The Fiji Electoral Law Reform Commission has been brutal in its criticism of Fiji’s current open-list proportional representation system. Every trade-off has been treated as a defect. Every perceived imperfection has been presented as a reason for replacement, even though no electoral system in the world is perfect. Having applied such an exacting standard to the present system, one would reasonably expect the Commission’s alternative to be vastly superior and capable of surviving the same level of scrutiny.
It does not. The proposed so-called Proportional Open List Constituency System, as publicly described, is mathematically incomplete, internally contradictory and incapable of reliably delivering the national proportionality it promises.
The fatal mathematical defect
The proposal combines 25 open constituency seats, 25 reserved women’s constituency seats and only 21 compensatory seats in a fixed 71-member Parliament. The first 50 seats would be decided through separate first-past-the-post contests. Only after those winner-take-all elections would the 21 list seats be used to try to restore proportionality. That is where the system begins to collapse.
The proposed formula, as detailed by commissioner Prof. Narsey is simple: a party’s list seats would equal its total proportional entitlement minus the constituency seats it has already won. In symbols, L = T − C. But that formula works only when a party’s constituency seats are fewer than, or equal to, its proportional entitlement. The proposal, as publicly articulated, provides no mechanism to ensure this.
Suppose a party wins 51 per cent of the national vote, spread relatively evenly across Fiji, and narrowly wins both contests in all 25 constituencies. It would hold all 50 constituency seats. Yet 51 per cent of a 71-member Parliament would entitle it to only about 36 seats. The formula would then produce minus 14 list seats. A political party cannot be allocated negative MPs.
There are only three possible responses. Remove 14 constituency winners, thereby overturning declared election results. Allow the party to retain all 50 seats, thereby abandoning proportionality. Or increase the size of Parliament through overhang and levelling seats, potentially very substantially. None of this is explained.
This is not a remote or fanciful possibility. It, in fact, is the central technical problem in mixed-member proportional systems. Countries using such systems have detailed rules governing overhang seats, levelling seats and adjustments to the size of Parliament. They also maintain a sufficiently large compensatory tier. Here, more than 70 per cent of Parliament would already be decided through winner-take-all contests before the correction mechanism even begins. Twenty-one compensatory seats cannot reliably repair distortions created by 50 constituency contests.
An unclear national party vote
There is also the question of how the national party vote would be calculated. The most obvious approach would be to add together all votes received by a party’s candidates in the open and women’s constituency contests. That is certainly possible. But once that approach is adopted, a series of further problems arises.
Each voter would effectively cast two party-linked votes. A voter could support Party A’s open candidate and Party B’s woman candidate, thereby contributing to the national totals of both parties. This may be intentional, but it means that there would be no single expression of the voter’s preferred party for the purpose of determining the composition of Parliament. The national result would instead be assembled from two candidate choices that may point in different political directions. So, how do you determine which party was the voter’s choice for the national seats?
The system would also need to decide whether both ballots carry equal weight. If they do, each voter effectively has two equal party votes. Yet the two contests may have different candidate fields, levels of competition, invalid-vote rates and turnout patterns. A strong candidate in one category could substantially increase a party’s national entitlement even where voters were primarily supporting the individual rather than the party. If the two ballots do not carry equal weight, the weighting would need to be justified and would introduce another layer of complexity.
Parties that do not contest every seat would face an obvious disadvantage. They would have fewer opportunities to accumulate votes towards their national entitlement than parties fielding candidates in all 50 contests. Their lower national total might therefore reflect incomplete candidate coverage rather than lower public support. Adjusting for this would require some form of weighting or standardisation, which would make the system considerably more complicated and could artificially inflate votes received in the constituencies they did contest.
A similar problem arises where a party contests only one category in a constituency. Voters supporting that party would be unable to vote for it on both ballots, while supporters of a party contesting both seats could contribute twice to that party’s national total. The parties would therefore not be competing on equal terms.
Independent candidates create a further difficulty. If votes cast for independents are included in the national denominator, they would reduce the proportional entitlements of political parties even though independents cannot receive compensatory list seats. If those votes are excluded, the effective value of a vote would depend on whether it was cast for a party candidate or an independent. An independent who wins a constituency seat would also occupy one of the fixed 71 seats, raising the question of whether party entitlements are calculated from the full Parliament or only from the seats remaining after independents are accommodated.
Uncontested constituencies would create another distortion. Where only one candidate is nominated, there may be no meaningful vote total to add to the party’s national result. If the candidate is declared elected without a poll, the party receives a constituency seat but no corresponding national votes. If a poll is nevertheless held, the party could receive an artificially large total in a contest offering voters no genuine choice. Either approach would undermine comparability between constituencies.
Votes cast for parties that fail to cross the national threshold must also be treated somehow. If those votes are excluded before proportional entitlements are calculated, the vote shares of qualifying parties will be increased. If they remain in the calculation, seats would be attributed proportionately to parties that are legally prohibited from receiving them. Because each voter casts two ballots, a voter could also support a qualifying party on one ballot and a non-qualifying party on the other, leaving half of that voter’s contribution included and the other half discarded.
These problems are not necessarily incapable of resolution. But every solution requires additional rules, and each rule affects the equality, meaning and weight of the vote. A compensatory electoral system cannot leave such matters to assumption. More importantly, even a perfectly specified method of combining the two ballots would not cure the proposal’s fatal overhang problem. A party could still win more constituency seats than its national vote entitlement, leaving the proposed formula mathematically incapable of restoring proportionality.
Not an open-list proportional system
The name of the proposed system is itself misleading. It is called a Proportional Open List Constituency System, but 50 of the 71 MPs would be elected through first-past-the-post. Only 21 would come through a compensatory process, and even those would not be selected from a conventional national open list. They would be losing constituency candidates ranked after the election. This is not open-list proportional representation in any meaningful sense. It is predominantly a first-past-the-post system with a small compensatory mechanism attached.
The “best losers” problem
The so-called “best losers” mechanism creates another set of problems. Ranking losing candidates by raw vote totals may sound simple, but raw totals are not comparable across constituencies of different sizes, turnout levels and voting populations.
A candidate who receives 7000 votes but only 30 per cent in a large urban constituency could outrank a candidate who receives 5000 votes and 49 per cent in a smaller constituency. The second candidate came much closer to winning, but the first would receive the list seat simply because more people lived in that constituency.
Maritime constituencies would be particularly disadvantaged. If they contain fewer voters than large Viti Levu constituencies, their losing candidates would struggle to compete on raw numbers. Yet if every constituency receives the same two MPs despite substantial population differences, the principle of equal voting power is weakened.
The proposal cannot simultaneously guarantee separate maritime representation, equal constituency populations and fair comparisons between losing candidates without more sophisticated rules. Calling standardisation confusing does not justify replacing it with an unfair calculation.
Misunderstanding candidate votes
The proposal also misunderstands candidate votes under the current OLPR system. One criticism is that a candidate with a lower personal vote can enter Parliament while a candidate from another party with thousands more votes does not. But proportional representation allocates seats to parties according to their collective support. Candidates then compete within their own parties for those seats. A candidate from a party receiving three per cent nationally cannot simply displace a candidate from a party entitled to 20 seats. That would transfer a seat from one party’s voters to another party.
Comparing personal votes across parties is therefore a category error. The proper comparison is party against party for the allocation of seats, and candidate against candidate within the same party for occupancy of those seats.
The “rockstar” problem, where a popular leader attracts a large share of the party’s total vote, can be addressed through replacing proportional representation with 50 first-past-the-post contests.
FPTP and wasted votes
The proposal also attacks the five per cent threshold for wasting votes while introducing 50 first-past-the-post elections in which every vote cast for a losing candidate fails to elect anyone. With a crowded field, a candidate could win a constituency with 25 or 30 per cent of the vote, even though a clear majority voted against that candidate. If the threshold is considered too high, reduce it. That issue can be debated on its own. It does not justify replacing a highly proportional national system with dozens of winner-take-all contests.
Gerrymandering still matters
The claim that gerrymandering would become irrelevant is also wrong. Constituency boundaries determine who wins local seats, where parties concentrate resources, which communities become marginal battlegrounds and where safe seats emerge.
The argument appears to be that winning an extra constituency seat would merely reduce a party’s list entitlement by one. But that is true only while the party’s constituency seats remain below its total proportional entitlement. Once the party crosses that point, constituency boundaries directly affect the size of the overhang and potentially the national composition of Parliament. Under this proposed 70:30 arrangement, boundaries could determine who governs.
Reviving ethnic majoritarianism
There are also troubling assumptions about ethnicity. Fiji’s single national constituency means that a vote cast by a dispersed minority has the same national value as a vote concentrated in a particular town or province. Parties have an incentive to appeal beyond narrow geographic and ethnic bases because every vote contributes to their national seat entitlement.
Constituency-based first-past-the-post politics creates the opposite incentive. Parties will study local ethnic arithmetic, search for safe seats, tailor candidates to communal majorities and focus divisive appeals on marginal constituencies. A system presented as a response to ethnic insecurity could easily revive local ethnic majoritarianism.
Women’s representation
The reserved seats for women pursue a legitimate and important objective, and women would remain free to contest the 25 open constituency seats. The problem is therefore not that women are confined to reserved seats. The concern is that the proposal creates three different routes into Parliament: 25 seats reserved for women, 25 open seats available to both women and men, and 21 compensatory seats drawn only from unsuccessful candidates in the open contests. If, as the proposal itself appears to assume, the open contests are likely to contain more male candidates, the compensatory tier may disproportionately benefit men.
Women could win open seats, but unsuccessful women contesting the reserved seats would apparently have no access to the compensatory list. A man losing an open contest could still enter Parliament through the list, while a woman narrowly losing a reserved-seat contest could not. The objective of increasing women’s representation is valid. The architecture chosen to achieve it, however, requires much more careful justification.
What the current system gets right
Fiji’s current OLPR system has one foundational strength of enormous importance: a vote cast anywhere in Fiji carries the same value in determining the national composition of Parliament. Parties receive representation broadly in proportion to the support they collectively earn.
In a country with a long history of race-based politics, this creates an incentive to seek support across communities, regions and ethnic groups. It rewards parties and candidates capable of building broad national appeal. That is precisely what an electoral system in a divided society should aspire to do.
The present system can still be refined. The threshold can be reviewed. Women’s representation can be strengthened through incentivizing political parties to put more women on their lists. Political parties can be made more democratic. Mechanisms for constituency service can be established without reviving winner-take-all parliamentary constituencies.
Globally, where the link between elected representatives and communities is considered weak, the appropriate response is to strengthen local government, decentralised institutions and constituency outreach. It is not to weaken the proportionality and representativeness of the national legislature to compensate for inadequate local governance, which some in Fiji appear to be pushing for.
A fatal design defect
But reform should preserve what already works. The proposed system does not do that. It sacrifices equal vote value and proven national proportionality for a complicated mixture of FPTP constituencies, reserved seats and an inadequate list tier. Its promise of proportionality disappears the moment a party wins more constituency seats than its national vote justifies.
This is not a small drafting problem that can be tidied up later but rather a fatal design defect. In its published form, the proposal is not a workable proportional electoral system. It is a predominantly first-past-the-post system with an insufficient and mathematically fragile correction mechanism attached to it. Fiji deserves electoral reform based on coherent principles, tested mathematics and global best practice.
NILESH LAL is the executive director of Dialogue Fiji, a civil society organisation working on strengthening democracy, human rights, and social cohesion in Fiji. A political scientist by training, he writes frequently on governance and development issues in Fiji.
