Reflections on the legislative election: regions, urbanization, and split-ticket voting

Everyone is focusing on the presidential results. Well, of course they are. The presidential results are important.

But just looking at the presidential results is deceptive in a few ways. First, Tsai won by 18%, which is a huge margin. As we’ll see, the DPP’s actual political foundation is not nearly that much bigger than the KMT’s. Second, the presidential vote gives the impression of stability. Tsai’s vote in 2020 was geographically very similar to that in 2016. She was a point better in some areas, a point worse in others, but overall it was pretty similar. The legislative results were a lot more turbulent. Third, Tsai’s vote looks like a solid block of 57.1% green voters. The legislative results make that coalition look much more fragmented and tenuous. There was a LOT of split-ticket voting this time. Fourth, the legislative results reveal just what a terrible candidate Han Kuo-yu was.


Let’s start by giving the official party results. This is the kind of data you might have seen in the newspapers. (Hint: This is not the right way to look at election results.) Note also: the nominal tier includes both the 73 single seat districts and the two indigenous districts.

Nominal Nominal List List Prez Prez
votes % votes % votes %
DPP 6383783 45.1% 4811241 34.0% 8170231 57.1%
KMT 5761995 40.7% 4723504 33.4% 5522119 38.6%
TPP 264478 1.9% 1588806 11.2%
NPP 141952 1.0% 1098100 7.8%
Statebuilding 141503 1.0% 447286 3.2%
Congress 81508 0.6% 40,331 0.3%
PFP 60614 0.4% 518921 3.7% 608590 4.3%
Green 39387 0.3% 341465 2.4%
other parties 190349 1.3% 590484 4.2%
independents 1086463 7.7% 4811241 34.0%
total 14152032 14160138 14300940


According to this, the DPP only beat the KMT by 4.4% in the nominal tier. It was close, but not that close. Remember, there were a few DPP candidates who ran as independents (like Su Chen-ching in Pingtung 2). The KMT had a couple as well. Also, if we are going to look at the broad balance of power, shouldn’t we think about all the minor parties as well? Here’s a second table looking at the big camps.

Nominal Nominal List List Prez Prez
votes % votes % votes %
All blue 6103580 43.1% 5422764 38.3% 6130709 42.9%
All green 7135019 50.4% 6921468 48.9% 8170231 57.1%
All white 325663 2.3% 1588806 11.2%
No camp 587770 4.2% 227100 1.6%

The blue camp includes the KMT, PFP, New, CUPP, and a couple independents (Li Weng Yueh-e [New Taipei 3], Lin Kuo-ching [Chiayi 2], Fu Kun-chi [Hualien], Kao Chin Su-mei [mountain indigenous]). The green camp includes the DPP, NPP, Statebuilding, Green, Taiwan Action, Formosa Alliance, and various independents (Freddy Lim [Taipei 5], Hung Tzu-yung [Taichung 3], Su Chen-ching [Pintung 2], Chao Cheng-yu [Taoyuan 6]). The white camp is the TPP plus two independents, Hsu Li-hsin [Taipei 5] and Lee Chin-ying [New Taipei 10]. I didn’t know what to do with Congress Party, since they come from the blue side but their leader endorsed Tsai, so I put them in the no camp bloc.

This grouping is not meant to imply that that these blocs are one team or that they will work together. Rather, the idea is that they draw from the same broad set of voters. For example, I’m not sure we will still consider the NPP as part of the green camp four years from now, but for right now, I think they best fit in that big pool of voters.

Looking at it this way, the green camp beats the blue camp by 7.3% in the nominal vote and 10.6% in the list vote. This is a much better indication of where the country is than the presidential results or the official party totals. The presidential results make the blue side look far too weak because (a) Han Kuo-yu was a disaster and drove every wavering voter away and (b) the white camp is absent.


Before moving on, let’s linger a bit on the TPP. This table says the white camp got 2.3% of the district vote, which looks pretty inconsequential. However, they didn’t run in many districts. Moreover, wherever they ran, they were not one of the two main candidates. It’s really hard to be the third candidate in a plurality race. Nevertheless, they did amazingly well. Look at this list of how white camp candidates performed.

district name votes %
Taipei 3 何景榮 5,730 2.6
Taitung 陳允萍 3,527 4.5
Tainan 1 顏耀星 8,793 4.9
Kaohsiung 3 莊貽量 12,099 5.2
Taichung 4 張渝江 13,434 5.7
Kaohsiung 8 敖博勝 14,043 6.0
Taipei 8 張幸松 12,111 6.3
Kaohsiung 1 羅鼎城 10,661 6.5
Taichung 5 謝文卿 18,768 7.3
Taichung 3 張睿倉 14,700 7.5
Kaohsiung 5 李佳玲 20,336 8.6
Taipei 7 蔡宜芳 17,435 9.5
Taoyuan 1 陳泓維 21,951 9.7
Hsinchu County 2 林碩彥 17,902 11.1
Miaoli 1 朱哲成 16,566 11.1
New Taipei 7 吳達偉 20,579 11.7
New Taipei 3 李旻蔚 35,843 18.2
Taipei 5 徐立信 22,208 12.2
New Taipei 10 李縉穎 38,977 18.8

Every one of them got votes. There was not a single white camp candidate who got completely marginalized. Third parties are not supposed to do this well. And remember, it isn’t the case that all of these are great politicians. Remember that guy I mocked in a previous post for his stance on stray dogs? That guy competed with two extremely famous people (Hung Tzu-yung and Yang Chiung-ying), got nearly 15000 votes, and almost certainly affected the outcome of the Taichung 3 race. There is a clear message here: Mayor Ko can transfer (a good deal of) his support to other people. The white camp is a serious force in Taiwan politics.


The New Power Party also had a very strong showing at the district level, though they didn’t nominate nearly as many candidates. Here is their list

district name votes %
New Taipei 10 賴嘉倫 13,563 7.3
New Taipei 1 張衞航 19,641 7.6
Kaohsiung 7 陳惠敏 17,466 7.8
Taoyuan 3 林佳瑋 17,512 8.3
Hsinchu City 高鈺婷 73,770 28.6

Again, their first four candidates did extremely well to get significant numbers of votes as the third candidate in a two-way race. It looks as though the NPP has solid support in society. However, since the NPP didn’t need to nominate ten people to qualify for the party list, it could concentrate on its best candidates. This list might look artificially impressive; it’s possible that the next five NPP candidates would have all been turkeys.

Now look at the fifth candidate. Kao Yu-ting was not a spoiler candidate. She was a major candidate in a true three-way race. This was absolutely shocking to me. I was mildly surprised by candidates like Lee Chin-ying in New Taipei 10 who got nearly 20%. However, Kao Yu-ting was on an entirely different level. The three main candidates in Hsinchu City got 95298, 82011, and 73770 votes. Those numbers are all within spitting distance of each other. When the NPP is that close, the DPP can no longer argue that a vote for the NPP is a wasted vote (or a de-facto vote for the KMT). It’s true that the green camp vote was split and the KMT won a seat that it didn’t deserve. However, it is not obvious that the NPP should have supported the DPP and not the other way around.


Compared to past years, there were a lot more significant third candidates in district elections. 15 districts had third candidates getting at least 10%, and 28 more candidates got between 5% and 10%. That is to say, most districts this year had a significant third- party presence. This is new.

The two big parties got 95.7% of the presidential vote, which makes it look as if they absolutely dominate Taiwanese politics. That is misleading. They only got 67.4% of the party list vote, which makes it look as if the party system is quite fragmented. That is also a bit misleading. The truth lies somewhere in between. The two big parties still sit atop the political structure, but their coalitions are a lot less solid than they might appear. This is especially true for the DPP, which succeeded in pulling together a massive presidential coalition from several forces who don’t really all want the same things.


Ok, let’s go back to the two big parties in the district elections. The best way to look at how they fared is not to look at either the party labels or the big camp blocs. Rather, the best way to look at them is by looking at who they asked people to vote for in each district. Each big party sponsored (officially or unofficially) one candidate in each seat. How did that go?

Note: I assume that Fu Kun-chi was the actual KMT candidate in Hualien. Someone else was the formal KMT nominee, but the party sent a pretty strong message through its VP candidate that Fu was the real KMT candidate. The only place without a major party candidate was the DPP in Kinmen. Chen Tsang-chiang dropped out of the DPP a few years ago, but I didn’t see any reference to the DPP encouraging people to vote for him this time. [Please comment if you know more about the DPP in Kinmen this time.]


So, how did the main green and main blue candidates do?

votes %
Main blue candidates 5987296 42.3%
Main green candidates 6892140 48.7%

The spread that really matters, the one that decided who would have power in the legislature, was 6.4%. That is not a small advantage, though it is nowhere near the 18% spread from the presidential election (or 14.2%, if you add Han’s and Soong’s votes together).

[Aside: The green side got 6.89 million votes. That number seems familiar…]

Unlike the gap in the presidential election, the gap between the two big blocs in the legislative races is a lot smaller than in 2016. Four years ago, the main green candidates got 12.3% more than the main blue candidates. This year, that gap was cut nearly in half.

2016 2020 2016 2020
Vote% Vote % seats Seats
Main Blue 39.7 42.3 25 27
Main Green 52.0 48.7 54 52

Four years ago, I assumed that things would never again be quite as perfect for the DPP. Even if Tsai was re-elected, it would inevitably be with a reduced margin and a reduced legislative majority, just as happened for President Ma in 2012. It didn’t work out that way in the presidential race because Han Kuo-yu was such a terrible candidate, but it did work out that way in the legislative election.

There is one hitch: the seat shares were nearly unaffected by the dramatically reduced gap in votes. The green side won 54 of the 79 seats with a 12.3% advantage in 2016 and 52 with only a 6.4% advantage in 2020. Why?

Before I answer that, let me editorialize. I kept telling people to stop focusing on the party lists, because the legislative majority would be decided by the districts. That is, in fact, exactly what happened. The DPP did terribly in the party list vote, but that did not cause it to lose its legislative majority. Because it is still the biggest party and the nominal tier is so majoritarian, it managed to win two-thirds of the nominal seats and hold an overall majority.

Geography mattered in this election. Based on polling data, I had speculated that Tsai was a bit better in the north, relative to past performance, and a little worse in most other areas. She was, but it wasn’t a very dramatic shift. The biggest change came in Taoyuan, where she was 3.8% higher than four years ago.

Tsai 2016 Tsai 2020 change
North 53.4% 55.1% 1.8%
North-central 49.0% 52.2% 3.3%
Central 55.1% 56.3% 1.2%
South-central 65.8% 65.3% -0.5%
South 63.4% 62.2% -1.2%
East & islands 47.1% 47.5% 0.4%
national 56.1% 57.1% 1.0%

What’s striking about this is not how much things have changed, but how much things have remained stable. It appears voters decided their presidential vote based on sovereignty, identity, and the other aspects of national-level politics.

However, once they cast their “responsible” vote for the presidency, many of them seem to have expressed somewhat different values with their legislative votes. In the legislative vote, as with the 2018 mayoral votes, people were free to register some unhappiness, try to restrain the DPP’s power, or vote on local or secondary issues.

If the geographic shifts in the presidential vote were mild, those in the legislative nominal tier were not. Again, this looks at the vote for the main green and main blue candidate in each district.

2020 main blue maingreen mb% mg%
north 1855480 2014050 43.8% 47.5%
north-central 968922 909486 45.1% 42.3%
central 1203565 1358526 44.5% 50.2%
south-central 741114 1143779 37.0% 57.1%
south 806146 1167737 36.6% 53.0%
East & islands 233322 246947 40.0% 42.4%
indigenous 178747 51615 67.0% 19.4%
national 5987296 6892140 42.3% 48.7%

Note: These are the traditional regional groupings. North: Taipei, New Taipei, Keelung. North-Central: Taoyuan, Hsinchu, Miaoli. Central: Taichung, Changhua, Nantou. South-central: Yunlin, Chiayi, Tainan. South: Kaohsiung, Pingtung. East & islands: everything else.

Compare this with the data for 2016:

2016 main blue maingreen mb% mg%
north 1537669 1808346 41.5% 48.8%
north-central 771873 774850 43.1% 43.3%
central 1046801 1160655 44.9% 49.8%
south-central 512981 1127577 29.7% 65.3%
south 645211 1149444 34.0% 60.6%
East & islands 174188 257415 35.5% 52.5%
indigenous 129300 33710 62.3% 16.2%
national 4818023 6311997 39.7% 52.0%

Let’s compare the change in the edge that the green camp holds over the blue camp in each area.

2016 (G-B) 2020 (G-B) change
north 7.3% 3.7% -3.6%
north-central 0.2% -2.8% -2.9%
central 4.9% 5.7% 0.8%
south-central 35.6% 20.1% -15.5%
south 26.6% 16.4% -10.1%
East & islands 17.0% 2.3% -14.6%
indigenous -46.1% -47.7% -1.6%
national 12.3% 6.4% -5.9%

Nationally, the green camp lost 5.9%, compared to 2016. However, they didn’t lose anything in central Taiwan, only had small losses in northern Taiwan, and they had massive losses in southern Taiwan.

Fortunately for the DPP, they suffered losses in areas that, because they had such a large cushion, they could afford to absorb some losses. Even with double digit losses throughout the south, the DPP still swept all 21 southern seats.

2016 2020 2016 2020
Blue seats Blue seats Green seats Green seats
north 7 7 14 14
north-central 5 8 5 3
central 6 5 8 9
south-central 0 0 10 11
south 0 0 12 10
East & islands 2 3 4 3
indigenous 5 4 1 2
national 25 27 54 52

You might notice that the green camp went from being even in North-central (+0.2%) to slightly worse than the blue camp (-2.8%). They were right at the tipping point, where they could not afford losses, and, not coincidentally, the north-central region was where the green camp lost the most ground in terms of seats. It’s also pretty remarkable that the green camp maintained a 14 to 7 advantage in the north, even though their lead was much smaller this time. Several races could have easily gone the other way.


In addition to the dramatic losses the DPP suffered in southern and eastern Taiwan, there was also an urban/rural pattern. The DPP lost a lot of ground in rural areas.

I divided Taiwan into three big categories, very urban, suburban, and rural. The urban category includes districts in the core areas of major cities, and most of them have population densities of over 10,000 people per km2. Some examples are New Taipei 6, Taichung 4, Chiayi City, and Kaohsiung 6. The suburban areas are usually thought of as part of the city, but they are just a little further out. Examples are New Taipei 10, Taichung 7, Taoyuan 1, Changhua 2, and Tainan 6. Rural areas generally only have smaller towns (such is the definition of rural in dense, dense Taiwan). Examples include Miaoli 1, Changhua 4, Tainan 1, Kaohsiung 2, and Yilan. This is a quick and dirty classification; there are no hard rules for these choices. It is entirely subjective. You’ll have to trust that it is mostly right. Anyway, the patterns that I’m going to show you probably don’t change much if you insist on moving Pingtung 1 from category 2 to category 3, for example.

In 2016, the green camp did very well in the most rural areas.

2016 main blue maingreen mb% mg%
most rural 1277007 1757475 38.8% 53.4%
suburban 1474932 1953120 39.6% 52.5%
most urban 2066084 2601402 40.2% 50.6%

In 2020, not so much.

2020 main blue maingreen mb% mg%
most rural 1636532 1724987 44.6% 47.0%
suburban 1909851 2324929 41.3% 50.3%
most urban 2440913 2842224 41.7% 48.5%

Here are the gaps and the change:

2016 (G-B) 2020 (G-B) change
most rural 14.6% 2.4% -12.2%
suburban 12.8% 9.0% -3.9%
most urban 10.4% 6.8% -3.6%


Of course, there is a lot of overlap between rural and south/south-central/east. I don’t have the energy right now to look into which factor is really the dominant one. I suspect they both matter.

Why did the DPP suffer such enormous losses in rural areas and in the south and east? I can think of several possibilities. First, this could be a protest about regional economic inequality. Second, this could be a backlash against same-sex marriage (and other progressive values) by the more conservative rural population. Third, it could have something to do with leadership styles, with Tsai being more of a Taipei personality and Han being a less suave, urbane type of guy.


There are two points that I want to close on. First, I want to reiterate the point I have been making throughout. The DPP coalition does not look nearly as solid this time as it did four years ago. This time, there are far more people who split their tickets, voting for Tsai but not the legislative candidates or parties endorsed by her.

Second, Han Kuo-yu severely underperformed. There were a lot more voters out there who were open to voting for the blue camp, but Han could not soak them up. To put it bluntly, Han was a drag on the ticket. He certainly did not pull any legislators to victory, and he probably dragged some down to defeat. He probably played a role in pushing some potential blue voters over to PFP or TPP legislative candidates. At any rate, he was demonstrably less popular than the rest of the party.

15 Responses to “Reflections on the legislative election: regions, urbanization, and split-ticket voting”

  1. Election Watcher Says:

    Great stuff!

    Could the urban/rural thing also be self-interest on pocketbook issues (i.e. a more specific version of concern about regional inequality)? I’m thinking of farmers wishing to sell their products to China and tourism operators that used to prosper on Chinese tours.

    It seems to me that a big problem of the KMT is that its economic message is directed at those groups, which are only a small part of the population (and more prevalent in rural areas). It’s also a tired message after Ma’s second term. The DPP’s housing and child care platform probably appeals more to urban voters, and that’s where the votes/seats are.

  2. Ian Lamont Says:

    I really would like to see an analysis of age and gender in this election vs. 2016. Is that data available (for instance, via exit polls)?

  3. MW Says:

    Wonderful discussion, thank you. I noted that eight (if including Taipei 5’s Hsu Li-hsin) of the eleven TPP candidates that ran against KMT candidates in northern and central Taiwan received a volume of votes that exceeded the gap between the KMT candidate and the DPP/Ind candidate. But there was no such phenomenon in the TPP races in the south or in Taitung. Thanks again.

  4. Frank Chen Says:

    First table second row from the last (under independents), the Prez votes columns were duplicated from DPP rows and should be removed.

  5. P Says:

    To what extent do you think age explains this? A number of polls had indicated that voters under the age of 40 overwhelmingly backed Tsai but had no loyalty to the DPP and thus planned on splitting their LY votes en masse. Should make for an interesting dynamic if this trend continues.

  6. Geer Says:

    Ko’s TPP is also being perceived more and more as a pan-blue party, despite trying to portray itself as neither-green-nor-blue. I suspect – and hope – this may have the effect of luring away some light blue votes in the future.

    The youth mock election done a few weeks ago showed that young Taiwanese (<20) only support the KMT at a rate of 3%, but they also dislike the DPP. It could be a huge boon for the TSP, NPP, etc. years down the road…….if those small parties can still survive until then.

  7. Irwin Says:

    I wonder if the urban/rural divide is really masking the age factor. We know younger voters in Taiwan are overwhelmingly liberal and pro sovereignty (note I didn’t say pro DPP). Could it be that what is manifesting as rural/urban divide is really a story about generation gap?

    • Mike Says:

      Oerhaps another compounding factor of the rural/ urban divide might potentially be the migration of younger voters moving out of rural areas and into urban areas for work/ education as well (or the “northward drift” that is frequently reported in the media).

  8. Taiwan voters reject Beijing, but not populism - taktik(z) Staging Says:

    […] vic­to­ry in the pres­i­den­tial elec­tion and retained its leg­isla­tive major­i­ty, it also lost seats to emer­gent third par­ties. The per­for­mance of Han’s Kuomintang (KMT) was far from […]

  9. 2020大選(上)-藍綠到底差異多大? | Says:

    […] 《Reflections on the legislative election: regions, urbanization, and split-ticket voting》(連結) […]

  10. 2020大選(下)-大黨板塊如何移動,小黨有多少空間? | Says:

    […] 《 Reflections on the legislative election: regions, urbanization, and split-ticket voting》(連結) […]

  11. The public’s view on same-sex marriage legalisation – Taiwan Insight Says:

    […] that this might encourage DPP legislators to support more progressive policies as the party was able to absorb these losses in most […]

  12. In Taiwanese politics, the DPP is taking over the KMT’s stable narrative niche – Sense Hofstede Says:

    […] scientist Nathan Batto helps understand the challenge to the structural support for the KMT. His analysis of the January legislative vote shows that the DPP’s vote is expanding in the urban north, where most Mainlanders live, whereas […]

  13. Can Tsai Ing-wen Avoid the Second Term Curse? – Taiwan Insight Says:

    […] on the road ahead. Much like Ma Ying-jeou, Tsai has started her second term with support that is broad but shallow, and it is easy to imagine scenarios where much of the diverse coalition that backed her reelection […]

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: