NEC Policy Paper No. 2 Establishing Electoral Districts

“Failure to address the issue of Election Districts will continue to contribute to a widening disparity in equality of the vote between and constituencies contrary to international standards”1

1. Context A country’s choice of election system can have a decisive bearing on which party secures the most seats or whether a party is represented in parliament. Since 2011, has used the First-Past-The-Post (FPTP) system for both House of Representative and Senate elections whereby the winning candidate requires a plurality of votes – also known as a simple majority – to be elected.2 Members of the two houses of parliament (MPs), are elected in single-seat constituencies (election districts). Four of the groups which observed the 2017 elections3 noted that there is a wide disparity in the number of voters registered among Liberia’s 73 House of Representatives electoral districts, with three stating that the election districting system in Liberia does not respect (or guarantee) the principle of equal suffrage.4 The European Union Election Observation Mission (EU EOM) stated that “Failure to address this issue will continue to contribute to a widening disparity in equality of the vote between counties and constituencies contrary to international standards.”5

While the number of seats allocated to five of Liberia’s fifteen counties were allocated in a way that approximately matches to their share of the population,6 two issues stand out:

1 The EU EOM to Liberia 2017, https://eeas.europa.eu/election-observation-missions/eom- liberia-2017/43557/final-report-eu-eom-liberia-2017-offers-23-recommendations-improvements-election- processes_en

2 The election system for House of Representative and Senate seats was changed through a referendum in August 2011. Previously, under the 1986 Constitution, in order to be elected candidates required over fifty per cent of the vote.

3 The Electoral Institute for Sustainable Democracy in Africa (EISA) the Election Observation Mission (EOM), the European Union (EU) EOM, the National Democratic Institute (NDI) EOM and the Carter Center (TCC) EOM.

4 The EU EOM noted a “perceptible inequality of the vote emerged on the basis of the numbers of registered voters in each for the 2017 elections.”. the Carter Center EOM stated that “the current constituency boundaries do not respect the principle of equal suffrage” https://www.cartercenter.org/resources/pdfs/news/peace_publications/election_reports/liberia-prelim- statement-101217.pdf, while the NDI EOM stated that the “the international principle of equal suffrage is not guaranteed”, while https://www.ndi.org/sites/default/files/ NDI%20Final%20Interational%20Election%20Observation%20Mission%20Report%20- %20Liberia%202017%20Presidential%20and%20Legislative%20Elections%20%282%29.pdf

5 EU EOM, op cit.

6 Bong; Grand Bassa; Grand Cape Mount, Grand Gedeh, Lofa and Maryland. 1 i) Seven counties7 are ‘over-represented’ and that most of these are among the most sparsely populated counties, and ii) Three counties8 are, based on the population numbers, under-represented with having significantly fewer seats than its proportional share of the population. Table 1 (below) sets out the actual allocation of seats at county level for the 2017 elections with additional columns for the number of districts a county would be ‘entitled to’ were the number of House seats to be allocated proportionally, based on i) voter registration totals9 and, ii) the 2008 Population and Housing Census.10

Table 1: Allocation of Election Districts within the

Number of Actual No. Number of districts districts if County Registered 2008 of if allocated allocated Voters Census Districts proportionally by proportionally by registration data census data Bomi 61,022 84,119 3 2 2 Bong 208,123 333,481 7 7 7 Gbarpolu 48,650 83,388 3 2 2 Grand Bassa 145,523 221,693 5 5 5 Grand Cape Mount 66,389 127,076 3 2 3 Grand Gedeh 63,202 125,258 3 2 3 Grand Kru 35,531 57,913 2 1 1 Lofa 167,427 276,863 5 6 6 Margibi 154,108 209,923 5 5 4 Maryland 57,140 135,938 3 2 3 Montserrado 778,291 1,118,241 17 26 23 Nimba 279,601 462,026 9 9 10 Rivergee 35,191 71,509 3 1 1 Rivercess 35,540 66,789 2 1 1 Sinoe 47,891 102,391 3 2 2 Tot al 2,183,62 3,476,60 73 73 73 9 8

Below the level of the county, the election district with the lowest voting population, River Gee No. 3 had 10,604 registered voters, while Montserrado No. 4 had 63,786 registered

7 Bomi; Gbarpolu; Grand Kru; Margibi, River Gee; Rivercess and Sinoe Counties.

8 Lofa, Montserrado and Nimba.

9 http://www.necliberia.org/pg_img/FinalVR_Report_2017.pdf

10 https://www.lisgis.net/pg_img/NPHC%202008%20Final%20Report.pdf 2 voters. This means that the voting weight of a voter in River Gee No. 3 is six times the weight of the vote of a voter in Montserrado No. 4. A full list of electoral districts ordered according to their voting weight (based on the number of registered voters), can be found at Annex 1. 2. International Instruments and Good Practice Guidelines 2.1 International Instruments Liberia ratified the International Covenant on Civil and Political Rights (ICCPR) in 2004. Article 25 of the (ICCPR) provides among other things that “every citizen of a state party to the ICCPR shall have the right and opportunity, without distinctions or unreasonable limitations […] to vote and to elected at genuine periodic elections, which shall be by universal and equal suffrage […]” (emphasis added). The ICCPR does not impose any specific type of electoral system on state parties11 but any system in operation must be compatible with the rights protected by Article 25 and must guarantee and give effect to the free expression of the will of the electors.12 In other words, the system should ensure among other things, equal suffrage. Equal suffrage is generally taken to mean that each voter has, in principle, one vote and where the electoral system provides voters with more than one vote, each voter has the same number of votes. In 1996, the UN Human Right Committee (UN HRC) issued General Comment 25 on Article 25 of the ICCPR. This stipulates that “the principle of ‘one person, one vote’, must apply, and within the framework of each State's electoral system, the vote of one elector should be equal to the vote of another” (emphasis added). General Comment 25 also states that “the drawing of electoral boundaries and the method of allocating votes should not distort the distribution of voters or discriminate against any group and should not exclude or restrict unreasonably the right of citizens to choose their representatives freely” (emphasis added). This is usually interpreted to mean that seats should be evenly distributed between the constituencies to ensure equal voting power between different electoral districts. 2.2 Good Practice Guidelines It is almost impossible to ensure precise mathematical equality in the number of electors between constituencies. Therefore, in many countries, legislation requires election districts to be ‘approximately equal’ and some even establish a permitted level of deviation from an average (also known as a ‘population quota’ or ‘threshold’). Usually, one of the following benchmarks is used to ensure a balanced distribution of seats among constituencies: population, number of resident nationals (including minors), number of registered voters, and on occasion the number of registered voters who voted in the last elections. However, the international and regional instruments offer little by way of guidance in establish what is a reasonable level of variation or deviation from the norm. In Europe, The Council of Europe has published a Code of Good Practice in Electoral Matters, which sets a benchmark figure: “[t]he permissible departure from the norm should not be more than 10% and should certainly not exceed 15% except in special circumstances”.13 The Code of Good Practice defines special circumstances as: i) the protection of a concentrated minority, and ii) in relation to a sparsely populated administrative entity.

11 A State party is a country which has ratified an international treaty, such as the ICCPR.

12 Paragraph 21 of General Comment 25 on Article 25, UN Human Rights Committee (1996).

13 The Council of Europe’s European Commission for Democracy through Law (Venice Commission), Code of Good Practice on Electoral Matters, 2002 (CDL-AD(2002)023rev2-cor). 3 In many countries, electoral laws establish additional criteria to ensure not just equal suffrage but also to avoid distorting the allocation of voters from communities to electoral districts. Commonly, the criteria for demarcating constituency boundaries include consideration of:14 ❖ Geographical features or natural barriers; ❖ Conformity of electoral districts with administrative divisions; ❖ Compactness e.g. prohibiting the establishment of constituencies with non-contiguous territory, and ❖ Historical, cultural and traditional boundaries (encompassing ‘communities of interest’). Electoral Boundaries should be reviewed periodically to reflect population changes. It is generally considered appropriate that boundaries should be reviewed at least every 10 years, and that the review preferably takes place outside election periods. Most commonly, responsibility for reviewing and/or redrawing boundaries is assigned to the legislator and/or the election commission, although some countries have established Boundary Commissions. Whichever body (or bodies) is/are involved in the delimitation of constituency boundaries, the process of re-drawing or re-defining constituency boundaries should be conducted: ❖ Impartially i.e. in a politically disinterested manner; ❖ In a non-discriminatory manner i.e. without detriment to national or other minorities; ❖ Using the most reliable available data; ❖ Ensuring expert input on technical, geographical, demographical and sociological aspects, and ❖ Inclusively and transparently e.g. involve consultation with political parties, civic associations and other interested parties. 3. Liberia’s Legal and Regulatory Framework and its Implementation 3.1 Legal and Regulatory Framework Article 80 of the Constitution of Liberia15 lays down several principles for demarcating electoral boundaries: 1. A clear requirement that the population of electoral districts should be approximately equal; 2. That the national census shall be used by the legislature as the basis for identifying population growth and movement and setting the population figure for electoral districts (threshold); 3. The number of electoral districts for the House of Representatives has an absolute ceiling of 100;

14 See ACE (the Electoral Knowledge Network) Project: http://aceproject.org/epic-en

15 Article 80 of the Constitution of Liberia stipulates that: “(d) Each constituency shall have an approximately equal population of 20,000, or such number of citizens as the Legislature shall prescribe in keeping with population growth and movements as revealed by a national census; provided that the total number of electoral constituencies in the Republic shall not exceed one hundred” and “(e) Immediately following a national census and before the next elections, the Elections Commission shall reapportion the constituencies in accordance with the new population figures so that every constituency shall have as close to the same population as possible; provided, however, that a constituency must be solely within a county.” 4 4. That while the Legislature prescribes the threshold, responsibility for reapportioning the electoral districts rests with the [National] Elections Commission (NEC), and 5. Electoral districts cannot be formed from parts of different counties. Liberia’s population has grown considerably since the adoption of the Constitution in 1986,16 and the figure of 20,000 and the maximum number of constituencies (100) would imply a maximum population of 2 million persons, whereas it is estimated that Liberia’s current population to be over 4.8 million. Therefore, the figure of 20,000 cannot be used and the Legislature is required to establish a population ‘threshold’. This threshold will, when divided by the total population, give the number of electoral districts that should be established by the NEC. Neither the Constitution of Liberia nor the legislation establishes how frequently the constituencies should be reviewed. In recent practice, they have been reviewed every twelve years, although the NEC is required to reapportion the election districts immediately after a census has been completed and the New Elections Law (NEL) provides that the NEC has the authority to do so “when deemed necessary and expedient”, although in practice it requires the Legislature’s cooperation in setting the threshold.17 The NEL also requires that any changes to constituency boundaries that will affect the subsequent election, must be published not later than 12 months before the election day unless the Commission announces, at least 120 days before the election day, that all necessary preparations can be made to allow the changes to apply to the election.18 Beyond these two provisions, the NEL does not provide any further guidance on the delineation of election districts boundaries. The NEC Regulations on ‘Establishing Electoral Districts’ provide the detail which is absent in the NEL.19 The Regulations provide that the following criteria are used for the purposes of demarcating electoral districts: ❖ No electoral district shall cross a county boundary, and electoral districts shall be contiguous as far as is practicable; ❖ The NEC shall endeavour, as much as is practicable, to have relatively equal numbers of registered voters in various electoral districts within each county; ❖ Additional criteria guiding the establishment of electoral districts may include, as much as practicable, geographical features, the existing administrative district, tribal/chiefdom boundaries, and communities of interest. The first and third provisions above are in line with international good practices. The second provision aims at achieving equal suffrage in electoral districts established within specific counties but, unlike the Constitution, does not aim at ensuring equality between electoral districts located in different counties.

16 According to the 1984 national census, Liberia had a population of approximately 2.1 million persons. This figure had grown to 3.4 million by 2008 and was estimated by the World Bank at 4.7 million in 2017: https://data.worldbank.org/country/liberia

17 NEL, §2.9, l

18 NEL, §4.1(3)

19 http://www.necliberia.org/document.php?&a4705305cd27e04fb1f66830e7e0ef9d=Mg%3D%3D 5 3.2 Implementation At the time of the 2005 elections, Liberia had not completed a census since 1984 and some clauses of the Constitution relating to elections were suspended20 thereby giving the NEC a legal base to establish electoral districts using voter registration data rather than population (census) data. For the 2005 House of Representatives elections, 64 electoral districts were established,21 However, these were never clearly defined by geographic boundaries. Instead, voting registration centers were simply amalgamated to form districts.22 In 2008, Liberia completed its first Population and Housing Census in 24 years. This established that Liberia had a population of 3,476,608. Montserrado had a population of 1,118,241 (or 32.2 per cent of Liberia’s entire population), while the smallest, Grand Kru had a population of 57,913 (or 1.7 per cent). Ahead of the 2011 elections, a protracted process took place to decide the threshold. Using the data of the 2008 Census, the NEC drafted a ‘Threshold Bill’ for consideration by the National Legislature,23 with the threshold set at 45,000, which implied 77 electoral districts (an increase of 14 electoral districts). In September 2009, the National Legislature set the threshold at 40,000, which implied 87 districts (an increase of 24). The Legislature also stipulated that no county should have less than two electoral districts. However, President Johnson Sirleaf declined to promulgate the law on the grounds that i) the corresponding number of seats in the House of Representatives (87) would place too heavy a burden on the nation’s finances and ii) because of a concern that guaranteeing each county with a minimum of two districts24 was not consistent with the Constitution.25 In response, the Legislature amended the Bill by removing the ‘two-district minimum’ clause, but retained the threshold at 40,000, leading to a second presidential veto. The delay in defining the threshold prevented the NEC from commencing its electoral preparations in a timely manner. The political impasse was ended in July 2010, through the National Legislature’s adoption of a Joint Resolution26 which, among other things, stated that the 2008 Census shall not be used as a basis for revising the number of Electoral Districts. The Joint Resolution did not however

20 The provisions were suspended by “The Electoral Reform Law - An Act Suspending Certain Provisions of the Constitution of Liberia and Amending Specific Sections of the New Elections Law 1986 and Approving New Provisions Relating to the Budget Appropriation of the National Elections Commission” of 17 December 2004. See also, NEC “Guidelines on the Establishment of Electoral Districts in Liberia” May 16, 2005.

21 Article 4.3 of the 2004 Electoral Reform Law (ERL) required each of the Liberia’s counties to be represented by not less than two seats in the House of Representatives.

22 “Statement on the Announcement of Final Electoral Districts for the 2011 Elections by Hon. James M. Fromayan, Chairman of the National Elections Commission (NEC) Friday, 8 July 2011 http:// www.necliberia.org/pg_img/Electoral%20Districts%20Final%20Electorla%20Districts.pdf

23 The House of Representatives and the Senate jointly assume responsibility to set the population threshold.

24 Article 4.3 of the 2004 Electoral Reform Law (ERL), required that each of the Liberia’s counties to be represented by not less than two seats in the House of Representatives. However, this law was not in force for the 2011 elections.

25 If the census data and the equality principle was applied, at least three counties (Grand Kru, Rivercess and Rivergee) could be allocated a single electoral district, and thus have fewer House of Representative seats (1) than Senate seats (2).

26 “Joint resolution LEG-002 (2010) of the Senate and the House of Representatives of the 52nd Legislature of the Republic of Liberia on the setting of an electoral threshold for the conduct of the 2011 Presidential and Legislative ” August 12, 2010. The Joint Resolution “mandates that sixty-four (64) electoral districts set up by the National Elections Commission (NEC) for the conduct of the 2005 legislative elections continue to be used but for the purposes of the 2011 Legislative elections, nine (9) additional electoral districts are prescribed and established making a total of seventy-three (73) electoral districts in the Republic of Liberia, which was approved by the and published into handbill.” 6 set a threshold but rather prescribed that the 64 Districts established by the NEC for the 2005 elections should continue to be used for the 2011 elections while stipulating that nine additional electoral districts should be established, bringing the total number to 73. In specifying the number of electoral districts rather than a threshold, the Legislature assumed the role assigned by the Constitution to the NEC.27 The wording of the Joint Resolution also meant that the NEC did not have the tools necessary to redraw the boundaries in a way that respected the equality of suffrage at national level between electoral districts. In April 2011, the NEC issued “Guidelines on the Establishment of Electoral Districts in Liberia” and a Regulation on “County Threshold for the Establishment of Electoral Districts”. The NEC Guidelines allocated the nine additional electoral districts to the most populous counties: ❖ Bong, Grand Bassa, Lofa and Margibi each receiving one additional seat (each increasing from four to five districts) ❖ Nimba received an additional two seats (increasing from seven to nine districts), and ❖ Montserrado received an additional three seats (increasing from fourteen to seventeen districts). However, the number of additional electoral districts (nine) was insufficient to achieve an equitable distribution at national level and Montserrado remained significantly under- represented. With seventy-three districts, the only way to remedy this would involve reducing the number of seats in the over-represented counties (Bomi, Gbarpolu, Grand Kru, Rivergee, Rivercess and Sinoe), and action that the NEC believed it was not, under the provisions of the Joint Resolution permitted to undertake. While the decision of the Legislature did not give the NEC the latitude necessary to ensure that the population size of electoral districts was approximately equal across the country, the Commission strived to establish districts with an approximately equal magnitude within counties. The NEC aimed for a maximum deviation of plus (+) or minus (-) 10% of the ‘county population threshold’.28 Following the adoption of this methodology, the Commission drew provisional electoral districts for each county using the delimitation criteria established. Annex 2 of this paper sets out the county threshold and the deviation from the county threshold in each election district. Of the 73 districts, 46 are within a range of 10% deviation from the county threshold. However, the Liberty Party (LP) and the Movement for Progressive Change challenged the authority of the NEC to alter the electoral district boundaries in all 15 counties. The political parties filed complaints with the Supreme Court arguing that the Joint Resolution only gave the NEC the authority to change the boundaries in the six counties that were allocated additional seats. In mid-June 2011 the Supreme Court dismissed the LP’s complaint and lifted a Stay Order which had caused the NEC to pause the demarcation process. As part of the boundary review process, the NEC also engaged in public consultation with traditional and local authorities and county elders across Liberia to identify if any serious concerns had been overlooked in its preparation of the provisional electoral districts. As a

27 On 8 July 2011, the former NEC Chair, The Hon. James M. Fromayan, stated that “the National Legislature prescribed 73 electoral districts for the 2011 elections, and as such, mandated the Commission to reapportion the districts accordingly, the Commission had no alternative, but to implement the mandate of the Joint Resolution”, Statement on the Announcement of Final Electoral Districts for the 2011 Elections, op cit.

28 The country threshold was calculated by dividing the total number of registered voters within the county by the number of seats allocated to the county. 7 result of the consultations, adjustments were made to five electoral districts.29 The NEC announced the final electoral districts for the October 2011 elections on 8 July 2011. As Joint Resolution LEG-002 was limited to the 2011 General Elections, on 19 July 2016, the NEC wrote the House of Representatives and the Senate seeking an electoral threshold on which it should conduct the 2017 General Elections. The NEC submitted that in order for it to satisfy the requirements of Article 80 (d & e) of the Constitution, and the principle of equal representation, the threshold for holding the 2017 General Elections should be based on census data and that if the census referred to in section 1(b) of the Joint Resolution LEG-002 will not be conducted before the holding of the 2017 General Elections, the NEC questioned whether the population figures from the 2008 census should be used. The Legislature declined to set a new electoral threshold for the 2017 elections on the grounds that a threshold should serve two regular periods of six years.30 4. Regional Situation Many countries that have an election system based on single seat electoral districts have difficulty in achieving an equitable balance of population and representation. In the ECOWAS region, six countries use the FPTP system (Côte D’Ivoire, the Gambia, Ghana, Liberia, Nigeria and Sierra Leone) while Mali uses the two-round system.31 The situation in Sierra Leone, Nigeria and Ghana is described below. Sierra Leone Article 38 of the Constitution of Sierra Leone provides detailed rules on the establishment of electoral districts. The Electoral Commission (EC) is mandated to conduct a revision of the electoral boundaries every five to seven years. The last census should be used to determine the population of Sierra Leone or if no census has been so held, “by reference to any available information, which in the opinion of the Electoral Commission best indicates the number of those inhabitants.”32 Sierra Leone’s Constitution requires that population of the electoral constituencies should be as close to the ‘population quota’ as is reasonably practicable. The ‘population quota’ as used in Sierra Leone and the ‘threshold’ as used in Liberia are different. The ‘population quota’ is obtained by dividing the number of inhabitants by the pre-determined number of constituencies into which the country is divided. In contrast, the ‘threshold’ determines the number of seats (electoral districts) by dividing the population number by the threshold. Changes to the election districting may be required if the parliament increases the number of Members of Parliament (MPs) and due to changes in the administrative division of the country. Any alteration to the election districts only comes into effect upon the next dissolution of Parliament after the alteration has been approved by Parliament. The Constitution of Sierra Leone also sets out the circumstances where the number of inhabitants of such a constituency may be greater or less than the population quota: “in order to take account of means of communications, geographical features, density of population, the distribution of different communities, the areas and boundaries of the

29 Nimba, Sinoe, Grand Cape Mount, Margibi and Lofa Counties.

30 The Chairman of the House Committee on Election and Inauguration, Rep. Gabriel B. Smith stated “After a careful examination of the request through a series of committee hearings and external consultations, the committee is convinced, that it would be legally unprecedented and politically inappropriate to prescribe a threshold on the basis of a census report that our legislative predecessors acted upon by a Joint Resolution; clearly instructing the NEC to maintain the sixty four electoral districts and add nine new seats reflecting an upward population in of the five counties,” https://www.liberianobserver.com/news/house-denies-new- threshold-for-2017-elections/

31 Guinea Niger and Senegal have parallel systems, which partly uses electoral districts.

32 Constitution of Sierra Leone, Section 38(7). 8 Chiefdoms and other administrative or traditional areas.”33 However, the legislation does not establish a permitted deviation range. Rather, this was set by the National Electoral Commission (NEC) at +/- 25 per cent. In 2016, the Sierra Leonian NEC revised election districts using the 2015 census data34 and in accordance with the parliament’s direction that there should be 132 seats; an increase of 20 from the previous election. The process was concluded in December 2016. However, at the end of January 2017, the government announced new administrative divisions. The Parliament approved the new divisions in March 2017. Consequently, the NEC had to completely rework its previously submitted constituency delimitation proposal. The revised delimitation process was finalized in July 2017. The EU EOM to Sierra Leone35 reported that “The number of parliamentary constituencies assigned to each [administrative] district based on the census population data using the largest remainder method was done correctly”. However, the EU EOM noted considerable discrepancies between the share of inhabitants and registered voters among individual wards and constituencies. These incongruities are also observed at district levels (which in Liberia correspond to counties).36 In other words, the numbers of voters registered did not correspond well with the census data. The EU EOM noted that “The deviations from the average number of registered voters (24,081), range from -57 per cent to +99 per cent.37 […] Had the number of constituencies been assigned to districts based on registered voters, instead of the census data, the capital and its suburbs would gain 9 additional parliamentary seats, at the expense of the Northern (-4), Eastern (-3), Southern (-1) and North-Western (-1) provinces.” In a similar vein to the 2017 EU EOM to Liberia, the EOM to Sierra Leone stated that “The current system produces significant discrepancies in the weight of the vote which contradicts the principle of the equal suffrage” and it goes on to recommend that “a new method for boundary delimitation, reflecting the number of registered voters, could be conceived”. Nigeria The Constitution of Nigeria provides that the House of Representatives shall consist of 360 members representing constituencies of nearly equal population as far as possible, provided that no constituencies shall fall in more than one State and that boundary delimitation is required every 10 years.38 Despite a census taking place in 2006, the constituencies for the 2015 elections have not been changed since 1996.39 The 2015 EU EOM reported that ‘there

33 Constitution of Sierra Leone, Section 38(3).

34 The 2015 census reported that Sierra Leone had at the time 7,092,113 inhabitants – a figure which was considerably higher than expected.

35 https://eeas.europa.eu/election-observation-missions/eom-sierra-leone-2018_en/46482/ EU%20EOM%20Sierra%20Leone% 202018%20Final %20Report,%2013%20June%202018

36 For instance, Western Area Urban, in which most of Freetown is located, accounts in the census for 14.9 per cent of the total population, yet its share of voters is 19.1 per cent. The share of registered voters on the overall estimated adult population varied significantly among districts, with the highest value of 127% in Western Area Rural (including the suburbs of Freetown) and the lowest share of 58% in Pujehun. Source EU EOM Final Report, op cit.

37 The EU EOM uses a different methodology to this paper in calculating the deviation to the average. If the same methodology is applied, there is a greater deviation between the largest and smallest population constituency in Liberia (46.9 per cent to 282.1 per cent) than in Sierra Leone (50.2 per cent to 229.3 per cent) – see Annex 1.

38 Constitution of Nigeria, Sections 49, 73 and 114.

39 Parliamentary sub-committees sat and/or INEC reports were published in 2002, 2008 and 2011 as well as for the 2015 elections. 9 are currently profound disparities in the size of constituencies […]. However precise calculation of constituency variations is not possible due to the lack of available population data (or registration data) broken down by constituency.”40 Nevertheless, EOM report cites examples of wide variations in the voting population of constituencies within States.

Ghana The Constitution of Ghana states that the number of inhabitants of each constituency should be, as nearly as possible, equal to the population quota.41 The Electoral Commission of Ghana (ECG) is required to delimit constituency boundaries at intervals of at least every seven years. During recent boundary reviews, the EC has tended to simply increase the number of constituencies each time by splitting existing constituencies rather than demarcating all constituencies with the objective of achieving equal suffrage.42 The ECG undertook a revision of boundaries just prior to the 2012 elections, increasing the number of constituencies from 230 to 275. In 2016, election observers noted marked disparities in the populations of constituencies e.g. Sekyere Afram Plains constituency in Ashanti Region has 13,976 registered voters and Ketu South in Volta Region has 142,008 registered voters.43 This situation in Greater Accra where 34 constituencies have at least two more registered voters than the quota, is similar to that in Montserrado where many constituencies are significantly under-represented. 5. Charting a Way Forward for Liberia 5.1 Using the Census Population Data vs Using Other Data Sources The delayed Population and Housing Census, which was due to take place in 2018, is expected to take place in late 2019. The conducting of the census on schedule, will allow the Legislature to set a new threshold and thereafter enable the NEC to undertake a review of the electoral district boundaries, most probably in the period 2020-2021. However, experience elsewhere has shown that significant and unexplained discrepancies between the census data and the number of registered voters can appear. These discrepancies can create controversies the effect of which is to lessen public confidence in the reliability of the data and the resulting boundary demarcation. As stated in the Constitution, and in conformity with international good practice, it is preferable to use census data, but only if the data is accepted as being reliable. However, if the Constitution is revised at some point in the future, it may be preferable to permit the election districting to be based on either census data or other data that is considered reliable e.g. voter registration data or data from the planned National Identity Register (NIR). This will mean that any failure to conduct a census, will not cause the electoral districts to be based on outdated information. 5.2 Equal Suffrage Two of the election observer groups that reported on the 2017 elections, proposed recommendations related to election districting:

40 http://eeas.europa.eu/archives/docs/nigeria/docs/eu-eom-nigeria-2015-final-report_en.pdf

41 Constitution of Ghana, Article 51.

42 Information referenced from the EU EOM Final Report on the 2016 Ghana elections, https:// eeas.europa.eu/headquarters/headquarters-homepage/21640/eu-eom-ghana-presidential-and- parliamentary-elections-2016-final-report_en.

43 EU EOM’s Final Report, op cit. 10 ❖ In redrawing legislative boundaries, adhere to the constitutional requirement that constituencies have an approximately equal population size (NDI). ❖ Delimitation of electoral constituency boundaries should be done in accordance with the Constitution and Liberia’s international obligations (EU). Both recommendations relate to the over-arching principle for establishing electoral districts, namely ensuring broadly equal suffrage. This principle is also firmly established in Liberia’s Constitution. While the NEC has strived to adhere to this principle in the apportioning of electoral districts within county boundaries, it can only achieve an equitable distribution of seats nationally, by either: 1) Reducing the current number of electoral districts in six less populated counties, which are currently ‘over-represented’, an action which could allow the number of districts (73) to remain unchanged, or 2) If election districts in these six counties cannot be reduced, then to allocate any additional seats to Montserrado. If the Legislature and the people of Liberia wish to ensure that less populated areas of the country are afforded minimum levels of representation in the House of Representatives, that this is clearly set out in legislation in a form that conforms to the Constitution, or if the Constitution is reviewed, that this requirement is clearly set out therein. To solidify the legal basis for the NEC’s actions, the Legislature could also consider establishing the periodicity for reviewing electoral boundaries and amending the NEL to establish clear criteria to guide the NEC in the apportionment of electoral districts, such as respect for administrative boundaries, contiguity of territory, respect for communities of interest. In the meantime, the NEC requests that after the census data has been publicly released the Legislator sets the threshold in a timely manner enabling the NEC to plan for re-districting to take place in 2021.44 If the threshold set by the Legislature allows for additional electoral districts to be established, the NEC is of the view that, depending on their number, most – if not all of these – should be allocated to Montserrado County as this action would significantly improve the equality of suffrage both in that county and at national level. Using 2008 census date, the following table illustrates how with 80 rather than 73 electoral districts the situation would be significantly improved through allocating a greater proportion to Montserrado:

Table 2: Example of Apportionment of Seats using 80 Electoral Districts

County 2008 Actual +/- to If 80 If 80 districts +/- to Census No. of proportiona Districts allocated proportio Districts l share proportionally nal share (73) Bomi 84,119 3 1 3 2 1 Bong 333,481 7 0 7 8 -1 Gbarpolu 83,388 3 1 3 2 1 Grand Bassa 221,693 5 0 5 5 0 Gr. Cape 127,076 3 0 3 3 0 Mount

44 According to NEL, §4.1(3). 11 Table 2: Example of Apportionment of Seats using 80 Electoral Districts

Grand Gedeh 125,258 3 0 3 3 0 Grand Kru 57,913 2 1 2 1 1 Lofa 276,863 5 -1 5 6 -1 Margibi 209,923 5 1 5 5 0 Maryland 135,938 3 0 3 3 0 Montserrado 1,118,241 17 -6 24 26 -2 Nimba 462,026 9 -1 9 11 -2 Rivergee 71,509 3 2 3 2 1 Rivercess 66,789 2 1 2 1 1 Sinoe 102,391 3 1 3 2 1 Tot al 3,476,60 73 0 80 80 0 8

If Liberia’s population is approximately 5 million at the time of the census, 80 seats would require the Legislature to set the population threshold at 62,500. 5.3 Temporary Special Measures International legal instruments45 permit temporary special measures to be applied to overcome de facto obstacles to the equal representation of women in legislatures (see Policy Paper No. 1). These may include the establishment of quotas and reserved seats. As such, the application of temporary special measures can have an impact on the election districting system. If the Affirmative Action for Equitable Participation and Representation Act of 2016 (known as the Equality Bill) is enacted into law, the Legislature should clearly set out the details on whether the reserved seats will be directly elected and, if so, how the elections for these seats should organized. This may require amendment of the NEL.

45 Notably, article 4 of the Convention on the Elimination of Discrimination against Women (CEDAW). 12 Annex 1: Election Districts Ordered according to their Voting Weight

13 District Registere Above/ Percentag Vote d Voters Below e of Weight average average Over River Gee 3 10604 19309 282.09% 2.82 - River Gee 1 12210 17703 244.99% 2.45 River Gee 2 12377 17536 241.68% 2.42 Repr Sinoe 3 14152 15761 211.37% 2.11 esent Sinoe 2 14820 15093 201.84% 2.02 ed Gbarpolu 1 15257 14656 196.06% 1.96 Grand Kru 1 15807 14106 189.24% 1.89 Gbarpolu 3 16167 13746 185.02% 1.85 Bomi 2 16889 13024 177.11% 1.77 Gbarpolu 2 17226 12687 173.65% 1.74 River Cess 1 17712 12201 168.88% 1.69 River Cess 2 17828 12085 167.79% 1.68 Maryland 3 17993 11920 166.25% 1.66 Maryland 1 18320 11593 163.28% 1.63 Sinoe 1 18919 10994 158.11% 1.58 Grand Cape Mount 1 19364 10549 154.48% 1.54 Grand Gedeh 3 19477 10436 153.58% 1.54 Grand Kru 2 19724 10189 151.66% 1.52 Grand Gedeh 2 20370 9543 146.85% 1.47 Maryland 2 20827 9086 143.62% 1.44 Bomi 3 21107 8806 141.72% 1.42 Bomi 1 23026 6887 129.91% 1.30 Grand Cape Mount 2 23027 6886 129.90% 1.30 Grand Gedeh 1 23355 6558 128.08% 1.28 Grand Cape Mount 3 23998 5915 124.65% 1.25 Margibi 5 25246 4667 118.49% 1.18 Repr Bong 6 26186 3727 114.23% 1.14 esent Nimba 5 26190 3723 114.21% 1.14 Nimba 9 26706 3207 112.01% 1.12 ed Bong 2 26855 3058 111.39% 1.11 Prop Grand Bassa 1 26959 2954 110.96% 1.11 ortio Grand Bassa 5 27089 2824 110.42% 1.10 nally 14 ortio nally Lofa 1 27348 2565 109.38% 1.09 Margibi 3 27377 2536 109.26% 1.09 Grand Bassa 2 27694 2219 108.01% 1.08 Nimba 6 27866 2047 107.34% 1.07 Bong 4 28323 1590 105.61% 1.06 Lofa 3 28356 1557 105.49% 1.05 Margibi 4 29356 557 101.90% 1.02 Grand Bassa 4 29468 445 101.51% 1.02 Bong 7 30377 -464 98.47% 0.98 Bong 5 30678 -765 97.51% 0.98 Nimba 8 31024 -1111 96.42% 0.96 Bong 1 31604 -1691 94.65% 0.95 Margibi 2 32514 -2601 92.00% 0.92 Nimba 2 32601 -2688 91.75% 0.92 Nimba 3 32927 -3014 90.85% 0.91 Nimba 7 33251 -3338 89.96% 0.90 Nimba 1 33583 -3670 89.07% 0.89 Lofa 5 33799 -3886 88.50% 0.89 Bong 3 34100 -4187 87.72% 0.88 Grand Bassa 3 34313 -4400 87.18% 0.87 Unde Nimba 4 35453 -5540 84.37% 0.84 r- Lofa 2 36319 -6406 82.36% 0.82 Montserrado 8 36826 -6913 81.23% 0.81 Repr Montserrado 13 38950 -9037 76.80% 0.77 esent Montserrado 15 39005 -9092 76.69% 0.77 ed Margibi 1 39615 -9702 75.51% 0.76 Montserrado 9 41516 -11603 72.05% 0.72 Lofa 4 41605 -11692 71.90% 0.72 Montserrado 1 41769 -11856 71.61% 0.72 Montserrado 16 41898 -11985 71.39% 0.71 Montserrado 10 42861 -12948 69.79% 0.70 Montserrado 7 43429 -13516 68.88% 0.69 Montserrado 11 44589 -14676 67.09% 0.67 Montserrado 3 45323 -15410 66.00% 0.66

15 Montserrado 14 45617 -15704 65.57% 0.66 Montserrado 12 46313 -16400 64.59% 0.65 Montserrado 5 46580 -16667 64.22% 0.64 Montserrado 2 50143 -20230 59.65% 0.60 Montserrado 6 53882 -23969 55.52% 0.56 Montserrado 17 55804 -25891 53.60% 0.54 Montserrado 4 63786 -33873 46.90% 0.47

The average number (arithmetic mean) of registered voters is 29,913. This figure is calculated by dividing the total number of registered voters in Liberia (2,183,629) by the number of Electoral Districts (73). The third column, ‘above or below average is calculated by subtracting the actual number of registered voters from the average. The percentage (column 4) is calculated by dividing the average by the number of registered voters. Similarly, column 5 expresses the percentage as a voting weight. At the extremes, in terms of representation, the weight of a vote in River Gee 3 is 2.82 times the average voting weight, while the weight of a vote in Montserrado 4 is 0.47 the national average. For the purposes of defining if voters in an electoral district are over-represented, proportionally represented or under-represented the benchmark of =/- 15 per cent of the average is used. On this measure, 26 (35.5%) districts are over-represented, 26 (35.5%) are represented proportionally and 21 are under-represented (29%).

16 Annex 2: Population by Election District

17 District Reg. +/- Deviation District Reg. +/- Deviatio Voters average Voters average n

Bomi 1 23,026 2,685 13.20% Gbarpolu 1 15,257 -960 -5.92% Bomi 2 16,889 -3,452 -16.97% Gbarpolu 2 17,226 1,009 6.22%

Bomi 3 21,107 766 3.77% Gbarpolu 3 16,167 -50 -0.31% Tot al 61,022 Tot al 48,650

County Threshold 20,341 County Threshold 16,217

Bong 1 31,604 1,872 6.30% Grand Bassa 1 26,959 -2,146 -7.37%

Bong 2 26,855 -2,877 -9.68% Grand Bassa 2 27,694 -1,411 -4.85% Bong 3 34,100 4,368 14.69% Grand Bassa 3 34,313 5,208 17.90%

Bong 4 28,323 -1,409 -4.74% Grand Bassa 4 29,468 363 1.25% Bong 5 30,678 946 3.18% Grand Bassa 5 27,089 -2,016 -6.93%

Bong 6 26,186 -3,546 -11.93% Tot al 145,523 Bong 7 30,377 645 2.17% County Threshold 29,105

Tot al 208,123 County Threshold 29,732

Gr. Cape Mount 1 19,364 -2,766 -12.50% Grand Gedeh 1 23,355 2,288 10.86%

Gr. Cape Mount 2 23,027 897 4.05% Grand Gedeh 2 20,370 -697 -3.31% Gr. Cape Mount 3 23,998 1,868 8.44% Grand Gedeh 3 19,477 -1,590 -7.55%

Tot al 66,389 Tot al 63,202 County Threshold 22,130 County Threshold 21,067

Grand Kru 1 15,807 -1,959 -11.02% Maryland 1 18,320 -727 -3.82%

Grand Kru 2 19,724 1,959 11.02% Maryland 2 20,827 1,780 9.35% Tot al 35,531 Maryland 3 17,993 -1,054 -5.53%

County Threshold 17,766 Tot al 57,140 County Threshold 19,047

Lofa 1 27,348 -6,137 -18.33% Margibi 1 39,615 8,793 28.53%

Lofa 2 36,319 2,834 8.46% Margibi 2 32,514 1,692 5.49% Lofa 3 28,356 -5,129 -15.32% Margibi 3 27,377 -3,445 -11.18%

Lofa 4 41,605 8,120 24.25% Margibi 4 29,356 -1,466 -4.76% Lofa 5 33,799 314 0.94% Margibi 5 25,246 -5,576 -18.09%

Tot al 167,427 Tot al 154,108 County Threshold 33,485 County Threshold 30,822

River Cess 1 17,712 -58 -0.33% River Gee 1 12,210 480 4.09%

18 District Reg. +/- Deviation District Reg. +/- Deviatio Voters average Voters average n

River Cess 2 17,828 58 0.33% River Gee 2 12,377 647 5.51% Tot al 35,540 River Gee 3 10,604 -1,126 -9.60%

County Threshold 17,770 Tot al 35,191 County Threshold 11,730

Montserrado 1 41,769 4,013 -8.77% Sinoe 1 18,919 2,955 18.51%

Montserrado 2 50,143 4,361 9.53% Sinoe 2 14,820 -1,144 -7.16% Montserrado 3 45,323 -459 -1.00% Sinoe 3 14,152 -1,812 -11.35%

Montserrado 4 63,786 18,004 39.33% Tot al 47,891 Montserrado 5 46,580 798 1.74% County Threshold 15,964

Montserrado 6 53,882 8,100 17.69% Montserrado 7 43,429 -2,353 -5.14% Nimba 1 33,583 2,516 8.10%

Montserrado 8 36,826 -8,956 -19.56% Nimba 2 32,601 1,534 4.94% Montserrado 9 41,516 -4,266 -9.32% Nimba 3 32,927 1,860 5.99%

Montserrado 10 42,861 -2,921 -6.38% Nimba 4 35,453 4,386 14.12% Montserrado 11 44,589 -1,193 -2.61% Nimba 5 26,190 -4,877 -15.70%

Montserrado 12 46,313 531 1.16% Nimba 6 27,866 -3,201 -10.30% Montserrado 13 38,950 -6,832 -14.92% Nimba 7 33,251 2,184 7.03%

Montserrado 14 45,617 -165 -0.36% Nimba 8 31,024 -43 -0.14% Montserrado 15 39,005 -6,777 -14.80% Nimba 9 26,706 -4,361 -14.04%

Montserrado 16 41,898 -3,884 -8.48% Tot al 279,601 Montserrado 17 55,804 10,022 21.89% County Threshold 31,067

Tot al 778,291 County Threshold 45,782

The county threshold is calculated by adding the number of all registered voters in the county and dividing by the number of seats. The figure +/- average denotes the number of registered voters in the district above or below the county average. The column ‘deviation’ is the percentage deviation from the average.

19