Stability is defined as the difference in seats between winner and second party

Unfairness is measured as the difference between share of seats and votes of unsuccessful party.

Assume that the other 6 MPs neutralise each other in terms support for A and B

The above table shows clearly how Sachs privileges stability while injecting a measure of fairness in the system. It recommended 30 compensatory PR. The other combinations are too unfair. With 10 seats, even the compensatory method yields an unfairness of 32%, as there are not enough seats to correct FPTP perversions. Stability is assured by a massive 50 seats difference. The parallel mode is very unfair at all levels (32%, 35% and 40% with 30, 20 and 10 seats respectively) . It should be noted that, in all cases, stability is strong. Even in the compensatory method, the winning party has a seat majority of 30 with 30 PR seats. Of course the seat-vote disproportionality has come down. But then this is the very objective of the reform. The problem with the parallel system is that it reinforces the huge disparity that exists after the FPTP mode. Not only has Party A been largely advantaged in the FPTP formula with 100% of the seats with 55% of the votes, but it also takes the lion?s share in the PR allotment. As a matter of fact the difference in the number of seats between the two parties increases with such allotment. For instance, with 30 PR seats, the difference of 60 seats with FPTP rises to 64 after the parallel allocation. It makes matters worse. As rightly stated by Dr Chu ?parallel PR hardly reduces the yawning gap between votes and overall seats when a party has a disproportionate share of FPTP seats. Worst still, it has the perverse effect of increasing the lead acquired via FPTP seat? as shown above.

If we want a parallel system to yield a relatively fair result (say a maximum of 15% difference between votes and seats), we need many more PR seats than 30 if we keep 70 FPTP seats. For instance a parallel system with 140 PR seats and 70 FPTP seats would give the party that polls 45% around 30% of seats if it has lost all constituency contests (60-0). This shows the degree of unfairness of a parallel mode if the number of seats is very low! As a matter of fact, a compensatory PR with 10 seats yields the same level of unfairness as a parallel one with 30 seats (32% in both cases). For fairness it is better to have 20 seats allotted by a compensatory mode than 30 using a parallel one.

In effect Model C has adopted what Sachs recommended. In turn, the latter has simply embraced what electoral specialists advocate in such a situation. It is still unfair with a 14% deviation. To lower this further would require PR seats to account for more than 30 % of the total. This is the case in Germany (50%) and New Zealand (45%) where the level of unfairness is lower than 10%. Lesotho has 80 FPTP and 40 PR seats that are allocated along a compensatory mode without any threshold compared to our 10%!

It is interesting to note that because of its acute unfairness, Sachs did not even rank

Model A as the second choice after Model C. As a matter of fact Model A was very far behind

Model C and Model B (compensating only the unsuccessful party for its underrepresentation). The report states:

?If there were no system which we regarded as manifestly superior, we would opt for PR Model B?. (Para 46). And the manifestly superior system is Model C!

As the choice of 70 FPTP seats out of a total of 100 signals a very high ranking of stability over fairness and as the other three features also support stability, the allotment of the 30 PR seats should use a compensatory formula that privileges fairness over stability. So as to ensure a measure of fairness and inclusivity in the overall system while still giving a big edge to stability. The first priority is stability. Hence the 70 % of FPTP seats, the 10 % threshold, the national list and d?Hondt. But the second objective of fairness must be met by the compensatory mode. Otherwise the system would be extremely biased and would still yield huge disproportion between votes and seats.

Taming the Rodrigues tiger

The result of the first elections held in Rodrigues with a mixed system has become an unfortunate fig-leaf in the debate on the merits of alternative formulae for electoral reform. Yet there is nothing apprehensive or mysterious in the outcome of these elections. It simply reflects the reality of political forces in Rodrigues and mirrors the geographical concentration of the two parties. Above all it is the direct result of the very fact that there are too few seats overall, leading to the problem of indivisibility. It is a fact that 58% and 42% of 6 PR seats yield the same figure (i.e. 3) when rounded to the nearest whole number, even if the difference in votes is a good 16%. Equally 52.7% of 18 is not different from 47.3% of 18! Both give 9. Such an outcome is very unlikely in Mauritius, as the number of seats is significantly higher at 100. 58% of 30 PR seats give 17 while 42% yield 13.Equally 52.7% of 100 makes 53 compared to 47.3% of 100 that produces 47.

The summary of the election results is as follows:

Some people are puzzled that the absolute difference between the two parties has shrunk from 4 after the FPTP elections to 2 subsequent to the allotment of the 6 PR seats. This is indeed the very essence of the formula. The OPR took 67% of the FPTP seats with 55 % of the votes (a deviation of + 12 %) while MR captured 33 % of these seats with 44% of votes (a deviation of -11%) .The OPR has thus a higher share of seats compared to its vote share and the reverse holds true for MR. With 55% of the vote, the winner of the FPTP contest remains the winner after the allotment of the 6 PR seats, but the wedge between the two parties is narrowed from 4 to 2 seats. This is due to the relatively higher compensation received by the second party in the PR mode, as it has been ?disadvantaged? in the FPTP mode. The degree of compensation depends on the percentage of votes polled by eligible parties. The award of 4 PR seats to MR and 2 to OPR reflects the twin facts that the OPR won 8 of the 12 constituency seats with 55% of the votes while the MR captured only 4 of these seats with 44%. Had OPR polled 58% of votes and MR 41%, the apportionment of the PR seats would have been different with three to each party. Overall the OPR would have obtained 11 seats and MR 7.Then the gap of 4 seats after the FPTP election would have remained constant after the application of the PR formula. But then 44% is a sizeable share of the vote that calls for fair representation.

It is sheer coincidence that the final results are very close to a pure PR mode of election. This is attributable to the very small number of seats and the geographical concentration of support for the two parties. With more seats in the system, there would still be a difference between votes polled and seats won (as in the case of Germany in the recent elections where the SDP-Green coalition won a majority of slightly over 50% of seats with around 45% of the votes). Usually the winner has an advantage with its share of seats outstripping its share of votes.

Grafting Rodrigues results on Mauritius

Judgements about the operation of a specific electoral system must be qualified by place. The same electoral system can have different consequences in different political cultures. Those who are concerned by the results in Rodrigues should take comfort from the fact that the reality of Mauritius is very different. Even in the worst-case scenario of a replication of the Rodrigues results in Mauritus, the problem of stability would not arise. This is not difficult to prove. The smaller the number of seats, the more likely the problem of indivisibility. With more seats, there would have been a difference. In Germany there is a difference of around 6 seats even if the percentage of votes is almost identical between the two parties, as there are 600 MPs. With 100 seats in Mauritius, the outcome would have been different.

Consider 60 FPTP MPs, 8 best losers and 30 PR MPs returned along a compensatory system. Assume as in Rodrigues that Party A wins two thirds of the FPTP seats with 55% of votes and Party B one third with 43.6% of votes and that Party A captures 5 best loser seats and Party B three. Using Model C, Party A would receive 10 PR seats and Party B 20. Overall Party A would receive 55 seats out of 98 (the other two are from Rodrigues) and Party B would obtain 43 MPs. Is that an unfair situation? Under a strict FPTP system with 68 seats, Party A would win 45 seats compared to 23 for Party B, thus giving Party A a majority of 22 seats. Under a mixed system, Party A would take 55 seats and Party B would end up with 43 Parliamentarians. While the result is fairer, it still ensures stability through a comfortable majority of 12 seats to the winner. Of course the disproportionality has come down, but then this is the very objective of the reform. The Rodrigues syndrome is caused by the smallness of the majority, which is due to the very low number of seats. A difference of 2 seat represents an 11% majority in an Assembly of 18.The same percentage would give a majority of 11 in a Parliament of 100. And 66 in Germany with 600 seats.

There is one inescapable truth, which many well-intentioned people seem to forget. Results under a mixed system are less unfair than under FPTP. Yet, at times, the very people who advocate reform to cure the unfairness of FPTP seem to show concern when this actually happens. They continue to refer back to FPTP and to its sizeable manufactured majority and its massive disproportionality! This is indeed unfair.