From Authoritarian Ruling Party to Electoral Competitor: Why Did the Kuomintang (KMT) Repeatedly Lose to the Democratic Progressive Party (DPP), 2000–2024?

Research Paper v1.0 | October 13, 2026 | Free China Movement

Completed from draft v0.1 (October 12, 2026) after fact-checking, data completion, and scholarly review. The original author's analytical framework and core hypotheses are preserved; all election figures have been rechecked against results published by the Central Election Commission (CEC) of the Republic of China (Taiwan).

Abstract

In 2000, the Republic of China (Taiwan) achieved its first change of ruling party at the central level, with Democratic Progressive Party candidate Chen Shui-bian elected President. Of the seven presidential elections since, the Kuomintang won only in 2008 and 2012 (Ma Ying-jeou), losing in 2000, 2004, 2016, 2020, and 2024. Using the seven presidential elections as the analytical spine, this paper systematically tests seven competing hypotheses: national-identity transformation, adaptation of historical narratives, cross-strait and security-policy evaluation, party organization and mobilization capacity, generational replacement, short-term campaign factors, and opposition-coordination costs.

The paper's core methodological claim is that presidential, legislative, and local elections must be analyzed as three distinct levels, and long-term structural change must be tested separately from short-term campaign effects. The empirical record supports no single determinism — neither "national-identity determinism" (the share identifying as Taiwanese has risen steadily, yet the Taiwan People's Party's breakthrough among young voters in 2024 shows identity does not equal party support), nor "irreversible KMT decline" (the KMT remained dominant in the 2018 and 2022 local elections and became the largest party in the 2024 Legislative Yuan). Lai Ching-te's 2024 victory with only 40.05% of the vote, alongside the DPP's simultaneous loss of its legislative majority, is direct evidence of split-ticket voting across electoral levels.

Keywords: Taiwan elections; Kuomintang; Democratic Progressive Party; party turnover; national identity; cross-strait relations; split-ticket voting; opposition coordination

1. Research Question and Conceptual Definitions

"Losing a long-term governing advantage" is not the same as "losing continuously," and neither equals "irreversible decline." Rigorous research must measure five dimensions separately: presidential outcomes, Legislative Yuan seat structure, local executive map, party identification, and the age composition of the electorate. Conflating these dimensions produces false causal inferences.

Two methodological warnings run through this paper:

1. Lien Chan's 23.10% in 2000 must not be read as the whole pan-blue base. The KMT split that year: James Soong ran as an independent and won 36.84%; the pan-blue camp totaled roughly 59.94%, exceeding Chen Shui-bian's 39.30%. Reading a split election's vote shares as a direct measure of party strength is a common misreading.

2. The 2024 three-way vote cannot be mechanically reduced to a two-way race. Ko Wen-je of the Taiwan People's Party won 26.46%, drawing heavily on young voters; simply adding Hou Yu-ih's 33.49% to Ko's share to infer that "opposition unity would have won" ignores second preferences, turnout effects, and candidate evaluations — the error of mechanical vote addition.

2. Seven Presidential Elections at a Glance (2000–2024)

All figures below have been rechecked against the CEC election database (https://db.cec.gov.tw/).

Year	Winner (party)	Votes	Share	Main opponents	Turnout
2000	Chen Shui-bian / Annette Lu (DPP)	4,977,697	39.30%	James Soong (ind.) 36.84%; Lien Chan (KMT) 23.10%	82.69%
2004	Chen Shui-bian / Annette Lu (DPP)	6,471,970	50.11%	Lien Chan / James Soong (KMT–PFP joint ticket) 49.89%	80.28%
2008	Ma Ying-jeou / Vincent Siew (KMT)	7,659,014	58.45%	Frank Hsieh (DPP) 41.55%	76.33%
2012	Ma Ying-jeou / Wu Den-yih (KMT)	6,891,139	51.60%	Tsai Ing-wen (DPP) 45.63%; James Soong (PFP) 2.77%	74.38%
2016	Tsai Ing-wen / Chen Chien-jen (DPP)	6,894,744	56.12%	Eric Chu (KMT) 31.04%; James Soong (PFP) 12.84%	66.27%
2020	Tsai Ing-wen / Lai Ching-te (DPP)	8,170,231	57.13%	Han Kuo-yu (KMT) 38.61%; James Soong (PFP) 4.26%	74.90%
2024	Lai Ching-te / Hsiao Bi-khim (DPP)	5,586,019	40.05%	Hou Yu-ih (KMT) 33.49%; Ko Wen-je (TPP) 26.46%	71.86%

Note: Chen Shui-bian's 2004 re-election margin — 29,518 votes, 0.22 percentage points — is the narrowest in the history of direct presidential elections in the ROC.

Four phases

2000–2004: Turnover and consolidation of competition. 2000 was a split election; in 2004 the KMT–PFP joint ticket still lost by a razor-thin margin, showing that opposition unity alone does not guarantee victory — candidate evaluation and campaign dynamics matter equally.
2008–2012: KMT resurgence. Ma's 58.45% landslide in 2008 and 51.60% re-election in 2012 prove the KMT could still win a national majority across social groups. This is a direct counterexample to any "structural determinism of defeat": if identity transformation doomed the KMT, the 2008 landslide is inexplicable.
2014–2020: Cross-strait contention and generational mobilization. The 2014 Sunflower Movement, the 2019 Hong Kong protests, and rising security anxiety reshaped voters' risk assessment of the KMT's cross-strait line. But this requires validation with polling and individual-level data — temporal coincidence alone does not establish causation. The KMT's 2018 local-election landslide in the same period shows voters evaluated the two parties differently across electoral levels.
2024: Three-party competition and split-ticket voting. Lai won with 40.05% — short of a majority — while the DPP simultaneously lost its legislative majority (51 seats to the KMT's 52 and the TPP's 8). A presidential victory coexisting with a hung legislature is direct evidence that voters choose differently at different levels.

3. The Legislative and Local Contrast: Why Presidential Elections Alone Mislead

Year	KMT seats	DPP seats	Others	Total	Control
2008	81	27	5	113	KMT
2012	64	40	9	113	KMT
2016	35	68	10	113	DPP
2020	38	61	14	113	DPP
2024	52	51	10 (TPP 8, ind. 2)	113	None (hung)

Local executives tell a different story again: the KMT won 15 of 22 county/city chief executives in 2018 and 14 in 2022 (including the delayed Chiayi City vote); the DPP won only 6 and 5 respectively. If "KMT decline" were true, its persistent local-level competitiveness would be inexplicable. The plausible reading: presidential elections, as national and highly symbolic contests, are more dominated by cross-strait and identity issues, while local elections turn more on governance performance, factional networks, and candidate quality — the traditional stronghold of KMT local organization.

4. Testing the Seven Hypotheses

H1 Identity transformation: Partially supported, not determinative

Long-running NCCU Election Study Center tracking shows self-identification as "Taiwanese only" rising from 17.6% (1992) to over 60% in the 2020s (64.3% in December 2020), while "Chinese only" fell to about 2–3%. This structural shift disadvantages a KMT narrative centered on Chinese identity.

But three counterpoints limit its explanatory power: first, Taiwanese identity does not equal DPP support or support for de jure independence — DPP party identification has long hovered at only 20–30%, far below the Taiwanese-identity share; second, Ma won 58.45% in 2008 when Taiwanese-only identity had already reached 48.4%; third, in 2024 the TPP led among voters aged 20–29 (consistent across post-election polls) — the very cohort with the highest Taiwanese-identity share. Conclusion: identity transformation is a background condition, not a sufficient explanation of outcomes.

H2 Adaptation of historical narratives: Awaiting testing

The speed with which the KMT adjusted its discourse on the triangle of "the Republic of China, Taiwan, and China" may have affected the breadth of its electoral coalition. The "1992 Consensus / one China, respective interpretations" formula successfully united deep-blue and centrist voters under Ma; its mobilizing power declined after 2016. But "speed of narrative adjustment" is hard to quantify; this hypothesis remains an analytical framework awaiting text-analysis and survey-experiment data.

H3 Security and cross-strait policy evaluation: Stronger for 2016–2024

Rising PRC military pressure on Taiwan and the 2019 Hong Kong crisis made "cross-strait exchange risk" a salient dimension in voters' evaluation of the KMT's cross-strait line. Tsai's 8,170,231 votes in 2020 — the highest raw total for any candidate since direct elections began — coincides with this period. Identification caveat: governing approval, candidate contrast (Han Kuo-yu's controversies), and other factors moved in the same window; coincidence must not be presented as causal proof.

H4 Organization and mobilization: Level-differentiated

The DPP holds advantages in candidate cultivation, social-movement linkages, and national mobilization networks, especially at the presidential level; KMT factional and organizational networks remain effective at the county/city level. The 2018 and 2022 local results and the KMT's 2024 plurality in the legislature (52 seats, the largest party) support "level-differentiated organizational advantage" rather than any blanket verdict that one party's organization is comprehensively superior.

H5 Generational replacement: Right direction, wrong beneficiary

Voters socialized after democratization do have different political experiences — but 2024 proves one cannot assume the young naturally support the DPP: youth votes flowed heavily to the TPP. What generational replacement changes is the issue agenda (housing justice, wage stagnation, and distributional issues rising in salience), not automatic support for any particular party.

H6 Short-term factors: Key to explaining variance

Economic performance, governing approval, and candidate image: in 2004 (razor-thin margin), 2008 (late-Chen corruption scandals), 2012 (Tsai's "Taiwan consensus" failing to win centrists), and 2020 (Han's personal controversies), short-term factors explain outcome variance better than structural ones. Any purely structural account that ignores candidates and campaigns is incomplete.

H7 Opposition-coordination costs: A central 2024 variable

The pan-blue splits of 2000 (Lien vs. Soong) and 2024 (failed KMT–TPP "blue-white" coordination) both accompanied DPP victories; the 2004 KMT–PFP joint ticket narrowed the gap to 0.22 pp. But mechanical addition of split-camp votes is a methodological error: had Lien and Soong united in 2000, Chen would not necessarily have lost (strategic-voting and turnout effects cannot be precisely estimated ex post); in 2024 a substantial share of Ko supporters told pre-election pollsters they would not consider voting for the KMT (the "never blue" response ran at roughly three in ten of his supporters across several polls). Coordination failure raises the probability of defeat but does not establish a counterfactual certain victory.

5. Competing Explanations and Identification Limits

1. Mutual causation: national-identity change and party preference may influence each other; cross-sectional data cannot identify causal direction.

2. Confounders: candidate matchups, economic conditions, the international security environment, and turnout vary across years and confound cross-year comparisons.

3. Party-label substitution: concrete policy positions must be compared; party labels must not stand in for measured positions.

4. Ecological fallacy: county-level correlations cannot be directly inferred as individual voting behavior; conclusions are limited to aggregate-level description absent individual survey data.

5. Unobservable counterfactuals: claims about "victory had the opposition united" are inherently unverifiable and can only be stated probabilistically.

6. Comparative Constitutional Significance

Taiwan's experience shows that a long-ruling authoritarian-era party can lose power after democratic transition (2000), win it back (2008), lose it again (2016), and remain competitive in parliament and local governance throughout. This trajectory has comparative value for studying the ROC constitutional system, party–state separation, military nationalization, and competitive elections. But this study's conclusions cannot be directly applied to mainland China — the two differ fundamentally in social structure, institutional conditions, and civil-society development, and any analogy requires a rigorous comparability argument.

Preliminary Conclusion

The electoral record supports a "multi-factor, phased" explanatory framework and rejects both a single-factor national-identity determinism and any thesis of irreversible party decline. The KMT has not lost competitiveness at all electoral levels; nor can the DPP's presidential advantage be read as permanent majority support. Full causal testing still requires original electoral databases, long-running tracking surveys, and replicable statistical models — the direction of this series' future research.

Methodology and Data Notes

Election data: CEC Election Database (https://db.cec.gov.tw/); candidate votes, electorate size, and turnout rechecked election by election for both presidential and legislative races.
Identity and party-preference series: NCCU Election Study Center long-running surveys (https://esc.nccu.edu.tw/).
Methods: descriptive time series, cross-level comparison (presidential / legislative / local), and counterexample testing (2008, 2018, 2022, 2024 split-ticket). No individual-level regression models are used; no causal claims are made.
Literature: see References for representative works on Taiwan electoral studies; specific attributions follow cited page numbers; unattributed judgments are the author's analysis.

Research Limitations

1. This is an interpretive synthesis, not causal inference from original individual-level survey data; all causal language should be read as "explanations consistent with the evidence," not verified causal effects.

2. Post-election survey data for 2024 were still being released at the time of writing; some generational and issue analysis relies on pre-election polls and is preliminary.

3. Campaign finance and media-exposure measures are not systematically included.

4. Testing H2 (narrative adaptation) requires systematic text analysis; this paper offers only the analytical framework.

References

Central Election Commission, Overview of Past Presidential and Vice-Presidential Elections, Election Database, https://db.cec.gov.tw/
Election Study Center, National Chengchi University, Trends in Taiwanese/Chinese Identity and Party Preference, https://esc.nccu.edu.tw/
Fell, Dafydd. Party Politics in Taiwan: Party Change and the Democratic Evolution of Taiwan, 1991–2004. London: Routledge, 2005.
Rigger, Shelley. Politics in Taiwan: Voting for Democracy. London: Routledge, 1999.
Chu, Yun-han, Larry Diamond, Andrew J. Nathan and Doh Chull Shin, eds. How East Asians View Democracy. New York: Columbia University Press, 2008.
Hsieh, John Fuh-sheng and David Newman, eds. How Asia Votes. New York: Chatham House, 2002. (Taiwan chapter)
Wang, Austin Horng-En, Yao-Yuan Yeh, Charles K.S. Wu and Fang-Yu Chen, "Why Does Taiwan Identity Decline?" Journal of Asian and African Studies, 2025. (Issue-ownership and hedging explanation for the 2016–2018 identity dip)

Suggested Citation

Free China Movement: "From Authoritarian Ruling Party to Electoral Competitor: Why Did the Kuomintang (KMT) Repeatedly Lose to the Democratic Progressive Party (DPP), 2000–2024?" ROC (Taiwan) and Mainland Democratic Transition Research series, Taiwan Democratic Development and Electoral History subseries, Research Paper v1.0, October 13, 2026. https://freechina.live/research-archive/research-library/roc-taiwan-mainland-democratic-transition-research/taiwan-democratic-development-electoral-history/kmt-dpp-electoral-competition-2000-2024/ (URL subject to actual publication)

Series Cross-References

DPP-01 (DPP study, forthcoming in this series): this paper's H1 and H4 judgments touching DPP organization and identity mobilization can be read against DPP-01 once published; the manuscripts are not merged.
ROC-CONST-01 (ROC constitutional studies): Section 6's comparative-constitutional discussion defers to ROC-CONST-01 for constitutional argumentation.
ROC-TRANS-02 (democratic transition studies): the party-turnover and democratic-consolidation discussion connects with ROC-TRANS-02; the transition-theory framework follows that volume.

Version History

v0.1 (October 12, 2026): Argument draft; seven-hypothesis framework; second-round research memo. Not peer-reviewed; not released.
v1.0 (October 13, 2026): Muse scholarly review completed. Added complete 2012, 2016, and 2020 presidential data; new Legislative Yuan seat table and local-executive comparison; H1–H7 tested one by one with evidence grades marked; added methodology, limitations, references, suggested citation, and version history; standardized on "President of the Republic of China (Taiwan)." The original author's framework and core arguments are preserved.

Author: Free China Movement/em>

Share this:
Facebook
X
Like this:
Like Loading…