geography of the swing

As we wake up this morning, we are still trying to figure out where the hell that tidal wave that blindsided us last night came from. The polls gave us fair warning about some of the races. The Taipei result was clearly within the realm of possibility. I actually scribbled down a few back-of-the-envelope numbers a few days ago for Sean Lien’s vote share and came up with 41%. Of course, I didn’t believe that and I actually put down 44.5% for our office prediction pool, but the point is that you could have seen that result from the polling data. Similarly, Keelung City wasn’t a big surprise, and I wasn’t shocked that the DPP easily won in Yunlin or that the Nantou race ended up being pretty close. Some of the other results, though, simply came out of the blue. In particular, Taoyuan and Hsinchu City were complete shocks. I hadn’t seen any hints that those races might even be competitive. Taichung City, Changhua, Chiayi City, and New Taipei City were surprising to me, though not as unfathomable as Taoyuan and Hsinchu City. In general, the pre-election polls simply didn’t prepare us for yesterday’s results. So what did the polls completely miss?

The simplest answer is that many undecided voters made up their mind at the last minute to vote against the KMT. I have serious doubts about this, but there is no way to test this hypothesis with the data available right now. The KMT should hope that this was simply lots of last-minute decisions, since that would imply that the wave against the KMT was rooted in very shallow disapproval and could easily be reversed by 2016.

I suspect what may have happened is the emergence of the new economic cleavage as a powerful electoral force. I see two big groups of people who might have been vastly underrepresented in the surveys: younger people who have completely cut their landline connections and lower-wage labor who rent apartments in one district but vote in another. We have good reasons to believe that both of these groups are especially dissatisfied with the current government, so it is plausible that they were the bulk of the surprise wave. Of course, this will need a lot of testing before it is anything more than mere speculation.

Let’s look at the wave to see a few trends. I’m comparing the KMT’s performance in yesterday’s elections to the KMT’s 2012 presidential vote share. I know that comparing presidential votes to mayoral votes is like comparing apples to oranges, so don’t bombard me with complaints about this. The purpose of using the presidential vote as a baseline is precisely to strip away the local factors to see what the partisan balance used to be. The difference between the two elections is a combination of (1) the national wave and (2) the local factors specific to that race.

I usually look at things from the DPP’s point of view because that usually provides a clearer picture. In this election, however, I think it is more appropriate to look at things from the KMT side. For one thing, I think that this was almost certainly an anti-KMT wave driven by dissatisfaction with the national government rather than a pro-DPP wave driven by widespread attraction to the DPP and its platform. For another, the KMT only had one serious split (Keelung) while the DPP arguably had three (Changhua, Hsinchu City,and maybe Miaoli).

The six direct municipalities are the least localized of the races. If we see national forces acting on the existing partisan structure, it should be most evident in these races. So in the six direct municipalities, here is how the KMT fared:

City 2012 president 2014 mayor Increase
Taipei 57.9% 40.8% -17.1%
New Taipei 53.7% 50.1% -3.6%
Taoyuan 57.2% 48.0% -9.2%
Taichung 52.2% 42.9% -9.3%
Tainan 39.8% 27.1% -12.7%
Kaohsiung 44.2% 30.9% -13.3%

Taipei City is by far the worst, but we expected that. Sean Lien was a special disaster, and the blue coalition broke apart in Taipei in a way that it did not anywhere else. In Tainan and Kaohsiung, extremely popular DPP incumbents were running against weak challengers. Here, the KMT dropped about 13%. In Taichung and Taoyuan, we had KMT incumbents running for re-election. In these two races, the KMT dropped 9%. New Taipei City is the obvious outlier. Eric Chu was also a KMT incumbent running for re-election, but the KMT vote in New Taipei dropped less than 4%. It was very surprising to many people (including me) that Chu made such a proud and aggressive victory speech last night. This is why. He can point out that he alone withstood the wave and, due to his personal popularity, held onto almost all of the previous KMT coalition. Even with his surprisingly narrow margin of victory, he is a clear winner in this election.

Another interesting facet of this wave is that it was much more powerful in the north than in the south.

City 2012 president 2014 mayor Increase
Taipei 57.9% 40.8% -17.1%
New Taipei 53.7% 50.1% -3.6%
Keelung 59.3% 43.7% -15.6%
Taoyuan 57.2% 48.0% -9.2%
Hsinchu City 57.4% 37.9% -19.5%
Hsinchu County 65.8% 46.9% -18.9%
Miaoli 63.9% 46.6% -17.3%
Taichung 52.2% 42.9% -9.3%
Changhua 50.6% 39.6% -11.0%
Nantou 54.6% 51.0% -3.6%
Yunlin 41.7% 43.0% +1.3%
Chiayi City 46.3% 45.5% -0.8%
Chiayi County 39.0% 34.0% -5.0%
Pingtung 42.9% 37.1% -5.8%
Penghu 49.8% 44.7% -5.1%
Yilan 44.9% 36.1% -8.8%
Taitung 39.8% 27.1% -12.7%

(Note: In Keelung, I included the votes for Huang Ching-tai and the official KMT nominee in the KMT’s vote share. I left off Hualien, Kinmen, and Lienchiang because those races did not run on party lines at all.)

In the north, there are specific local factors that depressed the KMT’s vote in nearly every race. Still, there was almost certainly a national effect underneath the local effects. The Taoyuan swing of -9% might be a fairly representative effect. In central Taiwan, both Taichung and Changhua also saw roughly 10% swings. However, once you get to southern Taiwan, things look different. Put aside Tainan and Kaohsiung (and their highly popular incumbents seeking re-election), and look at the mostly rural and more heavily agricultural south. The standard swing seems to be only about 5%, or about half that of northern and central Taiwan. I don’t want to cut things too fine here, but the crude pattern of a bigger swing in the north than in the south seems evident to me. (The fact that the KMT is stronger in the north and therefore has more votes to lose doesn’t account for the difference. If you divide the swing by the KMT’s 2012 vote share, the swing is still larger in the north than in the south.

Why? Again, these are all guesses at this point. I wonder if this has to do with urban labor forces. I think this pattern is compatible with the idea that the crucial group withdrawing support from the KMT is the lower income, renting, predominantly younger, wage labor or low-salaried labor force.

(Isn’t it intellectually thrilling to be bewildered!)

18 Responses to “geography of the swing”

  1. ジェームス (@jmstwn) Says:

    The city council results are really interesting too. Independents hold deciding votes in Taipei, New Taipei, Taichung, Taoyuan, Changhua, Yunlin, Pingtung, Penghu, Keelung, Hsinchu City, Yilan, and Chiayi City and County. The PFP can claim deciding votes in Taipei. The DPP got more city council votes than the KMT in Taipei but Ko ran way ahead of pan green. The DPP got way more New Taipei seats than the KMT.

    • Raj Says:

      Taipei – not quite true. If the PFP and New Party side with the KMT, they can outvote the DPP, TSU and independents. That doesn’t mean they will of course.

      New Taipei – no, the independents can only help the KMT deadlock the council, assuming the TSU sides with the DPP.

      Taoyuan – no, if the PFP side with the KMT they can outvote the rest.

      Taipei is an interesting situation. Will the KMT councillors just try to block everything Ko does? Will the non-KMT blues join in or be more reasonable? Hopefully they’ll be sensible and realise that if they veto popular measures they might get whacked by their electorates.

      • ジェームス (@jmstwn) Says:

        I have been lumping the PFP and greens together in the anti KMT block since they both opposed the services pact and think the KMT loss and Ko endorsement will encourage them to cross the color line more but perhaps the future council votes will prove me wrong?

      • Raj Says:

        Actually, you’re right re Taipei. The PFP weren’t hostile to Ko and he supported some (?) of their candidates. It would make sense to at least put them in the neutral camp. So the KMT can’t rely on having a majority in the council chamber.

  2. Holger Nygaard Says:

    I look forward to see your take on what this means for the next presidential election.

  3. Rust Says:

    Look forward to seeing more analysis from you!

    Meanwhile I will just throw in some of my own thoughts on the “surprises”, altho with focus more on individual campaigns.

    Surprise level: moderate

    With Lien’s campaign self-imploding in spectacular fashion & showing off nvr seen before desperation, it seems very likely he will go down in defeat. What was surprising rly was just the matter of scale. I was personally worried cuz anecdotal observation is that many who support Ko in Taipei may work there but don’t vote there. This & the historical pattern made the scale of Ko’s victory a bit of a surprise for me, altho a joke is that with Lien exposing himself as such a fraud the face that he still have so many vote should be the surprise!

    Another thing I think effected everywhere is the enormous attention given to Taipei’s race generated a lot of anti-princling sentiments throughout Taiwan, & might have effected John Wu & Eric Chu.

    New Taipei
    Surprise level: very

    I was expecting the race to be closer than those 250k~350k margins ppl are throwing around for Chu by the campaign was entering it’s final days. Since roughly a month prior Yu really started going negative as you suggest, albeit not to Chiu Yi level. His attack were rather well articulated, with primary focus on the Ting Xing land approval (Hsieh even help him start a class action lawsuit.) Alas, I felt his effort will be fruitless as judging by the clicks of Apple Daily’s online news articles, his attacks aren’t receiving too much attentions. I expected it to gain him at least some vote & motivate some of his base tho (& judging by Tsai’s last two performances, green side have upward to a million vote potential here!)

    However Chu started shooting himself in the foot by not approving any of Yu’s application for land use for his final event, which I think provoked a lot of angry or sympathetic votes for Yu. & this was just the biggest of many other abuse of government power that make Chu seems very petty (小肚雞腸). The fact that he was running a very low key campaign probably meant many of his own supporters stayed home for one reason or another too (& explains the great municipal councilor results for the DPP here). Lastly there’s the possible principling effect.

    Surprise level: extremely

    Like everyone I never expected this lol. Wu have some cases that provided ammunition for Zheng, but his campaign didn’t seems to be going anywhere. I guess the princling effect helped Zheng. Maybe many of the would be voters from the two northern municipalities carried their sentiment against the KMT candidates there to Taoyuan.

    Hsinchu City
    Surprise level: moderate

    Surprisingly I’m not too surprised about Hsinchu City lol
    Altho almost everyone was expecting this place to be a lost cause for the DPP, there was one party that kept saying this is a battleground, & that is the DPP itself! They focused a lot of resources here, sent in a lot heavyweight here, made fancy commercial here, & kept drumming up how well Lin is doing in the polls here. I was sure the DPP must indeed have good intel indicating a chance of victory here to dedicate so much resources in this campaign. It helps that from all that I heard of the incumbent Tsu have a terrible record.

    Surprise level: very

    I rly shouldn’t be surprised, the end result aligns rly well with what the polls were saying, & I have a high degree of trust in polls. However in the final two weeks there was indeed a mess of rumours indicated a closing race, & ppl that lives in Taichung just aren’t optimistic somehow lol. So when the race did turned out as the poll indicated, it was a surprise somehow. My surprise was only at the great margin tho, I was pretty sure Lin will win, & even more so after he resigned his legislator seat.

    Surprise level: very

    Like the above, I’m more surprised at the margin than anything. After the oil scandal Wei seems to have good momentum, & I was expecting a small margin of victory. However Wei did shockingly well (or Lin shockingly bad) in the end; Wei got more vote than Lin & Huang combined (& Huang was supposed to be splitting off the green votes!) The effort to lump the central Taiwan races together as a combined effort by the DPP might have helped too (& maybe elevated Nantou as well.) I was sure of a win after Wei resigned his seat here too.

    Chiayi City
    Surprise level: moderate

    This was always supposed to be a race in the bag for the DPP considering everything, but Tu simply did a bad job campaigning & Chen a very good one. Polls indicate that a moderate advantage for Tu have became a sizable Chen lead by the time polling is prohibited. I was expecting a lose at this point.

    However then the “A vote for Chen is supporting Ma” posters started appearing (& later appeared all over Taiwan too lol) & in my opinion Chen did a horrible job when she tried to distance herself from Ma, ultimately alienating both deep blue & neutral supporters. It also helped that Tsai Ing-wen nvr seems to have given up here, coming here to campaign with Tu in what feels like hundreds of times. Tu said himself in his concession speech “Every time when Tsai came onvr, the poll rise once.”

    • frozengarlic Says:

      Interesting comments. Just one quick response. It is interesting how many people are crediting Tsai. I have heard several people say, “Tsai visited XX City 23 times, and that was the difference.” This discourse is turning the result into a tremendous personal victory for Tsai, almost as the 1997 sweep was largely attributed to Chen Shui-bian’s extensive campaign efforts.

      • Rust Says:

        boy all the typos, I rly need to proofread before I post lol

        Another quick thing to add, in addition to Tsai I think this is also a great personal triumph for Yu despite it being a defeat in the end. Before he went on the attack he was widely dismissed even by green supporters. However Chu’s petty actions against his campaign & his much better than expected vote result what I see now instead are widespread sympathies & admiration poured upon him. For examples his FB like have increased by roughly 30k since the vote count, & an Apple Daily article on him have generated 400k clicks! This is an extraordinarily high number even compared to other articles with high amount of clicks.

        Said article:

  4. Pat Says:

    I think the DPP simply reached a ceiling in the south, hence the smaller swing. It won most of the counties there’s with over 60% of the vote.

    As for Chu, I’m not as impressed as you are. His race was in many races a parallel of the Kaoshiong race; popular incumbent running against a joke of an opponent. Even accounting for the DPP wave, he should’ve done far better, and he likely would’ve lost had Luo or Su been nominated instead of Yu.

  5. Joseph Wang Says:

    Just talking from someone in the deep blue camp…

    One thing that you should look at is how much of the withdrawal of KMT support comes from what are normally hard core KMT voters. Pretty much everyone that I know that is in the blue camp was pretty much disgusted with Ma Ying-Jeou and his incompetence and his terrible treatment of Wang Jin-Pyng. The people I know were in Taipei, and pretty much all of them were voting for Ko. Sean Lien’s last minute desperate appeals were likely triggered by the fact that he was seeing hard-core KMT voters defect to Ko.

    In the case of Taoyuan and Xinbei, one thing that might have happened is that because the polls gave the KMT incumbent such a large advantage, people were inclined to just sit this one out, or possibly vote for DPP as a protest vote.

    One thing that changed the dynamics is that it was pretty obvious that even if the KMT had done well, then nothing would have happened until Ma left office. The DPP had also done a very good job of moderating their campaign rhetoric not focus on symbolic issues.

    I’ve been trying to get data on the number of votes and voter turnout. One thing that goes against the idea that it was people not voting was that the vote totals in Taoyuan were higher than in 2009, and so it could have been a “sunflower” effect there.

    Something that could have effected the polls was deep-blue voters saying “well if he is going to win anyway why should I vote.”

    • Pat Says:

      John Wu’s vote total went up by 70k from the last election, so I doubt depressed turnout was much of a factor in Taoyuan. I also don’t see much logic behind blue voters abstaining in Xinbei to protest Ma when Chu is not really in his camp and has been lauded as the KMT’s savior for quite some time now. Giving him a big victory would allow him to take the reins from Ma and start turning things around. This close call has torpedoed his momentum.

      • Joseph Wang Says:

        It’s up from the last election but down from the 2012 Presidential election.

        I don’t think that either Taoyuan or Xinbei was an intentional anti-Ma vote. My theory is that because both Eric Chu and John Wu were so far ahead in the polls, some of the blue voters were inclined to stay at home or to cast a protest vote which they thought wouldn’t have made much difference.

        Everyone in the blue camp knew that this was going to be a bad and depressing election, but people assumed that both Chu and Wu were going to win, so there wasn’t a huge sense of urgency to getting out the vote. By contrast, the DPP was extremely energized in Taoyuan. They are probably kicking themselves that they had written off Xinbei.

        How big the effect of people coming and and people going out is going to make a big difference for campaign strategy.

        One pattern that I’m seeing is that the special local issues that causes the vote swing seems to be different from place to place, but I think what causes this to become systemic is that the DPP was very well organized this time, whereas the KMT was something of a mess internally.

      • Joseph Wang Says:

        One other theory for what happened was that the Lien disaster had knock-on effects. Pretty much everyone that I knew that was “light blue” was supporting Ko rather than Lien. What that meant was that you didn’t have a pool of KMT campaign volunteers to work in Taoyuan and Xinbei.

        By contrast because the DPP had secure races in Kaohsiung and Tainan, you could use campaign workers from those areas to work in surrounding areas.

        Something that I think made a big difference was that the DPP had a national strategy, and they had a single message and people moving from county to county. Essentially, the DPP did not care about individual races but they are all looking at 2016.

        By contrast, the KMT seemed to be fighting each election separately. Because you had a very unpopular leader, a recent civil war, and because the flagship race was something you wanted to keep separate, there wasn’t a coordinated national strategy, and so you had a general disaster.

        This means sense if you mix it with the effect of the sunflower students. Sunflower students may not be too motivated by local politics but would be convinced to vote if you invoke national issues. In addition, you have the effort of reaching out to the students, and this is much easier to do if you have a national strategy.

        One thing that the 9-in-1 elections did which I think the DPP figured out but the KMT didn’t was that by having all of the elections together, you end up with more than a collection of local elections.

  6. Joseph Wang Says:

    One other thing to look at is to compare the 2014 elections not so much with the 2012 elections but rather with the 2009 elections and with the 1997 election. The Taoyuan vote swings are much less.

    One thing about 2009/2010 is that the KMT won a lot of the elections by rather close margins.

    In looking at “who withdrew support.” I wouldn’t look just at the students and low income workers, but I’d also look at core KMT supporters who were disgusted and demoralized at Ma and decided not to vote.

  7. Joseph Wang Says:

    The experiment that I’d really like to do is to conduct a phone survey using the same methodology as the preelection phone survey and ask “did you vote and who did you vote for?”

    Also looking at the demographics of who voted and who didn’t vote would make a big difference.

  8. M Says:

    90% of the coverage on the national TV stations was on the Taipei race. Given that viewers were constantly bombarded with news from Taipei with very little coverage of the other races, the disastrous performance of the Lien camp must have had a spillover effect. Lien didn’t just lose Taipei for the KMT, he may also have lost them Taoyuan and nearly lost them New Taipei.

    • frozengarlic Says:

      I think there must have been some spillover effect. When Lien Chan started spouting orthodox ROC ideology from the 1960s about the Japanese colonial era, it had to have an effect in other races. The discourse about princelings probably also had a bigger effect on Chu and Wu than was anticipated. The difficult part is trying to figure out how big that spillover effect was, and I have no answers for that question.

  9. East Asia at the Centre Says:

    […] same young Taiwanese, who turned out en masse to cast their vote three weeks ago, have now sealed president Ma’s political fate. Traditional bastions of […]

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 )

Facebook photo

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

Connecting to %s

%d bloggers like this: