review insight



“We tried to look at our proposed divisions through the eyes of the elector and the candidate. The elector should see clearly why he is in one division rather than in another...the area should not be bigger than the elector would travel normally and with ease. He must see that he has a common interest with his fellow electors. From the point of view of the candidate, the division must not be too large in number or diverse in interest for him to canvass his electors or care for his constituents once elected.”
The committee’s terms of reference were to note centres of population, geography, communications, density and adequate representation of rural areas.


Changi Punggol-Tampines Serangoon Paya Lebar Whampoa Ulu Bedok Kampong Kapor Geylang Katong Stamford Havelock Telok Ayer Tanjong Pagar Tiong Bahru

Bukit Panjang


Bukit Timah


Farrer Park

Cairnhill Rochor Queenstown Pasir Panjang

300,000 electors
24 divisions 12 urban 5 mixed 7 rural

Southern Islands

THE disappearance of several high-profile constituencies over successive elections has bred cynicism among some voters about the process by which electoral boundaries are decided. Articulating commonly held concerns, constitutional law expert Thio Li-ann asks: “Was it to diffuse a voting bloc which was not sympathetic to the incumbent government? Was it really just an issue of changing demographics? People can draw their own inferences.” Insight checked the reports to see if there were demographic reasons behind these changes.

It disappeared in 1988 after it was won by Workers’ Party leader J.B. Jeyaretnam, first in the 1981 by-election and again in the 1984 General Election. In its report, the committee highlighted the “perennial problem of declining electorate in the Central Area constituencies”. There were seven, of which Anson was one. Of these seven, five had electorates that fell under the minimum size of 15,000. With 18,303 electors, Anson made the cut – and in fact had the biggest number of electors. Yet it was spliced into four pieces and scattered to the other constituencies, while Jalan Besar and Tanjong Pagar – with fewer than 15,000 electors each – survived.

Making sense of electoral boundaries
When boundaries were drawn in 1958, the constituencies were regular and compact in shape. Explanations were set out in a 42-page report, with one page for the PAP’s dissenting view. But in 2006, the report had just seven pages, with the reasons reduced to one paragraph. So, how did Tanglin evolve into Tanjong Pagar GRC? Insight finds out what happened over the years.
en-page study in minimalism. Of 23 constituencies, about half – 11 – had their boundaries changed. The committee offered a one-paragraph explanation: It had taken into account “population shifts and housing developments and balancing against other factors including the requirement to maintain a minimum of eight single-member constituencies (SMCs) and keeping, as much as feasible, the number of electors in each constituency within the respective ranges”. tives of HDB and the Singapore Land Authority. But given that the committee is usually convened for just a few months before each general election, much of the preparatory work in delineating proposed boundaries is actually done by its secretariat – the Elections Department, whose head serves as the committee’s secretary. The Elections Department is a division within the Prime Minister’s Office. Insight approached members of previous committees but they declined to be interviewed. What is known is that the Prime Minister issues the terms of reference for the committee’s work. These terms of reference may be based on stipulations in

is very much a discretionary process. So the only one privy to the relevant factors is he who determines what ‘relevancy’ is.”

Rise of odd-shaped GRCs with snaking tentacles
WHAT is clear is that in Singapore, the sands shift fast, both literally and metaphorically. For one thing, constant reclamation works necessitate electoral boundary changes to take in the changing borders. At the same time, Singapore’s population has increased steadily, and with it, the number of electors – from 300,000 in 1954 to 2.16 million in 2006. At least another 154,000 people will be added to the rolls by the next election, totalling 2,311,582 as of March this year. Factor in a fast-changing, fast-developing city-state where construction cranes demolish and build in ever-shorter cycles, and the committee has a numerical challenge on its hands. How, then, to slice the pie? One fundamental principle has been mathematical equality: that the divisions have a roughly equal number of voters. The clearest and most unequivocal articulation of this principle came in the 1958 report, which said: “Most important of all is the democratic principle that wherever he votes, one elector’s weight should carry equal weight with another. “There should therefore be the nearest approximation to equality in the number of electors registered in each division.” It explained why: The stakes are high when the number of electors differ. A candidate who polls 10,001 votes will win in a division of 20,000 electors but not so in another of 30,000. In subsequent years, this principle was regularly cited. In 1967, the committee was tasked to “recommend changes necessary in ensuring more equal representation throughout all constituencies in accordance with the register of citizens”. But the call disappeared by 1991, although in 2006, the committee gave a slight nod to the principle when it wrote that it had tried to keep “as much as feasible, the number of electors in each constituency within the respective ranges”. The number of electors divided by the number of MPs yields the average number of electors one division should have. A 20 per cent deviation rule was introduced in 1963 to allow for a range within which the number could fluctuate. In 1980, the committee increased it to 30 per cent. Outliers then generally have their boundaries adjusted. In 1984, for instance, burgeoning HDB new towns led to rapid population growth in the suburbs. With the 30 per cent rule, 19 of 75 constituencies fell outside the range. The three with the smallest number of electors – Bukit Ho Swee, Havelock and Katong – were eliminated. The rest had their

It disappeared in 1996 after a narrow People’s Action Party (PAP) victory of 52.4 per cent in the previous election in 1991. Indeed, it was too big. It contained 170,038 electors, far beyond the maximum size of 120,000 for a four-MP GRC. It was then divided among three GRCs. What was unexplained was why the other outlier was treated differently. Sembawang GRC, too big for a four-MP GRC, had two MPs added to it so that it fell within the range for a six-MP GRC.

It disappeared in 2001 after the PAP won with 54.82 per cent in the 1997 election. It was too big as a five-MP GRC, with 172,474 electors, outside the maximum size of 160,000. It was then divided among three GRCs. Again, other outliers were treated differently. Sembawang GRC was also too big but remained. Jalan Besar GRC and West Coast GRC, which were too small, absorbed parts of neighbouring constituencies so that their sizes fell into range. Just as curious was the treatment of opposition-held Potong Pasir. Its electorate size fell below the minimum in both 2001 and 2006 – but the constituency remained intact. PAP chairman Lim Boon Heng offered a hint in 2006: “Strictly speaking, by the criteria, it should have been absorbed elsewhere. So it would appear that the committee is also sensitive that there will be public criticism if that constituency is absorbed into the GRC.”


These SMCs were subsumed under GRCs in 2006, even though in terms of electorate size, neither was too big nor too small. Their veteran PAP MPs, Dr Tan Cheng Bock and Mr Wang Kai Yuen, stepped down just before the 2006 election. Their vote share in the 2001 election: 87.96 per cent and 77.37 per cent respectively.

Commenting on the vanishing constituencies, law professor Eugene Tan says: “I think there is a patent need to enhance the confidence level with regard to the committee’s work. Too often, Singaporeans view the redrawing of boundaries as calculated to benefit the PAP or to undermine the opposition.” He believes the electoral boundaries reports of today need to “better articulate the reasoning and justification for their recommendations”. He argues: “This is necessary for sustaining confidence and legitimacy, since the redrawing of electoral boundaries is an integral part of the electoral process.”

F IN 1958, residents of Kim Keat wondered why they were voting as part of the Kallang division rather than Toa Payoh, the reason was this: They were not farmers. Unlike their neighbours in Toa Payoh who lived in attap huts, reared pigs and grew rambutan trees, they lived in Singapore Improvement Trust flats in Kim Keat Avenue. “There did not seem adequate community of interest between those living in the Singapore Improvement Trust Estate in Kim Keat Avenue and the farmers of Toa Payoh,” explained the committee that reviewed the boundaries that year. Thus, it was more meaningful that Kim Keat residents should vote as part of the same constituency as the other public housing residents in Kallang. A constituency should after all be bound by a common, distinct interest, the committee said. Carefully reasoned out, the rationale can be found in Section 23 (iv) of the constituency boundary report released that year, available at the General Post Office for $1.50 – the price of a movie ticket. With 42 pages densely covered with population numbers, maps and boundary options, it was a tome by today’s standards. It went into detail on the composition of the committee – representatives of the four political parties, its principles, down to its deliberations – and disagreements. This included a page-long dissenting opinion by then People’s Action Party (PAP) chairman Toh Chin Chye on how the rural parts were partitioned. By contrast, the most recent report released ahead of the 2006 General Election was a slim, sev-

Left in the dark on thinking behind changes
OVER the years, the reports got progressively thinner. In 1976, the committee – then chaired by Housing Board chief executive officer Teh Cheang Wan, who later become Minister for National Development – stopped offering detailed explanations for changes it made. Since then, questions of why a constituency remains intact, is carved up, patched together with others, subsumed within another constituency, or disappears off the electoral map altogether, hang in the air after the release of each report. What the Government has said, time and again, is that population shifts are what drive the process. In March this year, Deputy Prime Minister Wong Kan Seng said in Parliament: “Generally, the committee will look at the population changes and then decide on where each of the constituencies would be, including the parts that need to be redrawn. “Why some are changed and why some are not changed, I think much depends on the configuration at the time. That’s very much left to the discretion of the committee.” Since 1980, the committee has been chaired by the Cabinet Secretary, and generally comprises civil servants well-versed in matters of land planning and demographics: a statistician, and the chief execu-


“Villages should not be divided between two or more electoral divisions and similarly villages without common interests should not be forced together.”
1958 committee

the Constitution and Parliamentary Elections Act, such as the prescription that there be a minimum of eight SMCs, which was part of the Act from 1996 to April this year. But guidelines that are not part of any legislation have also been issued. These ranged from the directive in 1957 to pay “due regard” to factors such as geography and communications, to that in 1991 to create constituencies “by wherever possible, the amalgamation of two or more constituencies”. Says constitutional law expert Thio Li-ann: “The problem is that there is no clear criteria for delimitation and so as I understand it, it