Which electoral system for Mauritius?
— Professor Singfat Chu
As more proposals for electoral reform and even more opinions are being circulated in mainstream and social media, I will discuss their pros and cons. The electoral system that we want in Mauritius must be simple and conducive to the formation of a stable government while delivering adequate representation to parties and our rainbow fabric straddling communities, gender, etc.
I begin with the proposal by Mr Joseph Tsang Man Kin for single member constituencies in the interview published in Mauritius Times on 5th October. He opines «… Dans ce nouveau système électoral, tout devient plus simple, plus démocratique, plus propre… ». Indeed, this proposal scores full marks on simplicity, transparency and its ability to generate a stable government. Unfortunately, the flagship “First Past the Post” (FPTP) system also scores very negatively with respect to the adequate representation of minorities be it opposition parties, community, gender, etc. No matter how Mauritius is carved into single member constituencies, it is a certainty that minority communities will be seriously underrepresented among the elects. This is the rationale that led the British to opt for multi-seat constituencies in the crucial 1967 Independence election.
The opposite extreme to single member constituencies is a single national constituency where voters choose among national Party Lists. This electoral system exemplifies Proportional Representation (PR) with parties, which meet a threshold vote level, dividing the seats among themselves according to their relative vote share. It scores highly on all the criteria stated above except that its seat-to-vote proportionality may not be conducive to the formation of a stable government in tight contests. Hence, parties do not favour it. Another danger is the possibility of Party Lists being made up entirely or in large part by candidates from a single community. This can be a danger to the rainbow fabric of Mauritius. We certainly do not want the “bottleneck” outcome which arises after almost every election in Israel, where PR seats are allocated based on the votes attracted by national Party Lists.
Last December, Professor Carcassonne proposed a regional PR variant with the country divided into say 10 constituencies with seats varying between 4 and 7. I pointed out the day after the release of the report that the fundamental flaw with the proposal is arithmetical in nature. With say 7 seats in a constituency, a party attracting say 50.1% of the votes will get 4 seats while another one getting 49.9% will get 3 seats. If there were 6 seats instead, 2 parties with say 43% and 57% will each get 3 seats! This follows because 43% of 6 amounts to 2.58 which rounds up to 3 seats. This can be confirmed on this online d’Hondt PR calculator, http://icon.cat/util/elections. These examples demonstrate that PR does not work well on a small number of seats. Margins of say 2% in party votes, which are likely in Mauritius and Rodrigues, can only de differentiated with 50 or more PR seats and this brings us back to the concept of national Party Lists discussed above.
I now turn to alternatives in between the above two unsuitable extremes. Our current system with 3-seat constituencies in Mauritius and a 2-seat constituency in Rodrigues launches the discussion in view of its familiarity. It has been employed in 10 general elections between 1967 and 2010. Its main shortcoming is that by virtue of its FPTP basis, it may not deliver an adequate representation of the Opposition. The ‘Seat Distribution’ table indicates that on average, the top party has been rewarded with an excess 21.3% of seats compared to its votes. Even the application of BLS hardly closed the gap to an average of 18.7%.
With regards to community representation, our current FPTP system has done quite well with the four communities defined in our Constitution splitting the seats 59.4% : 25.2% : 13.7% : 1.8% pre-BLS and 54.1% : 29.4% : 14.8% : 1.6% post-BLS. Thus BLS does not justify its raison d’être as it has not made significant differences in party or community representations. Apart from unequal number of voters in the different constituencies, another reason that our current FPTP delivers adequate community representation is the choice of 3 seats per constituency in Mauritius. Had there been 2 seats instead per constituency, a proposal which has considerable support among electoral reform discussants, my simulation indicates a significantly lower and inadequate representation of minority communities in the past 10 elections. Simply put, fewer seats imply fewer opportunities for minority candidates.
What essentially needs to be reformed in our current system therefore is the provision for a more adequate representation of the Opposition. A very simple way to do so, without any change whatsoever to the electoral process, is to “Vote 3 Elect 4” in the constituencies in Mauritius and “Vote 2 Elect 3” in Rodrigues i.e. implement a generalised Best Loser across all 21 constituencies. My simulation of the 10 past elections (see Table) indicates that this electoral system would have dramatically cut the seats-to-votes advantage to an average of 7.8%. Unfortunately, the community split would have been 58.6% : 26.5% : 13.4% : 1.6% and thus, minority under-representation would have been an issue. Another problem would be unhealthy rivalries between the candidates from the same party for the extra seat available.
A more effective way to enhance the representation of the Opposition and collaterally, communities, and gender is through a mixed PR system which couples Party Lists to our current FPTP system, as proposed by the Sachs Commission in 2002. However, there is an “Achilles’ heel” in its proposition to have voters cast 2 sets of votes i.e. 1 set of votes for the constituency candidates and 1 vote for a party. In the February 2012 elections in Rodrigues, the Front Patriotique Rodriguais (FPR) attracted 8.4% of the votes at the constituency level but 10.6% at the Regional or Party List level. The latter vote percentage qualified FPR for Party List seats and this led to a controversial electoral outcome.
Thus, I propose that entitlement to PR seats be based unambiguously on the votes attracted by party candidates at the constituency level in order to avoid similar problems. As for the threshold vote level needed to qualify for Party List seats, the higher it is the less chance there will be for a fringe or extreme party to claim these seats. In the past 10 elections, only PMSD managed to score above 7.5% as a third party, specifically scoring about 17% in 1976 and 8% in 1982. I think any threshold above 5% is reasonable.
There are two contending methods to allocate Party List seats. Method C proposed by the Sachs Commission can be interpreted to be based on the votes cumulated by all the candidates from a party. In January of this year, Dr Rama Sithanen proposed instead that the distribution of seats be based on the votes of the unelected or unreturned candidates of a party.
I will now illustrate the difference between the methods proposed by Sachs and Sithanen. Assume there are 20 Party Lists to be distributed. In the recent 2010 election, l’Alliance de l’Avenir had 41 FPTP elects against 18 for l’Alliance Du Coeur, 1 for FSM and 2 for OPR. Except for OPR, the total votes for each party were 1,001,903 (49.7%), 847,095 (42%), 51,161 (2.5%). Assuming that only the 2 alliances qualified for Party List seats, l’Alliance de L’Avenir would have ended with 43 (=.497 * (82 – 3) / [.497 + .42]) seats compared to 36 for l’Alliance Du Coeur.
Under the Sithanen’s method based on the votes of the unreturned candidates from the 2 alliances (i.e. 249,549 [26.4%] and 552,508 [58.5%]) as no other party would have qualified, l’Alliance de l’Avenir would have been assigned 6 (=.264 * 20 / [.264 + .585]) Party List seats thus bringing its total to 47 (=41 FPTP + 6 UVE) seats while l’Alliance du Coeur would have ended up with 32 (=18 FPTP + 14 UVE) seats.
The difference between Method C proposed by the Sachs Commission and the “Unreturned Votes Elect” (UVE) innovation of Dr Rama Sithanen is that the former attempts to align seats with votes as proportionately as possible while the latter is a parallel adjustment to FPTP results. Statistically then, UVE is more conducive to a stable government than Method C. Indeed, a simulation of the past 10 elections (see Table) shows an average votes-to-seat variance of 8.4% for UVE compared to a “too close for governmental stability comfort” 4.4% for Method C.
The Prime Minister has stated his wish for an electoral reform that will last for decades. At this juncture in our history, I submit that the minimal reform that can achieve this is a mixed PR system with 62 FPTP elects with another 20 elects from Party Lists. The justification for 20 Party list elects is that FPTP elects must make up at least 75% of the National Assembly in order not to deny an acclaimed party the right to amend our Constitution. For stable governments, the selection of the 20 Party List elects should preferably be done according to the innovative UVE concept of Dr Rama Sithanen.
MMM has proposed 62 FPTP + 20 Party List + 8 “Leader’s Selections”. I’ve argued above that there should not be more than 20 seats beyond the 62 FPTP seats. To reach a consensus, I’m agreeable to say 62 + 15 + 5 as there should not be too many “Leader’s Elections” to maintain the objectivity of the electoral system.
The Electoral Supervisory Commission will determine the split of the 20 (=15 + 5) extra seats to eligible parties. However, it will only name 15 elects from the top of Party Lists. I believe it is very hard for leaders to rank candidates at the bottom of Party Lists. Hence, I find it acceptable and certainly not controversial for them to reconsider who to pick among the remaining Party List candidates. This implies “soft” rankings near the bottom of Party Lists. This is exactly how soccer coaches choose substitutes according to how matches are progressing.
If a mixed PR system is implemented, we will also have to reform our mindset on the electoral outcome. Specifically, a party with the most FPTP seats need not be the overall winner. That was the case in 1976 when MMM won the most votes and FPTP seats but it was the post-election coalition of Independence Party and PMSD which formed the government. With say a total of 82 seats in a mixed PR system, a party is only assured of victory if it has captured 42 or more FPTP seats. If a party leads with less than 42 FTPT seats, its votes become crucial in the determination of the PR seats. Assuming a two-party contest with Party A leading Party B by 35 to 25 FPTP seats (with the other 2 seats going to other parties), it is possible for PR (under either Method C of Sachs, or UVE of Sithanen) to split the 20 extra seats say 4 to A and 16 to B thus bringing their total seats to 39 and 41 respectively. This implies that in a mixed PR system, both the FPTP seats AND the votes determine the overall winning party. Think of a mixed PR system as a soccer game with 2 halves where leading at half time (or FPTP seats) does not imply being the winner at the end of the second half (or when PR is applied).
Every electoral system has its pros and cons as demonstrated in this article. We must select one which is “balanced” on the stated criteria and which has the fewest insurmountable negatives. Reaching a consensus is essential towards establishing our brand of inclusive rainbow democracy.
Note: The 1976 figures (italicised & in bold types) refer to MMM which won the most votes and FPTP seats. Eventually, Independence Party and PMSD formed a majority coalition. To obtain the percentage of votes and seats for the Opposition (dominated mostly by 1 party) and Rodrigues-based parties (since 1982), subtract the above figures from 1
See Seat Distribution here