2020-04 Taiwan: everyone's electoral system?

TAIWAN ELECTIONS, 11.01.2020      OBSERVATION REPORT, SUPPLEMENT

THE DE BORDA INSTITUTE                 www.deborda.org

1          A Voters’ Profile 

Comparing the accuracy of different decision-making systems – from binary voting to preferential – is relatively easy.  Because of PR, however,[1] electoral systems vary enormously, and there are over 300 of them.  Suffice at this stage to say that, for any one electorate at any one time with any one set of preferences, a change in the electoral system might cause a huge difference in the results.

Consider, then, a democratic nation of just 240 voters: 12 little villages of 20 voters each, preparing to elect a parliament of 12 members: a simple setting which translate easily into a Taiwanese scenario.  And let it be assumed (a) that society is very homogenous – there are no ghettoes, rich suburbs, sectarian enclaves or other partisan concentrations of voters – and (b) that every village is the same, politically: 20 voters with 1^st preferences for the various parties/candidates – A, B, C and D – as shown in Table I. 

Table I           10  voters’ 1^st preferences 

	Number of voters
	8	3	4	5
1st preference	A	B	C	D

For those who believe that politics should be majoritarian, there is no majority in favour of any one party, so the corollary is also true: there is a majority against every party.  An accurate assessment of the voters’ collective will is therefore more likely to be achieved by an analysis of the voters’ preferences.  A full set is assumed to be as shown in Table II.

Table II          10  voters’ profile 

Preferences	Number of voters
	8	3	4	5
1st preference	A	B	C	D
2nd preference	B	C	B	B
3rd preference	C	D	D	C
4th preference	D	A	A	A

On the face of it, opinions on party/candidate A are very divided; those on D are rather less polarised; party/candidate C enjoys some overall support but, with a 1^st or 2^nd preference from every voter, B is obviously the most popular.  So with 12 MPs to be elected, maybe the fairest result would be 4 B, 4 C, 2 D and 2 A, or perhaps 4 B, 3 C, 3 D and 2 A.  But what happens in practice when different electoral systems are employed?   

2          The Setting

Non-PR systems take place in single-seat constituencies.  With PR, the minimum constituency size is normally taken to be a 3-seater.  This paper will examine various electoral systems in one 12-seater, two 6-seater and/or three 4-seater constituencies, as appropriate.  The analyses for each of several electoral systems in this hypothetical democracy of 240 voters are shown in Table III.

Taiwan’s parliamentary elections relate to 73 MPs under first-past-the-post, FPTP, and about half that number, 34, under a closed PR-list system with the Hare quota; in the following analysis, Taiwan’s numbers of 8 and 4 mps are therefore considered to be an accurate reflection.  Germany’s system is half-and-half – half FPTP and half PR-list – so this means 6 MPs would be elected under FPTP, and 6 with a PR-list system under d’Hondt.  Slovenia’s BC elections apply only to her ethnic minorities and is here analysed in three 4-seaters, while Denmark uses open PR-list to elect 175 MPs in 10 constituencies, so this best translates as one constituency of 12 MPs. 

3          The Analyses

It must be emphasised that the following results for the first three systems listed – FPTP, TRS and AV, the two-round system and the alternative vote – would apply only if society was, yes, homogenous, socially, economically, religiously, ethnically, but not necessarily politically.  Nevertheless, at least in this setting, FPTP, TRS and AV are all hopelessly inaccurate; single preference PR-list systems are also not the best, especially if conducted in relatively small constituencies, as in Taiwan, rather than a large one, as in Denmark, but this too is biased.  Taiwan’s semi-PR electoral system, in which such a major component is FPTP, is also liable to produce fake results; Germany’s all-PR system of MMP, multi-member proportional, is much better, because the A party, having gained so many seats in the FPTP part, doesn’t get any more in this analysis from the second part, the PR half. 

The BC is not PR, and nor for that matter is the more inclusive MBC; hence, of course, the introduction of the quota into the quota Borda system, QBS.  In a straight contest, therefore, if the Slovenian BC system were used, party A could nominate three x 4 = 12 candidates – A₁ + A₂ + A₃ + A₄, etc. – and thus win all three x 4 seats.

Ireland, South and North, uses PR-STV.  Southern Ireland is best compared in three 4-seater constituencies, the North to two 6-seaters.  In days gone by, some very large constituencies were quite common, but the counts therein were often pretty complicated, and the maximum in current usage is a 6-seater.  In the 4-seater analysis, A and B are elected on merit, C and D by default; in the 6-seater, A, B, C and D all get a quota of 1^st preferences, A actually gets two of them, and D gets a second success by default: so the final result is 2 A, 2 D, 1 B, 1 C. 

Table III        The Election Results

Electoral System		Used in?	PR or?	The constituencies			Scores per constituency		Results: MPs elected to parliament. Number of MPs of which party…
				No of?	No of voters in each?	No of MPs in each?			… in the single constituency	… overall
FPTP		uk	x	12	20	1	A-8. D-5, C-4, B-3		A	12 A
TRS		France	x	12	20	1	D-12, A-8		D	12 D
AV		Australia	x	12	20	1	C-12, A-8		C	12 C
Par- allel	FPTP	Taiwan	semi	8	20	1	A-8. D-5, C-4, B-3		8 A	9 A 1 D 1 C + 1 A/B*
	PR			1	240	4	1 A 1 D 1 C + 1 A/B*
MMP	FPTP	Germany	✓	6	40	1	A-8. D-5, C-4, B-3		6 A	6 A 2 D 2 C 2 B
	PR			1	240	6	3 A 1 D 1 C 1 B
BC		Slovenia	x	3	80	4	At best:	4 x (B-63, C-51, A-44, D-42)		3 B + 3 C + 3 A + 3 D
							At worst:	4 x (A1 + A2 + A3 + A4 ) = 4 A		12 A
PR-list		Denmark	✓	1	240	12	5 A + 3 D + 2 C + 2 B
PR-STV		Ireland	✓	3	80: q = 21	4	1 A + 1 B + 1 C + 1 D		3 A + 3 B + 3 C + 3 D
				2	120: q = 18	6	2 A + 2 D + 1 B + 1 C		4 A + 4 D + 2 B + 2 C
QBS			✓	3	80: q = 21	4	1 A + 1 B + 1 C + 1 D		3 A + 3 B + 3 D + 3 C
				2	120: q = 18	6	2 B + 2 C + 1 D + 1 A		4 B + 4 C + 2 D + 2 A

*          Parties A and B tie for the fourth seat.   q = quota

4          Conclusion

It is quite possible that despite the dominance of FPTP which can be so hopelessly inaccurate, the results of the recent Taiwanese elections would not have been very different from those declared on election day.  In the UK, however, if PR-STV or any other form of PR had been the norm, the results would have been very very different: with nearly 1 million votes, the Green Party won 1 seat; with 17 times as many votes, the Tory party won not 17 times but 365 times as many seats!  That’s outrageous, you might think.  Indeed, the choice of electoral system is often outrageous: parties invariably choose that which is in their vested interest.  Sadly, human rights lawyers and others rarely compare the various methodologies’ democratic credentials and/or their relative merits, and as stated in the main report, nor too do most international election observation missions.

In the above comparison at least, when compared on a scale of accuracy, the top electoral system is QBS, ideally, in six-seater constituencies.  Next comes PR-STV, then, maybe, a BC.  After that it’s Denmark’s, Germany’s and Taiwan’s systems, in that order.  The rest, as stated, are just hopelessly inaccurate.

5          Recommendation

That Taiwan gives consideration to the introduction of an electoral system which is both preferential and proportional.  The use of a non-binary and therefore multi-optional form of parliamentary decision-making might also help to create that which could serve as a brilliant example of an inclusive democracy, not least to its nearest neighbour.

Peter Emerson

Director, the de Borda Institute

Belfast BT14 7QQ

Northern Ireland

www.deborda.org

2020.01.19      Liaocheng, China.

Abbreviations

AV*      =          alternative vote             BC =          Borda count                 FPTP =          first-past-the-post

IRV*     =          instant run-off voting    MBC =          modified BC MMP =          multi-member proportional

PR =          proportional representation    PV*      =     preference voting     QBS =          quota Borda system               

STV *     =          single transferable vote    TRS =          two-round system

*          Four different names for the same system.  When used in single-seat constituencies, it is normally referred to as AV; in multi-member constituencies with PR, the system is called PR-STV; IRV is the American term; while in Australasia it is called PV, preference voting.

---

[1]           Invariably in decision-making, the final outcome can be only a sole decision, or just one prioritisation, or whatever, but always a singleton.  In contrast, when electing representatives, there may be just the one winner in a presidential contest or in a single-seat constituency in a parliamentary election, or there may be quite a few in a multi-member constituency, as in Ireland’s three- or four-seaters, or well over a hundred when the entire country is just the one constituency, as is the Dutch system.