The Councilor System: Why It Differs from the Legislator System, and SNTV's 30 Years in Local Politics

30-second overview: On January 12, 2008, Taiwan's seventh Legislative Yuan election adopted the "single-member district two-vote system" for the first time — one vote for a candidate, one vote for a party, with legislative seats slashed from 225 to 113 and the electoral method completely rewritten. Earlier that same year, the sixteenth city and county councilor elections still used the "Single Non-Transferable Vote" (SNTV) system inherited from the 1980s — a district electing five to ten members, voters could mark only one name, and excess votes could not be transferred to a same-party colleague. In the same year, the same pool of voters, two offices listed on the same election bulletin, running on two entirely different sets of rules. This is not a technical bug; it is the result of institutional reform deliberately stopping halfway. Behind it lie local factions' and the two major parties' entrenchment in the SNTV structure, the high cost of complementary reforms, and the difficulty of building cross-party consensus — and it is the most underrated unfinished business on Taiwan's 30-year road of democratic deepening.

The Same Year, Two Systems

On January 12, 2008, ballots were counted for the seventh Legislative Yuan election. This was the first time Taiwan's legislative elections used the "single-member district two-vote system" — the nation was divided into 73 single-member districts, each electing one representative; an additional 34 at-large seats were allocated by party vote; and 6 indigenous legislator seats were divided evenly between plainland and mountain indigenous constituencies, 3 each¹. Legislative seats were cut from the previous term's 225 to 113, colloquially known as "halving the legislature"².

On December 13 of the same year, city and county councilor elections were held in all counties and cities except Yunlin and Chiayi (whose schedules were offset due to a 2005 by-election)³. The councilor ballots those same voters received still followed the 1980s-era local council election system — a district electing five to ten members, voters could mark only one candidate, all candidates regardless of party were ranked by vote total, and the top vote-getters won seats⁴.

Two offices, the same year, the same pool of voters. Legislators operated under the "single-member district two-vote system" (Mixed-Member Majoritarian, MMM); councilors operated under the "Single Non-Transferable Vote" (SNTV). In a comparative electoral systems textbook, these two terms would belong in different chapters — one is a mixed system gradually adopted by OECD countries from the 1990s onward; the other is an old system Japan used from 1947 to 1994, later widely assessed by scholars as "encouraging factional politics and hindering party integration"⁵.

The legislative side was reformed. The councilor side was not.

This half-reformed state persisted through 2026 — a full 18 years after the legislative overhaul.

How SNTV Works

SNTV stands for "Single Non-Transferable Vote," translated into Chinese as 「複數選區單記非讓渡投票制」 — breaking down the three key terms:

Multi-member district: A district electing multiple seats (usually 3 or more). In the 2022 Taipei City councilor election, for example, the city was divided into 6 districts, with seats ranging from 8 to 12 per district, for a total of 63 seats citywide⁶. New Taipei City had 66 councilor seats across 12 districts, with 4 to 8 seats per district⁷.

Single-mark: A voter can mark only one candidate on a single ballot. Whether the district elects 5 seats or 12, your ballot allows you to "circle one box."

Non-transferable: Vote totals cannot be transferred between candidates of the same party. If the KMT nominates three candidates in a district and A gets 30,000 votes, B gets 10,000, and C gets 5,000 — A's "surplus votes" do not automatically transfer to B or C.

Combined, these three rules produce a counterintuitive result: the closer a candidate's vote total is to the "election threshold," the better; winning by a large margin is actually wasted votes.

Consider a simplified example. A district elects 5 seats with 120,000 valid votes total. The theoretical election threshold (Droop quota) is roughly 20,000 votes⁸. If a party nominates 2 candidates and A receives 50,000 votes while B receives 10,000 — A wins by a landslide but far exceeds the threshold, while B falls short and loses, so the party wins only 1 seat. If A and B each receive 30,000 votes, both clear the threshold and both win. The same 60,000 party votes, distributed differently, can mean a difference of one seat.

This is why, under SNTV, same-party candidates engage in "vote allocation" (配票) — before the election, supporters are divided into equal blocs by various methods (birth month, last digit of national ID number, geographic sub-district, household surname initial) to prevent vote concentration on any single candidate⁹. The KMT has commonly used "last-digit-of-ID vote allocation" in Taipei City councilor elections; the DPP has used "geographic sub-district allocation" in New Taipei City councilor elections; the Taiwan People's Party also attempted vote-allocation appeals in its 2022 electoral debut¹⁰.

Successful vote allocation maximizes the party's total seats. Failed vote allocation can result in the party with the most total votes actually winning fewer seats than its opponent. This is the core tactical dimension of the SNTV system — not "maximizing total party votes," but "precisely hitting the seat threshold."

Why SNTV Was Adopted in the 1980s

The institutional roots of SNTV in Taiwan run deeper than generally assumed. During the Japanese colonial period, the first prefectural and municipal council elections in 1935 used a limited-vote system (where the number of candidates a voter could select was fewer than the seats to be filled). After the war, the ROC government's local self-governance laws changed the electoral system to SNTV — a single vote for one candidate¹¹. From the 1950s onward, Taiwan Provincial Council, city and county council, and township council elections all used SNTV, a framework that persisted for over seventy years.

In the 1980s, Taiwanese society was still in the late authoritarian period. Local council elections served as a rehearsal ground for "limited democracy," but SNTV was highly conducive to the KMT's local faction operations — factional candidates did not need to win "majority public support"; they only needed to maintain a base within their cultivated townships, industries, or kinship networks to reach the election threshold. This structure made local factions the core nodes of KMT electoral mobilization and deeply rooted the campaign culture of "vote-broker politics" and "relationship cultivation"¹².

In 1992, when legislators were first fully directly elected, the institutional designers faced a choice: should the legislative election also be reformed to a new system? The prevailing political atmosphere prioritized "returning the legislature to the people," and debate over a major institutional overhaul was shelved. The direct election of legislators continued to use SNTV — the same system as councilor elections, just with larger districts and more seats.

This "legislators under SNTV" phase lasted 16 years (1992–2008, spanning five legislative terms). Every term brought criticism of SNTV: candidates "competed on base mobilization rather than policy platforms"; extreme positions could win easily (consolidating just 5–10% of die-hard supporters was enough to clear the threshold); intra-party competition was fiercer than competition against opponents; money politics and factional mobilization dominated. Academia labeled this the "pathology of SNTV"¹³.

The 2005 Constitutional Amendment: Legislators Were Reformed

The turning point for legislative reform was the seventh constitutional amendment in 2005.

Around the time of Chen Shui-bian's second presidential inauguration in 2004, cross-party criticism of "too many legislators," "poor legislative quality," and "insufficient proportional representation" reached a tipping point. The three main reform demands were: halving the number of legislators, single-member districts, and a party vote (two-vote system). In August 2004, the Legislative Yuan passed a constitutional reform bill combining all three and sent it to an ad-hoc National Assembly¹⁴.

In May 2005, 300 ad-hoc National Assembly delegates were elected nationwide specifically to deliberate on this constitutional reform bill. On June 7 of that year, the Assembly voted to pass the amendment to Article 4 of the Additional Articles of the Constitution — legislative seats were cut from 225 to 113, the system was changed to the single-member district two-vote system, the term was extended from 3 years to 4, and the National Assembly was abolished¹⁵.

Details of the new system:

• 73 district legislator seats: The nation was divided into 73 single-member districts based on population, each electing one representative (first-past-the-post, relative majority)
• 34 at-large legislator seats: Allocated by party vote share, with a 5% threshold
• 6 indigenous legislator seats: 3 for plainland indigenous and 3 for mountain indigenous (the multi-member SNTV system was retained here — the constitutional amendment did not touch this)
• Total 113 seats, 4-year term

The first election under the new system was held on January 12, 2008. The KMT won 81 seats (61 district + 20 at-large) and the DPP won 27 seats (13 district + 14 at-large), a seat ratio of roughly 3:1¹⁶. The first battle under the new system resulted in intensified two-party dominance — the 5% party vote threshold formed a high wall for small parties. But the direction of institutional reform at least shifted from "SNTV, which encouraged internal strife" toward "single-member districts + proportional representation, which more closely reflect majority preferences."

Legislators were reformed.

Councilors were not.

Why Councilors Were Not Reformed Together

The scope of the constitutional amendment was limited to the "central level" — the Legislative Yuan, the presidency, and the National Assembly. Councilor electoral systems are governed by the Local Institutions Act and the Civil Servants Election and Recall Act; they do not require a constitutional amendment, only legislative action. But this path was not completed in 18 years.

Several structural reasons:

Local faction opposition. SNTV is highly advantageous for locally entrenched factional candidates — they do not need to win broad public support, only to maintain their township or industry base. Switching to single-member districts would expand district boundaries, forcing candidates to face a broader electorate, diluting factional structures. From the perspective of both the KMT's and DPP's local political structures, this reform would harm their own people¹⁷.

Cross-party consensus is difficult. During legislative reform, the KMT and DPP shared consensus on "halving the legislature" (both parties agreed there were too many legislators) and could reluctantly agree on "single-member districts" (both major parties were competitive in single-member districts). But at the councilor level, the situation is different — SNTV represents a relatively stable equilibrium for both parties' local territories. Who would benefit and who would be harmed after a reform was unclear to both sides. No one had an incentive to move first.

High cost of complementary reforms. Councilor reform involves more complex complementary measures than legislative reform — should the number of councilors be halved? Should districts be redrawn? How should indigenous reserved seats be handled? How should women's reserved seats be addressed? Should reforms be applied simultaneously across all three tiers — special municipality councilors, county/city councilors, and township councilors? The district-redrawing engineering alone for 22 county and city councils plus 198 township councils across the nation is an order of magnitude larger than redrawing 73 legislative districts¹⁸.

Insufficient public momentum. Legislative reform had the simple slogan "halve the legislature" to mobilize public opinion. Councilor system reform lacks such a banner — ordinary voters do not feel the difference between SNTV and single-member districts, and they are highly sensitive to the costs of electoral reform. Councilor elections proceed normally every four years, with no crisis event of "systemic collapse."

With all four factors叠加, councilor system reform remained in place. Every few years, legislators or scholars would reintroduce proposals, only to have them shelved on the Legislative Yuan's legislative agenda¹⁹.

Structural Differences Between the Two Systems

The legislator system (single-member district two-vote system) and the councilor system (SNTV) produce structurally distinct outcomes in several notable ways.

Impact on small parties:

SNTV is "relatively friendly" to small parties — because the election threshold is a "relative threshold determined by the number of seats to be filled," not an "absolute majority." A district electing 10 seats has a theoretical threshold of roughly 9–10%; a district electing 5 seats has a threshold of roughly 17–18%. A small party that can push its vote share to around 10% in a given district has a chance of winning a seat. Small parties such as the Taiwan Statebuilding Party, the New Power Party, the Green Party, and the Social Democratic Party have all won seats at the city/county councilor level²⁰.

But the single-member district two-vote system imposes a stricter dual threshold on small parties — single-member district seats are "winner-take-all," making it nearly impossible for small parties to win them; they must rely on clearing the 5% party vote threshold for at-large seats to enter the legislature. The New Power Party won 5 legislative seats via at-large votes in 2016, only 3 in 2020, and was completely wiped out in 2024²¹. The Taiwan People's Party won 5 at-large seats in 2020 and, in 2024, won 0 district seats but 8 at-large seats²².

From this structural perspective, third forces actually find it easier to break through at the "councilor level" than at the "legislator level" — because of SNTV's lower threshold. But the councilor system lacks the "at-large" safety net; all councilor seats are produced through single-mark non-transferable competition in multi-member districts.

Impact on candidate behavior:

SNTV encourages candidates to cultivate "factions" over "issues" — because consolidating a base is enough to clear the threshold, there is no need to persuade a majority of voters. Single-member district systems encourage candidates to court "median voters" — because winning a relative majority within the district is necessary. The impact of the two systems on democratic quality is a long-standing academic debate that will not be elaborated here²³.

Impact on party competition:

Under SNTV, competition between same-party candidates can be fiercer than competition against opponents — because candidates must first secure the party nomination and then allocate votes. Under the single-member district legislator system, it is "one district, one candidate"; after the party primary, the nominee faces the opponent in a one-on-one contest.

Special Designs in Councilor Districts

Beyond the main SNTV framework, councilor districts have two unique reserved-seat designs that do not exist in the legislator system (of the 113 legislative seats, only 6 are indigenous seats; there are no gender-based reserved seats).

Indigenous reserved seats:

In city and county councilor elections, counties and cities whose indigenous population reaches a certain threshold establish indigenous councilor seats, divided into "plainland indigenous" and "mountain indigenous" categories, each with independent districts and independent vote counting. For example, of New Taipei City's 66 councilor seats, 1 is reserved for plainland indigenous and 1 for mountain indigenous²⁴; of Pingtung County's 55 councilor seats, 2 are reserved for plainland indigenous and 4 for mountain indigenous (Pingtung has seven mountain indigenous townships: Wutai, Majia, Taiwu, Laiyi, Chunri, Shizi, and Mudan)²⁵.

This design reflects the fact that local councils face the geographic distribution of ethnic populations more directly than the Legislative Yuan — the proportions of plainland and mountain indigenous peoples vary across different counties and cities, requiring localized seat guarantees.

Women's reserved seats:

The Civil Servants Election and Recall Act stipulates that in local councilor elections, "where the number of seats to be filled is four or more, the number of women elected shall not be less than one-fourth"²⁶. This means: for every 4 seats to be filled, at least 1 must be reserved for a woman.

How it works in practice: after the vote count, if the number of women elected in a district falls short of the reserved-seat threshold, the lowest-scoring male winner is "pushed out" and replaced by the highest-scoring female candidate who did not originally win. This design has been in place since the 1990s and is the key institutional guarantee behind the fact that Taiwan's female councilor ratio has remained above 35% for an extended period²⁷.

After the switch to single-member districts for legislators, women's reserved seats could no longer be applied (a single-member district elects only one person, making it impossible to reserve one-fourth for women), and the alternative adopted was "a 1/2 gender ratio requirement on at-large party lists." The two designs reflect the structural differences between the two systems — SNTV guarantees within multi-member districts, while the single-member district system guarantees within party proportional representation.

International Comparison: How Similar Systems Were Handled

SNTV is not unique to Taiwan. Several East Asian countries historically used it, and most have since reformed.

Japanese House of Councillors: The national district used SNTV from 1947 to 1982, then switched to a party-list proportional representation system in 1982. Local districts (prefectures) still primarily use SNTV today (some districts are single-member). The Japanese House of Representatives used SNTV for all seats from 1947 to 1993, then switched to a "single-member district proportional representation parallel system" in 1994 — nearly identical in direction to Taiwan's 2008 legislative reform²⁸.

South Korean National Assembly: Primarily single-member districts (253 seats), plus 47 proportional representation at-large seats. SNTV was never used²⁹.

German Bundestag: Mixed-Member Proportional (MMP) system — one vote for a candidate, one vote for a party, but final seat allocation is based on the party vote, with district winners drawing from the party vote allocation. This is widely regarded by scholars as the most small-party-friendly and proportionally accurate mixed system design. During Taiwan's 2005 constitutional amendment, the MMP system was considered, but the final choice was the "parallel system" (district seats and at-large seats do not affect each other)³⁰.

Singapore: Uses the "Group Representation Constituency" (GRC) system — multi-member districts but with parties nominating team slates, all-or-nothing. Completely different from SNTV.

From an international comparison, SNTV has been a minority system since the 1990s. Japan's 1994 reform is seen as one of the "farewell performances of SNTV." Taiwan's councilor system is one of the few national-level democratic systems still using SNTV in its entirety.

Observation Points for the 2026 Councilor Elections

On November 28, 2026, the nine-in-one elections will use SNTV for councilor races as before. Several structural issues worth watching:

Vote-allocation battles: Will the two major parties continue using vote-allocation appeals in hotly contested districts (such as Taipei City's sixth district, New Taipei City's third district, and Taichung City's sixth district)? Vote-allocation success rates were not high in 2022 — the Taiwan People's Party saw "lead candidate overvotes, running mate loses" vote-concentration problems in multiple districts. In 2026, will third forces (the Taiwan People's Party, the Statebuilding Party, remnants of the New Power Party, and emerging parties) use vote-allocation tools more skillfully?

Women's reserved seats: Of the 988 councilor seats nationwide in 2026, an estimated 250+ will be decided by the women's reservation clause or will displace originally elected men. The last seat in each district is often the critical seat determined by the gender reservation clause — this rule has rewritten the outcomes of thousands of elections over 30 years³¹.

Third-force seats: In the 2022 nine-in-one elections, the Taiwan Statebuilding Party, the Taiwan People's Party, and the New Power Party together won 30+ seats in councils nationwide. Whether third-force seats can break through the 50-seat mark in 2026 is an indicator of whether SNTV's "threshold dividend" for small parties continues to function.

No party vote: Councilor elections have no at-large system and no party vote; all seats come from SNTV competition in multi-member districts. This means there is only one path for small parties to win councilor seats — pushing past the election threshold in a given district. This structure makes the "party color" of councilor elections more blurred than that of legislative elections, with local factions, family operations, and personal charisma carrying greater weight.

The reform question remains on the table: Every few years, legislators introduce "councilor reform" bills. Will the new legislature after 2026 (a DPP minority government + KMT/TPP majority) bring this issue to the table? Will proposals such as "halving councilor seats," "single-member councilor districts," or "councilor at-large seats" emerge? This is one observation point for post-2026 Local Institutions Act reform.

The Significance of the Half-Reform

Legislators were reformed. Councilors were not. This half-reformed state has persisted for 18 years.

From an institutional design perspective, this is not the complete story — the complete story would be "after the single-member district two-vote system succeeded for legislators, councilors followed suit," or "the legislative reform proved SNTV was better, prompting a re-examination of the single-member district system." But Taiwan's political reality is that neither path was taken — legislators kept the new system, councilors kept the old system, and the two systems coexist.

This "institutional coexistence" itself can be interpreted in two ways.

One interpretation: a sign of democratic maturity. Different levels use different systems, each playing to its strengths — legislators use the single-member district two-vote system for national issues, while councilors use SNTV to safeguard local diverse representation. This interpretation treats "not reforming" as "should not have been reformed."

Another interpretation: the result of stalled reform. The electoral reform initiated in the 1990s stopped halfway after 2008, and the remaining half stayed in place because the political costs were too high and no one wanted to pay them. SNTV's persistence at the councilor level is not a "design choice" but "no one pushed for change." This interpretation treats "not reforming" as "should have been reformed but wasn't."

Between these two interpretations lies the arena of public opinion, political power, academic debate, and media attention. Taiwan.md does not take either side — but documenting this half-reformed state is essential groundwork for observing the 2026 elections.

On election day, the ballots voters receive will not include legislative races (legislative elections coincide with presidential elections, not held again until 2028), but the councilor ballots will still run on rules inherited from the 1980s. The same ballot, old and new systems coexisting — this is the most easily overlooked yet structurally deepest detail in Taiwan's democratic infrastructure.

For detailed institutional discussions, see: Politics Hub · 2026 九合一選舉 (2026 nine-in-one elections) · 九合一選舉是什麼 (What are the nine-in-one elections) · 中選會制度 (Central Election Commission system) · 村里長制度 (Village chief system) · 直轄市山地原住民區長 (Special municipality mountain indigenous district chiefs) · Taiwan's democratization · Taiwan elections and party politics

Further Reading

• Central Election Commission Election Database — Historical councilor election vote structures, vote-allocation records, and gender-reserved seat statistics
• Wang Yeh-lih (王業立), Comparative Electoral Systems (比較選舉制度) — The most comprehensive textbook on electoral systems in Taiwan; primary reference for SNTV and MMM system comparisons
• Shen Yu-chung (沈有忠), series of electoral systems papers — Political economy analysis of the 2005 constitutional amendment process and the single-member district two-vote system reform
• Legislative Yuan Gazette, August 2004 — Records of the third reading of the constitutional reform bill and cross-party negotiation process
• Ad-hoc National Assembly meeting minutes, June 2005 — Final resolution text on halving the legislature and single-member district reform