Sri Lankan Parliamentary Elections
Proportional Representation, First-Past-the-Post or Both?
A select committee of the Sri Lankan Parliament, headed by Dinesh Gunawardena, is considering reforms to the electoral system and election laws. Among other topics, the electoral system for Parliamentary Elections is under review.
Up to and including 1977, Sri Lankan Parliamentary Elections were conducted under the First-Past-the-Post (FPTP) system; since 1977 under Proportional Representation (PR). The said reforms might consider a return to FPTP, some mix of FPTP and PR, or retaining PR.
This article contains my own thoughts on these various options. As those of you who’ve read my past articles on Politics known, I’m no expert on Elections or Politics. I’m merely an interested citizen.
First-Past-the-Post (FPTP)
Until 1977, all our Parliamentary Elections were under FPTP. The country was divided into 160 electorates, each* electing one member to parliament. The candidate winning the most votes got elected.
[* except a small number of “multi-member” electorates. More on that later]
“Unrepresentative”
A serious problem of FPTP is that it can be “unrepresentative”; the final seat allocation is not proportional to the votes cast; because the winning party gets the seat, irrespective of how many votes it gets.
For example, if three parties contest a seat and get 34%, 33% and 33%, the first party gets the seat. This outcome is no different from a 100%, 0%, 0% vote division.
In 1977, the SLFP won ~30% of the votes, but only 8 out of 168 seats (~5%). After winning just over half the vote, the UNP won 140 of 168 seats (exactly 5/6ths).
In 1970 and July 1960, “unrepresentativeness” reversed the unfairness; the UNP received more votes than the SLFP but won fewer seats.
“Unrepresentativeness” is particularly sensitive when it favours certain groups (e.g. ethnic, economic or religious) over others. Unscrupulous officials can use techniques such as Gerrymandering electorate maps to exploit this weakness. For this reason, within the pre-1977* FPTP system, certain electorates (e.g. Colombo-Central and Nuwara-Eliya), deemed ethnically diverse, were assigned multiple members. Hence, FPTP was adapted to ensure more representativeness and, in this way, making the overall system (a little) more like PR.
[*By pre-1977, I actually mean “Up to and including 1977]
Proportional Representation (PR)
Under PR, the number of representatives elected was proportional to the number of votes received. Since (most of the) the pre-1977 electorates had only one member, and since it was not possible to “proportion” one member into “fractions”, the concept of “electoral districts” was introduced. An electoral district (ED) is a “super-electorate” with multiple seats. Each party wins a number of seats proportional to the number of votes it wins in the ED.
In the November 2020 Parliamentary Election, 12 Seats were assigned to the Kandy electoral District. The SLPP won 8 seats after winning 59% of the vote, while the SJB won 4 with 29% of the vote. If we used FPTP, the SLPP would have won all the seats in Kandy, having come top in all electorates.
“Working Majorities”
PR also has its problems. In 5 of the 8 Parliamentary Elections since 1977, the party with the most seats failed to win a majority in parliament and sought smaller parties’ support, often after dubious rounds of horse-trading. Hence, some claim that PR makes it difficult to form “working majorities”.
To me, this argument is weak. A parliament should be representative of the people, and if no party has the support of a majority of the population, none should enjoy a majority in parliament. More empirically, many countries (including many successful democracies) have “weak parties” forced to form coalitions with other weak parties. Often, weak parties lead to strong democracy — and vice versa.
However, there is a second, stronger argument.
Weakening the Voter-Representative Link
Consider my home ED, Colombo. In 2020, Colombo elected 19 MPs. Each party and independent group could nominate up to 22 candidates. With over 40 parties and groups contesting, over 800 candidates were contesting Colombo. It is almost impossible for the ordinary voter (myself included) to get one’s head around so many candidates. Voters either end up voting for a few candidates with ED spanning brands and bases (E.g. Ranil Wickramasinghe in Colombo, Mahinda Rajapakse in Kurunegala, Chandrika Bandaranaike in Gampaha, etc.) or else vote for only a party and no candidate.
It also made it impossible for smaller, more local parties to contest elections. Suppose I wanted to represent my home electorate of Borella. In 2020, the SJB came top in the Borella electorate with 20K votes. Hence, under FPTP, just 20K votes would have won me my home seat. However, under PR, since candidates and parties need to compete across the vast electoral district, spanning about 1.7M voters, a significantly larger number of votes are needed. The JBB, which won a single seat in Colombo, received 67K votes, marginally above the 5% cut-off (60K votes) to make a party eligible for seats.
Hence, PR shifts representation from concrete candidates to abstract (and especially large) parties. It becomes difficult for voters to hold their representatives accountable and for the said representatives to feel accountable towards their voters. The link between voter and representative is weakened or even broken. Small parties have no choice but to join larger parties in (often disadvantageous) collisions.
Mixed Options
In summary, FPTP can result in unrepresentativeness, while PR can weaken the Voter-Representative Link.
Can there be some compromise? Can we have it both ways? Can we have a system that is both representative and where the link between voter and representative is maintained?
There are several candidate solutions — none perfect. Here are three popular options.
Option 1: Mixed Member Majoritarian (MMM)
In a Mixed Member Majoritarian (MMM) system, a proportion of the seats are allocated to FPTP and the remainder to PR.
Consider the 2020 Parliamentary Election and the Kandy ED, which was assigned 12 seats. We could split these 12 seats 50%/50%, 6 to FPTP and 6 to PR. Since the SLPP won all the Kandy Electorates, it would get all 6 FPTP seats. If the remaining 6 PR seats were divided, the SLPP would win four and the SJB 2. The final result would be ten seats to the SLPP and two seats to the SJB.
While more “unrepresentative” than pure PR, MMM is less “unrepresentative” than pure FPTP.
MMM is used for Parliamentary Elections in many countries, and the FTPT/PR ratio tends to vary significantly. The PR fraction is 18.7% in South Korea, 30% in Taiwan, 37.5% in Japan and 68.7% in Armenia.
Option 2: Mixed-member proportional representation (MMPR)
Like MMM, in a Mixed-member proportional representation (MMPR) system, seats are split between FPTP and PR. However, unlike MMM, in MMPR, the PR seats are allocated in such a way as to make the final seat allocation proportional.
For example, in the previous Kandy case, of the 6 PR seats, the SLPP would get two, and the SJB would get four so that the final result would be as it was under PR.
Overhang
MMPR has the advantage of being as “representative” as pure PR. However, it has the problem of “Overhang seats”, where one party wins more FPTP seats than it would win under pure PR.
For example, consider the 2001 election. The Colombo ED had 20 seats, of which the UNP, UPFA and JHU won 9, 8 and 3, respectively. Of the 15 electorates, the UPFA and UNP won 9 and 6, respectively. If we assigned seats according to MMPR with an FPTP-PR ratio of 15–5 (75%-25%), then to achieve the same result as pure PR, we would need to assign the UNP and the JHU 3 seats each take-away one seat from the UPFA.
In practice, one of two solutions is applied to the overhang problem.
- The first is to ignore the overhang. For example, the 5 PR seats might be assigned 3 and 2 to the UNP and JHU, respectively, in our example. The disadvantage of this approach is imperfect proportionality, and hence some “unrepresentativeness”. In this example, the unrepresentativeness favours the UPFA against the JHU.
- The second is to add more seats to enable proportionality. For example, if Colombo had 23 seats, the PR assignment would be 10, 9 and 4 to the UNP, UPFA and JHU. After the 15 FPTP seats are assigned, the remaining 8 PR seats can be assigned 4 and 4 to the UNP and JHU, resulting in a match with perfect PR. The downside of this approach is that adding seats might result in parliament having a variable number of members. Also, adding seats to Colombo would mean that seats would have to be added to other EDs to ensure overall representativeness.
Option 3: Smaller EDs
The weaker link between voters and representatives under PR is worst in large EDs like Colombo and Gampaha, with 18 and 19 seats respectively, and a correspondingly large number of candidates per party and group.
It is less problematic in a smaller ED like Trincomalee, with only four seats. With fewer electorates and fewer candidates, the connection between voter and representative is stronger than a larger ED.
We can achieve the same effect in a large ED like Colombo by splitting it into smaller EDs. For example, the 19 seats could be split across five smaller EDs, with a maximum of 4 seats each.
While “Smaller Electoral Districts” is not strictly a mixed option and not referred to as one, I listed it here for the following reason: FPTP is an “extreme case” of PR where each electoral district has exactly one seat. General PR puts no limit on the size of the ED. Hence, PR with “Smaller EDs” is somewhere between FPTP and general PR.
On a historical note, the smaller EDs would not be dissimilar to the small number of Multi-Member Electorates that operate within the FPTP system. For example, in 1977, the Colombo-Central and Nuwara-Eliya Electorates (not to be confused with the Colombo and Nuwara-Eliya Districts) had three seats each. The three candidates getting the most number of votes won these seats.
Concluding Questions
The topics discussed above lead to the following questions:
Question 1: What would the electorate map look like if we returned to FPTP or adopted some mixed system?
If the Parliamentary Select Committee proposes a return to FPTP or a mixed system, we must redefine electorates and redraw the electoral map. This is because the current electoral districts vary significantly in size.
For example, at the 2012 census, Colombo-West had a population of ~54K, while Kaduwela had 252K or almost five times as large. It would be unrepresentative for both to have one seat each. Hence, Colombo-West would need to be merged into some other electorate, or Kaduwela is split into multiple electorates or both.
Question 2: How would we split large EDs like Colombo into smaller EDs?
The question would be relevant to Option 3.
Question 3: How would past election results be different, under different systems?
For example, if the 2020 Parliamentary Election was held under FPTP or some mixed system, would the SLPP have won a 2/3 majority?
I hope to answer these questions in future articles.
…
If you enjoyed reading this article, you might also like,
