Fr David Arms on the Open List System: Everything You Need To Know

Fr Arms (r.)

Fiji's Proposed New Open List PR System: Resolving Some Details

Version 3, by Fr D G Arms, 26th June 2012

0.0    Introduction
0.1    This paper is written as a follow-on to two previous papers prepared for the National Council for Building a Better Fiji (NCBBF) in 2008, namely The Urgency for a Change of Electoral System and Electoral Reform – the Way Forward. Although these papers went into a certain amount of detail regarding how an Open List system might operate in Fiji, this one goes into more detail again and takes up some of the more pragmatic issues of how the system might be administered.
0.2    The original proposals did not tie everything down for two reasons. Firstly, the main goal was to get acceptance in principle to the proposals. If too much detail was presented, readers (especially those not familiar with electoral matters) could easily get lost in the morass of issues to be resolved. Secondly, it seemed wise to leave certain issues open so that more people could make contributions towards finalizing a system suitable for Fiji.
0.3    As it happened, the consultation phase after the Charter had been completed did not throw up many questions needing resolution. This is partly because the consultation process was concerned with the whole Charter, so only limited time could be devoted to the electoral system. There were at the time some generalized criticisms of the system as being “confusing” or “cumbersome” or “divisive” but no back-up evidence or explanation for such remarks were provided. In any event, answers to queries that were raised about the system are contained in my two previous papers on the topic, plus this one. However, if there are further questions being consistently asked or that need to be resolved, the next stage would be to assemble such questions and provide a separate paper answering them.
0.4    This paper, then, will go into some detail as to how the Open List system being proposed for Fiji might operate in practice. As part of the electoral reform, the Charter process proposed a number of constitutional changes. This paper presumes on these changes being made. No comment will be made on them apart from what has already been said in the two papers mentioned above and what is said in this paragraph. The changes I am presuming will take place are:
The abolition of compulsory voting. While citizens should be strongly encouraged to vote, making it compulsory is undesirable and has (wisely) not been enforced. It is an unhealthy and unjustified borrowing from Australia.
Reduction of the minimum voting age from 21 to 18. This change got good endorsement from the Charter consultation process, which stands to reason. If people aged 18 can get married, pay taxes and serve in the military for their country, it is hard to see why they should be denied a vote.
Abolition of the communal seats. This was one of the main features of reform put forward immediately after the 2006 coup. The reasons have been extensively argued elsewhere, including the two papers referred to above.
Reduction of the size of the House of Representatives by 25%. During compilation of the Charter, a reduction of between 15% and 25% was suggested. I have adopted here the maximum of 25% because it seems the more logical figure. It yields a total of 54 MPs, which is very close to the number of MPs (52) in the Lower House prior to the 1990 Constitution. Even with this figure, each MP is representing only about 10,000 people, which is a very reasonable ratio.

1.0    Voter Registration
1.1    Voter registration will begin next month. Under current circumstances I have no problem with Government organizing and conducting this process through a hired company and involving advisors with expertise in this area, including the Elections Office. Nevertheless, it is unfortunate that the Electoral Commission and Supervisor of Elections are still not appointed. The task of voter registration is normally theirs, and it is surely likely to be conferred on them again by the new Constitution. Even if Government conducts the registration in this first instance (especially since the voter registration card might be useful for more than just elections), the Electoral Commission and Supervisor of Elections would benefit greatly from experiencing how it is done. They could draw lessons for further registration, and especially gain valuable insights into how they should organize the logistics and timing of the polling process. Their appointment now would be very useful in other ways too, for although the electoral system to be used must await the Constitution–making process, there are numerous other preparations that they could be doing, and even making contributions about to the Constitutional Commission (e.g. rules for political party funding, rules for election campaigning, rules for broadcasting, acquisition of more up-to-date ballot boxes).
1.2    The voter registration being done is a form of Electronic Voter Registration (EVR). A decision for EVR had already been made in principle by the last Electoral Commission. The objective of getting it all done in two months will be hard to achieve, but in any event, provision will need to be made for late registration. With elections not due until 2014, it would seem inevitable that, at the very least, some 16-year–olds might not get themselves registered now. The teams to carry out EVR (both late registration for 2014 and registration in the future for subsequent elections) should be of two types: mobile and fixed. The mobile teams would be deployed in the islands and rural areas, and the fixed teams in a few of the urban and/or peri-urban areas. If people miss out on registration when the mobile teams visit their island or area, they would be expected to go to one of the fixed centres.
1.3    The electoral reform proposals of the Charter advocate the abolition of compulsory voting but the retention of compulsory registration. I am not enthused about either of these being compulsory (though certainly they should both be strongly encouraged), but if the voter registration card is in fact to serve like an identity card for other purposes too, compulsory registration becomes more understandable. Certainly, penalizing people for not having the card, or not being registered, should proceed cautiously, at least until the populace becomes accustomed to the idea.
1.4    It has been proposed that, in general, people will vote only in designated polling stations, not in any polling station within the constituency (or even outside it), which was the practice under the previous voting system, the Alternative Vote (AV). It has been further proposed that registration stations be set up in the same locations as the polling stations will be. This would familiarize voters with their proper polling station. Having just one designated polling station for each voter would certainly have many administrative advantages (see Section 5 below), but this rule would need to be repeatedly and forcefully publicized, as it is quite different from the very free use of polling stations allowed under AV. However, making registration in these places also a strict requirement would seem to be too severe, since registration (unlike voting now) being compulsory. It is not good practice to impose a penalizable obligation and then make that obligation difficult to fulfil. Furthermore, looking ahead, it would be both more efficient and more convenient in the urban and peri-urban areas to set up a few centrally located registration centres manned by professionals, where people could come to register over an extended period.
1.5    The actual registration process should be kept as simple as possible. A person’s photo and thumb-print should be recorded, and only a limited amount of other data. On completing registration, the voter should be provided with a card containing his name, photo, and some of the other data, in particular the constituency, designated polling station and unique identity number. To cater for overseas voters, facilities should be set up at Nadi airport as soon as practicable to register voters as they leave the country. Special facilities may be made available overseas in some locations, e.g. for the RFMF and in major embassies or high commissions, but otherwise such people will miss out unless they have managed to register in Fiji prior to departing.
1.6    Once registration is completed, a roll will be produced. After a set period for corrections and late registration has expired, no further registrations should be made until after the election (for the following election). In doing registration at any time, the need for many or most 16- and 17-year-olds to register will need to be emphasized. Anybody who will turn 18 on or before voting day needs to be registered. If some 16- or 17-year olds wish to register whose 18th birthday occurs after the polling date, their registration should be accepted. The system for electronic registration will weed out those who cannot vote in the election concerned when printing out the roll. But there is no harm in being already in the database for later elections.

2.0    Constituencies
2.1    The current proposal of the NCBBF is to have between three and five multi-member constituencies based around the present four administrative Divisions. If there were four constituencies tied strictly to the Divisions, they would be of very different sizes, namely: Central 22 seats, Western 21, Northern 9, and Eastern 2.
2.2    The advantages of this arrangement are that the borders of these Divisions (constituencies) are already clear and accepted throughout the country. There would therefore be little opposition to them. There would also be so little work in confirming constituency boundaries that the Constituency Boundaries Commission (CBC) could be disbanded as redundant, thus reducing cost.
2.3    The disadvantages of this arrangement are that although the number of seats allotted per constituency is fair, the large difference in constituency size has an uneven effect on representation. With only two seats, the Eastern Division can hardly be said to be getting proportional representation at all. For a small party to get a seat there, it would need to be able to pull about 34% of the vote! In the Central Division on the other hand it would only need to pull about 4½%. This uneven effect is unfair to small parties and their supporters in the Eastern Division. The Northern Division also suffers in this way, but not nearly so badly.
2.4    A solution to this problem would be to join the Eastern Division to the Central Division. Although the area concerned would not have clearly separate representation, the lower effective threshold of 4% (see 3.3 below) for the Central Division and the fact that many people from the Eastern Division have migrated to the Central Division would ensure genuine representation for Eastern Division voters. In particular, Rotumans are far more likely to get a Rotuman candidate elected through being attached to the Central Division rather than to the Eastern.
2.5    There are two possible objections to the proposals above. One is that we still have three uneven constituencies with the problems for representation and fairness mentioned in 2.3. The second is that constituencies with 20 or more members are going to provide a large ballot paper with perhaps more than 100 names on it. Although still easy to fill out – just one name to be ticked – the large number of names may be daunting to some voters. I don’t feel there is much merit to this last objection. Many countries with populations less well educated than Fiji’s use large ballot papers without serious problems.
2.6    Nevertheless, a solution that attempts to address both objections is at least worth a look. If we left the Northern Division as it is but cut both the Western and Central Divisions in two, we could generate five constituencies with more even numbers of representatives, say between 8 and 14 each. The positive outcome of this re-division is that 1) parties have roughly equal chances in all five constituencies, 2) the effective threshold, though higher (between 6% & 11%), will stop excessive fragmenting of the parliament, and 3) the ballot paper will be of smaller and very manageable size.
2.7    However, rather convincing objections can be made to this five-constituency model too. Firstly, there is the loss of simplicity in determining the constituency boundaries. The CBC could not be dispensed with after all, and would have the job of dividing the two big Divisions in an acceptable way. But an even more serious difficulty comes with the threshold for election. The figure of 6% to 11% given in 2.6 is quite high, making things too difficult for smaller parties (see Section 3 below). All in all, the three-constituency model is definitely preferable, in my view.

3.0    Proportionality & Threshold
3.1    Proportional Representation (PR) provides parties with seats in accordance with their proportion of the vote, but that proportionality can differ according to the size of constituencies employed and the method used for calculating the proportions. The larger the constituency, the more proportional the results will be. Some countries (Holland, Israel, and Paraguay) therefore treat their whole country as one giant constituency. Even where a country is divided into several constituencies, it is possible to calculate the proportion of seats per party on the basis of the vote in the whole country, not merely on the basis of the respective proportions in each constituency. This is likely to mean a small increase in the allocation of seats to smaller parties. A practical way of catering for this in Fiji’s case would be to reserve 4 seats for such readjustments, dividing up only 50 (not 54) among the constituencies. The results for each party in the constituencies would be compared with what the proportions should be nationally, and the four top-up seats would be used to bring the final breakdown as close as possible to the national proportions (see my Paper 2, ¶ 6.8).
3.2    But there are counter-arguments too. Many people feel that by adhering to strict proportionality, a country can bring upon itself a parliament containing many small parties and independents. When it comes to forming a government and making policy decisions therefore, the negotiations can become extremely complex. A small group can have too much influence as against a much larger group. There are other people, of course, who disagree with this argumentation, regarding full representation and respect for minority views as paramount. The point is certainly arguable. In practice however, many countries have put in place a threshold, either explicitly or implicitly. That is, they have regulations that see to it that no party with less than 5%, say, of the vote will win a seat.
3.3    The system advocated here is the three-constituency model (as mentioned in 2.7). It does not have an explicit threshold but an implicit one (also known as an ‘effective’ threshold). It does not add the complication of calculating national proportions described above, but calculates the proportions on the basis of each constituency only. This yields an effective threshold of 4% for the Central-Eastern constituency, 5% for the Western constituency, and 10% for the Northern constituency. These are very reasonable figures. While the percentage for the Northern constituency may seem high, smaller parties with wide appeal will not be excluded, as they should be able to get someone elected at least through the larger two constituencies.
3.4    What the above discussion illustrates is how the various criteria need to be balanced off against each other. Proportionality (fairness to everybody) is one criterion, preventing excessive splintering of the parliament is another, and keeping the system simple is a third. Weighing together all such factors, the three-constituency model looks very attractive. The constituencies will not be excessively big, so the ballot paper will be of reasonable size. The effective threshold is quite moderate, assuring good proportional representation. The operation of the system is simple, and can be conducted by the respective Divisional administrations without the need for separate national calculations or consultation. The CBC can indeed be disbanded, as constituency decisions will be so straightforward that they can be handled by the Electoral Commission alone.
3.5    At the beginning of this section there was mention of the formula for calculating the proportions. The formula recommended here for allocating the number of seats to to be given to each party is the Modified Sainte Laguë (or Arithmetic Mean) method. This is a divisor method and ensures that results are monotonic and maximally fair, favouring neither small nor large parties.
3.6    In conducting the prior (see 3.7) process of allocating the right number of seats to each of the three constituencies, I recommend the unmodified or “Pure” Sainte Laguë method. It is good to retain basically the same method (Sainte Laguë) for determining the proportions in both cases.
3.7    The number of seats per constituency should be finalized just a month or two before the election. The calculation should be made on the basis of the number of registered voters in the constituency – hence the requirement in 1.6 that registration should not go on too long. Adding late registrants could, of course, change the proportions. Candidates need not be resident in the constituency for which they are running.

4.0    Ballot Papers
4.1    Two types of ballot paper can be suggested. The first type would normally consist of just one A3 page (possibly sometimes larger for the big constituencies, or it might be felt that two pages would be better). The voter instructions would be written in English, Fijian, Standard Hindi (in Devanagri script), Fiji Hindi (in Roman script) and Rotuman. They should be kept as short and simple as possible, such as: “Vote by writing one tick in the box to the right of one candidate’s name only”.
4.2    Strong black lines will act as a border inside which all the lists of candidates will be contained. The lists should be presented vertically with strong black lines between each list, and a much lighter line to the right of the names of each list. Light lines will also be drawn horizontally across the ballot paper under each name.  (The intersection of these lines will create the boxes where voters are to tick.) The name and symbol of each party will be placed on the top of the ballot paper at the head of their respective lists but outside the strong black border mentioned above. No box is to appear beside the party name or symbol. The order of the lists across the page should be random. (See the sample ballot paper in the Appendix.)
4.3    The candidates’ names on each list will be presented in alphabetical order. Random order would be marginally fairer, but it seems preferable to make it as easy as possible for voters to find the name they want. No party symbol will be placed next to the individual names (even though this has been the practice to date). The symbol at the top of the list should suffice. The names on the lists will not be numbered nor will photos of the candidates be included. It is important to keep the ballot paper as uncluttered as possible. A party may not run more candidates in a constituency than there are seats to be won. It may run less, in which case the later places on its list will be blank.
4.4    If a voter ticks a candidate’s name instead of the box to its right, it is accepted. If a voter ticks the party’s name or symbol (at the top of the list) instead of a candidate’s name, it will normally be accepted as a vote for the party – that is, it will count for determining the number of seats to be won by the party, but will not count for any particular candidate. However, only 5% of the total vote for that party in that constituency may be votes of this type. Any votes in excess of this percentage will be deemed invalid.
4.5    A tick entered in the left or right margin of the ballot paper is invalid. If it is entered in the top or bottom margin, it will be interpreted as a vote for the party, provided it is clearly over or under one list or other. In the case of lists containing less than the full number of permissible candidates, a vote below the list will be treated the same as a vote in the lower margin. If a voter enters two or more ticks for different candidates, the vote will be deemed invalid unless all the ticks are in the same list and/or are all above or below one party’s list (on the party name or symbol, or in the upper or lower margins), in which case they will be interpreted as one vote for the party. If the two or more ticks are placed beside one candidate’s name or are split between one candidate and the party name or symbol and/or the upper or lower margins, the vote will be interpreted as a vote for the candidate (as well as the party).
4.6    If a voter uses a symbol other than a tick, it (or they, provided only one type of symbol is used) will be interpreted as if they were a tick or ticks. However, if a voter mixes one or more ticks with one or more other symbols, only the tick(s) will be accepted if it (they) accord with the rules above. Note than any vote that is counted for the party only but not for a candidate, is only valid if it falls within the 5% limit prescribed in 4.4.
4.7    It may be preferred that a smaller ballot paper be used. This second type of ballot paper would be somewhat like that used in Finland. There would be just one ballot paper, of A5 size, used for all three constituencies. Each candidate is allotted a three-digit number by the Supervisor of Elections. The ballot paper contains an empty oblong box above which are written the voting instructions (using the same languages as indicated in 4.1) as follows: “Vote by writing the three-digit number of your preferred candidate in the box below.” (See the second sample ballot paper in the Appendix.) Three-digit numbers written outside the box, or half in and half outside, would in fact be accepted as valid if they were the only numbers on the ballot and it appeared to the leader of the count that they were intended as a vote.
4.8    The allocation of these three-digit numbers by the Supervisor of Elections is done as follows. Firstly, a hundred numbers each are assigned to the eight parties fielding most candidates, for instance numbers 200 to 299 to the Blue party, numbers 300 to 399 to the Red party, and so on. Numbers 100 to 199 would be allotted to the remaining smaller parties and Independents, allotting blocks of 10, 20, or even more numbers to each party so that the number of candidates the party is running can be accommodated. The particular set of a hundred (for each party) and the individual numbers (for each candidate) are drawn randomly (like lotto!). There is no hierarchy of numbers between parties or within a party, nor may a candidate (or party) seek to have their number (or series of numbers) changed, unless the above rules or their equivalent have not been followed.
4.9    There are advantages and disadvantages with both types of ballot paper. The large paper expresses clearly to the voters the realities of the system. They are presented with lists which are open, and they vote for their candidate by ticking the name, as in years past. The disadvantages come in the form of the ballot paper being so large. It is more expensive to produce and great care would be needed in proof-reading the paper to ensure all names were correctly spelt and correctly allocated to their party. Also, the area for counting the votes would need to be spacious (see 6.4). The three-digit number system has a clear advantage in terms of cost and ease of administration. The disadvantages of it are that it is more impersonal to the voter and care would need to be taken to explain the method clearly during voter education. Near election time, the numbers for each candidate and party would need to be clearly published. Each polling booth too would need to have a list of the candidates with their numbers pasted inside, so that voters can check when voting that they are writing the correct number for their candidate.

5.0    Polling [Electronic voting system will minimise many of these problems. Ed.]
5.1    The main polling should take place on a Saturday from 7am to 7pm. Fiji must adopt international practice in this regard and not continue with its practice of conducting polling over a period of eight days. If there is one thing that undermines faith in the system, it is surely the practice of holding ballot boxes for up to eight or so days before counting begins. There is no way of convincing voters that so many boxes can be totally secure for such a long period. The Elections Office has, I believe, researched the question and are convinced polling in one day can be done in Fiji. It will, of course, require a great number of personnel to man the polling stations. However, the simplicity of the Open List system will mean that the polling and counting of votes can be done by the same teams and on the one day. Furthermore, the counting teams at least can be smaller than previously. A total of eight members is proposed per team. As stated in 1.4, it has been proposed that the 1,150 proposed registration stations also be used as polling stations. The number of personnel required to run these polling (and counting) stations, therefore, would be 9,200. It could be less, if some registration stations are not used (see 5.3). This is a large number of people no doubt, but they are being hired for just one day’s work, not eight (polling teams) or up to four (counting teams) as previously. This represents a huge saving in costs.
5.2    There would be no more than about 600 voters per station, so polling could indeed be completed in one day. Each voter will have just one designated polling station where he is to vote on polling day. Checking on the identity etc. of voters would be much easier because of the Electronic Voter Registration; also because each polling station will be issued with a mini-roll for that particular station – namely, the names of the 600 or so voters entitled to vote there. It would be desirable to have the rolls used in polling stations printed out along with a small (but clear) photo of each voter.
5.3    The above-mentioned polling day would be the main (and the last possible) day for voting. However, facilities will be made for advance voting. Some voters, through being away from home in Fiji, may not be able to vote in their designated polling station on polling day.  They should be provided with the opportunity to vote over a limited period before polling day. This will be done at designated central places where there are professional electoral workers available to conduct the process and where there is reliable security for the votes cast. There will be similarly professional mobile teams with good security arrangements to visit hospitals and other such institutions. The Elections Office may deem it advisable for such mobile teams to visit some outer islands and remote rural areas too. Experience with getting to these locations for registration may indicate that advance voting would be better in their case. If so, this will need to be clearly publicized, so that voters in those locations can be present and vote when the mobile team comes, as they may have no alternative method of voting.
5.4    Postal voting will be abolished. Suitable security arrangements for it are very difficult to devise. Advance voting will take its place. Advance voting will also be the means employed for voters overseas. Special facilities will be made available as heretofore for some groups (e.g. the RFMF on peace-keeping duties) and similar facilities should be available in Fiji’s embassies or high commissions. Registered voters overseas should also be facilitated by being able to access their ballot paper on the internet and then faxing It into the Elections Offices, where due facilities will be set up to handle this form of voting.
5.5    Advance voting will need to terminate two or more weeks before polling day. This is not just to ensure that the votes are in position for the counting process (see Section 6 below). It is important that sufficient time be available so that those who have advance voted have their names especially marked on the rolls that will be circulated to the polling stations. This will ensure that such voters will not be able to vote again on polling day. To this end, secure and efficient arrangements will need to be made to get the names of those who have advance-voted to the Elections Office, and their votes to the appropriate central polling and counting stations.
5.6    The cards issued to voters at registration will indicate for each voter the designated polling station for that voter (as predetermined by the voting authorities, in consultation with the voter during registration, as being the most convenient to the residence of the voter concerned). Voting at one’s polling station is the expected form of voting. Although advance voting is provided for, it is only to be encouraged in the circumstances given above.
5.7    Multiple voting and impersonation have not been seen as serious problems in Fiji. They are even less likely to be so if voting is done in a single day. The process of inking the voter’s finger will be continued and would seem sufficient deterrent to people contemplating multiple voting or impersonation. Since the roll used at polling stations will hopefully include a photo of each voter on the roll, there should be little confusion as to whether a particular person, in possession of his registration card, is the voter on the roll or not. In a case of serious doubt, the team leader should consult higher authorities, and through them access the database compiled by the EVR to check on the voter’s thumb-print (and/or photo, if this is not at the polling station).

6.0    Counting [This section may need modification with the introduction of electronic voting system.Ed.]
6.1    The counting of votes should be done in each constituency on polling night, immediately after polling has ceased (7pm). The polling teams will conduct the count; they will therefore need to be trained in both the polling and counting processes. Advance votes will be counted in the electoral branch headquarters for each of the three constituencies. One agent from each party should be permitted in each polling and counting place, including those for advance voting (polling and counting) where possible. Another agent should be permitted to be nearby for substitution, or even for consultation if a problem arose.
6.2    The first process will be for the polling officials to tally the number of ballots issued, and then check, after emptying out the ballot box, how many votes have been cast. Without opening the votes, the counting team will count the votes, check that they are properly endorsed, and stack them into bundles. Any discrepancy must be accounted for by the team leader as far as possible.
6.3    Though simpler than Fiji’s past system, there are nevertheless two parts to the next stage of the counting process in the Open List system: totaling the votes for each party and totaling the votes for each candidate. (The same process would be used for either type of ballot paper suggested in Section 4). The current practice of showing each vote to the party agents will be discontinued as being not only too time-consuming but also unnecessary because of the new method of counting. Counting will be done following closely the ‘round-the-table’ method of counting used in New Zealand for MMP (see 6.4 below). Party agents will be able to hover and move round the table as the counting proceeds.
6.4    Firstly, all votes will be counted only according to what party was voted for. The members of the counting team will sit around a large table (or set of tables joined together) and each be given a bundle of votes by the team leader. These votes will be placed on his left. The team leader will also assign to each member of the team a party (or more than one party) to look after. Each team member will take each vote from the bundle on his left and check it for validity. If it is valid, he will check if the vote is for the party (or parties) he is taking care of. If so, he will place it on that party’s stack in front of him, otherwise he will place it to his right. If the vote has more than one tick on it, is dubious, or seems invalid, he will place it in the centre of the table. While the team member concerned (team member B) has been going through all the votes allotted him, the team member to his left (team member A) has been following the same process for the votes allotted to him. Team member A will have placed a bundle of votes to his right (which will be team member B’s left), so team member B now takes these votes too and treats them in the same way. Similarly, team member C (on team member B’s right) takes the votes that team member B placed to his right. In this way, votes pass around the table in an anti-clockwise direction until they arrive at their proper stack. When all votes have been handled, the team leader will take up the votes that have been placed in the centre of the table as having more than one tick, as dubious, or as seemingly invalid. He will make a decision on them on the basis of the rules provided in 4.4, 4.5, and 4.6. He will stamp them as valid or invalid accordingly. If there is serious discord over his interpretation, the count will go ahead according to his decision, but the vote(s) concerned should be additionally marked “disputed”. The votes for each party are then tallied, and it is checked that the totals for each party and of invalid votes add up to the total votes cast (see 6.2).
6.5    Then the team leader takes the votes of one party and divides them among the team members sitting around the table, placing them as before on each member’s left. He also assigns a party candidate (or, for the larger constituencies, two or three) for each team member to take care of. The same process as described in 6.4 is then followed to find out how many votes each candidate of that party holds. There should be no dubious or invalid votes, of course, at this stage, but if an irregularity is noticed or a vote has been placed in the pile of the wrong party, it is placed in the centre of the table. As in 6.4, when the normal counting is complete, the team leader resolves such issues, amending previous tallies if necessary. The process described in this paragraph is conducted as many times as there are parties in the contest.
6.6    At the completion of the count, the results of the counting process (6.2, 6.4, and 6.5) must be displayed in the polling station on a white-board (or equivalent) or projected onto a screen (if entered on computer) together with a clear indication from the team leader that this is the final result  for the polliing station. There should be a formal photograph taken of this image, and agents (and others) must be free to also photograph this final tally sheet. It is then communicated in full by email or fax to the central count centre of that constituency (in Labasa, Lautoka, or Suva). The entire counting process in the polling station should only take about two hours.
6.7    Declaring the final results in the central counting places, however, is likely to take a lot longer. The results from all polling stations and advance vote counting places (including votes faxed from overseas – see 5.4) will be tabulated for each constituency in the main counting room of the central counting place for that constituency. If the results from the polling stations are dispatched by email, it may be possible to have them automatically added to whatever totals have already accrued. Nevertheless, a hard-copy of each polling station’s results must be produced so that a physically separate record exists of all results. It is recommended that a very large screen be erected inside the main counting room of the three central locations so that agents (and others present, including TV coverage) can view the overall results as they are being tallied. As polling stations’ results come in, they should be announced and shown on screen, if possible, before being added to the overall totals.
6.8    When all results have been received and tallied for each candidate and for each party in the constituency, the number of seats in that constituency to be allotted to each party will be determined by the Modified Sainte Laguë method (see 3.5). When it is clear how many candidates each party should have, that number of candidates holding the highest number of votes for their party will be declared elected as the parliamentary representatives of that party. Any ties will be decided by lot.
6.9    If the votes of a party entitle it to more seats than it has candidates running, it cannot take up those seats. The Modified Sainte Laguë process for allotting seats is resumed to see which other party or parties should have the seat(s). This would also apply if an elected candidate died, resigned or was dismissed from parliament and there was no one else of that party left to replace him.
6.10    The Commissioner or his appointee for each constituency will make the declaration of election for each winning candidate. However, he will be assisted by two people appointed by the Supervisor of Elections with the approval of the Electoral Commission. These three people will jointly make the decisions referred to in 6.8 and 6.9 above. If they cannot reach agreement on something, or if serious problems have arisen at a particular polling and counting place which cannot be immediately resolved, the results declared will be clearly designated as “provisional” only. The Supervisor of Elections (who may take external advice on the matter) will ultimately investigate and decide the issue, which cannot be overturned except by the Court of Disputed Returns after the election. The Supervisor’s decision should be put in writing, any recounting or reassessing of votes being properly recorded. The decision should be made within 48 hours of the completion of the count, so that all results are final by that time.

7.0    Cumulation, ‘Panachage’, ‘Apparentement’
7.1    In some forms of list system, voters are given more than one vote, which they are able to spread over a number of candidates on a party’s list, or may cumulate on just one candidate. Other forms of list system allow the voter to even spread his votes over more than one party (this feature is known by the French word ‘panachage’).
7.2    Cumulation and ‘panachage’ are not provided for in the proposed Open List system. This is not just to keep the system simple. People rather too easily believe that, because the constituencies are multi-member ones, voters should have multiple votes as well. This is not true, and can actually work against the interests of the voter and the concept of proportional representation. It is better to stick to just ‘one person, one vote, one value’. This is enough for everyone. The multiplication of votes increases the likelihood of strategic voting and party or individual manipulation.
7.3    Another device sometimes used in list systems is ‘apparentement’ (another French word) whereby two or more parties are allowed to give their lists in association such that their lists will be treated as if they were just one list in the allocation of seats. Once the number of seats won by this association of parties is known, that number is re-divided out by the same divisor system among the parties concerned. For small parties in particular, this can lead to more likelihood of a seat, or more seats. ‘Apparentement’ too is not recommended in the proposed system. While small parties need to be given a fair go, the procedure can have the effect of over-protecting them (and is cumbersome).
7.4    Another possibility – a concept related to ‘apparentement’ – is that any party running less than half the full quota of candidates could be permitted (or required) to nominate one other party as its associate. If the first party gets more seats than it has candidates running, the associate party would pick up the extra seat(s). Similarly, if a candidate of the first party dies, resigns or is dismissed from parliament and there is no else on the original list of that party in the election who can take up the post, the associate party would do so. Again, however, I do not recommend this possibility. It is better to keep the system simple rather than clutter it with unnecessary extras. The parties should make proper provision for their own welfare and not rely on special concessions.

8.0    Campaigning
8.1    I present the topic of “campaigning” towards the end of this paper because readers (like political parties) need to consider how the entire system works in order to be able to understand (or devise) reasonable campaign strategies. For the first election or two under the proposed system, a bit of ‘trial and error’ may come into these strategies, but due consideration of the various aspects of the system should enable parties to focus appropriately.
8.2    The most obvious change, of course, if from single-member constituencies to large multi-member ones. The larger area to be covered by candidates on this account is not a serious problem. Indeed candidates for the three “Others” (or “General Electors”) seats of AV have already been dealing quite efficiently with areas closely corresponding to the proposed constituencies – and candidates for the Rotuman seat have had the whole country as their constituency! Parties running a few candidates may decide to carve up a constituency into sections for each candidate to campaign in. This would be a practical proposition up to a point, but these candidates would undoubtedly wish to appear in other sections too from time to time, since their ultimate election will depend on all sections of the constituency, not just one or a few. Parties will need to plan such matters considerately both to get the best out of the candidates on behalf of the party and also to present the party to the voters as a balanced and well-coordinated team.
8.3    There are, of course, those who will miss the more direct candidate-voter relationship that single-member constituencies provide, but there are definite advantages too with multi-member constituencies. Many countries in the world use list systems with their large constituencies and voters feel content and quite adequately represented. A danger for single-member constituencies in Fiji is that they are likely to accentuate ethnicity differences. Although there is an emphasis at the moment on getting people to think nationally rather than provincially or ethnically, old ways will die hard and the very geographical disposition of Fiji’s ethnic groups will tend to bring about that a vast majority of voters will vote for candidates of their own ethnicity. Single-member constituencies, therefore, are extremely likely to skew the ethnic proportionality of the electoral outcome.
8.4    It can be pointed out too that if in a single-member constituency a voter’s choice is not elected, he is being represented by just one person who is of a different political persuasion (and also ethnicity, if that matters to him). In a multi-member constituency on the other hand, the voter will almost certainly have a representative who is part of the government - and another who is part of the opposition. He can approach whatever representative attracts him best whether because of political beliefs, local understanding, or common language (and/or ethnicity).
8.5    Parties will need to ensure that they run enough candidates in a constituency. Failure to do so, of course, would result in missing out on one or more seats. An Independent who genuinely believes he has massive support would be thoroughly foolish, therefore, to run as such rather than form a party. If he failed to see this himself, his numerous supporters would surely persuade him otherwise.
8.6    Parties also need to think carefully as to whether they could suffer through running too many candidates. Running many candidates gives voters a lot of choice, but it can be costly for a small party. In addition the allotment of the party’s seats might turn into a bit of a lottery. A good example of this is provided in Paper 1, ¶2.21 to 2.23. Though the example (using the National Alliance Party) is entirely conjectural, it shows how a small party running eight candidates won only one seat. The candidate winning the seat does so with 1,400 votes out of a total of 5,000. If the party had run a more realistically low number of candidates, the ultimate winner of the seat would likely emerge with a more convincing degree of support. Nevertheless, a new party may indeed tend to field candidates a bit optimistically.
8.7    It is not just the number of candidates parties need to think about. They need to think about the range of candidates they are running. In single-member constituencies, parties will generally look for candidates who are most likely to attract a wide range of voters. This is often a middle-aged male professional. The advantage of multi-member constituencies is that these are not the only kind of people a party will want to run. Even if there are as few as 5 or so seats in a constituency, a party can appeal to various minor interests and choose candidates that will secure their vote even though not of particular appeal to other sections of their electorate (see Paper 1, ¶2.36). This clearly has much better potential for meeting voters’ needs. It is another way of bringing about accountability. Parties will need to be more attentive before elections as to the range of matters that are of concern to different voters and then run the sort of candidates that can best meet these concerns.
8.8    A big advantage of the Open List system for parties and voters alike is that voters can hold candidates more accountable. A non-performing candidate will be punished in the next election (if the party allows him to stand) by voters choosing to support a different candidate of the same party. In single-member constituencies on the other hand, voters have no choice but to vote for him again or to vote for another party, which they may not really want to do.
8.9    In pursuing certain goals, parties will need to truly gauge public opinion and not go overboard. For instance, if a party decided to really go after the Green vote, it might think that it would do well to run many Green-inclined candidates. Such a policy could well back-fire, providing unintended consequences. If in running a full quota of candidates in a 10-member constituency it ran 6 such Green candidates and if it in fact won 4 seats, it could in fact be the other 4 candidates on their list that get elected, not the Green ones. This would happen if the party had seriously over-estimated the degree of Green support. By dividing the support over 6 Green candidates the party would bring about their failure, whereas if they had run a lesser number, say 3, the support would have consolidated on those candidates and the party would have got one, two, or all three of them elected.
8.10    This point is of importance for the gender question. A party might want to appear very gender conscious and run 5 women and 5 men in a 10-member constituency. But unless the candidates are of good quality and the electorate is on board with the party’s approach, having this number of women may only serve to bring about their defeat, whereas if only 3 were run they might all get elected. Similarly, a party historically associated with Indigenous Fijians may wish to reach out to the Indo-Fijian population (or vice versa). Running 5 of them in a 10-member list, however, might only serve to exclude them.
8.11    The point is: Each party needs to understand its electorate, measure their support for various issues, and help educate them on values which the party believes are worth promoting. The Open List system is a very democratic one. Election depends on the people’s choice, and is less amenable to manipulation. Getting certain ideological positions accepted is not something to be forced, but something that the electorate needs to be persuaded of and genuinely give its assent to.

9.0    After the Elections
9.1    Being proportional, the Open List system will provide more parties with representation than has hitherto been the case. The largest parties, especially the SDL and the FLP, may be disposed to oppose it therefore, as it will mean that they get less seats than they have been getting to date. But this is an outcome that simply has to be accepted. Fair is fair. You cannot reapportion the cake without some getting less so that others get more. Fiji needs to accept and work with a variety of shades of opinion. That is a much more genuine democracy than one where majorities are manufactured by eliminating minority viewpoints. Such majorities are not truly representative of the people.
9.2    It is often adduced that for stability and decisive government it is best to have just one party in power and avoid coalitions. This is a specious argument, convenient for the larger parties. There are many countries using list systems in the world, and whether they yield single party majority or coalition governments, there is no systematic problem with them. Fiji has now had four coups, two under the First-Past-the-Post (FPP) system and two under the Alternative Vote (AV) system. Both these systems are touted as promoting single party governments, but it is manifestly true that this has not provided Fiji with stability.
9.3    The Open List system gives everybody a fair go. For instance, it caters evenly for ethnically-based or multi-ethnic parties, not attempting to advantage any group over another. The only way multi-ethnic parties might have an edge is that such parties are clearly attempting to appeal to the whole population whereas ethnic ones are appealing only to subsections of it. But such a theoretical advantage has to be balanced by realities on the ground.
9.4    What does need to be stressed on the ethnic question is this: If communal seats had been retained (which is certainly not being recommended here!), the Open List system could have been adapted to that situation. But now that communal seats are being done away with (as firmly recommended by the NCBBF), some sort of PR, preferably the Open List, is absolutely essential (cf. also Paper 1 ¶1.25). To do away with communal seats and retain the FPP or AV systems could be disastrous, as the severe distortions which they brought about in party representation in Fiji could be replicated in severe distortions of ethnic representation, with the intense resentment and unrest that this would generate.
9.5    It needs to be added however - indeed stressed - that when dealing with a PR system, the proportionality being referred to is the proportional representation of political parties not the proportional representation of ethnic groups, women, or any other group. The reason for PR being helpful in Fiji’s ethnic situation is that, if people are indeed voting along ethnic lines, it will provide good ethnic proportions. If the system ceases to do so, it will mean that the population is no longer so concerned about ethnicity. Conversely for women, one reason that women have been poorly represented in parliament is that the gender issue has not been seen as politically that important. As this viewpoint changes, Open List PR provides the opportunity for a much better proportion of women to be elected. Political parties that ignore this reality do so at their peril.
9.6    No matter what parties get representation, the next stage is the formation of a government. If a coalition is necessary, so be it. Just as the population lives together with its varied views, so its leaders, whatever their ethnicity or political agenda, must work together for the welfare of the nation. Fiji has had its experiences with the Korolevu Declaration and the Multi-Party Cabinet provisions. These protocols should not be discarded but need to be worked through to see what might be of use for future coalitions. There is also the sundry experience from various parts of the world which Fiji can draw upon to help it in mapping out a functioning and productive mode of operation. It may well be desirable to build some provisions on this matter into the new Constitution. These issues require further elaboration, but are beyond the parameters of the electoral system as such and the scope of this paper.
9.7    One additional issue that does require mention, however, is that of review of the new electoral system when it comes into operation. As mentioned in Paper 1 ¶3.5, the 1997 Constitution’s provisions for making changes to Fiji’s electoral system were “inappropriate and too demanding”. Though parliament should have a role in the process, it should not have the ultimate say, for it would thus be too easy for the parties comprising the parliament to change the electoral system for their own benefit without due concern for other parties or the interests of the people themselves.
9.8    I propose that an independent Electoral Review Commission be permanently established. It would be headed by a (retired) judge and be comprised of the Supervisor of Elections, a member of the Electoral Commission, one or two electoral experts and two or three impartial citizens of high reputation. The appointing body would be the Constitutional Offices Commission. The Electoral Review Commission would have the task after each election of reviewing the electoral system in virtue of experiences gained. It would seek input from the public and the political parties and publish its proposals for change (if any) within a year of the past election. After considering feedback received, it would make its final proposals for change within two years of the past election. The proposals would be put before Parliament but would be regarded as approved unless opposed by 60% of the members of parliament. The reason for this special majority is that it would not be good for a majority coalition to perpetually block reforms which might be aimed at cutting back on certain unfair advantages they may have been seen to have.
9.9    Electoral systems and their details should not be too readily changeable. The core elements, therefore, need to be in the Constitution. But there does need to be a means of making reasonable changes without excessive hassle (which would be the case if too much is in the Constitution). The above proposals about an Electoral Review Commission would seem to take care of this aspect, the Commission having the exclusive power to change the Electoral Act (with the above-mentioned approval of Parliament). But who sets up this Act in the first place?
9.10    This is a specific problem at the moment. The Constitution Commission and the Constituent Assembly are concerned only with the Constitution. If they do what I think they should do - that is, put only the core elements of the electoral system in the Constitution – then we still won’t have an electoral system in place by March 2013. An Electoral Act will need to be enacted, but by whom? Because of the importance of having the electoral system in place and agreed to, I would suggest that the terms of reference for the Constitution Commission and Constituent Assembly be extended to include the preparation and passing of an Electoral Act. An alternative solution would be to appoint in the next few months the Electoral Review Commission spoken of above. Their first task would be to draw up an Electoral Act, fleshing out the details of whatever electoral provisions are included in the Constitution by the Constitution Commission, and submitting that Act to the Constituent Assembly for finalization and approval along with the Constitution. No matter what process is decided upon, it is hoped that the details provided in this paper may be of some assistance in the total endeavour.



APPENDIX

Sample Ballot Papers

(See following two pages)



Sample List Ballot Paper – 9 Seats

Vote by writing a tick () in the box to the right of the name of one candidate only. [This instruction will appear in other languages also]


RED    BLUE    GREEN    SILVER    BLACK    PINK    ORANGE    BROWN    YELLOW      
                                      
Patemo Suisui        Ram Karan        Laisa Tagi        Luisa Vuli        Viliame Seseni        Barbara Lakepa        Bale Niu        Diwan Chandra        Alex Koi          
Hardip Narayan        Misaele Basu        Rup Chand        Kelepi Tui        Susan Pickering        Vineet Lateef                                  
Toni Suitu        Mitch Whippy        Hanisi Mua        Faria Nazli        Sharon Prasad        Justine Simpson                                  
Romit Kant        Harvey Hull        Isabel Simpson        Warwick Williams        Vilisi Vulaono                                          
Harry Williams        Nina Vuli        Alisi Levula        Tukaha Amoe        Sophie Shackley                                          
Jona Nagata        Veniana Buli        Sheetal Chand        Siteri Bakewa                                                  
Jope Tubia        Sanjana Narayan        Ana Eri        Kamlesh Kumar                                                  
Amoe Petero        Mitieli Seru        Seema Kewal                                                          
Alex Michel        Rama Murti                                                               


Sample List Ballot Paper
(Alternative Method)


Vote by writing the three-digit number of your preferred candidate in the box below.
[This instruction will appear in other languages also]