all but one. Still others will be needed because several candidates (at least 8) have died since nomination; it is not clear to the extent the deaths are natural or due to election violence.

The electoral system is mostly FPTP. The total size of the National Assembly is 342, and 272 of them are from single-seat districts via plurality. Others are reserved for women or ethnic minorities; some form of PR is supposedly used for these, but I do not have the details. Perhaps someone will enlighten us in the comments. This election is the first under a newly delimited constituency map.

Early results put Pakistan Tehreek-e-Insaf, led by Imran Khan, in the lead.


Unless I say so explicitly, mention of parties or candidates on this blog is not an endorsement. That is especially so when I have no clue what a party’s poster appearing with my entry even says!

Lebanon 2018

Lebanon votes on 6 May. The new electoral system was previously profiled here at F&V by Amal Hamdan. Now we get to see the system in action!

Ali Harb at Mideast Eye also has a discussion of the system, complete with a map of the districts, and notes a change in ballot format beyond the adoption of open lists:

Another change is that the interior ministry will be solely responsible for ballots on election day. In previous elections, ballots were handed to voters by campaigners for various political parties.

In past elections, less popular candidates were able to slip their names onto the lists of other parties in a phenomenon known as “booby-trapped ballots”.

Lebanon, welcome to open alliance lists!

Lebanon’s election is coming up (6 May), and the country is getting its first look at a new open-list proportional electoral system (profiled previously here by Amal Hamdan).

An interesting blog post from 12 March by Gino Raidy has just come to my attention*: “Why Political Parties are Terrified of Forming Lists”. The author discusses the perils for parties joining on an alliance list. Because the lists are open, it is possible for one party to help boost its partner’s seats but elect none of its own candidates. On the other hand, it is also possible that parties (and alliances) will want to recruit relatively independent figures who can appeal to a wider electorate.

These sorts of issues may be new to Lebanon, but they will be familiar to readers of this blog. I have talked about them before, most recently in a discussion of Brazil and Finland, where open alliance lists have been in place for some time.

____
*Thanks, Dan W.!

Lebanon’s New PR Electoral System: Undermining Proportional Outcomes in a Proportional Representation Electoral System

This is a guest post by Amal Hamdan

In May 2018, parliamentary elections are scheduled to be held in Lebanon using a PR electoral system for the first time. The last parliamentary elections in Lebanon were held in June 2009. Since then, parliament has extended its own term twice. After years of deadlock over electoral reform, Lebanon’s two main rival political alliances, the March 14 and March 8 blocs, passed a law in June 2017 abolishing the Block Vote [MNTV–ed.] electoral system, used since 1958 for legislative elections, and introduced Open List Proportional Representation (PR) (Law No.44). An analysis of key technical aspects of the new law – namely, the formula used to distribute seats to lists and an informal threshold for list eligibility – suggests that it was designed to enhance the chances of candidates within the March 14 and March 8 blocs to be elected and diminish the possibility of electing candidates outside these alliances. Lebanon has been politically polarized between the March 14 and March 8 blocs since 2005, following the assassination of former Prime Minister Rafic Hariri which led to an end of Syria’s hegemonic grip over Lebanese politics. The main political factions comprising the March 14 alliance are the Future Movement (FM), the Sunni community’s main political representative; Maronite Christian Samir Geagea’s Lebanese Forces (LF); and the Christian Kataeb party. The March 8 bloc comprises the main Shia political parties, Hizballah and the Amal Movement and their mainly Maronite Christian ally, the Free Patriotic Movement. Until 2009 Druze leader Walid Joumblatt’s Progressive Socialist Party (PSP) was also part of the March 14 bloc but has since withdrawn, further reinforcing Joumblatt’s ability to play political kingmaker in Lebanese politics.

128 parliamentarians will be elected in 15 new ‘major’ electoral districts. Many, but not all, of these 15 districts are comprised of minor constituencies [qada]. Each voter casts a ballot for a list of candidates; voters have the option to cast one preferential vote for their favorite candidate, as long as the candidate is on the same list they have chosen. There is a key restriction on candidates’ preferential vote: if the major electoral district is comprised of more than one minor constituency, voters can use their preferential vote only for candidates within their minor constituency, and not any candidate in the major electoral district.

The new electoral law adopts the Hare Quota Largest Remainder (HQLR) formula to distribute seats to lists. Under the HQLR, the ‘price’ of a seat, in the currency of votes, is determined by dividing the total valid votes cast in a district by district magnitude (number of seats allocated in the district). This provides the quota or price of a seat. Under Lebanon’s PR system, for every whole quota a list has won, it receives a seat. If there are unfilled seats, they are allocated to lists with the largest remaining votes.

Only lists that receive one full whole quota are eligible to seat allocation; any lists that receive less than 1.00 or a full simple whole number are disqualified. One full whole quota or the threshold for lists to qualify for seat allocation varies across districts from 7.69% of votes (Mount Lebanon 4) to 20% (South Lebanon One). Table 1.1 identifies the major constituencies and the effective threshold in each district for lists to qualify for seat allocation. The minor constituencies or qada within each major district are in brackets.

Table 1.1

Major Constituency Effective Threshold of List Votes
South Lebanon One (Sidon; Jezzine) 20%
Beqaa Two (Rashaya; West Beqaa) 16.67%
Mount Lebanon Three (Baabda) 16.67%
South Lebanon Two (Tyre; Zahrani) 14.29%
Beqaa One (Zahle) 14.29%
North Lebanon One (Akkar) 14.29%
Beirut One 12.50%
Mount Lebanon One (Jbeil; Kesrewan) 12.50%
Mount Lebanon Two (Metn) 12.50%
Beqaa Three (Baalbeck-Hermel) 10.00%
North Lebanon Three (Zgharta; Bcharri; Koura; Batroun) 10.00%
Beirut Two 9.09%
South Lebanon Three (Bint Jbeil; Nabatieh; Marjayoun-Hasbaya) 9.09%
North Lebanon Two (Tripoli; Minnieh-Dinnieh) 9.09%
Mount Lebanon Four (Chouf; Aley) 7.69%

 

Besides the threshold, an additional provision in Lebanon’s PR electoral law favors well-established alliances such as the March 14 and March 8 blocs and strongly reduces opportunities for less established, smaller parties or electoral alliances from winning seats.  Under the new law, the Hare quota will be calculated a second time excluding the votes won by a list or lists that did not achieve the electoral quotient. A hypothetical example of three lists competing in the newly-created South Lebanon One constituency, which has a total of five seats allocated to its minor constituencies, Sidon and Jezzine, demonstrates how provisions in Lebanon’s PR electoral law decreases chances of candidates not running on major ballots from winning. The following table provides seat and confessional allocation in South Lebanon One:

South Lebanon One ‘Major’ Constituency
Minor Constituency (Qada) No. Seats and Confessions
Sidon Qada 2 Sunni
Total Seats in Sidon Minor District 2
Jezzine Qada 2 Maronite
1 Greek Catholic
Total Seats in Jezzine Minor District 3
Total Seats in South Lebanon One 5

To calculate the threshold for a list to qualify for seat allocation, the total number of valid votes cast for all lists in the South One district are combined and divided by district magnitude. Assume there are three hypothetical lists competing in the South Lebanon One district. The number in the brackets next to each candidate indicates the candidate’s preferential votes.

Lists Contested in South Lebanon One Major Constituency
Minor Constituency and Confession List A List B List C
Sidon Qada      
Confessional Seat Sunni A1 (25,460) Sunni B1 (13,512) Sunni C1 (7,543)
Confessional Seat Sunni A2 (23,041) Sunni B2 (5,266) Sunni C2 (4,289)
Jezzine Qada
Confessional Seat Maronite A3 (10,792) Maronite B3 (15,648) Maronite C3 (7,399)
Confessional Seat Maronite A4 (5,403) Maronite B4 (13,285) Maronite C4 (4,338)
Confessional Seat Greek Catholic A5 (5,220) Greek Catholic B5

(14,914)

Greek Catholic C5

(6,498)

List’s Total Votes 73,917 64,826 33,168

Note: The total votes won by each list exceeds its total preferential votes because voters have the option of voting for the list without casting a preferential vote for a candidate (Article 98, Clause 1)

 

The first step is to calculate whether lists are eligible to qualify for seat distribution; only those with at least one whole quotient will be eligible.

Step 1. Calculate the electoral quotient:

Electoral Quotient: (73,917 List A votes + 64,826 List B votes + 33,168 List C votes) = 171,911 ÷ 5 seats = 34, 382

Step 2. Calculate if lists won a whole quotient by dividing each list’s total votes by the electoral quotient:

List A: 73,917 list votes ÷ 34, 382 electoral quotient = 2.15.

List B: 64,826 list votes ÷ 34, 382 electoral quotient = 1.88.

List C: 33,168 list votes ÷ 34, 382 electoral quotient = 0.96.

Since List C won less than one whole electoral quotient it is disqualified – even though this list won 19.29% of total votes cast. Since a list has been disqualified, a second electoral quotient using the HQLR formula must be calculated to distribute seats to eligible lists.

 Step 3. Calculate the second quotient:

 Formula for second quotient: Qualifying Lists’ Votes ÷ District Magnitude.

Second Quotient: (73,917 List A votes+ 64,826 List B votes) = 138,743 ÷ 5 seats = 27,749.

The price of a seat dropped from 34,382 to 27,749 votes. Arguably, the provision to calculate the price of a seat twice is aimed at lowering the cost and enhancing the March 14 and March 8’s blocs chances of sharing seats (presuming these alliances remain in tact).

Step 4. Distribute seats to qualifying lists.

This is determined by dividing the qualifying lists’ total votes by the second electoral quotient.

 List A: 73,917 ÷ 27,749 = 2.66 = 3 seats.

List B: 64,826 ÷ 27,749 = 2.33 = 2 seats.

The next step is to distribute seats to candidates. Seat distribution to candidates across lists will not be straightforward in Lebanon since PR will be implemented alongside a confessional quota. The next section demonstrates how implementing a PR electoral system alongside a confessional quota will likely lead to anomalies in seat distribution and consequently, anomalies in representation.

The Potential Anomalies in Seat Distribution under Lebanon’s PR Electoral System

Lebanon is comprised of 18 officially recognized religious communities, known in Lebanese jargon as confessions or sects. These 18 sects are mainly Muslim and Christian denominations, although there remains a small Jewish minority. None of these 18 confessions are a majority, making Lebanon a country of minorities. All 128 parliamentary seats are reserved for 10 Muslim and Christian confessional communities. One seat is reserved for ‘minorities’, meant to represent the remaining communities not designated seats.

To distribute seats to candidates in qualifying lists, each candidates’ percentage of preferential votes is calculated; then all candidates are ranked in a single list from highest to lowest percentage and seats are distributed accordingly. However – seat distribution will also need to take into account the confessional allotment of seats and their allocation to minor constituencies: if all the seats reserved for a confession have been filled, a candidate can be disqualified even if he or she is ranking higher than their opponent. A candidate may also be disqualified if all the seats allotted to their minor constituency have been filled. Drawing on the hypothetical example from the South Lebanon One district where List A won 3 seats and List B won 2 seats clarifies these points and demonstrates how potential anomalies in representation could arise in Lebanon’s new PR electoral system.

Step 1.  Calculate each candidates’ percentage of preferential votes.

This is done by dividing the number of preferential votes won by a candidate by the total preferential votes cast for all candidates, regardless their confession, from qualifying lists in each qada separately. For example, the formula to calculate candidate Sunni A1’s percentage of preferential votes in the Sidon minor constituency is calculated by dividing their preferential votes by the total preferential votes won by candidates in qualifying lists in Sidon alone, rather than the major constituency:

% of Preferential Votes: Candidate in Qada ÷ Total preferential votes for all candidates in qualifying Lists in Qada.

Sunni A’s % percentage of preferential votes: Sunni A preferential votes ÷ Total preferential votes from qualifying lists in Sidon = 25,460 ÷ 67,279 = 37.84%

The following table calculates candidate’s percentage of preferential votes in each minor constituency in the South Lebanon One electoral district.

South Lebanon One District
Minor Constituency (Qada) List A List B
  Candidate No. Preferential Votes % Preferential Votes Candidate No. Preferential Votes % Preferential Votes
Sidon Sunni A1 25,460 37.84% Sunni B1 13,512 20.10%
Sunni A2 23,041 34.25% Sunni B2 5,266 7.83%
Total Preferential Votes in Sidon Qada  

67,279

 

 

Jezzine Maronite A3 10,792 16.54% Maronite B3 15,648 23.98%
Maronite A4 5,403 8.28% Maronite B4 13,285 20.36%
Greek Catholic A5 5,220 8.00% Greek Catholic B5 14,914 22.85%
Total Preferential Votes in Jezzine Qada  

 

65,262

 

Step 2. Rank all candidates from both constituencies from the highest percentage of preferential votes to the lowest.

Candidates’ Ranking According to Percentage of Preferential Votes, South Lebanon One District
Rank Candidate Minor Constituency % Preferential Votes
1 Sunni A1 Sidon 37.84%
2 Sunni A2 Sidon 34.25%
3 Maronite B3   Jezzine 23.98%
4 Greek Catholic B5   Jezzine 22.85%
5 Maronite B4   Jezzine 20.36%
6 Sunni B1 Sidon 20.10%
7 Maronite A3   Jezzine 16.54%
8 Maronite A4   Jezzine 8.28%
9 Greek Catholic A5   Jezzine 8.00%
10 Sunni B2 Sidon 7.83%

 

Step 3: Distribute seats to candidates.

Seats are distributed to candidates in descending order of their percentage of preferential votes, but with these critical caveats:

  • If the seats for a confession in a minor constituency have been filled, the remaining candidates from that confession are excluded, even if they have won higher percentages of preferential votes than candidates from other confessions;
  • once the seats allocated to a list have been filled, the remaining candidates for that list are disqualified, even if they have higher percentages of preferential votes from other lists.
  • If candidates are tied for preferential votes and are both eligible because the confessional quota in their district hasn’t been filled, the older candidate wins the seat.

In the hypothetical example, the distribution of seats to candidates according to minor constituency and confessions is the following:

List A: 3 Seats List B: 2 Seats
Candidate A1 (Sunni/Sidon)  B3 (Maronite/Jezzine)
Candidate A2 (Sunni/Sidon) B5 (Greek Catholic/Jezzine)
Candidate A3 (Maronite/Jezzine)

 

This demonstrates exactly the potential anomalies in representation under Lebanon’s new PR system: although the candidate in rank 5, Maronite Candidate B4 won 20.36% preferential votes, he or she was disqualified because List B’s two seats had been filled and consequently, the seat was awarded to the candidate in rank 7, Maronite Candidate A3, who received 16.54% of preferential votes. The candidate in rank 6, Sunni B1, was also excluded because all the Sunni seats in Sidon were filled; even if they had not been filled, List B also already been allocated its two seats.

Lebanon’s new electoral system

This is a guest post by Steven Verbanck

Lebanon finally has rewritten its electoral so that parliament can be renewed in 2018 (the sitting parliament was elected in 2009). The best explanations I’ve found so far are on Blog Baladi, Moulahazat, and Executive Magazine. Until I’ve found a translation of the text of the law itself, I have to rely on these sources.

1. Lebanon makes a major electoral system change by changing from multiple non-transferable vote (MNTV) to (open) list PR with average magnitude of 8,5 (128 seats in 15 districts, range 5 to 11).

It’s in line with a more general trend: many countries once started with MNTV but when elections become more partisan, search for ways to make the outcome less risky. One way to do this is to reduce district magnitude, with many countries ending up with M=1 (FPTP). In Lebanon the endpoint was the qada as district (law of 1960 and Doha law of 2008: 26 districts). Designing even smaller districts is too difficult given the predetermined allocation of every seat to a specific religious community.

The districting still bears the mark of majoritarian thinking: they avoided merging districts that differ too much in their religious makeup.

Introducing a proportional system makes the outcome less risky in competitive districts (51%-49% -> 3-2 instead of 5-0) but on the other hand, in districts who were until now won overwhelmingly, competition become viable (80%-20% -> 4-1 instead of 5-0)

Specifically, they have chosen single quota largest remainder as allocation mechanism between lists. The quota is also threshold and the quota is recalculated with the vote total of the lists still in the running.

2. The intraparty allocation of seats to candidates is rather complex because they combine proportionality between parties with the prefixed allocation of seats to religious communities and to sub-districts (= nada).

A voter is not restricted to candidates of his own religious community but can only give a personal vote to a candidate of his sub-district. It’s unclear to me if a voter has the option of a list vote without a personal vote for a candidate of that list.

All candidates (irrespective of their list or their religious community & sub-district) are ordered according to their personal vote share in their sub-district (not the absolute number of personal votes to balance sub-districts of different sizes).

Seats are then awarded one by one to the candidates in that order. As soon as a list is awarded all its seats, all lower candidates of that list are defeated and as soon as a religious community & sub-district is awarded all its seats, all lower candidates of that religious community & sub-district are defeated. Or put in another way: a candidate loses if enough higher placed candidates of his list have already won to fill the vacancies of that list or if enough higher placed candidates of his religious community & sub-district have already won to fill the vacancies of that religious community & sub-district.

For lower-placed candidates this can be somewhat erratic and unpredictable: you can end up defeated because your list or religious community & sub-district has already enough winners, while lower placed candidates overtake you because their list and their religious community & sub-district still have vacancies. (This will even more be the case if personal votes are concentrated on the first candidate on the list: the differences in personal votes between lower candidates then becomes less meaningful but stays relevant for the allocation.)

For the last vacancy, there can be only one winner because there is only one list left with one vacancy and only one religious community & sub-district left with one vacancy, no matter how few votes that candidate received. (Compare with Romania 2008 where the Hungarian party won the last seat, the only seat of the expatriate district for Africa & Middle-East with only 2% of the votes in that district.)

I have no idea what happens if there is no such candidate: does the seat remain empty? (Compare with Mauritius where sometimes no candidate meets the requirements for the best loser seats.) This problem, proportionality in two dimensions (rows and columns, parties and preset religious communities & sub-districts) can be solved in a better way .)