Gerrymandering

Jun 24, 2019

Finally, in order to discern irregularities at the level of districts and single-member constituencies (SMC), we shall analyze two fundamental questions of this transition, i.e. (1) connection between distribution of votes on party lists and for candidates in SMC and (2) specific nature of constituency boundaries.

Split of the votes from political party to single-member constituency candidate.  

When the voter has two ballots instead of one (as it has been in Moldova before 2019) it results in fundamentally new electoral construction with its own specificity and patterns. Very often in such case the very logic of electoral process is different, and so irregularities are different too. In contrast to the usual situation of voting on party lists, amount of irregularities isn’t the only thing that matters, as “quality” and “targeting” of irregularities matter too.

  1. Difference between results in single-member constituencies (SMC) and results of voting on party lists exists always and everywhere, in every country, region or autonomous district where such kind of election takes place.
  2. This difference can be very high. The Seim of Lithuania and The Senate of Czech Republic election can serve as examples, as in both cases there is two-round elections and in both cases independent candidates rarely enter parliament .
  3. In Ukraine (one-round election) situation is drastically different. There are very many independent MP, which are usually supported by one of main parties or by the regional administrations. They win most often in regions with less defined political preferences (i.e. central region of the country). Even in the 2014 election this pattern held .
  4. Finally, in Germany the difference between results of voting on party lists and of for SMC is negligible. The German model rather connects deputy with his constituency than creates some new trend .
  5. However, in Moldova at the first mixed system elections none of aforementioned models seems to be applicable.

It’s generally considered that the following consistent patterns hold in the transition from party lists to a single-member constituency, especially in the case of one-round election.

  1. In the constituencies where there are pronounced political preferences, they tend to remain in first-past-the-post voting (FPTP) elections.
  2. In the most cases (there are very few exceptions) the situation in a single-member constituency is less complex than the results of voting on party lists. There is rational explanation: voters understand who can win in their constituency and they act in accordance to the situation. Furthermore, the distribution of votes given for obviously unwinnable candidates also should meet some rules.
  3. Probability of winning in the first round is 10-15% for candidates who has got the third place and less than 3% for candidates who has got the fourth one. Actually, such constituencies (which are sometimes called “balanced” constituencies) occur rarely and so they are subject of a separate study.

We’ll formulate the main rule of transition from proportional system to the FPTP one. The votes redistribute for two (less often three) leading candidates and generally from small parties to stronger individual candidates. The severity of this redistribution is unique in each case, but it decreases as overlapping between party lists and SMC candidates increases. Reverse is also true – existence of strong independent candidates can totally change the overall statistics.

In case of Moldova 2019 elections these rules apply very selectively. For example, there are unusually many violations of rule outlined in p1. It’s even more strange, as 4 parties which have obtained seats in the parliament have gained 90% of total votes. Especially taking into account that there had been leading parties’ candidates in almost all SMC. So, the winner changes in spite of almost complete absence of “free” electoral space. We’ll look at this issue most thoroughly in the following discussion of situation in certain constituencies. We should stress that the foregoing assumptions have a clear quantifiable indicator. In comparison, in the case of 2014 Ukrainian parliamentary elections, top-4 parties got 64% of the total votes, while the six parties that have obtained places in the parliament got 77%. Thus, there is free space for winner swing in a constituency in Ukraine and in Moldova there isn’t. Meanwhile, the magnitude of this swing (in percent) is about the same.

The rules outlined in p2 still apply in Moldova but in a very strange way. I.e. they apply properly where DPM wasn’t interested in the situation, but otherwise they aren’t just violated – they are reversed!

We’ve already considered situation in the Chișinău and have found that it’s irregular but the amount of these irregularities is limited for a number of reasons. Below we’ll discuss the two most similar cases, i.e Bălți and suburbs of Chișinău. This is even more important because in Bălți there is one more (fifth) political force, i.e. “Our party”, so there’s opportunity to test our assumptions on a larger selection of examples.

 

  DPM Acum PSRM Sor Other right
  list const. differ. list const. differ. list const. differ. list const. list const.
c32 (overall) 19.0 19,5   40.7 37.0 -3.7 23.8 24.7   4.5 4.2 6.5 11.4
Parts c32  
Cricova 25.3 20.3 -5 29.1 30.9 +1.8 30.0 33.3 +3.3 5.1 4.4 3.7 7.4
Vadul-l-Voda 12.9 11.5 -1.4 26.5 27.0   46.9 50.1 +3.2 4.6 3.7 2.7 4.3
Vatra 12.8 12.0   28.7 26.7 -2 41.4 43,7 +2.3 6.4 6.3 4.7 8.6
Cioresu 15.6 16.0   40.5 38.4 -2.1 29.3 30.8 +1.5 4.4 4.4 4.3 6.7
   
c33 (overall) 26.2 29.2 +3 44.7 46.3 +1.6 17.8 16.5 -1.3 3.2 2.8 3.9 5.3
Parts c33  
Codru 21.4 23.6 +2.4 44.5 46.4 +1.9 23.0 21.8 -1.2 3.5 2.9 3.7 5.8
Durlesti 21.5 23.7 +2.2 47.4 49.4 +2 19,7 21.4 +1.7 2.8 2.5 4.4 5.5

Table 1. Vote swing from party list to single candidates in Chișinău suburbs from c32 and c33 constituencies (difference less than 1% is not indicated).

 

  DPM Acum PSRM Sor Partidul Nostru
  list const. differ list const. differ list const. differ list const. differ list const. differ
c9 (overall) 12.3 12.2   13.5 12.7   36.4 33.4 -3 6.4 6.5   26.3 31.3 +5
Parts c9  
Part of Bălți (c9) 10.3 9.3   13.3 12.4   37.0 34.0 -3 6.3 6.4   28.1 33.9 +5.8
Falesti district part (c9) 26.9 29.5 +2.6 17.3 17.3   29.8 30.1   6.6 6.7   12.1 12.7  
   
C10 (overall) 13.4 12.6   14.0 13.5   36.5 32.3 -4.2 8.5 8.6   22.2 26.7 +4.5
Parts c10  
Part of Bălți (c10) 8.9 8.3   15.0 14.8   37.6 32.8 -4.8 6.4 6.9   27.5 35.1 +7.6
Singerei dist  part (c10) 21.1 21.3   10.7 11.0   36.8 38.0 +2 13.6 14.1   10.4 11.6 +1.2
Suburb part (c10) 33.5 33.2   17.2 20.1 +2.9 18.6 18.1   8.4 8.7   15.2 17.7 +2.5

Table 2. Vote swing from party list to single candidates in Bălți and nearest areas from c9 and c10 constituencies (difference less than 1% is not indicated).

 

Therefore, we’ve considered four most urbanized constituencies (except nine constituencies in the Chișinău which we’ve considered earlier) and we have got is an almost perfect illustration of transition from proportional to majority system.

  1. All four parties that have obtained places in the parliament tend to small transitions from one system to another. Most of these shifts are caused not even by the system change but presence of the “fifth force”, i.e. Dorin Chirtoaca (ex-mayor of Chișinău ) in constituency c32 and the “Our party” candidates in the constituencies c9 and c10.
  2. Our main rule precisely describes the situation. “Our party” and Chirtoaca get votes in a SMC while their parties lose votes in the proportional system. Chirtoaca takes votes from Acum and “Our party” takes them from PSRM.
  3. PSRM gains votes in constituency c32 (where it has got 2nd place) while losing votes in constituency c33 (where it has got only 3rd place).
  4. Now consider several important exceptions. Without apparent reason in the part of constituency с9 which is located in Falesti district DPM has got extra 2.6% of total votes. We stress this fact because the DPM candidate has got only the 4th place – even not the 3rd one. I.e. actually he hadn’t struggle for leadership and the voters (including those ones living in close rural/suburban area) had known that. In such circumstances this votes swing should imply existence of irregularities. In comparison, in the constituency c10 (where rural part is bigger) things described above didn’t happen.
  5. Therefore, in the constituencies c9, c10, c32 and c33 irregularities were relatively small, statistics was almost proper and there are no reasons to assume irregularities to influence election results in SMC. No reasons except one very important.
  6. The geography of two constituencies in Bălți (i.e. c9 and c10) has directly influenced the election results. It’s easy to see from the table 2 that is the election had been held only in the urban area, “Our party” candidates could easily win. So addition of parts of Falesti and Singerei districts had changed circumstances of the election in these constituencies in many ways.

One should also mention that the constituencies in Bălți are higher by 10-15%, than ones in Chișinău and higher than average around the country. So merge of parts of Falesti and Singerei districts to Bălți was a biased decision which could (or even was meant to) change conditions of electoral struggle in the constituencies involved.

Such actions obviously are responsible for some part of irregularities in election. These irregularities are created before the election with an apparent purpose of changing election results.

Moldovan Gerrymandering.

 

Gerrymandering is an intentional distortion of political geography aimed to create the most advantageous political configuration for a certain party. This notion exists in the strict sense as biased demarcation of constituencies. In the broader sense it can be any other kind of intentional distortion of political geography. In the case of Moldova, it’s all mixed up. Every possible kind of gerrymandering and even some kinds specific for Moldova have been used by the ruling party. For example, 11 constituencies in Chișinău (with suburbs) were completely separated from all neighboring districts. The same happened in Gagauzia (it could be agreeable as Gagauzia is an autonomous region and constituencies there are larger than in the rest of the country) and in Taraclia district (unlike the previous case it’s clear violation of voters’ rights). At the same time, in no other district such demarcation changes have occurred .

What’s even more weird is that Chișinău and Bălți while having same administrative status have become subjects to different approaches, i.e:

  1. Constituencies in Bălți had been larger by an average of 10.
  2. In Chișinău the integrity of municipal unit had been preserved, while in Bălți it had been violated.
  3. In Chișinău suburbs hadn’t been merged with urban constituencies, while in Bălți they had.

That alone is enough to raise reasonable doubt about the transparency of election, as there is violation of a fundamental right of any voter – equal conditions; however other violations are even more important. We’ll study some of them in more detail below.

 

Tricks with number.

 

Seemingly, what could be easier than dividing the country into 43 regular constituencies (that’s what will remain without the Left bank, diasporas, Gagauzia and Taraclia) and take them natural geographic shape? We’ll study shape in the next chapter, now let’s consider their size.

According to the CEC data average number of voters for all 43 regular constituencies is about 61.5 thousand. However, in case of different geographic zones this size can vary a lot.

  Number

of constituencies

 Mean, thousands Winners
DPM PSRM Acum Sor
North 10 65 3 7    
Center – North 6 62 5   1  
Ungheni – Calarasi 2 63 2      
Orhei 2 63       2
Chișinău 11 56,5   5 6  
Center – South 12 61 8 1 3  
  43 61,5 18 13 10 2

Table 3. Number of the voters in the different geographical areas of the Republic of Moldova.

 

Now we are able to ask a number of questions concerning “principles” that have determined policy of designers of this demarcation, and subsequently answer them. We’ll begin from discussion of two peculiars of the data presented in the table above. Ungheni and Orhei have been intentionally dealt with as separate categories. Orhei, which is under control of Șor Party for 4 years has therefore become a separate political region. Ungheni, which is located at the border between central and northern regions of Moldova is too. It is where one can discern principles that make possible to conclude such irregularities had been occurring in the election since the moment of the demarcation of the constituencies. We’ll study this issue more thoroughly in the end of this chapter.

  1. The sizes of the constituencies are random, although their volatility is geographically explicit, and that’s inacceptable. It’s possible that constituencies slightly vary in size as result of agreement of main parties. However, it’s absolutely unacceptable that difference of size of constituencies is related to the part of country it’s located in, while these parts vary in their political preferences.
  2. Chișinău is separated from the surrounding area, thus contradicting all rules of the constituency demarcation. Indeed, constituencies of Bălți aren’t separated from neighboring districts. Voters are on equal footing if borders of constituencies are the same than those of districts; otherwise they aren’t.
  3. Constituencies in Chișinău are smaller by 10% than country average and smaller by 15% than in the north of Moldova. So, 650 thousand of voters in the north for 10 constituencies while 625 thousand in Chișinău form 11 constituencies.

4.Why? The answer is obvious. In all the parts of country where DPM had pretended to victory average size of a constituency was smaller than in the north where it hadn’t.

  1. Chișinău is located such a manner that every addition of neighboring districts would make DPM unable to win in any of these “composite” constituencies. So, Chișinău had got one extra constituency (at the cost of north, where PSRM had been the favorite anyway). As a result, DPM won in 4 out of 7 constituencies located around Chișinău. In the case of merge of this constituencies with Chișinău this number would be much smaller. Most likely, there would be no such constituencies at all.
  2. Ungheni district have a high population (enough for 1.5 constituencies), as Ungheni is large (on local standards) city. In this city the DPM candidate was only 3rd. There is trend for some part of rural area of this district to have political preferences like north of the country. In the Călăraşi district Acum had been a certain favorite and unsurprisingly won. As Ungheni is situated near country’s border the only neighboring districts are Falesti and Călăraşi. Merge of Falesti and Ungheni would make PSRM undoubted favorite. So, very “precise” demarcation become needed. As a result, constituency c16 become the place of the tightest race in the whole country – difference between the winner (DPM) and party that got the 3rd place (Acum) was just 5%. A small change in demarcation could easily lead to a different winner.
  3. Orhei district could be composed from 3 constituencies instead of 2, especially taking into account that the only district of the left bank (i.e. Dubăsari district) was separated between two different constituencies. In this case, Șor Party would become a favorite just as in the constituencies c18 and c19. As a result, Orhei district was cut into parts of 4 different constituencies. In the two of them (c14 and c20) share of Orhei district was intentionally reduced to prevent Șor Party victory.

 

Conclusions.

 

  1. The approach to the constituencies demarcation described above is apparently biased and unacceptable in terms of voters’ equal rights. International observers should certainly had capture these violations at the very moment of their creation, i.e. long before the election.

 

  1. Simply by violation of quantitative equality of constituencies’ size DPM gained 3 or 4 additional mandates (1-2 at the expense of PSRM), one another at the expense of Șor party. Acum which won only in the center of country and in Chișinău had suffered less; however, if demarcation in the Chișinău had been less biased, Acum could potentially get one or two extra mandates.

 

  1. The violation of quantitative equality of constituencies’ is inherently connected to the electoral geography of this constituencies and districts. There was more bias in the constituencies where Acum and PSRM had been favorites.

 

Tricks with boundaries

 

We’ll consider two similar but not exactly the same kinds of gerrymandering which were used in Moldova at the same time.

In the first one (which was relatively rare) districts having electoral preferences undesirable for DPM were simply eliminated as a separate constituency. Due to the fact that regular (i.e. all 30 constituencies outside Chișinău  and Bălți) are usually larger than an average district, it was conceivable for the most of districts (or parts of them) to be merged with other ones. However, each of these cases provide an example of a biased decision, which inevitably leads to violation of the rights of voters.

 

There are only two correct solutions:

  1. Districts, municipalities and communities which have more than half of the population of the district don’t really matter at all. Size and borders should be negotiated by the main parties, that take part in the election in accordance with their shared understanding of equitable demarcation.

 

  1. Each of administrative territorial entities gets its own constituency. If needed, remaining parts of other districts can be added to equalize number of voters. In some countries rule of obligatory representation of each district (or any other territorial entity of highest rank lower than national) is provided for by low.

 

  1. For example, such rule exists in the USA, Russia, Ukraine, ets. Indeed, it’s hard to imagine that there wouldn’t be delegate from some certain state in the House of Representatives. Although the number of voters must be strictly equal and so a special committee revises constituencies’ borders before election. They often are merged or new ones are created; however, the total number of constituencies remain the same.

As a result of the CEC (and the ruling party) policy, 5 districts ceased to exist as separate constituencies. The results of voting on the proportional lists in these districts is presented in the table 4.

 

District Number of valid votes (by list) Number of fragments

(after division)

 Acum DPM PSRM Sor Party Winners in the constituencies
Ocnita 20858 2 6.7 20.9 55.8 7.9 PSRM, PSRM
Donduseni 17270 3 10.9 26.6 43.8 9.1 PSRM, PSRM, DPM
Soldanesti 16568 2 16.8 42.4 17.9 14.9 DPM, Acum
Dubasari 12093 2 19.2 17.2 34.2 17.0 Sor, Acum
Basarabeasca 9424 1 18.8 23.7 39.4 5.6 DPM

Table 4. Vote in the districts, dissolved by surrounding constituencies.

 

  1. In all the districts except Basarabeasca the number of voters significantly (in Ocniţa, Dondușeni and Şoldăneşti) or slightly (in Dubăsari) exceeds half of the number of voters in the whole constituency. So only in Basarabeasca elimination of the constituency can be justified by need to equalize the number of voters.

 

  1. PSRM gained a landslide victory in 4 districts out of 5 with a margin of at least 1.5 in each of them, which is usually enough to ensure victory in the whole constituency.

 

  1. As a result of gerrymandering, PSRM won only in 5 out of 10 constituencies formed in these districts, thus losing at least two mandates.

 

In the case of the second kind of gerrymandering, district isn’t completely eliminated. Instead, it’s divided and re-merged with the neighboring districts to change the winner. It’s more promising kind of gerrymandering, as it’s possible to exchange parts between 3 or even 4 districts instead of two in the previous case. We’ll define were these demarcation change biased or not. The cases of Ungheni and Orhei have been already studied in the previous chapter, so now we’ll consider other districts.

The scheme is simple. An increment (part of another district) gets merged with the main district on the basis of which a constituency is formed. The ratio of sizes of this increment and the main district is always 20 to 35%. Why? If lower, it could not influence the ultimate result. If higher, it could but only if the situation in this district is already flexile. The situation when increment is comparable in size with the main district is prone to getting out of control, so this way is used only if all other have failed.

There are special situations with small increments. This strange enclaves gets very dependent of the influence peddling. That’s because political activists dislike such pieces and are very reluctant to act in the alien constituency. In this situation the local authority (which had been controlled by DPM at the moment of the election) gets a monopoly for promotion of their candidate. It’s worth mentioning that difference of increment size can easily be controlled if one needs, for example, to outnumber their competitors greater. If modest advantage is enough, increment can be made smaller thus “cleaning up” the statistics. That’s why the CEC website wasn’t working during the night of election, and ultimate results hadn’t been available until one week since the election. I.e. “results first – and only then protocols”.

That way DPM was able to change results of election in 5 districts or even 7 CMS, which had served as “base” for constituencies demarcation. That’s the number of districts where the winner on the proportional system lost the election under the majority system to the DPM candidate.

 

Electoral system Acum DPM PSRM Sor Party Total
Winners by party lists in the boundaries of the single-member constituencies 5 11 12 2 30
Winners in the single-member constituencies 4 18 6 2 30
Difference -1 +7 -6 0 0

Table 5. Swing of the winner between party and single-member systems within the boundaries of constituencies (30 constituencies in area without Chișinău , Bălți, Left Bank, diaspora and Southern region (Gagauzia and Taraclia district)).

 

So, the swing of the winner happened in 7 out of 30 constituencies which had been created in the districts. It’s enormous, as it’s considered to be normal if it doesn’t exceed 10% (i.e. 3 constituencies in our case) not taking into account the independent candidates. In our case there had been none (Ion Groza in constituency c43 was actually a “covert” DPM candidate).

Such statistical violations in itself prove the existence of irregularities on the election. We’ll study only connection between districts and constituencies not present in the previous tables in analogy to the table 4. It’s easy to see that there’s only one real exception i.e. constituency c3 (entire Edineţ district) PSRM won on the proportional lists with 1.5% advantage over DPM while the DPM candidate won under the majority system with 3.5%. In this constituency there also had been irregularities but they were related to the voting process itself instead of gerrymandering.

 

District Number of constituencies Winner in district Winners in the constituencies
  within district without increment party advantage sum of the single  constituencies in the boundaries of district
Cahul 2 1 PSRM 8,1% Individual + DPM DPM, DPM (ind)
Calarasi 2 – Acum 3,2% Acum DPM, Sor
Drochia 2 – PSRM 7,1% PSRM DPM, PSRM
Singerei 2 – PSRM 3,1% PSRM DPM, PSRM (within c10 Bălți)

Table 6. Swing of the winner in the districts with not regular partitions and increments.

 

What we’ve got is an absolutely irregular picture. It should be accounted that both constituencies where PSRM won has their own specificity. Part of Sîngerei district is part of the constituency c10 of Bălți and constituency c4 (Rîşcani, Drochia and Dondușeni) is the only constituency where the winner changed after the results of the election had already been announced. One should also take into account that parties that won in these districts had a large advantage in this districts and significant in other ones.

Using the examples of Sîngerei and Drochia districts we’ll track the usage of this scheme of electoral irregularities. It’s worth noting that in these two districts PSRM won both on the proportional lists and under the majority system. In Drochia the advantage of PSRM under the majority system was smaller (3%) but still significant while in the case of Sîngerei it was the same as on the proportional lists.

This scheme acts very resembling this one described above. Usually in the Drochia district the leftists get much more votes than in Sîngerei. So creation of a constituency that incorporates the entire territory of the district had been an inherently very poor idea for DPM. Instead, it should had been divided that way that every increment would be able to change the election result. As a result, there had been created two constituencies with very strange geography – constituency c4 (which is made up of parts of Rîşcani district – 63%, 30% Drochia –and 7% completely dismantled Dondușeni district) – and c6 (it consists of 78% of parts of Drochia district, 20% of Soroca district and 2% of Dondușeni district). In the former case, ultimately the PSRM candidate had been declared winner. In the latter case, all the DPM advantage had been due to the “hidden” part of the Soroca district, in spite of PSRM landslide victory in the town of Soroca itself. So, DPM success had been based on advantageous demarcation of the Drochia district (which has 70000 voters i.e. much more than in an average constituency) and the irregular increment).

In the Sîngerei district, the demarcation was different although still advantageous for DPM due to the same principle. The smaller part of the district (which commonly votes for the leftists) was added to the constituency c10 while the larger one (with a 17% increment of the Florești district) became a separate constituency c12. Both parts had been composed of “DPM-friendly” communities and the district center. In case of the most unfavorable situation, the constituency c12 had been made with some “reserve”, thus assuring DPM advantage. Ultimately, we can see that PSRM had won in both districts (both under the majority system and on the proportional lists) while losing both of the constituencies where communities of the corresponding districts were located. In the Rîşcani district the situation is almost the same. The only difference was PSRM winning in the “base” constituency while losing in the constituency c5 (Glodeni) where Rîşcani district had served as an increment. So in spite of minor differences the situation was almost the same.

Conclusions.

  1. In the 2019 Moldova election all kinds of gerrymandering whether one could use at all in such a small country has been used, such as quantitative gerrymandering, biased demarcation of the constituencies, and the preventive elimination of constituencies potentially “hostile” to DPM.
  2. Simply by beneficial constituencies demarcation, DPM contributed for party success in constituencies c5 (Glodeni with Rîşcani as an increment, the losing side was PSRM), c6 (Drochia with Soroca and Dondușeni as an increment, the losing side was PSRM), c15(Călăraşi with Ungheni as an increment, the losing side was Acum), c16(Ungheni, the losing side was Acum or PSRM), c42(Cantemir with Cahul as an increment, the losing side was PSRM), c43(Cahul, the losing side was PSRM or Acum). It’s 6 constituencies out of 30 formed in all the districts, i.e. 20% only as a result of tricks with geography.
  3. In the highly urbanized constituencies DPM has encountered some difficulties. For example, in the most tossup constituency in areas (c16) PSRM and Acum got almost equal amount of votes thus leading to DPM victory with great difficulty. In the constituency c43 influence peddling in favor to the quasi-independent head of the district had been quite enough.

4.In addition, other 1.5 constituencies have been taken from PSRM by eliminating (as a separate election units) districts with a large number of PSRM voters in the north of country, and 0.5 from Șor Party due to division of Orhei district into 4 parts.

  1. In two other constituencies c3 (Edineţ) and c40 (Cimișlia and Basarabeasca) DPM has won because of direct influence peddling instead of gerrymandering.Partition of c40 is not absolutely clear, but may be possible.
  2. Simply by the apparent irregularities DPM has contributed 10 (out of 30 formed in all the areas) extra mandates in the SMC. It leads us to the conclusion about irregular character of the election in general.
  3. Everything described above is just the part of the irregularities which had been “preset” at the stage of constituencies demarcation. Another part which is connected directly to the voting process needs to be studied separately. It will be subject of the next chapter.

 

 

 

 

Share This